Synchrotron Techniques in Interfacial Electrochemistry

NATO ASI Series Advanced Science Institutes Series

ASeries presenting the resu/ts of activities sponsored by the NATO Science Committee, which aims at the dissemination of advanced scientific and technological knowledge, with a view to strengthening links between scientific communities.

The Series is published by an international board of publishers in conjunction with the NATO Scientific Affairs Division

A Life Sciences B Physics

C Mathematical and Physical Sciences

o Behavioural and Social Sciences E Applied Sciences

F Computer and Systems Sciences G Ecological Sciences H Ce" Biology I Global Environmental Change

NATO-PCO-DATA BASE

Plenum Publishing Corporation London and New York

Kluwer Academic Publishers Dordrecht, Boston and London

Springer-Verlag Berlin, Heidelberg, New York, London, Paris and Tokyo

The electronic index to the NATO ASI Series provides full bibliographical references (with keywords and/or abstracts) to more than 30000 contributions from international scientists published in all sections of the NATO ASI Series. Access to the NATO-PCO-DATA BASE is possible in two ways:

- via online FILE 128 (NATO-PCO-DATA BASE) hosted by ESRIN, Via Galileo Galilei, 1-00044 Frascati, Italy.

- via CD-ROM "NATO-PCO-DATA BASE" with user-friendly retrieval software in English, French and German (© WTV GmbH and DATAWARE Technologies Inc. 1989).

The CD-ROM can be ordered through any member of the Board of Publishers or through NATO-PCO, Overijse, Belgium.

Series C: Mathematical and Physical Sciences - Vol. 432

Synchrotron Techniques in Interfacial Electrochemistry edited by

c. A. Melendres Materials Science and Chemical Technology Divisions, Argonne National Laboratory, Argonne, IIlinois, U.SA

and

A. Tadjeddine CNRS, Laboratoire d'Electrochimie Interfaciale (Meudon) and LURE (Orsay), France

SPRINGER-SCIENCE+BUSINESS MEDIA, BV.

Proceedings of the NATO Advanced Research Workshop on Synchrotron Techniques in Interfacial Electrochemistry Funchal, Madeira, Portugal December 14-18,1992

A C.I.P. Catalogue record for this book is available from the Library of Congress.

ISBN 978-90-481-4406-8 ISBN 978-94-017-3200-0 (eBook) DOI 10.1007/978-94-017-3200-0

Printed on acid-free paper

AII Rights Reserved © 1994 Springer Science+Business Media Oordrecht Originally published by Kluwer Academic Publishers in 1994 No part of the material protected by this copyright notice may be reproduced or utilized in any form or by any means, electronic or mechanical, including photo­copying, recording or by any information storage and retrieval system, without written permission from the copyright owner.

CONTENTS

Preface .................................................... ix

List of Participants ............................................ xi

Synchrotron Radiation and Instrumentation J. Robinson .................................................. 1

The Electrode/Solution Interphase: Problems for Synchrotron Radiation R. Parsons ......................•........................... 21

Nature of Surface Films J. Kruger ................................................... 33

Theory of the X-ray Scattering from Surfaces and Interfaces R. A. Cowley ................................................ 67

X-ray Diffuse Scattering as a Probe for thin Film and Interface Structure S. K. Sinha . . . . . . . . . . . . . . . . . . . . . • . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85

Surface Morphology Characterization with X-ray Scattering Techniques C. Thompson ................................................ 97

Studies of Electrodes by In-Situ X-ray Scattering M. F. Toney ................................................. 109

Surface Structu re of the Au (111) Electrode B. M. Ocko and J. Wang . . . • . . . . . . • . . . . . . . . . . . . . . . . . . . . . . . . . . .. 127

In Situ X-ray Diffraction Studies of the EJectrodeposition of Pb Monolayers on Au(100) Single Crystals K. M. Robinson and W. E. O'Grady . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 157

Oxidation of Mo(OO1) Surfaces I. K. Robinson .............................................. 171

Extended X-ray Absorption Fine Structure: Physical Principles and Data Analysis D. C. Koningsberger . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 181

The Use of X-ray Techniques in the In-Situ Study of Corrosion H. S. Isaacs .. . . . . . . . • . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 199

v

vi

In Situ X-ray Absorption Spectroscopy Investigation of UPD Meta! Monolayers A. Tadjeddine ............................................... 215

In Situ X-ray Absorption Spectroscopy of Nickel Oxide Electrodes W.E. O'Grady and K.1. Pandya .................................. 247

The UPD of Copper on Pt (100) In-Situ EXAFS and Ex Situ Structural LEED Investigations D. Aberdam, Y. Gauthier, R. Durand and R. Faure .................... 263

Characterization of New Systems for the Catalytic Electroreduction of Oxygen by Electrochemistry and X-ray Absorption Spectroscopy M. C. Martins Alves, J. P. Dodelet, D. Guay, M. Ladouceur and G. Tourillon ............................................. 281

In Situ and Ex Situ Examination of Passivating Cu20 Layers with EXAFS and REFLEXAFS H. H. Strehblow, P. Borthen and P. Druska ......................... 295

In-Situ and Ex-Situ Spectroelectrochemical and X-ray Absorption Studies on Rechargeable, Chemically-Modified and other Mn02 Materials B. E. Conway, D. Qu and J. McBreen ............................. 311

EXAFS Studies of Film Coated Electrodes R. C. Eider, L. R. Sharpe, D. H. Igo, R. O. Rigney and W. R. Heineman .... 335

Electrode-Electrolyte Interfaces Investigated with X-ray Standing Waves: Cu(III}/Pb, Ti J. Zegenhagen, G. Materlik, J. P. Dirks and M. Schmäh ................ 349

X-ray Standing Wave Studies of Underpotentially Deposited Metal Monolayers G. M. Bommarito, D. Acevedo, J. F. Rodrrquez, H. D. Abrufla, T. Gog and G. Materlik .............................................. 371

The Application of Infrared Synchrotron Radiation to the Study of Interfa­cial Vibrational Modes C. J. Hirschmugl and G. P. Williams .............................. 387

Fourier Transform Infrared Combined with Synchrotron Radiation for Probing the Electrochemical Interface Y. L. Mathis, K. Murakoshi, A. Tadjeddine and P. Roy .................. 401

Far Infrared Synchrotron Radiation and the Electrochemicallnterface A. E. Russell and W. O'Grady ................................... 421

The Adsorption of CO and H20 on Polycrystalline Gold as Studied by Synchrotron Infrared Spectroscopy

vii

B. Beden, C. A. Melendres, G. A. Bowmaker, C. Liu and V. A. Maroni ...... 433

Layered Semiconductor/Electrolyte Model Interfaces Investigated in UHV by Synchrotron Induced Photoelectron Spectroscopy T. Mayer and W. Jaegermann ................................... 451

Future Prospects for the Application of Synchrotron Techniques to Interfacial Electrochemistry C.A. Melendres . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 469

Index ................................. '" ................. 475

PREFACE

The structure of the interface between a solution and a solid electrode continues to be of great interest not only trom a practical but also a theoretical standpoint. The rate of an electrochemical reaction is completely intertwined with interfacial structure so that an understanding of the latter is basic to the control of many industrial processes, e.g., electrolysis, electrocoating, energy conversion. The limited information obtainable from classical current-potential measurements has given impetus to the development of spectroscopic, diffraction, and imaging techniques that allow examination of electrodes "in-situ" and which provide atomic and molecular level structural information. The last decade has seen a rapid growth in the development of synchrotron based spectroscopic and scattering techniques that can be applied for the "in-situ" interrogation of the electrochemical interface. The versatility and uniqueness of techniques IIke EXAFS, XANES, X-ray reflectivity, X-ray diffraction, standing waves, etc. have been amply demonstrated. The advent of more powerful synchrotron radiation sources like those being constructed in Argonne (Illinois, USA), Grenoble (France), Japan and even smaller machines planned in Korea, Taiwan, Brazil, etc. promises to provide the tools necessary not only to examine classic problems that have remained unsolved but also open up new fields of investigations in electrochemical science and technology.

A NATO Advanced Research Workshop entitled "Synchrotron Techniques in Interfacial Electrochemistry" was held in Funchal in the Portuguese island of Madeira on December 14-19, 1992 in order to bring together people interested in the use of synchrotron techniques for the investigation of interfacial electrochemical problems. The Organizing Committee of this workshop consisted of: Prof. D. Kolb (Ulm, Germany), Dr. W. E. O'Grady, (Washington, DC, USA), Dr. J. Robinson (Coventry, UK), Prof. M. G. S. Ferreira (Usboa, Portugal) who was co-director for local organization, Dr. A. Tadjeddine (Meudon, France) who was cO-director, and Dr. C. A. Melendres (Argonne, IL, USA) who acted as director. Among the objectives of the meeting were to examine the state of knowledge and recent advances in the field, as weil as discuss the prospects forfuture application of such techniques. The workshop consisted of plenary lecturers by leading authorities in each experimental technique, as weil as, by theoreticians who could put the results on a more solid foundation. Lectures on the structure of the electrode/solution interface and the nature of surface films reviewed some of the outstanding problems in interfacial electrochemistry that may be amenable to the various synChrotron techniques presently available and those under development. Other keynote lectures dealt on the theory of x-ray scattering from surfaces and interfaces, surface x-ray diffraction, x-ray reflectivity, diffuse x-ray scattering, x-ray standing waves, EXAF and XANES, as weil as infrared spectroscopy with a synchrotron source. Presentations by other invited contributors covered the results of recent work. Round table discussions addressed what the participants envisioned to be the future relevance and prospects for using synchrotron methods for the solution of problems in interfacial electrochemical science and technology.

ix

x

The participants consisted of an even mix of electrochemists, physicists and materials scientists. There was good interaction among the participants and heated discussions at times (as is to be expected when physicists and chemists get together). Motivating physicists and electrochemists to talk to each other (especially following the saga of Cold Fusion) is no easy task and we are glad to have accomplished this. The workshop was very successful in bringing together such a diverse group to discuss problems in interfacial electrochemistry. The excellent facilities of the Savoy Hotel and the idyllic setting of the island of Madeira no doubt contributed immensely to the success of the workshop. The workshop was truly a pleasant leaming experience for everyone.

This proceedings volume is a collection of most of the keynote lectures and invited talks that were presented at the meeting. Unfortunately, there are always some UHPs (Unrepentant Habitual Procrastinators) who, for one reason or another, elected not to have their lectures in print in this volume; thus we will regretfully miss them. Our purpose in this book is to help electrochemists, who are uninitiated in the use of synchrotron techniques, get a start in the use of new and truly versatile tools to solve their problems. It is our sincere hope that both novices and experts alike, not only in electrochemistry but other fields of science as weil, would find this book informative and useful in their pursuits. All the toils and long hours spent by the authors in writing their lectures will have been justly rewarded.

The organizers acknowledge with great gratitude the financial support of the NATO Science Committee through its Office of Advanced Research Workshops Programme (Prof. L. Sertorio, Director), which made the conduct of this meeting possible. The supplementary support of the U. S. Office of Naval Research (Dr. R. J. Nowak, Program Manager) and the Portuguese Council for Science and Technology (JNICT) is also gratefully recognized. Finally, sincere thanks are due to all the contributors to this vOlume, to our co-organizers of the workshop, and to all the participants with whom we had a truly wonderful time.

C. A. Melendres

A. Tadjeddine

LIST OF PARTICIPANTS

Or. Oaniel Aberdam (S) *

Laboratoire de Spectrometrie Physique, CNRS-UA8 Universite Joseph Fourier B.P.87-38402 Saint Martin d'Heres Cedex France

Prof. Luisa Abrantes University of Usbon Faculty of Sciences Lisbon, Portugal

Prof. Hector O. Abruna (L) Oepartrnent of Chemistry Comell University Ithaca, NY 14853, USA

Ms. Maria Martins Alves CNRS, LURE Batiment 2090 91405 Orsay Cedex France

Prof. Bemard Beden (S)

*

Laboratoire de Chimie I, UA-CNRS 350 University of Poitiers Poitiers, France

Prof. Antonio Bianconi (L) Oepartment of Physics University of Rome "La Sapienza" Rome,ltaly

Mr. Ahmet Bulut Oepartment of Physics University of Warwick Coventry CV47AL U.K.

xi

xii

Dr. Marcella Cappadonia Institut fur Energie Verfahrens Technik Forchungszentrum Julich GmbH Postfach 913 D-5170 Julich Germany

Prof. Brian E. Conway (S) Department of Chemistry University of Ottawa Ottawa, Ontario K1 N 6N5 Canada

Dr. Robert Cortes (L) CNRS, Physique des Liquides et Electrochimie 4 Place Jussieu 75230 Paris Cedex 05 France

Prof. Roger A. Cowley, FRS (L) Department of Physics Oxtord University Clarendon Laboratory Parks Road, Oxford, OX1 3PU UK

Dr. Manuel Da Cunha Balo CNRS, Lab. de Chirnie Metallurgique 15 Rue George Urbain 94407 Vitry-sur-Seine France

Prof. Moshe Deutsch (S) Bar lIam University Department of Physics Ramat-Gan 52900 Israel

Prof. Richard Eider (S) Department of Chemistry University of Cincinnati Cincinnati, OH 45221 USA

Prof. Tulay Eskikaya Department of Chemistry Istanbul Technical University Istanbul, Turkey

Prof. Mario G. S. Ferreira (0)* Instituto Superior Tecnico Department of Chemical Engineering 1096 Usboa Codex, Portugal

Mr. Egil Gulbrandsen Department of Chemistry University of Oslo N-03150s10 Norway

Dr. Claudio Gutierrez Instituto de Quimica Fisica "Rocasolano", CSIC Serrano 119 Madrid Spain

Dr. Antoinette Hamlin CNRS, Laboratoire d'Electrochimie Interfaciale 1 Place Aristide Briand F92195 Meudon Cedex France

Mr. Gerard Hastie Department of Chemistry Strathclyde University Glasgow GL 1xl, Scotland UK

Dr. Hugh S. Isaacs (L) Department of Applied Science Brookhaven National Laboratory Upton, NY 11973 USA

Dr. Wolfram Jaegermann (I) Hahn-Meitner Institute Glienicker Strasse 100 0-1000 Berlin Germany

xiii

xiv

Prof. D. C. Koningsberger (L) Department of Inorganic Chemistry University of Utrecht Utrecht, The Netherlands

Prof. Albert D. Kowalak Department of Chemistry University of Massachusetts Lowell, MA 01854

Prof. Jerome Kruger (L) Department of Materials Science and Engineering Johns Hopkins University Baltimore, MD 21218 USA

Dr. Andrew S. Lin Chemistry Division, Code 6170 Naval Research Laboratory Washington, DC 20375 USA

Dr. Carlos A. Melendres (0) Materials Science and Chemical Technology Divisions Argonne National Laboratory Argonne, IL 60439 USA

Dr. Kei Murakoshi Laboratoire d' Electrochimie Interfaciale, CNRS 1 Place A. Briand 92195 Meudon Cedex France

Dr. Ben M. Ocko (L) Department of Physics Brookhaven National Laboratory Upton, NY 11973 USA

Dr. William E. O'Grady (L, 0) Chemistry Division, Code 6170 Naval Research laboratory Washington, DC 20375 USA

Dr. Carlos Paliteiro Department of Chemistry University of Coimbra 3049-Coimbra Portugal

Dr. N. Papadopoulos Department of Chemistry University of Thessaloniki Thessaloniki 54006 Greece

Prof. Roger Parsons, FRS (L) Department of Chemistry University of Southampton Highfield, Southampton S095NH UK

Prof. lan Robinson (S) Department of Physics University of lilinois Urbana, IL 61801 USA

Dr. James Robinson (L, 0) Department of Physics University of Warwick Coventry CV47AL UK

Dr. Karl M. Robinson (S) Naval Research Laboratory Chemistry Division, Code 6170 Washington, DC 20375 USA

xv

xvi

Dr. Pascale Roy (S) CNRS, LURE Batiment 209D 914050rsay France

Dr. Andrea Russell (S) Department of Chemistry University of Liverpool Liverpool, UK L693BX

Prof. A. Sezai Sarae Department of Chemistry Istanbul Technical University Istanbul, Turkey

Mr. Gerhard Scherb Department of Electroehemistry University of Ulm D-.7900 Ulm, Germany

Prof. Daniel Scherson (S) Department of Chemistry Case Westem Reserve University Cleveland, OH 44106 USA

Dr. Alda Simoes Instituto Superior Tecnico Department of Chemieal Engineering 1096 Usboa Cedex Portugal

Dr. Sunil K. Sinha (L) Exxon Research and Engineering Company Route 22 East Annandale, NJ 08801 USA

Dr. Ricardo M. Souto Universidad De La Laguna Departamento De Quimiea Fisica La Laguna, Tenerife Spain

Prof. H. Henning Strehblow (S) Institute für Physikalische Chemie and Electrochemie Heinrich-Heine Universität Dusseldorf D-4000 Dusseldorf Germany

Dr. A. Tadjeddine (L, 0) Laboratoire d'Electrochimie Interfaciale, CNRS 1 Place Aristide Briand F92195 Meudon Cedex France

Dr. Carol Thompson (S) Physics Department Polytechnic University Brooklyn, NY 11201

Dr. Michael F. Toney (L) IBM Almaden Research Center San Jose, CA 95120-6099

Prof. Michael Weaver (L) Department of Chemistry Purdue University West Lafayette, IN 47907 USA

Dr. Gwyn P. Williams (L) National Synchrotron Light Source Brookhaven National Laboratory Upton, NY 11973 USA

Dr. Kozo Yoshikawa Takasago Research and Development Center Mitsubishi Heavy Industries Inc. 2-1-1 Shinhama, Arai-cho Takasago, Hyogo Prefecture 676 Japan

xvii

xviii

Dr. Jorg Zegenhagen (L) Max Planck Institut für Festkörperförschung Heisenbergstr. 1 D-7oo0 8tuttgart 80 Germany

* L), (0), (8) indicates principallecturer, organizer, andlor invited speaker, respectively.

SYNCHROTRON RADIATION AND INSTRUMENTATION

J. ROBINSON Department 0/ Physics University 0/ Warwick Coventry CV4 7AL Great Britain

ABSTRACT. Abrief history of the development of synchrotron radiation sources is presented. The basic design criteria for storage rings as sources of synchrotron radiation are described, the utilization of insertion devices is outlined, and the properties of synchrotron radiation that make it such a useful and unique source are discussed. Associated instrumentation including monochromators, mirrors and detectors is also considered.

1. Introduction

We are inclined to regard synchrotron radiation (SR) to be a modern development since it is associated with high technology installations such as partiele accelerators. In fact it has been around far longer than any of uso SR is produced when any charged partiele, travelling at a velo city elose to the speed of light, experiences a radial acceleration, and travels along a path with radius of curvature that is sufficiently large that quantum mechanical effects are negligible. These conditions can exist in out er space, for example in a super nova, however, for us to be able to take advantage of the unique properties of SR the source has to be positioned here on earth.

This artiele is concerned primarily with a discussion of some of the design criteria for SR sourees, and of the properties of the radiation that have lead to its widespread and ever increasing application as an experimental tool. Some of the ancillary instrumentation that is required to make effective use of SR will also be described. Before presenting this material abrief outline of the historical developments will be given.

2. The Historical Development of Synchrotron Radiation

The development of a theory of SR effectively has its origin at the turn of the century when Larmor [1] proposed a theoretical treatment for the emission of electromagnetic radiation by accelerated charged partieles, and more specifically when Schott [2] showed that an electron moving on a circular orbit is a strong sour ce of electromagnetic radiation. This latter observation was actually part of an attempt to develop a elassical model for a stable atom but it was forgotten when Bohr's model was developed. Interest was re-kindled in the 1940's, when it was realised by Ivanenko and Pomeranchuk [3] that SR would limit the maximum energy that could be achieved in a partiele accelerator, such as a betatron, and

C. A. Melendres and A. Tadjeddine (eds.), Synchrotron Techniques in Interfacial Electrochemistry 1-19. © 1994 Kluwer Academic Publishers.

2

many theoretical developments followed. The first experimental evidence far SR was in fact indirect, when Blewett [4] in 1946 measured the contraction of the orbit in a betatron, caused by the loss of energy to SR. Unfortunately he was unable to see the SR, as there were no appropriate windows in the vacuum vessel. The first sighting occurred a year later at the General Electric Laboratories in Schenectady, on one of the first electron synchrotrons [5], and hence the name of synchrotron radiation.

At this point one might have expected SR research to take-off and many applications of SR to be rapidly developed. However, for about 10 more years SR continued to be considered at best a curiosity and at worst a serious problem, in that it limited the behaviour of electron accelerators. It was not until 1956 that the potential of SR as a light source for far-UV jsoft-x-ray spectroscopy was recognised by Tomboulian and Hartman [6]. It then took a further 5 years for this potential to be realised [7]. At about the same time as these developments the opportunities for the use of SR in the hard-x-ray region also began to be appreciated.

With hindsight this progress appears to have been very slow. Whilst there are probably many reasons far this, two stand out as the principal ones. Firstly, the accelerators then available were designed and built for the particle physicists, and particle physics research, and therefore SR was only available as a by-product of this use of these machines (the so­calied parasitic mode of operation). This did not encourage applications, since machines and operating procedures optimised for the particle physicists were generally not appropriate for SR use. Secondly, the research communities that were eventually to become SR users were not accustomed to working away from horne on big-science machines, in the way that particle physicists were, and had to be convinced of the potential of this new light source befare they could be dragged away from the comforts of their horne laboratory bench.

From the early 1960's SR user communities of early converts began to develop at a nu mb er of facilities around the world, but it was not until ab out 1970 that the first storage ring dedicated to SR, TANTALUS in Wisconsin, USA, became available, though even this machine was not originally specifically designed for SR use. A few years later the first dedicated, purpose built, SR sour ces became available (the so-called second-generation sour ces ) and developments have continued ever since, so that there are now many machines around the world, serving an extremely diverse user community of many thousands of experimentalists. With the experiences gained from these second-generation machines new design strategies for SR sources have been developed and we are now moving on to third­generation machines such as the ESRF in Europe and the APS in USA. As will be explained shortly, these new machines have progressed from a reliance on the bending magnets as SR sources to the use of insertion devices, such as wigglers and undulators, tailored to very specific parts of the electromagnetic spectrum, and to particular types of experiment. When these new machines become available the vastly increased light flux at the sampie will make possible a whole new range of experiments, particularly in the areas of the very dilute systems and time resolved studies which, of course, are particularly interesting to electrochemists.

Before discussing the generation of SR a comment about costs is probably appropriate. As anyone presenting a talk about the use of SR to a group of electrochemists will be aware one of the first quest ions asked will be, "but how much does it all cost". The original cost of building an SR source is of course very high, but when one considers that the machine will, hopefuliy, last far several decades, and serve a large number of simultaneous users, the actual costs per experiment become much more reasonable. In reality these costs are not significantly different from those associated with running an electron microscope or a SIMS machine. On a more practical note the facilities are already in existence and, at least for academic users, the user is not generally charged far beamtime, so electrochemists might

.-- experimental station ~

L-J'"-=~~

bending magnet? ----.p

booster electron path

Figure 1: A schematic diagram of a storage ring synchrotron radiation source.

3

just as weil get out there and use their share of SR! The message should be that this is small, not big, science.

3. Synchrotron Radiation Sources

As we have already said the first observed source of SR was an electron synchrotron. As a practical source the synchrotron suffers from a major problem in that the accelerated electron beam, and hence the associated SR, decays very rapidly with time. This is really not acceptable for most types of experiment and therefore SR sources are now exclusively based on storage rings, where the charged particles (usually electrons) are held in a closed orbit for many hours.

3.1. THE ELECTRON STORAGE RING

A schematic diagram of a storage ring is shown in Figure 1. For simplicity it will be assumed in this discussion that the charged particles circulating in the storage ring are electrons. This is the most common situation though as will be seen later there are actually some advantages, and disadvantages, to using positrons, though heavier particles are unsuitable.

The basic components of this storage ring are:

• A source of electrons.

4

• An injection system that accelerates these electrons, and injects them into the storage ring.

• An ultra-high vacuum system that encloses the electron path in both the injector and the storage ring.

• Dipole magnets which bend the electron beam into the required circulatory orbit.

• Various other magnetic devices to provide focusing and steering of the electron beam.

• A radio frequency cavity which serves to restore the energy that the electrons have lost because of the emission of SR.

• Radiation shielding.

• A sophisticated computerised control system.

• Beamlines to take the SR from the ring to the experimental areas.

3.1.1. The Particle Source. Let us first see why it is that electrons and their anti-particles, positrons, are the favoured charged particles for the generation of SR. The first requirement is clearly that the particle should be stable, or it would decay into other partides in a short time, with a consequent loss of SR. This realistically limits the choice to electrons, protons, and their corresponding anti-particles. If we now consider the radiated power, P, emitted by a charged particle of mass, m, and energy, E, following a circular orbit of radius, r, the choice becomes dear. This radiated power is given by

cq2 E 4 p=-----'-----

67rEor2m4c8 (1)

where Eo is the permittivity of free space, cis the velo city of light and q is the charge on the particle. The inverse dependence on the fourth power of the particle mass means that the radiation intensity from protons will be 18364 less than from electrons under identical conditions: clearly an overwhelming factar.

The choice between electrons and positrons is not so clear cut. If the production of these two particles was equally efficient, then positrons would be favoured as they interact less with residual particles in the vacuum chamber of the storage ring, resulting in significantly longer beam lifetimes. Unfortunately, whilst electrons are found in every solid and an electron beam is easily generated from a heated cathode, positrons are only created in a pair-production process when accelerated electrons are collided with matter. This process is not very efficient and therefore it is difficult to obtain large positron currents. For this reason most storage rings operate with electrons, and the electron source is simply a heated cathode.

3.1.2. Particle Accelemtion. The next stage in the process of obtaining SR is to accelerate the electrons up to relativistic velocities. As we will see shortly most storage rings operate at energies in the few GeV range and therefore the electrons must be accelerated up to this value. For practical reasons this is usually a multi-stage process.

The simplest form of accelerator is a linear accelemtor, or LINAC, in which the electrons are simply accelerated in a straight line by an electric field. For technical reasons the acceleration using a static field is limited to a few MeV and so for energies greater than

5

this radio frequency LINACS are used [8]. With these devices it is impossible to inject directly into a storage ring at, or even near, its operating energy because the LINAC becomes prohibitively long at such energies. One way of resolving this is to use a microtron which is effectively an RF LINAC wrapped around on itself [8]. Using this device final energies of up to 100 MeV can be achieved in a reasonably compact device. This energy may be sufficient for injection into a storage ring designed for vacuum ultra-violet (VUV) and soft x-ray radiation, though the electrons will still need accelerating further in the storage ring. This process is known as ramping, when the storage ring operates rat her like a synchrotron, and the magnetic fields in the bending magnets are increased in synchronony with the input of energy via a radio frequency cavity. Once the final energy is reached the magnetic fields remain constant, and energy is only input at a rate sufficient to replace the energy lost to SR.

For storage rings designed to provide hard x-rays the injection energy has to be higher than can be achieved with a microtron, and so extra acceleration is provided by a booster synchrotron, where the electrons are usually accelerated up to the operating energy of the storage ring, prior to injection. Injecting at the energy of operation of the storage ring has the advantage that in principle, at least, the storage ring can be "topped up" during operation to replace electrons that have been lost.

3.1.3. The Vacuum System. The electrons circulate around the storage ring in a narrow stainless steel tube, or beampipe. One of the principal factors that limits the lifetime of this electron beam is scattering by residual gas particles and therefore the desirability of maintaining a good ultra-high vacuum (UHV) within the beampipe is obvious. In conven­tional UHV work it is usual to aim for apressure in the 10-11 Torr range and with a typical surface science vacuum system, for example, such apressure can be achieved fairly rapidly (a couple of days). With a storage ring it may take several weeks (or even longer), due to its greater complexity, greater size, and the effects of SR, which stimulates desorption of particles adsorbed on the inside of the vacuum vessel. It is dear that once this vacuum is achieved it should be retained as long as possible, and therefore it is weil protected by fast valves, so that, at least for most types of experiment, it is effectively impossible for a SR experimentalist to do anything that will cause the vacuum to be lost. In the early days of parasitic use this was not always so and a number of experimentalists have been severely embarrassed and others terribly aggravated!

3.1.4. The Bending Magnets. The beampipe of the storage ring does not foilow a circular path, as has perhaps been implied, but consists of a number of straight sections joined together by bends, often, though not necessarily, to form an approximate cirde. At each bend in the beampipe the path of the electron beam is changed by a dipole magnet, which is constructed in the form of a C, as shown in Figure 2, so that the beampipe can pass through the gap. For relativistic electrons the bending of the electron beam by the magnet is described by,

r

1 0.2998B = E

(2)

where r is the radius of curvature in metres, Eis the energy in GeV and B is the field in Tesla. The magnetic fields typically used are around 1 or 2 Tesla, which can be achieved with conventional, Le. non-superconducting, electromagnets. Recently however, some thought has been given to using superconducting bending magnets, even where conven­tional ones are available, because of their compactness which would permit an experimental station to be placed much doser to the source, thereby increasing the available fiux density at a sampie.

6

i ron yoke

~I - ---

~I

- ---~ ,

, . l __ --"/COI S /

Figure 2: A schematic cross section through a bending magnet.

3.1.5. Focusing Magnets. A small source size is an important parameter for many SR applications and this requires that the electron beam be focussed. This is achieved by placing quadrupole magnets at appropriate points around the ring. Unfortunately whilst a single quadrupole focuses in one plane, e.g. the horizontal, it will defocus in the other. This can be overcome, and net focusing be achieved, by using pairs of quadrupoles, one focusing and the other de-focussing in a given plane. If the focallength of the former is I and of the latter - I, then, if they are a distance d apart the net focusing, Itotall is given by

J2 Itotal = d (3)

Within the electron beam there will always be a small spread of energies and therefore the focal length of the quadrupoles, which depends on the beam energy (chromaticity), will vary for different electrons. This can lead to instabilities in the electron beam, which must be corrected by the use of sextupoles. In a typical storage ring there are a number of these quadrupole pairs and correcting sextupoles distributed around the ring. The configuration of the bending dipole magnets, and the other non-dipole magnets, is referred to as the lattice of the ring.

3.2. STORAGE RING PARAMETERS

The are very many parameters of storage rings which, to a lesser or greater extent, affect the generation, and properties, of the SR emitted. Some of the more important of these will now be discussed.

3.2.1. The Energy. The operating energy is one of the most fundamental parameters that characterizes a storage ring and for the purposes of the production of SR an energy of about 10 GeV can be regarded to be the maximum practical value. There are really two reasons for this, both associated with SR production. Firstly, the energy lost by the circulating electrons to SR must be replaced in the RF cavity, and for energies in excess of 10 GeV this will require megawatts of power at a prohibitive cost. Secondly, the SR power, as shown in Equation 1, increases as the fourth power of the energy. Much of this power is not used in experiments, but must be absorbed and removed as heat, which can prove

7

to be difficult. This problem can be eased to some extent by increasing the size of the ring thereby distributing the power dissipation, but since the improvement only goes as the square of the radius (see Equation 1), the required ring size increases very rapidly with increasing energy. It is generally thought that LEP at CERN, at 100 GeV and with a circumference of 27 km, will probably be the largest storage ring for electrons ever buHt.

3.2.2. The Ring Gurrent. From an SR experimentalist 's point of view it would appear that the higher the current the better, as this should lead to higher light intensities. From the storage ring design standpoint it is not so simple. As we saw above in the discussion of the ring energy, increased synchrotron radiation power can lead to problems in dissipating the heat. This, however, is not usuaily the limiting factor as far as the maximum ring current is concerned. Instead it is the problems associated with keeping the tightly packed bunches of electrons circulating around the ring stable, since large currents can in du ce undesirable osciilations which feed back to the beam, are amplified, and eventually result in beam loss. Current SR sources typically have maximum currents of up to about 500 mA.

3.2.3. Bunch Size and Number. The electron beam in the storage ring is not continuous but consists of one or more bunches of electrons. This is a natural consequence of the way the energy is restored to the beam via the RF cavity: the RF field effectively establishes aseries of circulating potential weils, or buckets, in which the electrons are stored stably. The field strength is usually chosen so that an electron that emits photons (to SR), and hence changes its moment um , cannot escape from its bucket before returning to the RF cavity, thus maintaining the ring current and not adversely affecting the lifetime. The total number of buckets that can exist is determined simply by dividing the ring circumference by the wavelength of the RF, and is typicaily up to ab out 100. For kinetic work the storage ring may be run with as few as one bunch, and if a narrow potential weil is used light pulses as short as a 10-12 s can be obtained. For normal multi-bunch use the size of the potential weils is chosen to maximise the beam lifetime and the light source is generally treated as continuous.

3.2.4. Beam Position Stability. For many experiments, particularly those using small photon spot sizes, it is vital that over periods of several hours (preferably a complete fiil cycle), the beam does not move significantly with respect to the sample. When we realize that the experiment may be many 10's of met res from the source we can see that this is a severe requirement. Beam movements can arise from a number of thermal, vibrational and electrical causes but careful design of buildings, power supplies, temperature control systems etc. can do much to aileviate problems. However, for the utmost stability, active regulation measures, utilizing beam position monitors and steering magnets, are required and in this way it is currently possible to stabilize the source point to ab out 10 fJ,m and 1 fJ,rad.

3.2.5. Beam Lifetime. The stored beam current in a storage ring typicaily undergoes an exponential decay. The lifetime of this decay is therefore an important parameter to the experimentalist, as it determines how rapidly the photon flux will change during an experiment, and how long it will be before the flux is insufficient to be useful, the latter situation usually resulting in the beam being dumped and the ring being re-filled.

The two most important effects limiting beam lifetime are: interactions of the electron beam with residual particles in the vacuum vessel, and electron-electron scattering in the electron bunches (the Toushek effect). As we have already seen the residual particle effect is reduced by retaining a good vacuum. The Touschek effect results in changes in the electron

8

momentum, much like SR emission, so can to some extent be compensated for within the RF cavity. For electron storage rings in optimum condition, lifetimes are typically in the range of a few, to a few tens of hours. To provide optimal conditions for experiments this means that the beam may be dumped and re-filled up to two or three times over each 24-hour period, each refill typically taking up to about an hour (this varies greatly between storage rings depending in part on the precision of beam positioning and also on the number and complexity of insertion devices).

One of the great frustrations in SR experimentation is unscheduled beam loss. Whilst much has been done in the way of complex computer control systems to minimize this problem, nothing can be done to prevent, for example, brief losses of power due to lightning strikes, resulting in the inevitable loss of the beam. Given their complexity it is perhaps surprising how reliable modern SR sources are.

3.2.6. Beam Size, Beam Divergence and Emittance. It has already been pointed out that the beamsize is an important parameter. If everything behaved ideally, then all electrons would follow the same path with the same energy, and the effective source would be infinitesimally small, but of course the real world is not like this; there will always be a spread of energies, whilst various osdllations result in a spadal spread. A convenient way to describe the quality of an electron beam is by its emittance, which is given by 1/7r times the area of the phase space diagram in which the particles of the beam are located. The emittance, which is fixed for all points on the electron trajectory, is a useful parameter for comparing different storage rings as it has a significant influence on properties of the SR, such as brightness. It is proportional to the square of the energy of the ring and also depends strongly on the magnetic lattice. For most purposes, everything else being equal, the smaller the emittance the better.

The SR experimentalist is not directly interested in the emittance, but is interested in beam size and divergence. These two parameters, unlike the emittance, are not fixed for the ring but also depend on the local optics. It is worth noting that the way the electron optics of the ring interact is very complex, and it is generally impossible to tailor the divergence, or beam size, at a particular point without affecting the values elsewhere on the ring.

4. Synchrotron Radiation

The previous discussion clearly shows what a complex and expensive process it is to generate SR, so why is it done? The answer is simply that SR has a number of unique properties that make it an extremely powerful and versatile experimental tool. The most important of these properties are

• It is a very intense light source.

• It has a broad, useful, spectral range extending from the infra-red to the hard x-ray region.

• It is highly collimated.

• In the plane of the storage ring it is 100% plane polarized, whilst above, or below, this plane it exhibits elliptical polarization.

9

electron orbit electron orbit - '"

radial acceleration radial acceleration

a b

1/y

Figure 3: The radiation pattern from: (a) a non-relativistic and (b) a relativistic radially accelerated electron

4.1. THE SPECTRAL DISTRIBUTION

To understand why SR has a broad spectral distribution we must consider the generation process. If we consider first of all a non-relativistic electron experiencing a radial accel­eration, then the distribution of the dipole radiation is given by P oe sin2 (J, in the rest frame of the electron, as shown in Figure 3a. For a relativistic electron this pattern must be transformed by the Lorentz transformation into the laboratory frame. As shown in Fig­ure 3b the result of this transformation is a small forward pointing cone of opening angle (J ,...., 1/1, where 'Y is the energy of the electron in units of its rest energy, i.e.

'Y = E/Eo (4)

where Eo is 0.511 MeV. To understand the frequency spectrum we must now imagine looking down a beamline

whilst this cone of radiation sweeps past. The cone will pass the observer in a time of approximately r /l3c, where r is the radius of curvature and c is the velocity of light. For an imaginary 2 GeV machine '"'I is approximately 4000 whilst r might be 4 m, then the light passing the observer will appear as a 0.25 x 10-18 s wide pulse. To obtain the spectral distribution in the frequency domain we must take the Fourier transform of this short pulse and we find that it contains the harmonics of the orbit frequency up to v ~ '"'I3c/r , or in our example 4 x 1018 Hz (,....,4 keV). In a real storage ring there is not one electron but many, and we obtain a continuous spectrum over the frequency range, rather like that for black body radiation, as opposed to a line spectrum.

This simplistic qualitative treatment also explains the excellent collimation of SR as being due to the narrow opening angle of the radiation cone.

4.2. A COMPARlSON WITH OTHER LIGHT SOURCES

In view of its unique properties it is not immediately apparent how best to compare the intensity of SR with that of other forms of radiation. However, in the final analysis, it is the performance in a particular experiment that matters, and therefore adefinition of intensity that permits comparison of different sources in this respect is desired. The figure of merit

10

oS 1013

'0 'ii

1012 -g <'CI ..c

1017

t-t laser

~ 1011 .....

ci CuK

........ .... §

1010 ........ ....

1 109 ........

(.J

x-ray tubes

CI) CIl CK ........ CIl

= AlK 0 108 Ö ..c: Cl.

107

10-1 100 102 H1"

wavelength I A

Figure 4: A comparison between the spectral brightness of SR and various other light sources

that is used is the spectral brightness (sometimes incorrectly called brilliance), since this parameter is not affected by ideal optical components in the experimental configuration. It is measured in photons per second, per mm2 source area, per mrad2 source divergence, per 0.1% spectral bandwidth:

I(x,z, 8, "p,E, t)[Phot./(s mm2 mrad2 O.l%tl.E/ E] (5)

Using this definition comparisons can be made with other types of light source as shown in Figure 4. It is immediately apparent that, with the exception of the laser, SR is by a significant factor the brightest source. In the x-ray region, for example, exceeding the Bremsstrahlung from a conventional fixed anode x-ray tube by some 5 orders of magnitude, and the Ka lines by 2.

It can be seen from Equation 5 that brightness is defined per unit source area. For experiments that can effectively utilize high brightness it is therefore clear that if everything else remains constant a small source size is desirable. In general a small source has high brightness. For experiments requiring high brightness it is also important that the beam divergence is low, or large and expensive optical components will be required to col1ect all of the light.

11

For some experiments it is not possible to take advantage of high brightness, since, for example, concentrating too many photons in a small area may damage the sampie. In such cases a more useful figure of merit for comparisons is the spectral flux, which is the brightness integrated over the source area, and over all vertical angles. Clearly this parameter is insensitive to source area and beam divergence.

4.3. PROPERTIES OF HENDING MAGNET RADIATION

We have considered some of the properties of SR in very general and qualitative terms. We will now look at some of them in more detail.

4.3.1. The Radiated Power. Equation 1 describes some of the most important properties of SR generated in a bending magnet. When we consider a large number of electrons circulating around the storage ring then the total power in kW, Ptotal, is given by

n _ 88.5E4[ rtotal -

r (6)

where Eis the energy in GeV, r is the bending radius in metres and I is the circulating current in amperes. Or we can use Equation 2 to eliminate the bending radius to obtain

Ptotal = 26.5E3 I B (7)

where B is the bending magnet field in Tesla. As an example if we consider the SRS at Daresbury, which operates at 2.0 GeV with 1.2 Tesla magnets and with a typical current of 200 mA, then the total radiated power is approximately 50 kW. The SRS is designed primarily for x-ray use. For a VUV I soft x-ray machine the operating energy is more likely to be about 1.0 GeV and therefore the radiated power will be lower. These figures give some idea of the total power output from typical current machines. The output from the next generation rings such as the ESRF will be much higher.

The above discussion describes the total power radiated but says not hing about the spectral or angular distribution. We have already looked qualitatively at these properties but we will now attempt to quantify them a little.

4.3.2. The Radiation Spectrum. In describing the radiated spectrum an important para­meter is the critical wavelength, Ac. This is defined as the wavelength that divides the power spectrum into two equal parts, and is given by

18.6 5.6r Ac ~ (BE2) ~ E3

It is now possible to show that the spectral flux distribution, N(A), is given by

N(A) ~ 2.5 X 1014EG1I (:J where G1 is a universal function given by

G1(y) = y loo K!(x)dx

(8)

(9)

(10)

and where K A (x) is a modified Hessel function of the second kind. The actual form of Equation 9 is 3 shown in Figure 5 in what is known as the universal flux curve, where the

12

-5 "0 1011 .if; "0 C cu

,l:l

~ 1010

6 -.. > <U C) -..

109 "0 cu ... E .......

~ ....... 108 C) QJ CI)

....... CI)

s:: 0 "0 107 ..c Q..

10-1 10° 101 102 103 lQ4

wavelength (/.../')..)

Figure 5: The universal distribution of photon flux from a bending magnet.

flux has been normalized to unit energy (GeV) and current (mA), and a bandpass ofO.1%. This universal curve enables values for any specific bending magnet to be obtained readily.

Useful flux is normally assumed to extend down to a wavelength of 0.1 Ac and peaks at 4 Ac. At longer wavelengths the flux falls off slowly as shown in Figure 5. Again using our example ofthe SRS at Daresbury, for which Ac is 3.9 A, the peak photon flux under normal operating conditions is approximately 1013 photons S-l mrad-1 in a 0.1 % bandpass. This is a typical figure for a bending magnet source and Figure 6 shows a comparison of several such sources, and of some wiggler beamlines.

4.3.3. Polarization. For many experiments the polarization of the light beam is important, either because it is actually used in the experiment, or because it affects the way that data are analysed. In the horizontal plane the beam is 100% plane polarized but as we move above and below it there is also a vertically polarized component as shown in Figure 7.

The degree of polarization depends strongly on the wavelength of the radiation; at a given vertical angle the fraction of parallel polarized light increases with decreasing wavelength. It should also be noted that the parallel and perpendicularly polarized components are phase correlated, and therefore the emission out of the plane of the ring is elliptically polarized.

5. Insertion Devices

In the early days of SR light was obtained exclusively from the bending magnets. This

o

1 -.. U IIJ CI> -.. ~ o 'ö -a

10.1

13

SRS (wiggler) ! ,,- .. ..::::-.- ---.. . ......"... SRS (bm)

/ . . ~ .. ~~\ / ..-- - :--.... /~DCI ---

./ j/ "/

/1

4--""':'/ '--ESRF (bm)

wavelength / Ä

Figure 6: A comparison of photon fluxes from various synchrotron radiation sour ces (the SRS wiggler is a wavelength shifter whilst that on the ESRF is a flux enhancer).

meant that the spectral properties of the light entering each beamline on a given storage ring were very similar- It was then realised, that by passing the electron beam through periodic magnetic structures in the straight sections of the ring, it was possible to create new types of light source. These structures, which are known as insertion devices, come in two distinct forms; wigglers and undulators. Hut whilst their spectral properties are very different, structurally they are very similar. One of the great benefits of using insertion devices is that light sources can be tailor-made to specific applications, without significant interaction with other sources in the ring. It is for this reason that the next generation of SR sources will rely very heavily on such devices.

An insertion device is shown schematically in Figure 8. It consists of a periodic array of magnetic poles through which the electron beam passes without any net deviation. The magnetic array is usually arranged so that the electron beam experiences a sinusoidal magnetic field in the vertical plane, and therefore its trajectory varies sinusoidally in the horizontal one. An important parameter in characterizing an insertion device is the wiggler parameter, J(, given by

J( = 93.4AoB o (11)

where Ao is the period of the magnetic array in met res and Bo is the peak magnetic field.

14

1.0 horizontal

.-.,

.0 .. ..s '-'

.S 0.5 jg cu .5

0

0 2 4 6 8

vertical angle Y'l'

Figure 7: The relative intensities of horizontally and vertically polarized components of bending magnet radiation as a function vertical angle, for A = lOOAc

Figure 8: A schematic diagram of an undulator or wiggler. The curve shows the electron trajectory in the horizontal plane.

15

In terms of 1(, the maximum angular deflection of the electron beam, {j, is

(12)

where , you will recall is the energy of the electrons in units of their rest energy. For the case where I( < 1, {j is less than the opening angle of the radiation cone (see

Figure 3), and strong interference is exhibited between successive periods ofthe oscillations of the electron beam. This is the basis of the operation of an undulator. When J( > > 1, interference effects are less important and the device is called a wiggler. It is of course also possible to construct devices where the peak field is variable, and thus by altering the field change from one type of device to the other. As with ben ding magnet radiation the light emitted from an insertion device is linearly polarized in the horizontal plane. Out of the plane the light remains linearly polarized but the direction of polarization changes in a complicated way.

5.1. WIGGLERS

When I( is large, and the angular deviations of the electron beam are large, each point on the electron path emits a spectrum similar to that of a bending magnet with a critical wavelength of

A _ 18.6 c - E2B (13)

where B is the magnetic field at that point. These individual sources add together inco­herently and the wiggler emits a cone of radIation with a horizontal opening angle of 2{j.

The spectral flux emitted by one half period of the wiggler oscillation is approximately the same as that emitted by a ben ding magnet with the same Ac as given by Equation 13. Consequently if there are N periods in the wiggler the flux is nearly 2N times that of the bending magnet. Thus a use of a multipole wiggler is to increase the flux.

A second use of wigglers is to shift the value of Ac to smaller values, thus extending the usable wavelength range. For this purpose few poles are used (often only 3) and superconducting magnets are utilized to achieve the high fields required. In this way the critical wavelength can be brought down to a value that would require a much higher energy storage ring if only bending magnets were available. Figure 6 shows examples of the use of wigglers for both flux enhancement and wavelength shifting.

5.2. UNDULATORS

When I( is small, and interference effects are significant the emISSIOn spectrum of the insertion device is no longer continuous, but exhibits sharp peaks at the fundamental frequency, VI, and its odd harmonics. VI is given by

2q2( 1(2 )-1 VI Co: -- 1 + - + ,2(J2

Ao 2 (14)

where (J is the viewing angle in the horizontal plane, with respect to the axis of the un­dulator. This dependence on (J means that the device is not strictly monochromatic. The relative bandwidth of the spectral peaks is given by

ßVn ~ _1_ nN

(15)

16

1019

2.3

.s 1 "0 1018

·i "0

\ \

§ \ \ ..c \ \ ~ \ ,

1017 \ \ \ s:! \ \ \ ....

"0 \ ~ 1

..... I

~ 1016 I ..... I u I QJ

8.0-1 '" I ..... I

~ 101S I I I

I 5.5 I • I I I -a

I I 1~2.3 I

3.5 ~ I I I

1014

10.1 100 101 1~

photon energy / ke V

Figure 9: The tuning range of some undulators eonsidered for the ESRF. The solid eurves refer to the fundamental and the broken ones to the third harmonie. The numbers on the eurves refer to the period of the undulator in em.

where n is the harmonie, whilst the intensity varies as N 2 • These properties make it desirable to eonstruet undulators with as many periods as possible, but this is limited by teehnieal problems associated with the limited length of the straight seetions in the ring, and the minimum period that ean be eonstrueted.

In order to tune the radiation from the undulator one of the parameters in Equation 14 must be ehanged. The only praetieal one to vary dynamieally is K whieh ean be ehanged by varying the vertieal gap between the magnetie poles; the so-ealled clamshell undulator. Clearly it is not possible to vary the gap over a wide range and therefore the degree of tunability is limited. As an example Figure 9 shows the tunability ranges of some undulators that were eonsidered for the ESRF during its design stage.

We have already seen that the radiation from an undulator is linearly polarized. An interesting deviee ean be eonstrueted by effeetively having two undulators orthogonal to one another and that are spacially phase shifted by 11"/2. This arrangement, a helical undulator, produces intense cireularlY polarized light [9].

17

5.3. THE FREE ELECTRON LASER

Laser radiation as we all know is monochromatic, highly collimated and coherent. In a conventionallaser, it is generated by the amplification of certain characteristic frequencies of electromagnetic radiation through a stimulated emission process in an appropriate medium. In a free electron laser (FEL) electrons travel through a helical undulator along with a plane electromagnetic wave. Under the resonance condition

\ _ >'0 ( J(2) 1\-- 1+-

2,2 2 (16)

where >. is the wavelength of the electromagnetic radiation and >'0 is the length of the undulator period, energy can be transferred to the electromagnetic wave. The details of this process are beyond the scope of this article but have been presented elsewhere [10].

In re cent years FEL's have been developed to operate in the infra-red region of the spectrum, where they provide very intense sources. Operation in the soft x-ray region is also feasible but operation with hard x-rays appears to be unlikely.

6. Beamline Optics

It would be possible to write a book on the topic of beamline optics, e.g. monochromators and mirrors. The aim of this section therefore will simply be to give a very brief overview of the components that are required to get the light from the storage ring to the experiment in an optimal manner. After the light has interacted with the sampie it will then need to be detected, but the requirements here are very closely tied to the type of experiment being conducted and therefore discussion of detectors in all but very general terms is inappropriate.

The light path from the bending magnet, or insertion device, is enclosed in an evacuated pipe, and as we have already seen the vacuum of the storage ring is usually protected from failures in the beamline by fast valves. The front-end of the beamline mayaiso contain appropriate windows, e.g. of beryllium in an x-ray beamline, which can filter out undesirable parts of the spectral range. It is also likely that the light from a particular sour ce point is divided between a number of experimental stations rather than going to a single one. The beamline will then contain the relevant mirrors and monochromator to deliver light onto the sampie. Most electrochemical SR users utilize x-ray radiation. For x-ray beamlines the experimental station is a radiation proof hutch which is interlocked for safety reasons. If it is desired to change anything, e.g. electrode potential, whilst the sampie is being irradiated, this must be done remotely.

Great effort and expense has been devoted to the generation of SR and therefore it is important that the experimenter makes optimal use of it. In electrochemical applications this generally means maximizing the fiux on the sampie (electrode) without sacrificing resolution. As so often when combining scattering and spectroscopic techniques with elec­trochemistry, cell design is a major problem and the other chapters in this book will show some solutions. For most applications the requirements of the electrochemical SR user, as far as the light is concerned, are not significantly different from those of other user communities performing similar experiments.

18

6.1. MONOCHROMATORS

Almost every experiment performed on a storage ring requires the use of a monochromator. In the low energy region from the infra-red to the soft x-ray region this will usually mean using a grating, whilst crystal monochromators can be used throughout the x-ray region. The actual design of monochromator depends a great deal on the type of experiment. For example in x-ray diffraction most work will be done at a fixed wavelength, and even if it is desired to change the value, e.g. for anomalous scattering measurements, there is no need to be able to do so rapidly. On the other hand in x-ray absorption measurements it is necessary to be able to scan accurately over energy ranges of 1 or more ke V, and to do so quickly.

For x-ray experiments the most widely used monochromator crystals are silicon and germanium, and it is usual to use a double crystal configuration because of the convenience of maintaining the optics in line. This also means that the second crystal can be de­tuned slightly to minimize the effect of harmonics. The first crystal in the monochromator experiences a high thermal load and therefore requires cooling if its properties are to remain stable. Such crystal heating is likely to become a major problem with the next generation of machines, and may ultimately limit the" available Hux densities. This has lead to new crystal materials being sought and of these diamond looks to be a promising candidate paticularly since it has a 15 times higher thermal conductivity than silicon.

Outside the hard x-ray region most monochromators are based on diffraction gratings, usually operating in the reHection mode. Most of the developments with this type of monochromator have been directed at the VUV part of the spectrum which is of little interest to electrochemists and therefore they will not be discussed here, which is a pity as some of them have such evocative names, e.g Grasshopper.

6.2. MIRRORS

To be able to fully utilize the photons emitted from the storage ring we endeavour to match the emittance of the ring with the acceptance of the monochromator, which requires focusing optics. In the x-ray part of the spectrum lenses are not available and therefore focusing mirrors must be used instead.

The refractive index for x-rays in a vacuum is greater than that in a solid, such as a metal. To obtain total external reHection at a vacuum/metal interface therefore requires angles of incidence less than the critical angle, which may be as small as 0.1 degree. This means that x-ray mirrors must be extremely Hat if they are not to introducespurious effects. The usual mirror substrate is highly polished fused silica; which is then coated with a thin film of a high atomi-c number metal, such as platinum. Focusing is achieved by bending the mirror in one, or more, planes. Usually the bending remains fixed during an experiment but can sometimes be varied dynamically for example during an energy scan. As with monochromator crystals thermalloading can become a problem.

6.3. DETECTORS

As we have already seen the new light sources promise greatly increased Hux densities, and total Huxes. The thermalloading problems in mirrors and monochromators have also been identified. Assuming that these are overcome thereremains one weak link and that is in detectors. Many types of position sensitive detector fpr x-ray scattering experiments are already suffering from saturation effects, as are the solid state detectors used in the Huorescence detection of x-ray absorption. H there is one area where significant further

19

work is required (if our hopes for these new sources are to be realised), it is in detector development.

7. Closure

It is certainly possible to successfully conduct experiments using SR without any under­standing of how the radiation is produced, and brought to the experimental area. However most experimentalists would find this situation to be rat her unsatisfactory; the storage ring is rat her a large object to treat as a black box! The aim of this article has been to provide abrief introduction into the area of machine physics and hopefully after reading it the generation of SR is less of a mystery.

8. Referenees

[1] J. Larmor, Phil. Mag. 44(1897)503.

[2] G.A. Schott, Ann. Phys. 24(1907)635

[3] D. Ivanenko and J. Pomeranchuk, Phys. Rev. 65(1944)343

[4] J.P. Blewett, Phys. Rev. 69(1946)87

[5] F.R. EIder, A.M. Gurewitsch, R.V. Langmuir and H.C. Pollock Phys. Rev. 71(1947)829

[6] D.H. Tomboulian and P.L. Harman, Phys. Rev. 102(1956)1423

[7] R.P. Madden and K. Codling, Phys. Rev. Lett. 10(1963)516

[8] W. Scharf,"Particle Accelerators and Their Uses", Harwood Academic Press, New York,1986.

[9] B.M. Kincaid, J. Appl. Phys. 48(1977)2684

[10] S. Krinsky, M.L. Perlman and R.E. Watson, "Handbook on Synchrotron Radiation", Ed E.E. Koch, North Holland, Amsterdam, 1983, chapter 2.

1HE ELECTRODE/SOLUTION INfERPHASE: PROBlEMS FOR SYNCHR01RON RADIATION

R. PARSONS Department 0/ Chemistry University 0/ Southampton Southampton, S09 5NH United Kingdom

ABS1RACT. Abrief outline is given of present ideas about the structure of charged interphases, predominantly those between a metal and an electrolyte. An attempt is made to indicate particular aspects where verification of models or more direct information is necessary and where the use of synchrotron radiation may provide this information.

1. Introduction

Tbe structure ofthe interfacial region between a metal and an electrolyte has been studied for many years usinggiobal probes and thermodynamic methods, particularly the use of electrocapillary curves(interfacial tension y as a function of electrode potential E) for mercury and other liquid metals and interfacial capacity (C) as a function of E. Data of this type may be analysed thermodynamically by the methods introduced by Gibbs [1], see alsö [2, 3]. Tbe capacity curves mayaiso be used directly to relate to models of the interphase. In recent years, a large variety of new techniques has brought new information about charged interphase which is more molecular in nature and has led to greater insight into their structure. Arecent review [4], gives a summary of some of this worl<, together with many references. New techniques are described in several recent compilations [5-7]. Tbe aim of the present paper is to bring out aspects of current models of the interphase which need confirrnation by more direct molecular-Ievel methods.

2. The Simplest Form of Meta1!Electrolyte Interphase

Tbe simplest situation occurs when the electrolyte is completely dissociated into ions which interact with the electrode only in a simple electrostatic way. Tbis is achieved experimentally with metals like Hg, Au and Ag in contact with strongiy solvated electrolytes like alkali fluorides in high permittivity solvents. In such systems the differential capacity (C)-potential (E) curve is found to be concentration dependent in the region where the charge density (0) on the metal is small. Tbe curves are more (Ag [8], Au [9)), or less (Hg [10)) symmetrical about the potential of zero charge pzc). Such data have been interpreted using a simple model in which the ions are attracted or repelled by the excess charge on the metal surface. Since they have a finite size (usually including the solvation shell, they can approach the electrode ünly up to a distance Xz. Tbe interphase on the electrolyte side is then divided into two regions by a plane parallel to the metal surface at a distance X2 from it. Tbis is known as the outer Helmholtz plane. Tbe region from the metal surface to Xz is known as the inner layer while that beyond Xz is known as

21

C. A. Melendres and A. Tadjeddine (eds.), Synchrotron Techniques in lnterfacial Electrochemistry 21-32. © 1994 Kluwer Academic Publishers.

22

the diffuse layer since the excess population of ions of one sign gradually decays as the distance to the metal surface increases.

The first model ofthe diffuse layerwas proposed by Gouy in 1910 and Chapman in 1913. It is essentially the same model used later for the Debye-Rückel theory of electrolyte activity coefficients and for the space charge in a semiconductor. Thus, the ions are assumed to be point charges (except in their approach to the metal surface) and they are assumed to move in a structureless medium having a constant permittivity. This model is elearly oversimplified, but experimental tests, though rather Iimited in scope [11], te nd to suggest that it is a good approximation. More detailed tests have been carried out by comparing the Gouy-Chapman model with results obtained for hard sphere ions in a structureless medium, using the Monte Carlo method, notably by Torrie and Valleau [12]. This showed good agreement for 1:1 electrolytes, but later work [13] showed substantial deviations in the case of unsymmetrical electrolytes.

The deviations from the Gouy-Chapman model appear most strongly in the potential and ion density profile, which do not affect greatly the integral quantities accessible in conventional experiments. The ion density profile is, however, accessible by the X-ray standing wave technique and, in fact, such measurements have been made recently in membrane-like systems [14], but so far no direct comparison with the simulation has been made.

The inner layer region between the metal and the outer Heimholtz plane is probably about one solvent molecule thick [15, 16] and the structure is probably perturbed substantially as compared with the bulk. The capacity of this layer is in series with the capacity of the diffuse layer and so may be obtained from the experimental value by using the usual formula for capacitors in series and assuming the Gouy-Chapman model for the diffuse layer. The result is a capacitance which varies strongly with charge, somewhat with temperature, but is independent of electrolyte concentration to a good approximation [8, 9, 10]. Molecular interpretation of these properties began [17, 18] with the assumption that the component of the solvent dipole perpendicular to the interface could take up two opposite orientations. The population ofthese two orientations was equal elose to the pzc but one or other became predominant as the charge increased in either direction. Greater elaboration of this type of model followed. Probably the most plausible physically is that of Guidelli [19] who considered a variety of orientations of water molecules determined by the possibility of hydrogen bonding between them. This can account for most of the observable properties. More recently Torrie and Patey [20] have used the reference hypernetted chain (RRNC) integral equation approximation for an assembly of water molecules ne ar a charged wall. The molecules are represented by a hard spheres with embedded point multipoles. This approach has worked weil for bulk water and leads to some interesting results for the structure elose to the wall, such as an ice-like arrangement at the uncharged surface. A direct test of models of this type would be desirable and an in situ X-ray diffraction method appears possible. The distribution in the direction perpendicular to the wall shows that the perturbation due to the wall dies away over two ar three molecular diameters. Similar conelusions can be drawn from molecular dynamics simulation [21].

This work [20] has been extended to systems containing ions, with at least one surprising result: a marked difference between the spatial distribution of K+ and Na + perpendicular to the uncharged wall surface. This would result in a marked shift of the pzc between the two systems or in the surface potential of the free surface of water, neither of which is observed [22, 23]. The X-ray standing wave method should provide direct experimental evidence far the actual distribution.

An earlier statistical mechanical approach using the mean spherical approximation [24, 25] and the hypernetted chain approximation for an assembly of hard spheres having charges (ions) or dipoles (solvent molecules) has been used to treat the electrolyte side of the interphase as a whole. The result is formally comparable with the division into two

23

regions as described above, but, whi!e one reduces simply to the Gouy-Chapman model at low concentrations, the other depends on the polarisation of the dipolar solvent molecules throughout the interphase, not just in the first layer.

So far, the models described have treated the metal as a structureless plane wall which is infinitely polarisable. In the last 15 years, the early suggestion [26] that the electron distribution in the metal sUlface is important has been revived on the basis of the jellium model of a meta!. This has been successful for the free surface of a metal, particularly in the calculation of the electronic work function [27, 28]. The positive cores of the metal atoms are modelled as a uniform background of positive charge in which the valence­electron gas is free to move. At the surface, there is a sharp boundary to this positive background charge, but the electrons can penetrate outside it to a limited extent as a result of their low mass and their kinetic energy. Thus, while the bulk of the metal is electrically neutral, the surface has dipol ar layer, the electrons outside the positive background giving rise to a negative outermost layer and the deficiency of electrons in the adjacent region within the positive background to the compensating positive layer. The combination of this with a model of the solution [29, 30] leads to a higher total capacity both because the charge on the metal is doser to the solution and because the mean position of this charge can vary with the excess charge applied to either side of the interphase. An experimental determination of the location of the mean position of this 'electron tai!' would cJearly provide a direct verification of this mode!. However, it is questionable whether X-ray reflectivity would be able to distinguish this contribution from others at the interphase.

3. Interphase in Which Ions Interact More Strongly With The Metal: Specific Adsorption.

In the majority of systems the simple model described above does not hold. This is due to the fact that the metal-ion interaction cannot be described in terms of acharge interacting with a field. As the ion approaches the metal surface, it may lose so me or all of its solvation shell and form a chemical bond with the metal surface. It is customary to ascribe all deviations from the adsorption calculated from the Gouy-Chapman model to 'specific ads01ption', i.e. adsorption that is specific to the nature of the ions; adsorption in the Gouy-Chapman model being dependent on the ions' charge alone. The amount of this specific adsorption usually corresponds to less than a monolayer and since the forces involved are short-range, it is usually assumed that these species form a partially occupied monolayer with the ion centres in a plane called the inner HeImholtz plane. The outer Helmholtz plane and the diffuse layer then lie further from the metal surface. If a chemical bond is formed with the metal, then it would be expected that there should be aredistribution of charge (electrons) between the ion and the meta!. This concept was introduced into surface physics by Langmuir [31], but was not used in electrochemistry until the work of Lorenz [32]. In both types of system there is great difficulty in ascertaining the exact charge distribution. It is possible to measure an effective dipole moment change as the result of adsorption, but this can be a consequence of several different types of charge rearrangement. It has been shown by a direct comparison of similar systems [33], that the behaviour of the metal/vacuum system is dosely similar to that of the metal/electrolyte system. The effective dipole moment was found to be almost identical when Br- was adsorbed on Ag(110) from aqueous solution and when Br was adsorbed on the same surface from vacuum, provided that in the lauer a monolayer of water was gradually replaced by Br (interpolation from a set of experiments). This work shows that the outer part of the interphase plays very little part in the charge distribution in the inner layer and that the role of the solvent is smal!. However, the major concJusion is that the effective dipole moment in both systems is very much smaller (1/20) than would be expected for Br- carrying its fuH charge inducing a positive image charge in the meta!.

24

At first sight, this suggests that in the solution experiment the Br- ion transfers most ofits charge to the metal and in the vacuum experiment very little charge transfer occurs. However, Lang [34] has shown that these small effective dipoles can also arise without charge transfer to or from the adsorbing species by the movement of electrons in the metal, resulting in the screening of the charge on the adsorbed ion. The fact that very small effective dipoles are found with many different adsorbing ions in solution [35] including some forwhich charge transfer seems improbable(CI04-, S042-, N03-) suggests that the screening model may be the more appropriate. It is clear that a more direct method for the study of electron distribution around an adsorbed ion would be very desirable. In the vacuum experiment, it may be possible to get some information from the so-called PAX method [36], using the XPS of Xe physisorbed on the surface with chemisorbed species, but this is clearly impossible for the electrochemical experiment. It may be that an EXAFS experiment would be more revealing.

These effective dipole moments are determined at low degrees of coverage by the adsorbing ion. As the coverage increases there may be two-dimensional order of the adsorbate, although in the majority of systems where anions are adsorbed, whether on liquid or solid metals, the adsorption isotherms are rather featureless. Nevertheless, the marked dependence of the adsorption on single crystal surfaces (for example cr on the basal planes of Ag [37]) suggests that there must be ordering related to the structure of the metal surface. Similar results are found for neutral molecules [38]. X-ray diffraction would provide direct evidence for these structures. It might also settle the question of the orientation of long chain ions at a metal surface. It has long been assumed that such species adsorbed with the hydrophilic ionic groups towards the solution while the hydrocarbon tails pack together towards the metal and perpendicular to its surface. Recent work [39] has led to the suggestion that these molecules lie along the surface.

4. Other Interfaces

The metal!molten salt interphase appears to be the simplest of metal!electrolyte interphases. However, the high ionic concentration leads to difficulties in its theoretical description. The U-shaped capacity-potential curves [40] have been interpreted in terms of a diffuse layer in which the ionic charge has an oscillating form which dies out in the direction of the bulk [41] or in terms of an incipient faradaic reaction [42]. The ionic distribution perpendicular to the interface could be determined from standing wave X-ray experiments.

At the semiconductor/electrolyte interphase, the behaviour observable in electrochemical experiments is dominated by the space charge in the semiconductor unless it is highly doped. This means that it is more difficult to explore the region comparable to that described above for the metal!electrolyte interphase. This region is important in that species adsorbed from the solution may give rise to surface states which modify the electrochemical properties of the interface. The nature and distribution of species in the semiconductor/electrolyte interphase pose interesting problems for molecular probes.

Similar problems to those discussed so far exist at the air/electrolyte interphase where distributions into the solution, orientation of molecules, etc. require direct probing.

5. Adsorption of Atoms

5.1 GENERAL REMARKS

The adsorption of hydrogen and of many metal atoms differs from that described above in that there is reasonable evidence [43] that there is a complele transfer of charge in this

25

process, Le. the adsorption can be described by

(1)

where ~;ol) is a solvated ion, e(metal) is an electron in the metal, and Mads is an adsorbed atom. This me ans, first that the equilibrium of such a process can be described by an analogue of the Nernst equation:

(2)

where EO is the standard electrode potential of re action (1), n is the number of electrons transferred in the reaction, and the a j are activities of the subscripted species. This me ans that the primary variable controlling adsorption in an equilibrium system of this type is the electrode potential, in contrast to the systems discussed here in previous sections where it is the charge.

The second consequence of total charge transfer is that the amount adsorbed may be measured directiy from the charge passed during adsorption, apart from a sm all correction due to the redistribution of charge in the ionic population of the interface.

These two properties have led to the popularity of linear sweep voltammetry for the study of this type of system (as weil as for other types). The electrode potential is changed linearly with time and the current passing through the electrode is recorded. The resulting plot of I against E, the voltammogram, shows one or more peaks corresponding to regions of potential where the adsorption changes most rapidly with electrode potential. Since the E axis is also a time axis, integration of the peaks gives the amount adsorbed. In the simplest case, the adsorption process is exactiy symmetrical to the desorption process, which occurs when the sweep direction is reversed. . In this case, the system is at equilibrium at each point, equation (2) is obeyed, and the potential scale gives the Gibbs energy of adsorption. The stability of such an electrode's behaviour can be verified by repeating the potential cyde - a cyclic voltammogram (CV). Since the current is given by

1= dq/dt (3)

where A is the electrode area and q the charge flowing into the electrode interphase, it follows directiy that the electrode capacitance

C = (dq/dE)/A = I/vA (4)

where v is the potential sweep rate dE/dt. Thus, this method can give the same information for the systems described in previous sections as the direct measurement of the differential capacitance, though it is not usually as accurate.

There are also many systems for which a CV symmetrical about the zero current line is not found. This results from some kinetic effect such as a slow electron transfer, surface rearrangement, etc. In such cases, the amount of charge transferred can still be determined from the voltammogram, but there is no longer a simple relation between the potential and the Gibbs energy.

5.2 THE ADSORPTION OF HYDROGEN

The adsorption of atomic hydrogen on Pt occurs by the complete neutralization of protons from the solvated state [44,45] and is a rapid process. Thus voltammograms obtained at a moderate sweep speed (around 100 mV S-I) represent this re action at equilibrium in the range of a few hundred millivolts positive of the equilibrium hydrogen potential. The

26

detailed form of this voltammogram is highly sensitive to surface structure and in a long series of publications Clavilier [46, 47] has shown that a great deal of information about the structure of a Pt surface can be obtained by this method. In particular, the high degree of mobility of the surface under a variety of electrochemical and pre-treatment conditions has been illustrated. High temperature annealing tends to produce a highly ordered surface, which is disrupted by the adsorption of oxygen, either thermally or electrochemically, to produce a random stepped structure. The electrochemical evidence for such changes has been confirmed by ex situ LEED measurements [48, 49], but in situ measurements are only recently available. These come from S1M studies by Itaya et al. [50], and from second harmonic generation (SHG) studies by Com et al.[51]. Nevertheless, direct confirrnation by in situ X-ray diffraction would be desirable. This would, of course, be able to detect only structural changes in the Pt surface; direct information about the hydrogen atoms themselves seems at present, out of the question.

A notable feature of hydrogen adsorption on Pt is the marked influence of the nature and concentration of anionic species. The more strongly adsorbed anions appear to displace the hydrogen and to shift the region of hydrogen adsorption to more negative potentials, for example, in the series CI04-, SOr, cr, in which the adsorbability on Pt increases. There is also evidence for a strong effect of the structure of the surface of the Pt on the adsorbability. This has been confirmed by recent measurement using radio­labelIed ions [52]. It is suggested that the stronger adsorption of SO/- on Pt(111) compared with Pt(100) is due to the matching of the structure of the ion to the three-fold symmetry of sites on the Pt(111). It seems that an EXAFS study in situ could provide definitive evidence on this point.

The main weakness of electrochemical methods like voltammetry is that the measurement of charge cannot lead to identification of species. This has led to controversy, notably in the case of a well-ordered Pt(111) surface, where a peak originally [46] attributed to hydrogen adsorbed with an exceptionally high adsorption energy was also attributed to the adsorption of SO/- [53]. Radio-tracer experiments giving the amount ofsulphate adsorbed [54] suggest that the original interpretation is more probable. A particular point is the origin of a very sharp spike in this region at the most positive potential. This type of spike is usually associated with a two dimensional phase transition. The occurrence of such a phase transition in the adsorbed sulphate is suggested by an abrupt change in the frequency of an IR peak at around 1200 cm-1 [55]. The presence of such a phase transition could be confirmed by in situ X-ray diffraction and its nature investigated in more detail. Direct information as to whether the spike is due to SO/- alone or the effect of the phase transition on the hydrogen adsorption is clearly not so easy to obtain.

5.3 THE ADSORPTION OF METAL ATOMS

Many metal atoms are adsorbed on metal electrodes under conditions closely similar to those described above for hydrogen. Adsorption occurs on polycrystalline surfaces, often in several steps; for example, Pb on Au shows a broad peak at low coverages (more positive potentials) and a sharper one at higher coverages (more negative potential) [56]. Adsorption is rapid and again the voltammogram corresponds to the equilibrium condition. It is tempting to ascribe these different states to different types of sites on the polycrystalline surface. However, studies of the low index planes of Au show substantially more complex behaviour [57] including differences in the course of adsorption and desorption. Since the adsorption of Pb on these surfaces has also been studied in the gas phase, following the ordering of the Pb by use of LEED [58] the attempt has been made to correlate the structures observed in vacuum with the peaks observed at the electrodes [57] as weIl as to use stepped surfaces in identifying surface structures. Perhaps the most detailed attempt to use electrochemical information to identify the surface structures in metal atom deposition was carried out [59, 60] using a Au substrate with (111) terraces

27

about 10 atoms wide, using charges obtained under strictly equilibrium conditions (constant current = constant rate of metal deposition). Although very precise matching of the experimental charge with that calculated from the assumed overlayer structures was obtained, it would be much more convincing if direct structural information were obtained. Tbis seems to be quite possible using X-ray diffraction. Tbe ingenious interference technique combined with multiple scanning and sampling used by Chabala and Rayment [61], which has been used to measure the distance of Pb atoms from the surface plane of the substrate in situ, seems to show promise for extension to the meaSurement of overlayer structures of tbis type.

Although in many metal monolayer systems, the transfer of charge seems to be complete, in those for which it is not, the same problems occur as described above for the adsorption of anions and a direct method far the determination of electron distribution would be valuable. Further complications arise because of the co-adsorption of anions with the metal atoms. Tbe nature of this interaction could be explored by EXAFS.

5.4 THE EARLY STAGES OF METAL OXIDATION

On Pt, as on all noble metal surfaces, oxidation occurs as the electrode potential is made more positive. By controlling the potential, it is possible to limit this process to a fraction of a monolayer, or to allow it to proceed to multilayers [62]. Tbe very first stages of formation of a fractional monolayer on Pt or Au can be made reversible by careful control of conditions [63], but at longer times and/or higher potentials the desorption process occurs at much lower potentials than the adsorption process. Tbis change in the nature of the process is attributed to a place exchange mechanism [63, 64] in which Pt atoms exchange places with 0 or OH species, a step which may be regarded as a preliminary to the growth of the phase oxide. Tbe general form of this behaviour occurs also on single crystal surfaces although there are some differences in detail. More work has been done so far on Au surfaces [65,66] than on Pt, but again the differences are in detail.

Apart from direct evidence on the structures produced by place exchange, more information is needed on the chemical state of the metal and oxide during the progress of oxidation. Ex situ EELS studies [67] have provided evidence for the presence of OH, 0 and bulk oxide in heavily oxidized electrodes heated to various temperatures in vacuum. Tbe oxidation state of Pt has also been studied ex situ using XPS [68, 69] but, especially at low coverages, there are uncertainties resulting from the transfer process as weIl as from the lack of standard sampies of species like PtO. Up to the present there has been little work of this type on single crystals. In view of the inevitable surface disruption this may not be a great disadvantage. As the thicker oxides are grown at high potentials, or as the electrodes are cycled to form hydrous oxides [70], there are many problems of short and long range structures which might be attacked by EXAFS or X-ray diffraction, although the dependence of the growth on time must inevitably cause problems with these rather slow techniques. At least at the monolayer level a solution to this may be found with the cycling and sampling technique [61] even with these irreversible but repeatable processes.

6. Thicker Layers

Tbe initial stage of the growth of phase oxides and other new phases in the electrochemical environment raises questions about the mechanism of this process. At what stage can the high field growth mechanism described by Mott and Cabrera [71] set in? What is the relation of the structure of thenew phase to that of the underlying metal? Such questions may be answered by X-ray diffraction and related techniques. _

A newer area which raises somewhat similar problems is that of modified electrodes [72], or the related polymer modified electrodes [73], particularly in the growth of the latter

28

[74]. However, the mechanism of switching from a non-conducting to a conducting state and the propagation of this process through the film is a problem where the ability to determine ionic distribution through the film might provide a solution to this problem. So far it has been attacked by global techniques such as variation of the ionic species in solution and the quartz microbalance [75]. Tbe structure of the metal!polymer and polymer electrolyte interphases at these electrodes has not yet been explored Tbis problem is also related to that of the metal!solid electrolyte interphase where models describing the charge distribution have been set up [76], but no direct verification has so far been possible.

7. Conclusion

Tbere are many problems in the interfacial structure at a metal!electrolyte boundary which might be solved using synchrotron radiation. Some of these have been treated in the papers in this workshop. Here the ones mentioned in this introduction are summarized. It should be emphasized that these are problems in the understanding of the structure, not in the application of the systems in practice.

8.

[1]

[2]

[3]

[4]

[5]

[6]

Distribution of ions in the diffuse layer. Water and other solvent structures in the interphase. Electron distribution on the metal surface. Charge distribution on adsorbed ions and molecules. Solvation of adsorbed ions. Ordering of adsorbed ions and atoms on single crystal surfaces. Orientation of long chain and other large molecules. lonic distribution in a molten salt electrolyte elose to the metal surface. Interaction of adsorbed metal atoms and other solution species especially anions. Valence states and structure in their growing films such as oxides. Mechanism of switching in polymer modified electrode. Nature of surface states on semiconductors due to adsorbed ions. Charge distribution in metal/solid electrolyte interphase.

References

J.W. Gibbs, Collected Works, Yale University Press, New Haven, 1948, Vol. 1, pp 230,331.

D.M. Mohilner in Electroanalytical Chemistry, (Ed.) AJ. Bard, Dekker, New York, 1966, Vol. 1, Chapter 4.

R Parsons in Comprehensive Treatise of Electrochemistry, (Eds.) J .O'M. Bockris, B.B. Conway and E. Yeager, Plenum, New York, 1980, Voll, Chapter 1.

R Parsons, Chem. Rev., 90 (1990) 813.

RJ. Gale, (Ed.) Spectroelectrochemistry, Tbeory and Practice, Plenum, New York, 1988.

Ravi Varma andJ.R SeIman, (Eds.) Techniques for CharacterizationofElectrodes and Electrochemical Processes, Tbe Electrochemical Society/John Wiley, New York,1991.

29

[7] H.D. Abruna, (Ed.) Electrochernical Interfaces, VCH Publishers, New York, 1991.

[8] G. Valette, J. Electroanal. Chern., 122 (1981) 285.

[9] A. Harnelin, J. Electroanal. Chern., 138 (1982) 395.

[10] D.C. Graharne, J. Amer. Chern. Soc., 76 (1954) 4819.

[11] R. Parsons and S. Trasatti, Trans. Faraday Soc., 65 (1969) 3314.

[12] G.M. Torrie and J.P. Valleau, J. Chern. Physics, 73 (1980) 5807.

[13] G.M. Torrie and J.P. Valleau, J. Phys. Chern., 86 (1982) 325.

[14] M. McCaffery, Faraday Discussions 94 (1992) in press.

[15] R. Parsons, Proc. Roy. Soc., A 261 (1961) 79.

[16] R. Parsons and F.G.R. Zobel, J. Electroanal. Chern.,2 (1965) 333.

[17] R.J. Watts-Tobin, Phil. Mag., 6 (1961) 133.

[18] J.O'M. Bockris, M.A.V. Devanathan and K. Müller, Proc. Roy. Soc., A 274 (1963) 55.

[19] R. Guidelli, J. Electroanal. Chern., 123 (1981) 59, 197 (1986) 77.

[20] G.M. Torrie and G.N. Patey, Electrochirn. Acta, 36 (1991) 1677.

[21] E. Spohr and K. Heinzinger, Electrochirn. Acta, 33 (1988) 1211.

[22] D.C. Graharne, E.M. Coffin, J.I. Curnrnings and M.A. Poth, J. Amer. Chern. Soc., 74 (1952) 1207.

[23] J.E.B. Randles, Physical Chernistry of Liquids, 1 (1977) 107.

[24] S.L. Carnie and D.Y.C. Chan, J. Chern. Physics, 73 (1980) 2949.

[25] L. Blurn and D. Henderson, J. Chern. Physics, 74 (1981) 1902.

[26] O.K. Rice, J. Physical Chern., 30 (1926) 150l.

[27] N.D. Lang and W. Kohn, Phys. Rev. B 1 (1970) 4555.

[28] N.D. Lang and W. Kohn, Phys. Rev. B 3 (1971) 1215.

[29] J.P. Badiali, M.L. Rosinberg and J. Goodisrnan, J. Electroanal. Chern., 143 (1983) 73.

[30] D. Henderson and W. Schrnickler, J. Chern. Physics, 82 (1985) 2925.

[31] I. Langmuir, J. Amer. Chern. Soc., 54 (1932) 1798.

30

[32]

[33]

[34]

[35]

[36]

[37]

[38]

[39]

[40]

[41]

[42]

[43]

[44]

[45]

[46]

[47]

[48]

[49]

[50]

[51]

W. Larenz and G. Salie, Zeit. physikal Chern., 218 (1961) 259.

K. Bange, B. Strachler, J.K. Sass and R. Parsons, J. Electroanal. Chern., 229 (1987) 87.

N.D. Lang, Surface Sci., 127 (1983) L 118.

R. Parsons, BuH. Electrochern., 2 (1990) 566.

J. Kuppers, H. Michel, F. Nitschke, K. Wandelt and G. Erd, Surface Sei., 89 (1971) 361.

G. Valette, A Harnelin and R. Parsons, Zeit. für physikal Chern., 113 (1978) 71.

D.F. Yang, L. Stolberg, J. Lipkowski and D.E. Irish, J. Electroanal. Chern., 329 (1992) 259.

M.R. Moncelli and M.L. Foresti, J. Electroanal. Chern., in press.

AD. Graves, G.J. Hills and D. Inrnan in Adv. Electrochern. and Electrochern. Eng., (Eds.) P. Delahay, C.W. Tobias, Interscience/Wiley, New York, 1966, Vol 4, 152.

R.R. Dogonadze and Yu. A Chizmadzhev, Doklady Akad. Nauk. SSSR, 157 (1964) 944.

AD. Graves and D. Inrnan, J. Electroanal. Chern., 25 (1970) 349, 357.

D.M. Kolb in Advances in Electrochern. and Electrochern. Eng., (Eds.) H. Gerischer and C.W. Tobias, Wiley!lnterscience, New York, 1977, Vol11, Chapter 2.

R. Durand, J. Electioanal. Chern., 97 (1979) 293.

B.M. Grafov, E.V. Pekar and O.A. Petry, J. Electroanal. Chern., 40 (1972) 179.

J. Clavilier, R. Faure, C. Guinet and R. Durand, J. Electroanal. Chern., 107 (1980) 205.

J. Clavilier, A Rodes, K. EI Achi, M.A Zarnakhchari, J. Chirn. Physique, 88 (1991) 1291.

D. Aberdarn, C. Corotte, D. Dufayaid, R. Durand, R. Faure and G. Guinet, Proc. 4th Int. Conf. Solid Surf., Cannes, 1980, p622.

F.T. Wagner and P.N. Ross, J. Electroanal. Chern., 150 (1983) 141.

K. Itaya, S. Suguwara, K. Sashikata and N. Furuya, J. Vac. Sci. Technol., 148 (1990) 515.

M.L. Lynch, B.J. Barner and R.M. Corn, J. Electroanal. Chern., 300 (1991) 447.

[52]

[53]

[54]

[55]

[56]

[57]

[58]

[59]

[60]

[61]

[62]

[63]

[64]

[65]

[66]

[67]

[68]

[69]

[70]

[71]

[72]

[73]

31

M.E. Garnboa-Aldeco, E. Herrero, P. Zeleny and A Wieckowski, J. Electroanal. Chern., in press.

DA Scherson and D.M. Kolb, J. Electroanal. Chern., 176 (1984) 353.

E.K. Krauskopf, L.M. Rice and A Wieckowski, J. Electroanal. Chern., 244 (1988) 347.

RJ. Nicholls, Thesis, Southarnpton, 1988.

RR. Adzic, E.B. Yeager and B. Cahan, J. Electrochern. Soc., 121 (1974) 474.

AHarnelin and J. Lipkowski, J. Electroanal. Chern., 171 (1984) 317.

J.P. Biberian, Surface Sei., 74 (1978) 437.

J.P. Ganon and J. ClaviIier, Surface Sei., 145 (1984) 487.

J.P. Ganon and J. ClaviIier, Surface Sei., 147 (1984) 583.

E.D. Chabala and T. Rayrnent, Faraday Discussions, 94 (1992) in press.

R Woods in Electroanalytical Chernistry, (Ed.) AJ. Bard, Dekker, New York, 1976, Vol. 9, P 59.

H. Angerstein-Kozlowska, B.E. Conway and W.B.A Sharp, J. Electroanal. Chern., 43 (1973) 9.

P. Stonehart, H. Angerstein-Kozlowska and B.E. Conway, Proc. Roy. Soc., A 310 (1969) 547.

D. Dickertrnann, J.W. Schultze and K.J. Vetter, J. ElectroanaI. Chern., 55 (1974) 429.

H. Angerstein-Kozlowska, B.E. Conway, AHarnelin and L. Stoicovieiu, Electrochirn. Acta, 31 (1986) 1051.

M. Peuckert and H. Ibach, Surface Seience, 136 (1984) 319.

J.S. Harnmond and N. Winograd, J. Electroanal. Chern., 78 (1977) 55.

M. Peuckert, EP. Coenen and H.P. Bonzel, Electrochirn. Acta, 29 (1984) 1305.

L.D. Burke and M.E.O. Lyons in Modem Aspects of Electrochern., (Eds.) RE. White, J.O'M. Bockris and B.E. Conway, Plenum Press, New York, 1986, Vol. 18, pp 169-248.

N.F. Mott and N. Cabrera, Report, Progress in Physics, 12 (1949) 163.

RW. Murray, Phil. Trans. Roy. Soc. A, 302 (1981) 253.

AR. HilIman in Electrochernical Seience and Technology of Polymers, (Ed.) R Linford, Elsevier, London, 1987, Vol. 1, pp 103-292.

32

[74] A.l. Downard and D. Pletcher, l. Electroanal. Chern., 206 (1986) 139, 147.

[75] D.A. Buttry in Ref. 7, Chapter 10.

[76] D.O. Raieigh in Electroanalytical Chernistry, (Ed.) A.l. Bard, Dekker, New York, 1973, Vol. 6, P 87.

NATURE OF SURFACE FILMS

J. KRUGER The lohns Hopkins University

Materials Science and Engineering 3400 N. Charles Street Baltimore, MD 21218 U.SA.

ABSTRACf. Virtually all eIectrode processes are affected by the films that exist or form on the surfaces of eIectrodes. Consequently, the nature of these surface films playa critical role in affecting the electrochemistry taking place at a surface. The properties to be discussed that determine the nature of surface films are the following: (a)thickness, (b) composition, (c) structure, (d) defects, (e) eIectronic, optical, and magnetic properties, and (t) mechanical properties.

All metals (perhaps gold is an exception) react with the environment and thereby have present on their surfaces a thin protective reaction product film. This film, usually an oxide, is responsible for metallic materials remaining metallic and not reverting back to the thermodynamically stable condition of their origin - the ores from whieh they were produced. These surface films, in many instances, control the electrochemistry occurring at the metal C or semiconductor) / electrolyte interface. Consequently, in order to gain a better understanding of the processes that take place on an electrode surface requires an examination of the properties that determine the nature of surface films. Synchrotron techniques offer many opportunities to examine the properties of the surface films that control the electrode processes that strongly influence corrosion, electrodeposition, electrocatalysis, electrochemical energy technologies, and solid state electronics.

This paper will discuss the following: Ca) thickness, Cb) composition, Cc) structure, Cd) defects, Ce) electronic, magnetic and optieal properties, and Cf) mechanical properties. Wherever appropriate, the discussion will provide instances where synchrotron techniques have contributed to our knowledge of many of these properties of surface films. Moreover, because of the emphasis of this NATO Advanced Research Workshop on techniques, this paper will also highlight techniques in general that have been used to study surface films. This discussion will only consider the nature of reaction product surface films and will not cover adsorbed layers. It will rely heavily on reviews by the author [1-3] and a monograph [4] that are concerned with the special, but relevant, surface films,

33

C. A. Melendres and A. Tadjeddine (eds.), Synchrotron Techniques in lnteifacial Electrochemistry 33--66. © 1994 Kluwer Academic Publishers.

34

the passive films, that control the electrochemistry of corrosion processes. Moreover, surface films on iron and iron alloys will provide most, but not all, of the examples used in the discussion. Particular emphasis will also be placed on the role played by the nature of surface films in controlling the breakdown of films, a process that leads to reactions at discrete sites on electrode surfaces.

1. Thickness

Thickness is the property of a surface film that determines the amount of film that is necessary to exert an influence on the electrochemical processes that take place on a metal surface. One important aspect strongly affected by the thickness of a surface film is the phenomenon of breakdown , the means whereby aggressive ions breach the film and reach the metal surface to initiate interaction with the surface at localized sites. Thickness as an aspect of the nature of surface films became an issue around the time of Faraday's researches into passivity [5]. It has sparked a lively controversy that strongly affects the various conceptions of the mechanisms of breakdown. The controversy was between those, for example Uhlig [6], who favored a two-dimensional film, an adsorbed monolayer or less than a monolayer of oxygen which retarded surface reaction rates, and those, for example, Evans [7], who proposed a three-dimensional film, a phase oxide that had a thickness greater than one unit cell so that it could serve as a barrier to the diffusion of metal cations into the solution. Today, through the use of a host of techniques the controversy appears to be resolved because, depending on conditions, it has been shown that surface films can be either two or three dimensional. Obviously, all of the work directed at resolving the thickness question cannot be cited. Instead, a few experiments establishing the three-dimensional picture (thickness > monolayer) and those establishing the two dimensional (thickness :5 monolayer) will be described.

1.1 TRICKNESS GREATER TRAN A MONOLAYER

Using an ultra-high vacuum system to prepare as bare an iron surface as possible (necessary in order to assure that an initial adsorbed layer was not present) and then introducing a passivating solution (nitrite), after first measuring the optical parameters of the bare iron surface, Kruger [8] showed using ellipsometry that a film greater than a monolayer in thickness was present on iron whenpotentials weH into the passive region were measured. These results were verified by cathodic reduction (coulometric) measurements by Draper [9] using iron single crystal surfaces made bare by cathodic reduction. Similar thicknesses were also found for passive films on iron formed by anodic oxidation through the use of ellipsometry [10] and cathodic reduction (coulometric) techniques [11].

The thickness of three dimensional films can be influenced by many factors. For example, Cortes et al.[12] found by ReflEXAFS measurements that for NiMo

35

alloys, the thickness decreases when the films on these alloys undergo a recrystallization process.

1.2 THICKNESS LESS THAN A MONOLA YER

Using electrochemical techniques, coverages of less than a monolayer were found by Kubanov et al.[13] for iron in dilute NaOH and by Frankenthai for iron in borate-buffer [14] and iron-chromium (24% Cr) in 1N H2S04 [15]. Frankenthal's studies actually provide a link between two-dimensional and three-dimensional films. He found that at low potentials in the passive region (-0.4 to -0.1 SHE) the film measured was less than a unit cell for Fe30 4 or Fe20 3 (around 0.84 nm). This film may be considered to be chemisorbed oxygen. Above these potentials he measured thicknesses greater than the unit cell for the phase oxide with Fe(I1I), i.e., yFe20 3• Extrapolations of ellipsometric studies, for example [10], to potentials in the region where FrankenthaI found a two-dimensional film also gave thicknesses of less than 0.84 nm. FrankenthaI found similar results for Fe-24%Cr [15]. Ellipsometric studies [16] on easily passivated metals like chromium also show a quite wide potential region (as much as 600 mV wide) where passive films exhibit thicknesses less than one unit cell of a phase oxide both in neutral and acidic solutions.

1.3 THICKNESS AND BREAKDOWN

Thickness can affect breakdown in at least three ways:

1.3.1 Determine mechanism 0/ breakdown It is not too surpnsmg that the mechanism of breakdown will depend on whether a passive film is an adsorbed monolayer of oxygen or a phase oxide. If the film is a monolayer, the mechanism of breakdown involves the replacement of the adsorbed oxygen by an aggressive anion such as the chloride ion [17]. If the film is a phase oxide, breakdown requires that an aggressive anion must somehow (a number of mechanisms exist) traverse the thickness of the oxide film to the metal-oxide interface. Ambrose and Kruger [18] showed (see Fig. 1) that it is necessary that chloride ions penetrate the entire passive film on iron to the metal-oxide interface at a potential above a critical potential before breakdown can occur. BardweIl et al.[19] found that the breakdown of the passive film on Fe depends on a "certain critical oxide thickness" which depends on the nature of the anion (Cl" or Br-) and its concentration. It is the "critical oxide thickness" rather than the anodic potential (which is related to thickness) that direct1y controls the breakdown process.

1.3.2 Determine the induction period tor breakdown which is related to susceptibility As was pointed out in (a) above, the thickness of a passive film must be traversed by an aggressive anion before breakdown can occur. This journey of a chloride ion across the thickness of surface films to the metal-oxide interface, whatever

36

c:alhodic reduc:lion of pallive film

inalanlaneoua breakdown

Fig. 1. Schematic representation of the results of Ambrose and Kruger [18] showing that a new film will form at anodic potentials when the outer portion of a passive film is cathodically reduced leaving only the inner portion to prevent chloride ions from reaching a bare iron surface. When the entire passive flIm is reduced, a1lowing chloride ions to interact with the iron surface, breakdown is instantaneous.

mechanism is proposed (via pores [20], defects [21] or thinned areas [22] or other modes of transport), requires time. This required transit time is probably responsible for the existence of an induction period, 1, before breakdown occurs. McBee and Kruger [21] have shown that, indeed, a direct relationship between 1 and film thickness (Fig. 2) exists for surface films on iron. Bohni and Uhlig [17], however, found no dependence of pitting susceptibility on thickness. Sugimoto et al.[23] using microellipsometry to measure passive film thicknesses on different grains of 18-8 stainless steel, which exhibited different thicknesses, could not show a clear relationship between thickness and breakdown susceptibility. However, recent studies using Dynamic Imaging Microellipsometry [24] found that different thicknesses on two adjacent phases resulted in breakdown at the grain boundary between the two phases.

1.3.3 Affect the rate 0/ repassivation Repassivation rate (the rate at which a surface film regrows after breakdown) has been suggested as an important aspect in various mechanisms of breakdown [21,23,25,26]. The effectiveness of the repassivation process depends strongly on the limiting thickness of the regrown film, the smaller the limiting thickness, the more rapid is the repassivation process and the more resistant is a given passive system to breakdown. It has been shown, [27] that the rate of repassivation is greater for stainless steel than for iron and that the limiting thickness on the former is less than that on the latter. Moreover, Kruger and McBee [28] among others, have shown that a stainless steel is also more resistant to breakdown than iron. This is, perhaps, a consequence of the higher repassivation rate for the Fe-Cr and Fe-Cr-Mo alloys they studied, as weH as being due to the structure and composition of surface films on these aHoys; these issues that will be discussed in the next two sections.

• b

In t

Fig. 2. Plots showing the dependence of the time of breakdown (tind) on the thickness of the passive film on iron. (a) thickness vs. hand (b) thickness vs. Int. From McBee and Kruger [20].

2. Composition

37

The controversy surrounding the thickness of what constitutes an effective barrier similarly extends to the composition of such films. An example of such an effective film, that has been extensively studied, is the surface film on iron. Cohen [29] has compiled most of the proposed compositions (see Fig. 3) for surface films on iron. The proposed compositions, many of which are non-stoichiometric, involve either one or more layers containing the following oxides, hydroxides or oxy-hydroxides: Fe304' yFe20 3, yFeOOH, a polymerie layered [Fe(OH)2]x [30], a non-stoichiometric cation deficient y Fe20 3 containing varying amounts of protons Fe2_xHx0 3 [31] and a cation deficient y Fe203(Fe2-2x0x03) [29]. Cahan and Chen [32] suggest that the chemical composition can be characterized as a "highly protonated, trivalent iron oxy-hydroxide capable of existing over a relatively wide range of stoichiometry." However, in some cases stoichiometric oxide films are formed. For example, Weisler, Toney et al.[33], using X-ray reflectivity found the oxide films formed on Nb and Ti single crystal surfaces to be stoichiometric.

Larramona and Gutierrez [34,35] examined the role of the aqueous solution environment (pH) on the composition of the surface films on Fe and Ni using the in-situ potential modulation reflectance (PMR) technique. They found that the composition of surface films on Fe was not influenced by pH, observing the composition of the film to contain Fe(II) at all potentials with a top layer of FeOOH that becomes dehydrated at higher potentials to a and y Fe203' It should be pointed out that Nagayama and Cohen [36] found that FeOOH is usually formed by precipitation of Fe(II) from solution and is not the protective (passive)

38

A B C y-Fe2 0 3 y-Fe 2_x Hx0 3

~ Fe304 Fe304 Fe Fe Fe

D E F y -FeOOH

[(Fe (OH12] " Fe2-211 Ox 03

7 - Fe20 3 Fe203 Fe304

Fe Fe Fe

Fig.3. A double layer, yFe203-yF~04; B double layer with H in outer layer; C single layer, yFe20 3; D double layer, yFeOOH­yFe20 3; E polymerie layered [Fe(OH)2]x; F non-stoiehiometric cation deficient yFe203 with varying number of portons. Proposed models of passive film on iron. FlOm Cohen [29].

film that forms on Fe at potentials in the passive region. Results on Ni [35] were found to be similar to those of Fe with respect to the lack of influence of pH on the formation of NiO at all potentials; higher Ni oxides were only found in alkaline solutions at high er potentials.

Four aspects come to mind when we consider the important general issues that bear on the effect of composition on the properties of surface films that affect interfacial electrochemistry: 1. single or multilayer compositions, 2. hydrogen in the film, 3. alloying elements in the film, and 4. impurity elements in the film.

2.1 SINGLE OR MULTILAYER COMPOSmON

One of the early proposals that surface films on iron had a two layer composition like that shown as model A in Fig. 3 was from Vetter [37] who argued for its validity on thermodynamic grounds. Cohen [29] suggested that the boundary may not be as sharp as Model A shows but is a cubic lattice with decreasing concentration of iron as the solution interface is approached. Its main experimental basis comes from such studies by Cohen and co-workers as that by Nagayama and Cohen [11] who used a cathodic reduction technique (see Fig. 4) coupled with analytical chemical measurements of the composition of the solution in which the reduction was carried out, as well as electron diffraction measurements of the films at different reduction stages.

-400

> E -.i -600 « i= -700 z w I-0

-800 0.

-900

-1000

cothodic r~uction curv~ 'or lh onodis~d sp~cim~n 01 .600 mV

10

colhodic curr~nl 10 !JA cm-2 bor.ic acid - borol~

QUANTITY OF ELECTRICITY • mC

A

50

Fig. 4. Cathodic reduction curve of passive film on iron showing two waves corresponding to reduction of Fe20 3 (near -600 mV) and F~04 (near -900 mV). From Nagayama and Cohen [11].

39

Using the optical technique of ellipsometry, Ord and De Smet [38] followed oxidation and reduction cycles of iron in borate buffer solutions. They showed an abrupt change in optical constants during reduction which they interpreted as indicating a distinct change for yFe20 3 to Fe30 4• Other studies involving ellipsometry and a.c. impedance measurements by Cahan and Chen [32] were interpreted as showing no bi-Iayer. Electron diffraction studies by Foley, Kruger and Bechtoldt [39] also support the ideas of the existence of a layer of y Fe203 and a layer of Fe30 4• They claimed to be able to find differences between these very similar spinels, the Fe20 3 giving extra weak reflections and possessing a lower lattice parameter than the Fep4' They found that when the potential of film formation was in the passive region, yFe20 3 was found (along with Fe30 4) but at potentials outside the passive region only Fe30 4 or oxy-hydroxides were observed. More recent electron diffraction studies [40] were not able to find any differences between yFe20 3 and Fe30 4 but concluded from ellipsometric and a.c. impedance studies that only yFe20 3 was present. Sato and co-workers [41] found (using ellipsometric and potentiostatic techniques) the type A (Fig. 3) film in acid solutions but in neutral solutions a different bilayered film is produced (Type D), where the inner layer (the barrier layer) is affected by the anions in the solution used. An iron (III) oxide forms in a borate solution and a mixed iron (II) - iron (III) oxide in a phosphate solution. Sato attributes these differences to the effect of anions on the barrier layer and to the ion selectivity of the outer precipitated or deposited layer, yFeOOH, it being anion selective in borate solution and cation selective in phosphate. Cohen [29] has questioned these results saying that Sato's experimental conditions promoted the formation of the deposited film. Nagayama and Cohen [36] showed that this deposited film yFeOOH is not related to passivity and therefore not part of a passive film.

40

2.2 HYDROGEN IN FILM

It is the consideration of this aspect and the question of structure, to be considered later, that many workers believe requires the application of in situ techniques and provides the impetus for the development and use of new ones. This is so because as Uhlig [6] said in his review of the history of passivity ''There is no assurance ... that any film isolation techniques [or other ex situ techniques] deal with surface films in situ rather than an altered or decomposed passive film ... since H20 is a component of the film substance on at least some metals, there is no guarantee that such water remains within the film during irradiation or electron bombardment in vacuum". Proponents of the use of in situ techniques such as electron diffraction, Auger electron spectroscopy (AES), x-ray photoelectron spectroscopy (XPS), and others [29,40] counter such criticisms by pointing out that these techniques cause minimal heating, 20 0 C, which is not enough driving force to create the highly oriented crystalline films found by electron diffraction [39,40]. It should also be pointed out that when the films on Fe-Cr alloys are examined, evidence for "bound water" can be found using electron diffraction [42].

Even if exposure to a vacuum and heating by electrons do not change surface films structurally, recent in situ techniques, Mössbauer spectroscopy [43] and EXAFS [44,45], have revealed that removal from an aqueous solution and the change or elimination of the potential which the passive surface was held at has changed the composition of surface films on iron, presumably eliminating some hydrogen containing species. Moreover, the use of radiotracer experiments [46,47] (using tritiated water), without drying out the films produced in these experiments during measurement of the radioactive hydrogen retained in the film, has demonstrated the presence of some hydrogen containing species, H +, OH-, or HP, in the in situ passive film on iron.

These in situ or quasi-in situ experiments, each in its own way, showed the following:

2.2.1 Mössbauer Spectroscopy - O'Grady [43] found that the Mössbauer spectroscopic parameters found for the in situ passive film on iron correlated best with the parameters found in the literature for amorphous iron (III) oxides, Le., iron containing polymerie chains bonded together by bi-nuc1ear iron compounds containing di-oxy and di-hydroxy bridging bonds between the iron atoms. He interpreted his results as showing that these chains are linked together by water. When the in situ (passive) films are extensively and irreversibly dried, yFep3 is observed by Mössbauer spectroscopy. Similar results were obtained on dried passive films on iron by Simmons, Kellerman and Leidheiser [48] using emission Mössbauer spectroscopy but they interpreted. their results for the in situ film as showing the film to be hydrated superparamagnetic yFe20 3• Both studies and a

41

more recent one [49], however, provide evidence for the existence of hydrogen (presumably water) in surface films on iron.

2.2.2 EXAFS - This synchrotron technique can be used for in situ measurements on thin films and provide compositional and structural data. Both in situ [44] and non-vacuum ex situ [44,45] studies of surface films on iron have been carried out. In Fig. 5 the results of the ex situ are compared to the in situ. It shows that x-ray absorption spectra taken in the region near the K-edge for iron differs markedly between the films measured in an aqueous environment and those measured in the in situ experiments in a He atmosphere.

-... >­... o ...

o

- --- - chromate (ex situ ___ nitrite (ex situ)

( a)

2 3 r • .8.

o 2 r . .8.

Fig. 5. Magnitudes of Fourier transforms of EXAFS spectra obtained from passive films on iron. From [1].

The structural implications of these and other EXAFS results and of the Mössbauer spectroscopic results will be discussed when the structure of surface films is considered. It is dear, however, with regard to the issue of the composition of the film, that these results indicate that the in situ films differ from the ex situ and therefore the composition of the in situ films presumably must involve some hydrogen species.

42

2.2.3 Radiotracer Studies - These studies are only quasi-in situ. They have, however, revealed a quite different role for hydrogen in affecting the composition of surface films on iron. This role, suggested by Bloom and Goldenberg [31], is based on their findings that show that y Fe203 is not merely a spinel structure containing vacancies but is a modification of the Fe30 4 structure in which hydrogen atoms (protons) are substituted for some of the ferrous ions and that yFe20 3, or rather HFe50 s, is the end result of such a substitution when all the ferrous ions have been replaced. Yolken, Kruger, and Calvert [47] sought to test this idea by forming passive films on iron in solutions containing tritiated water and measuring the counting rate when the radioactive water adsorbed on the film's surface is removed by rinsing in non-radioactive water as the outer portions of surface films were removed by lowering the potential to values where the outer layers of surface films start to dissolve. It was found that the majority of the hydrogen introduced by forming surface films in a solution containing tritium was found in the outer parts of surface films on iron. The amount of hydrogen found in the outer portion of surface films in this experiment and others was of the same order of magnitude as that calculated by Bloom and Goldenberg.

2.3 ALLOYING ELEMENTS IN FILM

This discussion will mainly use iron-chromium alloys. These alloys were first introduced as a means for improving the ability of iron to form a more protective passive film and are therefore a good model system to use in considering the effect of alloying in a· simple manner. A number of techniques have shown that the composition of surface films on iron-chromium alloys does indeed contain chromium. Out of these techniques have come the following facts:

(a) The outer part of surface films on Fe-Cr alloys is enriched in Cr. This was shown by AES [50,51], XPS [52,53], EXAFS [44], gamma spectrometry [54] which makes an in situ determination of trace amounts of alloying elements that enter the solution during dissolution, soft x-ray spectroscopy (SXS) [55] and ion scattering spectrometry (ISS)[56]. (b) The film is made up of a mixture of iron and chromium oxides, the chromium probably being a hydrated chromium oxy-hydroxide. This was shown by XPS [32,57] and RHEED [42]. (c) The oxygen present in surface films on Fe-Cr alloys exists in two different binding states as M-OH or M-OOH and M-O. This was shown by XPS [52,57]. (d) The valence of alloying elements such as Cr can have different values. XANES studies by Long et al.[58] and Bardwell et al.[59] of FeCr alloys found evidence for Cr(III) and/or Cr(VI) ions in the surface films on the alloys. Another in-situ study by Davenport et al.[60] of the films on AlCr alloys found similar results.

43

2.4 MINOR IMPURITY ELEMENTS IN FILM

A good illustration of the effects of the introduction of minor impurity elements into the composition of surface films has been provided by Marcus and Grimal [61]. Using radiotracer and XPS techniques to investigate the effects of the impurity element sulfur on the passivation of NiCrFe alloys, they found that alloying with Cr had a strong beneficial effect on the passivation of Ni and NiFe alloys in the presence of S. With no Cr in the alloy, adsorbed S on the surface blocks the formation of a protective surface film. When the Ni or the NiFe alloy is alloyed with Cr, the detrimental effect of S is "strongly counteracted." Marcus and Grimal propose that the beneficial effect of Cr results from a competition between the growth of nickel sulfide and chromium oxide islands on the metal surface. This prevents the spread of the sulfide, allowing· the formation of a protective film of chromium oxide.

2.5 COMPOSITION AND BREAKDOWN

Compositional aspects can have consequences for affecting the breakdown process. Three aspects can be mentioned where breakdown has either altered the composition of the film or the composition of the passive layer has affected the breakdown process:

2.5.1 Breakdown affects composition Foley, Kruger and Bechtoldt [39] found in electron diffraction studies ofthe films formed on iron under various conditions that only spinel oxides with preferred orientation were observed when the film was formed in the passive region of the anodic polarization curve. If the film was formed in the pre-passive or transpassive regions, only oxy-hydroxides constituted the composition of the films. When chloride ions, which bring about breakdown, were introduced, even und er conditions where passivity conditions prevailed, oxy-hydroxides were observed. This suggests, but doesnot prove, that the breakdown process promoted by the chloride ions caused the conversion of the spinel oxides to oxy-hydroxides probably by a· dissolution-precipitation process.

Foley and Kruger [21] found evidence, using spectroscopic ellipsometry, that the composition of passive films on Fe was altered by the breakdown process due to introduction of Cl- into thefilm. Schneider et al.[62] found, however, that the breakdown of the films on FeCr and Fe Mo alloy surfaces involved no incorporation of CI- in the films and that the Cr and Mo content of the passive layers was not alteredby the presence of Cl-.

2.5.2 Hydrogen is involved in breakdown The radiotracer experiments [47] described earlier strongly support the existence of hydrogen or hydrogen containing species in the passive layer on iron. Does this hydrogen playa role in breakdown? Another result from these radiotracer experiments suggests that it

44

does. It was found that when a passive film formed in a passivating solution (O.lN NaN02) containing tritiated water was allowed to undergo breakdown in chloride solution, the chloride ions extracted the same quantity of hydrogen that they had measured in the outer layers of surface films; this strongly suggests that the breakdown process may involve the replacement or expulsion of hydrogen containing species by the aggressive species responsible for breakdown. McBee and Kruger [21] have proposed the following exchange process:

(CI-)solution + 00- + 2(OH-)mm -+ (CI-)mm +200w +2(OH-)solution

2.5.3 Composition affects the repassivation rate A major element in the breakdown process is believed to be the repassivation (film repair) ability of a metal substrate. Virtanen and Böhni [64] found in their study of FeCrP metallic glasses that the presence of the metalloid element P provided a high repassivation ability. They also found that the P played another role. When it was incorporated into the outer part of the chromium oxide surface film as a phosphate, it made the film cation-selective, thereby hindering the penetration of breakdown inducing Cl- into the surface film.

3. Structure

Since composition determines structure, these two aspects of the nature of surface films are closely tied together. The controversy surrounding the thickness of surface films on iron extends, of course, to its structure. Hence, much of the discussion on chemical composition is relevant to structure. This is especially so with regard to the different results obtained by ex situ and in situ techniques. As pointed out in the previous section, the composition of passive films examined under ex situ conditions differ from those measured in situ. The same considerations, apply to the structure of the film.

3.1 DETERMINATION OF STRUcrURE

Electron diffraction has been the main technique utilized in the determination of the structure of surface films. This ex situ technique has determined that surface films on iron has a cubic spinel structure [39,40] which can include the structurally very similar yFe20 3 and Fe30 4• Foley, Kruger and Bechtoldt [39] found that extra lines could be observed for yFe20 3 but Kuroda et al. [40] were not able to distinguish between the two spineis. A number of doubts have been expressed [39] that the structure of surface films determined by electron diffraction is that of the film that is formed in an aqueous solution while others [29,40] argue that electron diffraction does provide the actual structure of surface films on iron because the conditions in the electron microscope (heating by electron bombardment and dehydration by vacuum) are not sufficient to create the highly oriented crystalline films observed.

45

EXAFS [65] was applied to structural studies ofpassive films formed in aqueous passivating solutions (nitrite and chromate) by Long, Kruger and co-workers as a means to determine the structure of passive layers on iron. Their ex situ non-vacuum EXAFS measurements found spinel-like structures for surface films, but they also found that surface films exhibited Fe-Fe distances that differed from the anhydrous crystalline spineis. Also the sharpness of the peaks of the EXAFS spectra of surface films (especially the film formed in chromate) were not as good as the crystalline yFez03 and Fe304 used as standards. This suggests a lesser degree of order in surface films. A more re cent in-situ EXAFS study by Robinson [66] concluded that the passive film on Fe had a basic structure similar to that of y-FeOOH but with the significant differences that the second shell radius was shorter and that the structure of the film was significantly more disordered.

The EXAFS technique can also measure actual bond distances, e.g., Fe-O and Fe-Fe distances to better than 0.005 nm. An in situ EXAFS technique was used to compare the structures found by the non-vacuum ex situ EXAFS measurements described above [44]. The in situ results differ markedly from the ex situ and these differences reflect the effect of water (or other hydrogen containing species) on the structure of surface films. The Fe-O and Fe-Fe distances were obtained for the four types of films measured. These values, along with the averaged values for stoichiometric crystalline oxide, yFezÜ3' are given in Table I.

The increase in the Fe-O distance in comparing y Fe20 3 to the in situ passive film formed in chromate support the suggestion of Revesz and Kruger [67] that the hydrogen introduced in the in situ film "can lead to increased structural flexibility by forming M-OH bonds in addition to M-O bonds." The introduction of hydrogen in the in situ film may thereby increase the average Fe-O nearest neighbor distance in such a film. Moreover, the increased structural flexibility resulting from H bonding promotes, according to Revesz and Kruger, a tendency towards a more glass-like structure for the in situ film.

3.2ALWYING

An argument similar to that given for hydrogen in films can be made for the incorporation of an alloying element, e.g. Cr, into the film. The decrease in the Fe-Fe bond distance shown in Table I for the in situ chromate formed passive film when compared to the crystalline Fe-O suggests the same tendency. This is due to the fact that chromium is a good "glass former" because its oxides have good bond flexibility [67]. This increased flexibility due to the Cr and to the H introduced in the in situ film will result in Fe atoms exhibiting shorter bond distances in the in situ chromate formed film than those in the crystalline yFe20 3. This results from chromium being incorporated into surface films and apparently promoting the formation of a more glass-like structure, an indication of the effect of alloying iron with chromium.

46

TABLE I: EXAFS determined bond distances. From [63].

Fe-O Fe-Fe nrn (±O.OO2) nrn (±O.OO5)

yFe20 3 0.201 0.332

nitrite, er situ 0.200 0.327

chromate, er situ 0.204 0.322

nitrite, in situ 0.201 0.302

chromate, in situ 0.208 0.305

The trends shown in Table 1 will be discussed in Section 7. Besides bond distances, another observation can be made from Fig. 5. The structures of chromate~formed films are more disordered than the structures of the films formed in nitrite. This can be deduced from the Fourier resolution of the transforms [63]. As mentioned earlier, this results from chromium being incorporated into surface films and apparently promoting the formation of a more glass-like structure, an indication of the effect of alloying iron with chromium.

McBee and Kruger [68] found support for this idea in an electron diffraction study of surface films on Fe-Cr alloys. They found that as the amount of chromium in an Fe-Cr alloy goes up, the structure of surface films becomes more non-crystalline and the thickness goes down. This is shown in Table 11.

Even though these were er situ experiments, at high concentrations of chromium the noncrystalline structure is apparently stable enough outside of the aqueous solution.

The effect of alloying on surface film structure has also been found for Mg alloys using glancing angle reflection EXAFS, i.e., reflEXAFS [69]. ReflEXAFS provides the following kinds of structural information: 1) the nearest neighbor distances (eg. for the first shell the O-Mg distance, rl); 2) the coordination number, nl , for the first shell; and 3) the Debye-Waller factor, a2, a measure of the degree of structural and thermal disorder. The results obtained for pure Mg and a number of alloys and the two standards, MgO and Mg(OH)2' are given in Table 111. They show that there are clearly two types of surface film structures involved. The first group, representedby pure Mg and AZ61 (Mg-6%Al-1 %Zn), have films whose structures and compositions are closer to MgO. AZ61, however, has a lower coordination number, suggesting greater static disorder in the film.

T ABLE II: Tbe structural characteristics of anodic passive films on Fe-Cr alloys formed in IN H2S04 at potentials

in the passive region. From [67].

%Cr Structure Limiting Film Thickness (nm)

0 weIl oriented spinel 3.6

5 weIl oriented spinel 2.7

12 poorly oriented spinel 2.1

19 mainly amorphous 1.9

24 completely amorphous 1.8

47

Tbe second group Mg-15%Al, Mg-30%Al, and an extruded rapidly solidified alloy obtained from Allied-Signal Corp., appear to have surface films more like Mg(OH)2 because of the larger O-Mg distances. Tbe greatest degree of dis order was observed for the film on the RSP alloy with the Mg-15%Al alloy not far behind. Tbis interpretation of the results is suggested by theoretical studies which tied degree of noncrystallinity to corrosion resistance [67]. A greater tendency to exhibit structures that are even less crystalline than the films on Mg was found for Al by Kobayashi and Niioka [70]. Tbey found that the anodic films on high purity Al had a vitreous short range order structure that was similar to that of air-formed films on electropolished Al.

T ABLE III. ReflEXAFS data for Mg and Mg alloys and MgO and Mg(OH)2 standards. From Ref. [69].

r t (nm) n t

MgO 2.106 6.0

Mg(OH)2 2.173 3.0

Pure Mg 2.09 2.9

AZ6l 2.09 2.3

Mg15Al 2.12 3.1

Mg30Al 2.12 2.8

RS Mg Alloy 2.14 1.3

a2

0.0054

0.0056

0.0134

0.0056

0.0112

48

There are a number of studies in the literature [39,71,72] which show that the structure of the substrate influences the structure of the surface film upon which it is growing because of epitaxy. For example, Weisler et al.[71] found by an X­ray scattering technique that the anodic films formed on a Ti(1120) surface was crystalline and stoichiometric.

The structure of the substrate apparently also affects the properties of surface films. Clayton et al.[73] found that an ion implanted stainless steel surface became amorphous and this resulted in the formation on that surface of an amorphous passive layer with superior properties. However, in an examination of the role of metal-metalloid glass surfaces on the surface film properties, Janik­Czachor [74] found that the glass-like structure of these surfaces played a minor role in affecting the stability and, hence, the protective quality of surface films. Rather, the composition of the metallic glasses, especially the presence of the metalloids, strongly influenced the ability of the surface film to lower the reactivity of an electrode surface. She found that the metalloid phosphorous had the most beneficial effect on the ability of surface films to provide an effective barrier to electrochemical reactions at an electrode surface.

3.3 STRUCTURE AND BREAKDOWN

Do these changes from a crystalline towards a more noncrystalline glassy structure as chromium is introduced into a passive layer on iron, whether by alloying additions or by a chromate inhibitor, affect breakdown? McBee and Kruger [28] found that, indeed, if one compares the breakdown behavior of iron with that of

an Fe-20 Cr alloy, three effects can be observed: (a) the glass-like structure of surface films on Fe-Cr is more difficult to breakdown; (b) the effects of the breakdown process on optical changes occurring in its passive layer are considerably less; and (c) there is no recovery of the passive state, as there is for pure iron, upon removal of chloride ions from the environment. Obviously, these effects, besides being influenced by the structure of surface films, are also affected by the composition of the film, wh ich, as pointed out earlier, also influences the structure.

The second large influence on breakdown by the structure of surface films results from the fact that the substrate that bears this film can affect its structure. For example, Kruger [75] examined the role of crystallographic orientation of iron surfaces on breakdown tendencies and found that the tendency to pit go es up as the surface approaches the {llO} orientation (Fig. 6). He also found that the density of breakdown sites was highest for the {llO} orientation [8]. This role played by the {llO} orientation can perhaps be explained by the results of Foley, Kruger, and Bechtoldt [39] who found that surface films formed on {110} oriented iron surfaces had different epitaxial relationships with the substratethan

( 111 »

• PITTING

o NONPITTING o

• • (100' l-------.....---<.) (1101

Fig. 6. Stereographic triangle showing the effect of orientation of iron grains on tendency to pit. From Kruger [75].

49

non-passive films. They speculated that such differences promoted different epitaxial strains that made the film growing on {110} surfaces less protective.

Another influence of the surface upon which a passive film can reside can be inferred from the effect found by McBee and Kruger [68] for passive layers in Fe-Cr alloys containing different percentages of chromium. They found that, as the chromium conte nt was increased, fewer surface orientations were observed to form crystalline passive films, a 24% chromium alloy exhibiting noncrystalline films for all surface orientations studied. As pointed out previously, such passive layers that have a noncrystalline or glassy structure have a lower breakdown tendency.

What happens when the surface bearing a passive film has a glassy structure? This question was examined by Bertocci and Kruger [76] using an electrochemical noise technique. They compared the electrochemical noise spectra obtained in 1M H2S04 for the metallic glass Fe32Ni36Cr14P12B6 and the same alloy recrystallized. They observed that the noise currents of the metal glass surface was over two orders of magnitude lower than that found for the crystalline alloy surface. This was taken to indicate that the film on the crystalline surface has a much greater tendency to breakdown and suffer localized attack. Interestingly, their noise measurements revealed that the more breakdown resistant film on the metallic glass surface was not the result of the film's resistance to uniform corrosion because the overall current density observed for this film was not greatly different than that found for the passive layer on the crystallized alloy surface. Bertocci and Kruger suggest that the film on the glassy alloy was more resistant to breakdown than that on the crystallized alloy surface because its

50

structure is probably noncrystalline and therefore has the ability to inhibit the dynamic processes (probably breakdown-heal events) that are resposible for electrochemical noise and for localized breakdown. The effect of substrate structure on the properties of surface films that affect breakdown was also examined by Clayton et al.[73]. They found that a surface that had been ion implanted became amorphous and this resulted in the formation on that surface of an amorphous passive layer with superior properties. Natishan et al.[77], however, found that altering the surface structure of Al by implanting Al ions did not markedly affect susceptibility to breakdown.

4. Dereets

Two kinds of structural defects that affect the breakdown process will be considered here: (a) defects that result from the surface films having a nonstoichiometric composition and (b) cation vacancies.

4.1 DEFECTS FROM NON-STOICHIOMETRIC COMPOSmON

There are many indications that the passive layer on iron is non-stoichiometric, a condition that requires that such films contain defects arising from cation or anion vacancies. Ambrose and Kruger [18] and McBee and Kruger [21] sought to examine how the defect concentration affected the breakdown of surface films on iron by altering the concentration of these defects. They did this byexamining the effect of time and temperature of film growth process on breakdown. Table IV lists the induction times for breakdown, tind' observed by Ambrose and Kruger for various passive film growth times and temperatures.

The effect of thickness was also explored and Table IV shows that thickness was not the sole factor influencing the time to breakdown. For example, the time to breakdown for a film grown at + 0.64 V for 100 minutes was comparable to that for a thicker film formed initially at + 1.04 V for one minute and grown for 100 minutes at + 0.64 V.

The effect of growth time on tind can be interpreted using the results of Nagayama and Cohen [36] who showed that the concentration of defects in surface films changes with time of growth. They showed that the concentration of iron in the outer portions of surface films decreased with an increase in the time of film growth. They interpreted this to mean an increase in vacancies and Fe +6 ions. Another interpretation could ascribe this change in iron concentration to the oxidation of Fe+2 to Fe+3 and to the filling of the resulting vacancies with protons as suggested by BIoom and Goldenberg [31].

51

TABLE IV: Effect of growth time and heating on breakdown of passive film formed in borate-buffer solution (pR 8.4) and broken down in 7.5xlO-3N NaCI

solution. From Ambrose and Kruger [18].

Potential Growth Time Thickness Heated to Average volts,SHE min. Ä 80C Induction

Time tind,min.

+0.64 100 42.0 NO < 1

+0.64 1000 46.8 NO > 100

+1.04 100 52.4 NO < 10

+0.6~ 0.5 .... 0.5 .... 45.0 NO < 16 + 1.04-+ + 0.64 99

+0.64 100 47.2 YES > 100

The effect of temperature of film growth on tind is explained by the work of Vermilyea [78] who showed that one can lower the number of defects in anodic films on Ta and thereby decrease the rate at which these films will dissolve in hydrofluoric acid. It was reasoned that at the higher temperature, enhanced diffusion would effect a more rapid removal of defects in the film. Films treated in this manner at 65 C or higher with growth times amounting to no more than a total of 100 minutes were then exposed to high concentrations of chloride ions at room temperature. The breakdown times were comparable to those observed for films grown for a thousand minutes or longer at room temperature.

4.2 CATION VACANCIES

Using the point defect model [79] and the solvent-vacancy interaction model [80], Urquidi-Macdonald and Macdonald [81] have been able to examine how cation vacancy concentration influences the effect of minor alloying elements on the breakdown process. They were able to show how cation vacancy concentration, n.., affects the frequency distribution of the breakdown parameters, critical breakdown potential and induction time. Fig. 7 shows how the distribution of induction times depends "substantially" on 1ly for an Fe, Cr, Ni containing alloy that has a passive film with a 0.25 cation percent of the minor alloying element Mo. The theoretical results given in Fig. 7 imply that as 1ly decreases, tind increases and thereby increases the resistance to breakdown.

52

1.0

o~~~----~~--~----~----~ o 0.056 0.112 0.168

lind (s)

Pig.7. Effect of cation vacancy concentration per cm3 (numbers on curves) on the cumulative distribution function, P(tind), for the induction time, tind, for a passive fIlm containing 6+ - 3- complexes. Prom Urquidi and Macdonald [81].

5. Electronic, Optical and Magnetic Properties

Electronic and optical properties are intimately connected and will discussed together. The magnetic properties of surface films of relevance to interfacial electrochemistry have not been extensively studied, especially when determined by in situ measurements. A mention will be made of some recent work that has attempted to make such measurements.

5.1 ELECTRONIC AND OPTICAL PROPERTIES

An early example of studies of surface films concerned with optical and electronic properties using asynchrotron technique is that of Long, Kruger et al.[82] who obtained information about the electronic properties of passive films on iron from ex situ and in situ x-ray absorption measurements (XANES). XANES provides information on the shifts of x-ray absorption edges which depends on such factors as the valency, the degree of ionicity and the coordination number [82]. Besides edge shifts XANES provides information on electronic transition probabilities. A convenient method for studying the detailed shapes of the XANES spectra is to take the derivative of each spectrum.

5.1.1 Electronic conductivity One major concern in the literature regarding electronic properties has been with electronic conductivity, because the kinetics of passive film formation may be controlled by electronic conduction, ionic conduction, surface reactions, or a combination of all three. Iron, unlike valve metals such as Al, Ti, Ta, forms a very thin passive layer (less than 10 nm). The

53

valve metals whose films are good insulators can support large electric fields and by so doing form quite thick films (hundreds of nanometers). Oxygen cannot be evolved from valve metals. Iron, when high potentials are applied evolves oxygen instead of continuing to grow a thicker film. It is for this reason that many workers have suggested that surface films on iron is a good electronic conductor, or at least a semiconductor [83].

Derivative XANES spectra obtained from iron surfaces [82] are shown in Figs. 8 and 9. These spectra compare the detailed structure of the near edge regions for ex situ and in situ films formed in nitrite and chromate solutions. Fig. 8 contains the derivative spectra for the ex situ films. The shapes are quite similar to one another and also to the shape of yFeZ0 3• The entire spectrum for the chromate-formed film is shifted to lower energies as described in earlier work [65], although by a smaller amount than reported initially. The tendency toward a lower effective coordination charge, however, is again found for the chromate-formed film, indicating greater covalency in bonding in this film.

In Fig. 9, the derivative spectra are shown for the in situ passive films. The change in the intensity of the ls to 3d (see Fig. 8) between the in situ and ex situ spectra suggests the incorporation of hydrogen or water into the structure. The enhancement in this transition prob ability can be interpreted as a structural change toward a less symmetrie structure in the in situ case. This is consistent with the trends found in the nearest neighbor distances (Table I). The third peak, which is shifted far to higher energies (to about 7132 eV), may be taken as evidence that the next higher allowed state (np like) occurs after a rather large gap, indicating a change in the in situ semiconducting passive film toward becoming insulating.

This aspect of the nature of surface films is quite important since it is a significant factor in determining the mechanisms of film formation, breakdown, and the rate of metal dissolution. This is so because these processes involve the movement of electrons and ions from the metal surface through surface films or from the solution into the film. Moreover, electron transfer reactions that occur on surfaces with passive films depend strongly on the electronic properties of such films.

Long, Kruger et al.[82] used high resolution XANES to study this aspect. A number of in situ experimental techniques have also been applied by those who are interested in studying electron transfer reactions at passive film surfaces. These studies have shown passive films to have the properties of semiconductors. There are a number of examples of such studies. SchuItze and Stimming [84] obtained Tafel plots (potential vs.log current) for the redox reaction on passivated iron surfaces in aborate buffer solution for the [Fe(CN)6]4-/3- redox system and found large cathodic transfer coefficients which indicate according to their

54

I I I I.

I , , I JlOO ,". 1120 '124 '121 '132

!ntrn "\1

Fig. 8. Derivative spectra for the ex-situ nitrite-formed (solid line) and the chromate-formed (dashed line) passive films. From Lang, Kruger et al. [82].

711Z l1U 7121 7132 1131 71U

rnervr 'e\1

Fig.9. Derivative spectra for the in-situ nitrite formed (circles) and the chromate-formed (triangles) passive films. From Lang, Krüger et al. [82].

55

interpretation that the outer part of surfaee films, assumed by them to be yFe20 3, is an n-type semieonduetor. Similar studies were earried out by Delnick and Hackerman [85]. Using a photoeleetric potential technique, Oshe, Rosenfeld and Doroskenko also found n-type semiconducting properties for passivated iron surfaces [86]. Schmickler [87] likewise considers surface films on iron to be semiconductors but proposes a different tunnelling mechanism from that proposed by Schultze and Stimming, i.e., resonance tunnelling rather than direct coherent elastic tunnelling. An indication that passivated Ni also has a surface film with semiconductive properties comes from Hoppe and Strehblow [88] whose XPS and UPS studies led them to interpret that a change from Ni(OHh to NiOOH in the transpassive range in alkaline solutions could be interpreted on the basis of a simple semiconductor mode. The semiconductor properties of the surface films on Zn, ZnO single crystal, and Zn alloy electrode surfaces were also studied by Vilche et al.[89]. They found that the flat band potentials were practically the same whether measured on an oxidized Zn surface or the surface of a ZnO single crystal. The alloying elements Co or Ni generated deep donor levels.

In contradiction to the results just described, that indicate a semiconducting film on passive iron and other metals, are studies suggesting that the film has low electronic conductivity. This is based on a mechanism that fits the kinetics of film growth that suggests field assisted ionic migration which requires a film that is an insulator to support the large fields expected. Moreover, because yFe203 or HFeSOg (if it constitutes surface films or part of it) is an insulator, other workers [90,91] consider surface films or part of the film on iron to be an insulator.

An attempt has been made by Cahan and Chen [32] to reconcile these opposing findings. They propose that surface films on iron is neither a semiconductor nor an insulator but a combination of both - a "chemi-conductor". They define their new term as follows: "A chemi-conductor is defined here as a material whose stoichiometry can be varied by oxidative and/or reductive valency state changes. This nonstoichiometry can then modify the loeal electronic (and/or ionic) conductivity of the film. This is equivalent to a variable doping with defects rather than foreign species". Using a variety of techniques, they sought to show that surface films on iron goes from an insulator to a semiconductor as defects are added or removed. They interpreted their impedance measurements as showing that at low potentials in the passive region, the film does not have Fe30 4 at the metal interface (they claim to have ruled out a bi-Iayered film) but Fe2+ balanced for charge neutrality by excess protons or oxygen deficiencies. The material they propose as constituting the film along with the excess protons and the Fe2+ is y FeOOH. There is a "spread" of Fe2+ concentration from the metal film interface to the solution-film interface. At higher potentials, the Fe2+ ions are "pushed back towards the metal interface by the removal of excess protons until the film mainly consists of Fe3+, the outer part of the film thus becomes a stoichiometric Fe3+

oxide and an insulator. At the potentials where oxygen evolution is observed, they propose that the removal ofprotons induces the formation ofFe4 + ions. Although

56

there are many eontroversial aspeets to their model, it does explain the existenee of both semieondueting and insulating properties in surfaee films on iron de~ending on potential as brought about by the injection or removal of protons, Fe + and Fe4+ species. It is important to point out that the presenee or absence of water andj or protons in the film ean have a big influenee on eleetronie eonduetivity. For example, DiQuarto et al.[92] examined the semieonductor properties of surface films on er by a photocurrent spectroscopie technique and found that the value of the band gap of the films was related to the water content of the film.

The conductivity of the Si02 surface (passive) film on Si whieh is usually a good insulator was shown by Sugano [93] to be affected by the amount ofwater present in the oxidizing atmosphere producing these films. He found that the density of electron traps increases with an increase of water in the oxidizing atmosphere.

Not surprisingly, alloying affects the electronic conductivity of surface films. Kloppers et al.[94] found, using a photoelectrochemieal technique, from flat band potential measurements on the surface films of Fe-Cr alloys that the films become increasingly n-type as Cr content in the alloy goes up. This indicates doping of the film by Cr in the 4+ or 6+ valence state.

5.1.2 Valence and bonding state Adetermination of the optieal properties of surfaee films ean provide information on the valence state of species in the film. A potential-modulated reflectance study by Gutierrez et al.[95] of the valence state of Ir in anodic iridium oxide films found that optieal transitions (electroehromic effect) were caused by different valence states at different potentials.

The type of bonding determines electronic and optical properties. Thus, the edge shifts measured by EXAFS give an indieation of the degree of covalency in bonding [65]. Surface film, especially the ones formed in chromate, showed a greater degree of covalency than the crystalline oxides measured. This is also an indieation that surface films on iron have a less crystalline ordered structure than the oxides used as standards [65]. Thus, passive films on iron formed in passivating solutions resemble the structure of the crystalline cubie spineis but are less ordered than the crystalline anhydrous oxides, exhibiting a tendency toward more vitreous structures.

5.2 MAGNETIC PROPERTIES

Only one example will be given of studies aimed at determining the magnetic properties of surface films that are relevant to electrochemical processes because there exist few, if any, such studies. Krebs et al.[96] have applied the new (with respeet to in situ studies of surface films) technique of neutron reflectivity

57

measurements in the presence of a magnetic field. This study has been concerned with resolving the issue as to whether the passive film on Fe is the ferromagnetic yFe20 3 or the non-ferromagnetic yFeOOH. Preliminary indications suggest that the surface film is ferromagnetic.

5.3 ELECfRONIC AND OPTICAL PROPERTIES AND BREAKDOWN

Changes in the electronic properties of passive layers result when the breakdown process is initiated. This has been studied using the optical technique of ellipsometry which measures the effect of the interaction of chloride ions with passive layers on the films optical and, therefore, the closely related electronic properties. McBee and Kruger [21] found that in the visible spectral region, the relative phase retardation, ä, for surface films differed from the same film exposed to chloride ions at only three wavelengths for films formed at potentials above the critical potential far breakdown (Fig. 10). Observations at any of the wavelengths where breakdown was accompanied by optical (i.e. electronic) changes occurred showed extensive changes occurring during the breakdown process for surface films on iron. When the same measurements were made on an Fe-20Cr alloy [28] (which also showed breakdown induced changes at only three wavelengths), the optical changes in the ellipsometric parameters ä and 1jr (relative amplitude reduction) were considerably less pronounced. It appears then. that breakdown is more difficult because of the glassy structure of surface films on the Fe-Cr alloy. Moreover, smaller changes in the electronic properties (as reflected in the optical changes) for the film remain the same when chloride is removed befare the breakdown process is completed; this is unlike the film on iron which can be made to recover. This tighter bonding of the chloride in surface films on the alloy may be due to the more covalent character of the film on the Fe-Cr alloy as indicated by the XANES studies described previously.

6. Mechanical Properties

Mechanical properties of surface films are important because they playa role in the breakdown process. Less is known about them than the other properties covered above because of the great difficulty encountered when trying to deterrnine such properties for films in the 1 to 10 mm range. However, some early attempts have been made by Bubar and Verrnilyea [97,98] using an innovative electrochernical-mechanical technique to study a number of metals that form passive films. Leach and Neufeld [99] have also been able to measure the mechanical properties of passive layers on aluminum showing that an applied potential can affect the deformation of anodic films in this metal. Bubar and Vermilyea showed that the ductility of the passive layer on 304 stainless steel was higher than the film on unalloyed iron. Revesz and Kruger [67] have suggested that this may be a consequence of the film on the chrornium alloy being more

58

Fig. 10. The spectrum for the passive film at + 1000 mV (SHE) on {221} Fe in borate is shown by the solid curve. The dotted curve indicates the measurements brought about by .OO5N chloride addition after film growth. From McBee and Kruger [21].

noncrystalline than that on iron. However, they caution that the Leach and Neufeld results indicate that other factors mayaiso affect mechanical properties.

The mechanical property, residual stress, in oxide films on Ti and Al was found by Nelson and Oriani [100] to be generated during anodic oxidation. They found that more slowly grown surface films developed higher compressive stresses than those grown rapidly. Tensile stresses developed at the oxide film/metal interface because of the volume differences between the oxide and the meta!. Studies by Nelson and Oriani [101] of the mechanical stresses existing in anodic films on Ni showed that the films formed in the presence of Ct exhibited higher stress than those produced in Cl--free solutions. Potential steps in the film forming potential region produced stress variations that correlate with volume changes as a result of valence changes within the oxide film.

7. Concepts of the Nature of Surface Films Suggested by Their Properties

The properties of surface films as discussed above suggest the consideration and emphasis of three issues that bear strongly on the ability of surface films to affect interfacial electrochemistry: the role of crystallinity, the effect of alloying on crystallinity, and the role of hydrogen.

59

7.1 ROLE OF CRYSTALLINITY

T.P. Hoar [102] in his Palladium Medal address anticipated the importance of promoting a noncrystalline structure that would constitute a "superior" passive film. He set forth guidelines for producing the ideal structure, a "cross-linked monolithic amorphous structure not containing the extremely defective grain boundary regions of polycrystalline materials". Revesz and Kruger [67], following the ideas of Hoar and others [103,104] formulated some general considerations of factors influencing the formation of passive films with the proper noncrystalline structures:

(a) The structure should be noncrystalline (vitreous) and also have short range order. Being noncrystalline is not sufficient. For example fused silica and SiOz films have excellent short range order and are rather perfect vitreous solids whereas vacuum deposited SiO is non-crystalline solid without short range order, an indieation of chemical disorder producing an imperfect solid with a profusion of defects. The vitreous structures are like large inorganic polymerie molecules without the grain boundary paths of diffusion that characterize polycrystalline solids.

(b) The formation of vitreous structures is promoted by the existence of a large number of crystalline polymorphs which have almost the same standard free enthalpy of formation and the same short range order. This is called structural flexibility. An example of such asolid of importance to passivity is aluminum oxide (corundum) whieh has five polymorphs whose standard free enthalpies are within 2.5% of each other [67].

(c) Films with vitreous structures are promoted by bond flexibility. Such bond flexibility means that many different bond angles can exist in a vitreous structure because a large number of Iocal conformations will possess almost the same energy. This situation arises because the bonds in vitreous solids are usually partially ionie and partially covalent. In the most studied and most ideal vitreous solid Si02 the proportion of ionie to covalent is 40% to 60% for the Si-O bond. Bond flexibility arises when this proportion can vary widely without affecting the total bond energy. This situation promotes the possibility of a Iarge number of different local conformations (and hence bond angIes) of almost the same energy. Ex situ non-vacuum EXAFS studies cited earlier [65] showed that surface films on iron have a high er proportion of covalent bonding than crystalline iron oxides thereby indieating greater possibilities for the bond flexibility that promotes vitreous structures.

7.2 ROLE OF ALLOYING

When, as described above, factors exist that lead to an ideal vitreous structure for a surface film, they are influenced by alloying. Alloying with chromium provides the most important example of this. Chromium is a good "glass former" because its oxides have good bond flexibility [3]. In this instance the bond flexibility arises, unlike SiOz, out of the fact that chromium exhibits different oxidation states but, most importantly, the standard free enthalpies for these states are dose in value,

60

-141 and -136 kcaljmole for Cr03 and Crp3 respectively. Because ofthis, various structural conformations of nearly the same energy may exist and thereby promote the formation of vitreous passive films on chromium alloys. (See Table I where results are given for Cr added to the film by the environment rather than from an alloy, an equivalent effect.)

7.3 ROLE OF HYDROGEN

As pointed out by Revesz and Kruger [67], the hydrogen in surface films "can lead to increased structural flexibility by forming M-OH bonds in addition to M-O bonds." If the corresponding metal hydroxide has various polymorphs as, for instance, AlOOH [105], then various confirmations of M-OH may exist in the H-containing passive film; this results in an increased tendency towards noncrystallinity. The M-OH groups can be linked together via hydrogen bonds as in the case of AlOOH [105]; this facilitates the formation of a noncrystalline network (e.g. H-bonding in organic polymers)." Revesz and Kruger, however, also caution that H or OH do not form hydrogen bonded polymers for some metals, e.g., Fe, Ni or Cu [106] but do so for such metals as Cr, Al, Ta, Si etc. Hence a possible synergistic role for Cr in Fe-Cr alloys. Okamato [104] has described another roIe for hydrogen or water in passivity, the promotion of film repair where it may be involved, as Revesz and Kruger have suggested for silicon, in reducing reactive sites by tieing up dangling bonds [67]. It is the bond flexibility and structural flexibility introduced by the addition of hydrogen that are responsible for the trends observed in Table I.

8. Acknowledgement

I gratefully acknowledge partial support of this work by the National Science Foundation.

9. References

1. Kruger, J. (1988) Int. Mat. Rev., 33, p. 113.

2. Kruger, J. (1990) in "Advances in Localized Corrosion," H.S. Isaacs, U. Bertocci, J. Kruger, and S.S. Smialowska, Editors, Nat. Assoc. Corros. Engrs., Houston, p. 1.

3. Kruger, J. (1976) in "Passivity and Its Breakdown on Iron and lron Base Alloys," R.W. Staehle and H. Okada, Editors, Nat. Assoc. Corros. Engrs., Houston, p. 131.

4. "Passivity of Metals," R.P. Frankenthai and J. Kruger, Editors, Electrochern. Soc., Princeton (1978).

61

5. Faraday, M. (1884) Experimental Researches in Electricity, 2, Dover, Pub. (Reprinted 1965) New York, p. 250.

6. Uhlig, H.H. in Ref. 4.

7. Evans, U.R (1927) J. Chem. Soc., p. 1020.

8. Kruger, J. (1963) J. Electrochem. Soc., 110, p. 654.

9. Draper, P.M.G. (1967) Corros. Sci., 7, p. 91.

10. Kruger, J. and Calvert, J.P. (1967) J. Electrochem. Soc., 114, p. 43.

11. Nagayama, M. and Cohen, M. (1962) ibid., 109, p. 781.

12. Cortes, R, Froment, M., Hugot-Legoff, A. and Joiret, S. (1990) Corros. Sci., 31, p. 121.

13. Kubanov, K, Burstein, R, and Frumkin, A. (1947) Discuss. Faraday Soc., 1, p. 259.

14. Frankenthai, RP. (1971) Electrochim. Acta, 16, p. 1845.

15. Frankenthai, RP. (1967) J. Electrochem. Soc., 114, p. 542.

16. Genshaw, M.A. and Siroka, RS. (1971) J. Electrochem. Soc., 118, p. 1558.

17. Böhni, H. and Uhlig, H.H. (1969) J. Electrochem. Soc., 116, p. 906.

18. Ambrose, J.R and Kruger, J. (1972) in "Proc. 4th Int. Cong. on Met. Carr.," Nat. Assoc. Corros. Engrs., Houston, p. 698.

19. Bardwell, J.A., Fraser, J.W., MacDougall, B., and Graham, M.J. (1992) J. Electrochem. Soc., 139, p. 366.

20. Richardson, J.A. and Wood, G.c. (1973) ibid., 120, p. 193.

21. McBee, c.L. and Kruger, J. (1974) in "Localized Corrosion," Staehle, Brown, Kruger, and Agrawal, Editors, Nat. Assoc. Corros. Engrs., p. 252.

22. Heusler, KG. and Fischer, L. (1976) Werkstoff und Korrosion, 27, p. 551.

23. Sugimoto, K, Matsuda, S., Ogivara, Y., and Kitamura, K (1985) J. Electrochem. Soc., 132, p. 1791.

62

24. Kruger, J., Lillard, RS., Streinz, e.e., and Moran, P.J. (in press) Faraday Discussion 94, in "The Liquid-Solid Interface at High Resolution."

25. Videm, K, (1974) Kjeller Report KR-149, Institutt for Atomenergi, Kjeller, Norway, p. 67.

26. Galvele, J.R in Ref. 4, p. 285.

27. Kruger, J. (1987) in "Environment Sensitive Fracture of Metals and Alloys," RP. Wei, D.J. Duquette, T.W. Crooker, and AJ. Sedriks, Editors, Office of Naval Research, Arlington, VA, p. 133.

28. McBee, C.L. and Kruger, J. (1976) in "Passivity and Its Breakdown on Iron and Iron Base Alloys," RW. Staehle and H. Okada, Editors, Nat. Assoc. Coros. Engrs., Houston, p. 131.

29. Cohen, M. in Ref. 4, p. 521.

30. Revie, RW., Baker, B.G., and Bockris, J.O.M. (1975) J. Electrochem. Soc., 122, p. 1460.

31. Bloom, M.e. and Goldenberg, L., (1965) Corros. Sci., 5, p. 623.

32. Cahan, B.D. and Chen, C-T. (1982) J. Electrochem. Soc., 129, p. 921.

33. Wiesler, D.G., Toney, M.F., McMillan, e.S., and Smyrl, W.H. (1991) in ''The Application of Surface Analysis Methods to Environmental/Materials Interactions," D.R Baer, C.R Clayton, and G.D. Davis, Editors, Electrochem. Soc., Pennington, NJ, p. 440.

34. Larramona, G. and Gutierrez, e. (1989) J. Electrochem. Soc., 136, p. 2171.

35. Larramona, G. and Gutierrez, C. (1990) ibid., 137, p. 428.

36. Nagayama, M. and Cohen, M. (1963) ibid., 110, p. 670.

37. Vetter, KS. (1958) Z. Elektrochem., 62, p. 642.

38. Ord, J.L. and DeSmet, D.J. (1976) J. Electrochem. Soc., 123, p. 1876.

39. Foley, C.L., Kruger, J. and Bechtoldt, C.J. (1967) ibid., 114, p. 944.

40. Kuroda, K, Cahan, B.D., Nazri, e., Yeager, E., and Mitchell, T.E. (1982) ibid., 129, p. 2163.

63

41. Sato, N., Kudo, K, and Nishimura, R (1976) ibid., 123, p. 1419.

42. Clayton, C.R, Doss, K, and Warren, J.B. (1983) in "Passivity of Metals," M. Froment, Editor, Elsevier, Amsterdam, p. 585.

43. O'Grady, W.E. (l980) J. Eleetroehem. Soe., 127, p. 555.

44. Kruger, J., Lang, G.G., Kuriyama, M., and Goldman, AI. (1983) in "Passivity of Metals and Semiconduetors," M. Frament, Editor, Elsevier, Amsterdam, p. 163.

45. Lang, G.G., Kruger, J., Blaek, D.R, and Kuriyama, M. (1983) J. Eleetroehem. Soe., 130, p. 240.

46. Okamoto, G. and Shibata, T. (1965) Nature, 206, p. 1350.

47. Yolken, H.T., Kruger, J., and Calvert, J.P. (l968) Corras. Sei., 8, p. 103.

48. Simmons, G.W., Kellerman, E., and Leidheiser, Jr., H. (1976) J. Eleetroehem. Soe., 123, p. 1276.

49. Eldridge, J., Kordesch, M.E. and Hoffman, RW. (1982) J. Vac. Sei. Teeh., 20, p. 934.

50. Yaniv, AE., Lumsden, J.B. and Staehle, RW. (1977) J. Electrochem. Soc., 124, p. 490.

51. Okada, H., Ogawa, H., Itoh, I. and Omata, H. (1976) in "Passivity and 115 Breakdown on Iran and Iran Base Alloys," RW. Staehle and H. Okada, Editors, Nat. Assoe. Corras. Engrs., Houston, p. 82.

52. Asami, K, Hashimoto, K, Musumoto, T., and Shimodaira, S. (1976) Corros. Sei., 16, p. 909.

53. Okamoto, G., Taehibana, K, Shibata, T., and Hoshino, K (1974) J. Jpn. Inst. Met., 38, p. 117.

54. Kolotyrkin, Y.M. and Knyazheva, V.M. in Ref. 4, p. 678.

55. Holliday, J.E. and Frankenthal, RP. (1972) J. Eleetrochem. Soe., 119, p. 1190.

56. FrankenthaI, RP. and Malm, D.L. (1976) ibid., 123, p. 186.

57. Tjong, S.C., Hoffman, RW., and Yeager, E.B. (1982) ibid., 129, p. 1662.

64

58. Long, G.G., Kruger, J., and Tanaka, D.K (1987) ibid., 134, p. 264.

59. BardweIl, J.A., Sproule, G.!., MacDougall, B., Graham, M.J., Davenport, AJ., and Isaacs, H.S. (1992) ibid., 139, p. 371.

60. Davenport, AJ., Isaacs, H.S., FrankeI, G.S., Schrott, AG., Jahnes, C.V., and Russak, M.A (1992) in "X-Ray Methods in Corrosion and Interfaeial Electro-chemistry," A Davenport and J.G. Gordon II, Editors, Electrochem. Soc., Pennington, NJ, Vol. 92-1.

61. Marcus, P. and Grimal, J.M. (1990) Corros. Sci., 31, p. 377.

62. Foley, c.L., Kruger, J., and Bechtoldt, c.J. (1967) J. Electrochem. Soc., 114, p.944.

63. Long, G.G. and Kruger, J. (1991) in "Techniques for the Characterization of Electrodes and Electrochemical Processes," R. Varma and J.R. SeIman, Editors, John Wiley and Sons, NY, p. 167.

64. Virtanen, S. and Böhni, H. (1990) Corros. Sei., 31, p. 333.

65. Long, G.G., Kruger, J., Black, D.R., and Kuriyama, M (1983) J. Electroanal. Chem., 150, p. 603.

66. Robinson, J. in Ref. 60, p. 239.

67. Revesz, AG. and Kruger, J. in Ref. 4, p. 137.

68. McBee, c.L. and Kruger, J. (1972) Electrochimica Acta, 17, p. 1337.

69. Kruger, J., Long, G.G., Zhang, Z., and Tanaka, D.K (1990) Corros. Sei., 31, p. 111.

70. Kobayashi, M. and Niioka, Y. (1990) Corros. Sei., 31, p. 237.

71. Wiesler, D.G., Toney, M.F., Samant, M.G., Melroy, O.R., McMillan, C.S., and Smyrl, W.H. (1992) Surf. Sei., 268, p. 57.

72. Kruger, J. (1961) J. Electrochem. Soc., 108, p. 503.

73. Clayton, c.R., Doss, KG.K Wang, Y.F., Wauen, J.B., and Hubler, G.K (1982) in "Ion Implantation into Metals," V. Ashworth, Editor, Pergamon, Oxford, p. 67.

74. Janik-Czachor, M. (1990) Corros. Sci., 31, p. 325.

65

75. Kruger, J. (1959) J. Electrochern. Soc., 106, p. 736.

76. Bertocei, U. and Kruger, J. (1980) Surf. Sei., 101, p. 608.

77. Natishan, P.M., McCafferty, E., and Hubler, G.K. (1986) J. Electroehern. Soe., 133, p. 1061.

78. Verrnilyea, D. (1957) ibid., 104, p. 485.

79. Un, L.F., Chao, C.Y., and Maedonald, D.D. (1981) ibid., 128, p. 1194.

80. Macdonald, D.D., and Urquidi, M. (1985) ibid., 132, p. 555.

81. Urquidi-Maedonald, M. and Maedonald, D.D. in Ref. 2, p. 33.

82. Long, G.G., Kruger, J., Kuriyarna, M., Blaek, D.R, Farabaugh, E., Saunders, D.M., and Goldrnan, A.I. (1984) in "Proe. 9th lnt. Cong. on Met. Corr.," National Research Council, Ottawa, Vol. 3, p. 419.

83. Vetter, KJ. (1963) J. Eleetroehern. Soe., 110, p. 597.

84. Shultze, W. and Stimming, U. (1957) Z. Physik. Chern. N.F., 98, p. 285.

85. Delnick, F.M., and Hackerrnan, N. in Ref. 4, p. 116.

86. Oshe, E.K., Rosenfeld, J.L. and Doroskenko, V.c. (1970) Kokl. Akad. Nauk. SSSR, 194, p. 612.

87. Schmiekler, W. in Ref. 4, p. 102.

88. Hoppe, H.-W and Strehblow, H.-H. (1990) Corros. Sei., 31, p. 167.

89. Vilehe, J.R, Jüttner, K., Lorenz, W.J., Kautek, W., Paatsch, W., Dean, M.H., and Stirnming, U. (1990) ibid., 31, p. 679.

90. Ord, J. and DeSrnet, D.J. (1966) J. Electrochern. Soe., 113, p. 1258.

91. Moshtev, RV. (1971) Eleetroehirn. Acta, 16, p. 2039.

92. DiQuarto, F., Piazza, S., and Sunseri, C. (1990) Corros. Sei., 31, p. 721.

93. Sugano, T. (1990) ibid., 31, p. 21.

94. Kloppers, M.J., Bellucei, F., and Latanisiion, RM. (1992) Corrosion, 48, p.229.

66

95. Gutierrez, e., Sänchez, M.A, Pefia, J.I., Martinez, e., and Martinez, M.A (1987) J. Electrochem. Soc., 134, p. 2119.

96. Krebs, L.A., Kruger, J., Lang, G.G., Ankner, J.F., Majkrzak, C.F., Sutija, S.K., and Wiesler, D.G. (1992) in "Oxide Films on Metals and Alloys," B.R MacDougall, RS. Alwitt, and T.A Ramanarayanan, Editors, Electrochem. Soc., Pennington, NJ, Vol. 92-22, p. 580.

97. Bubar, S.F. and Vermilyea, D.A (1966) J. Electrochem. Soc., 113, p. 892.

98. Bubar, S.F. and Vermilyea, D.A (1967) ibid., 114, p. 882.

99. Leach, J.S.L. and Neufeld, P. (1969) Coros. Sei., 9, p. 225.

100. Nelson, J.C. and Oriani, RA, ibid., in press.

101. Nelson, J.C. and Oriani, RA Electrochimica Acta, in press.

102. Hoar, T.P. (1970) J. Electrochem. Soc., 117, p. 17e.

103. McBee, e.L. and Kruger, J. (1972) Electrochim. Acta, 17, p. 1337.

104. Okamoto, G. (1973) Corros. Sei., 13, p. 471.

105. Leach, J.S.L. (1975) Surf. Sci., 53, p. 257.

106. Wells, AF. (1975) in "Structural Inorganic Chemistry," Clarendon Press, Oxford, p. 459.

107. Phillips, C.S.G. and Williams, RJ.P. (1965) in "Inorganic Chemistry," Oxford University Press, NY, Vol. 1, p. 533.

THEORY OF THE X-RAY SCATTERING FROM SURFACES AND INTERFACES

R.A. COWLEY Oxford Physics Clarendon Laboratory Parks Road Oxford OXI 3PU UK

ABSTRACT. The theory of the X-ray scattering from surfaces and interfaces is developed using the kinematical approximation. Initially the scattering is obtained for a simple flat surface and then for a simple overlayer. The relationship between the scattering around Bragg reflections and the reflectivity at small incident angles is then developed. Finally the effect of surface roughening on the scattering is described. Throughout reference is made to some typical experimental results obtained both with rotating anode and synchrotron sources of X-rays.

1. INTRODUCTION

The past 10 years has seen the development of X-ray scattering techniques for the study of the structures of surfaces, and thin layers. At first this seems a surprising development because X-ray scattering has been used for 90 years to obtain the structures ofbulk materials, while low energy electron techniques, LEED and RHEED, have been very successfully used to probe surface structures. The reasons for the development of the X-ray techniques are at least three-fold. Low energy electrons or helium atoms are strongly scattered by atoms and so the results can only be quantitatively interpreted if fuH and complex multiple scattering calculations are performed. In contrast X-rays are weakly scattered by individual atoms, and so the intensities can mostly be readily obtained from the lowest order of scattering theory known as the kinematical theory. This leads to an easy and direct interpretation of the experimental results in terms of the surface structure.

The second reason is the development of the X-ray scattering techniques. Clearly the use of high intensity X-ray beams from synchrotron sources has been important, but it is now realised that with the appropriate diffractometers it is possible to perform many measurements with conventional X-ray sources such as rotating anodes. There is no real reason why surfaces could not have been studied with X-rays in the 1930s! The third reason for the use of X-rays is that low energy electrons and helium atoms can only be used in high vacua. The greater penetrability of X-rays enables them

67

C. A. Melentires anti A. Tadjeddine (eds.), Synchrotron Techniques in Inter/acial Electrochemistry 67-84. © 1994 Kluwer Academic Publishers.

68

to be used in gaseous and liquid environments, and also to study buried surfaces and interfaces. This is, of course, of crucial importance for the in-situ study of electrochemistry .

The purpose of the rest of this paper is to show the important features of the X­ray scattering from surfaces and thin layers. This is developed in detail for some idealised models because unfortunately the relevant theory is not yet included in standard solid state text books. The results of the theory will be illustrated by experiments on a variety of simple systems to illustrate various aspects of the theory. We leave to other lectures the description of the results which have been obtained for real electrochemical systems.

2. SCATTERING FROM A FLAT SURFACE

Initially we consider the scattering from a simple cubic crystal which has a flat surface at the plane f = 0, as shown in fig. la. The incident beam of X-rays has wavevector, k, and is incident at an angle (}l to the surface. The scattered beam emerges from the surface at an angle (}2. Throughout we assume that the crystal is much larger than the X-ray beam; extended face geometry, unlike the case for the most bulk crystal structure determinations for which the crystal is usually much smaller than the beam. Within the kinematical approximation the scattering amplitude is obtained by adding the scattered amplitude from each atom with the appropriate phase factors, and neglecting all the multiple scattering. If the scattering from each unit cell is b the scattered amplitude from the whole crystal is

A=

where Q is the wavevector transfer while a1 and a2 are the lattice parameters within the planes. The scattering amplitude b is given by (e2/mc2)F(Q) n1.n2 where F(Q) is the structure factor of the unit cell and n1 and n2 are the polarisation vectors of the incident and scattered X-rays. Eqn (2.1) is the usual X-ray scattering equation except that the sum over f is taken from 0 to 00 instead of - 00 to 00, and we have included the effects of p., the absorption coefficient associated with each plane. These changes imply that the effect of the surface is explicitly included. The scattering cross-section can then be obtained in the usual way by performing the summations over mh m2 and f to give

2 I b r Ao S(Q) = 47t - -.-axy sm61

1 ----- A(Q)A(Q) . 1 +e -z"-2e -"cos(Qza)

(2.2)

69

o

2

Figure 1. Part (a) shows a simple square lattice and the scattering geometry. Part (b) shows a model adsorbed layer and the consequent distortion of the first bulk layer.

In this expression Ll(QJ Ll(Qy} is zero unless Qx and Qy are reciprocaI lattice vectors; Tx, Ty ' while iixy is the area of one unit cell in the x - y plane. Ao is the cross-sectionaI area of the beam and Ao/sin81 is the area of the beam footprint on the crystaI. Usually the attenuation is smaII, p. < < 1, when eqn (2.2) becomes

(2.3)

Eqn (2.3) is crucial to the development of the rest of this chapter. It differs from the

70

normal three-dimensional Bragg reflection formula in that it corresponds to streaks in reciprocal space for wavevectors, such that Qx and Qyare Tx and Ty , as illustrated in fig. 2a.

a

bll.,x

Fig. 2. Part (a) shows the reciprocal lattice for fig. la and the surface streaks are shown by solid lines. Part (b) is a similar figure for fig. lb showing the bulk streaks and the streaks due just to the adsorbate (dotted).

These streaks or lines in reciprocal space are the scattering manifestation of the surface at f = 0 in fig. la. The intensity of these streaks or crystal truncation rods is large whenever Qz = Tz at three dimensional Bragg reflections. elose to these reflections, Qz = Tz + q, eqn (2.2) becomes;

Zlbl2 Ao I S(q) = 41t - -. - 2-.. t:..(Q) t:..(Q), v 8mB l q2 + (Illa) )

(2.4)

71

where v = axy a is the volume of the unit cel1. When this expression is integrated over q, the intensity becomes proportional to Ao/(sin 81 p.) which is the illuminated surface area times the penetration depth. The intensity is therefore proportional to the bulk of the crystal. In contrast when qa > p., the intensity is proportional to Ao/(sin ( 1), the surface area, and the scattering is a surface effect. This scattering does not occur in the usual derivation because the sum over t is from - 00 to 00 and p. is neglected so that the surface effects are complete1y negligible compared with bulk effects. In real crystals they are not negligible.

Before discussing more complex situations we must discuss the relationship between S(Q) and the scattering intensity I(Q) observed in an experiment. Frequently the scans are performed along the streaks by varying Qz while keeping Qx and Qy fixed,

f • oS

10000

1000

100

10

0.01 0.1

WQy."e~tor "'p'

Fig. 3. Intensity versus wavevector for Si (111) slices with different thicknesses of oxide films. The straight lines are to q-a with ~ = 1.95 ±O.07 [2].

72

i::' -;;; c u C :; 10' u

~ .. u

.~

\ I

" '" - -rer"endicular moment um tr3n~fcr. 11 r.l.u.1

Fig.4. The scattered intensity [3] along the %(2 + f, 2 + f, 4 + f) streak of a (111) Ge surface. The dotted line is calculated for a smooth bulk surface, and the solid line inc1udes surface reconstruction.

and using the finite width of the detector to perform the integration over the delta functions in eqn (2.3). If the x direction is in the scattering plane, and the angle of acceptance of the detector is (XI then dQx = (Xl k sin 82, while if y is perpendicular to the scattering plane, dQy = (X2 k. The observed intensity is then proportional to eqn (2.3) but with the delta functions replaced by 1/(12 sin 82),

The observation of the intensity variation of the surface streaks along Q. is now the basis of many determinations of surface structure [1]. Surface streaks have been measured using semiconductor slices, and the results, presented in fig. 3, show data obtained with a rotating anode source [2], and c1early illustrate the q.2 behaviour dose to the Bragg peaks. The synchrotron data [3] shown in fig. 4 shows that the streaks extend from one Bragg peak to another and give a result very dose to that expected for a flat surface. The small deviations are important but surface reconstructions are needed to explain them.

3. AN IDEAL OVERLAYER

A simple model of an overlayer is shown in fig. lb. The crystal of fig. 1a is rotated to give a (110) surface, and on top of this surface is deposited a half-fi1led monolayer of possibly a larger atomic species. The amplitude of the scattering by the overlayer

73

is taken to be 2bB, and the layer is situated a distance a l above the bulk as shown in fig. Ib. As a result of the overlayer the top layer of the bulk is distorted by a lateral displacement ua, and is a vertical displacement from the bulk of a2' The scattering amplitude can then be obtained by adding to eqn (2.1) the scattering from the overlayer and top deformed layer to give the scattering as:

(3.1)

where

(3.2)

and Cl = cos Qx(a/2 + au), C2 = COS Qx a/2 while the attenuation I-' = O. The first term in eqn (3.2) gives the scattering from the undistorted layers which are rods of intensity in the z - direction with an intensity increasing as q-2 close to the Bragg reflections. The remaining terms cancel if B = 0, u = 0 and a2 = a/2, when the crystal has an undistort~d surface. The interesting results in eqn (3.2) are in the final term which gives rise to rods of scattering between the bulk truncation rods, fig. 2b. The scattering in these rods depends only on the two top layers and so does not have 1/q2 singularities. Indeed if u = 0 and a2 = 1/2a their intensity is proportional to B2 and their contribution to eqn (3.2) is independent of Q. This is because then only the top layer has the larger periodicity, and the scattering from a single layer is independent of Qz. If the topmost layer of the crystal is also distorted, the Q. dependence of eqn (3.2) reflects the interference between the two distorted layers and the bulk. They give rise to asymmetries in the intensities about the Bragg reflections.

A detailed study of the Qz dependence of the intensities of the different types of rods enables in principle the structure of surface layers to be deduced as illustrated in part in fig. 4 and also in fig. 5, wh ich illustrates the effect of changing the coverage of the top layer of a Ge crystal [4]. Unfortunately it is very difficult and time consuming to determine the Qz dependence of many of the rods of scattering, and furthermore it is only in the last few years that sufficiently flexible instrumentation has become available. It is not surprising therefore that in practice the fuH power of X-ray scattering for determining the structure of surfaces and layers has not been exploited.

74

.,'/ //

// //

,// ."/

." / ,~ /

.' /

/" .' , 6t = 0.75 I

i / 0,1 = 0.35 " I

'I

Fig. 5. The scattered intensity [4] from a Ge (I 1 f) streak showing a clean surface and after sputtering half a layer (open points).

4. REFLECTIVITY

Glancing angle reflectivity is a special example of the scattering discussed above because it occurs around the Bragg reflection at the origin of reciprocal space Tx = Ty = Tz = 0, and so occurs for all materials; crystals, polycrystals, amorphous materials and even liquids. For X-ray scattering, glancing angle reflectivity occurs when 81 and 82 are small, so that the wavevector transfer q = k(81 + 82) or if 81 = 82 = 8, q = 2k8. Under these conditions eqn (2.4) becomes, neglecting the absorption,

S(q) = 81t21 !!.12 Aok ä(Q,) ä(Q) v q3

(4.1)

The reflectivity is then given by integrating over the delta functions with the detector aperture as discussed in section 2 to give

R(q) = 161t2 I !!. r ' q4 V

(4.2)

75

wh ich can be written as

(4.3)

with

(4.4)

This result shows that the reflectivity decreases as q-4, unlike the q-2 factor obtained near Bragg reflections. The additional factor of q-2 arises from the sin 01 in eqn (2.4) which is the change in the illuminated surface area as (JI varies and from a similar sin (J2 factor arising in the integral over ~(QJ in eqn (4.1).

For small wavevectors, q, the kinematical approximation on which these results depend fails and eqn (4.3) is incorrect. The failure arises because, if the reflectivity is large, a substantial fraction of the beam is reflected at each plane, and it is necessary to take account of the multiple scattering between the incident and scattered beams. A second effect is that the theory developed so far has assumed that the electric field within the crystal is the same as that incident on the crystal. This is not the case when 0 is small, and it is necessary to take account of the difference in the theory. It is then possible to obtain the reflectivity by using the dynamical theory of X-ray diffraction [5]. An alternative approach is to use the well-known Fresnel formula [6] for the reflection of electromagnetic waves which gives the same result, namely;

(4.5)

1

where the Fresnel critical angle is (Je = qR/(2k). There is a similar failure of the kinematical approximation near to a Bragg

reflection, when q becomes very small. The multiple scattering of the dynamical theory gives a correct description of the scattering and it is different from the kinematical theory when q < 3qo where qe = 47r/k I b/v I (sin (JI sin (J2Y'h. It is however worth commenting that the q-2 behaviour near Bragg reflections was first obtained using the dynamical theory for the scattering from a slab of crystal [5], but for many years it was not appreciated that the behaviour arose from the surface and not from multiple scattering in the bulk.

In practice the failure of the kinematical theory near Bragg reflections is usually

76

unimportant because qc - 10-4 A-l, and the quality of the crystal and the instrumental resolution usually prohibit accurate measurements of the intensities for these small wavevectors, q. The kinematical theory of the reflectivity fails if q < 'h and 'h is much larger - 0.02 A-l, and so it is frequently necessary to make Use of the dynamical theory in interpreting reflectivity measurements. It is, however, the case that the measurements are difficult to make reliably at very small q and 0, and that often the most important part of the reflectivity curve is that for which the kinematical theory is appropriate.

Most reflectivity measurements are not made of materials with perfectly clean surfaces. Almost invariably there are oxide layers or layers which have been deliberately deposited. The theory of the reflectivity including layers can within the kinematical theory be developed as described in section 3, except that it is essential to consider all the layers as every material has a reflectivity profile. The effect of different layers is then to give rise to interference between the scattering from the

T

1)-1

2 3 IP (deg)

4

Fig. 6. The reflectivity T = RcfJ4 from Si slices with nominally 30 A and 50 A of oxide [2]. The solid lines give fits which show the thicknesses were 35 and 50 A and a roughness of about 5 A.

77

different layers, and the shape of the reflectivity gives information about the thickness of the layers (period of the oscillations), and the difference in the electron density in the layers (the amplitude of the oscillations). This is illustrated in fig. 6 which shows the reflectivity multiplied by q4 for some silicon slices on which oxide layers have been grown of different thickness [2]. The oscillatory form gives the oxide thickness accurately. It is interesting to compare fig. 6 with fig. 3 for the same silicon samples. The oscillations from the interference between the oxide and substrate are absent near the Bragg reflection because the oxide is amorphous and does not have Bragg reflections. Fig. 3 is therefore sensitive only to the silicon crystal surface, while the reflectivity, fig. 6, provides information about both the silicon and silicon oxide layer. The decrease in the reflectivity at large q in fig. 6 falls off more rapidly than q-4 because the top surface is rough as will be described in the next section.

5. SURFACE ROUGHENING

Most surfaces encountered experimentally are not the flat surfaces discussed so far in this article. It is therefore essential that the effect of surface roughness on the scattering be considered. In many ways the results are similar to the effects of thermal motion on the scattering by crystals, giving rise to thermal diffuse scattering, and Debye Waller factors, if the motion becomes too large the lack of long range order. The effect of surface roughness can be included in the development if the surface at a position R = (x,y) = (al ml , a2 m2) is given by h(R). Then eqn (2.3) becomes

S(Q) I b f F(Q) (5.1) \12 + 4sin2(Qz a/2)

with

F(Q) =E exp i Qz(h (R1)-h (~)exp i Qp .r (5.2) 1,2

where r = R l - R 2 and Qp = (Qx , Qy). The scattering is given by an ensemble average over F(Q), and the structure is very dependent upon the behaviour of heR) as R ~ 00. If the heights are bounded, the surface is said to be smooth, and there is a sharp delta function component to the scattering

F(Q) =1 D(Q) 12 !J.(Q) !J.(Q), (5.3)

78

where

D(Q)=< exp(i Qz h(R» >

If h(R) is a random number of steps away from the average plane, a, and n is a Poisson random variable with variance rr

D(Q) = exp( - 0 2 8in2 (Qz a/2) ) (5.4)

Close to a Bragg reflection this is very similar to a Debye Waller factor

and so reduces the intensity at large q as shown in fig. 6 for the reflectivity. There is also diffuse scattering, but since there is no correlation between the heights in this model, this diffuse scattering is spread over all Qx and Qy and cannot be distinguished from bulk scattering and background.

A more realistic model of the roughness is one in which the roughness is correlated, and the h(Rl) and h(R2) are similar when R 1 and R2 are neighbouring points. A number of models of this type of surface have been proposed, and they all give similar results for the truncation or Bragg-like rod, eqns. (5.1), (5.2) and (5.4). The diffuse scattering depends on the nature of the correlation in the roughness [7]. The basic features are that the scattering as a function of Ox and Qy is peaked at the reciprocal lattice point, under the Bragg-like rod, but the width of the peak and its dependence on Qz are dependent upon the detailed form of the correlation function between the heights.

Rough surfaces are ones in which the difference in the height of two points diverges with increasing distance, r.

0( I h(o) - h(r) f > = A r 2y (5.5)

In general the form of F(Q) cannot then be found, but if'Y = 1/2

79

(5.6)

while if 'Y = 1

F(Q) (5.7)

where p = (Qx- Tx ' Qy - Ty ).

Both of these forms are peaked as p -+ 0, but do not have the delta funetion eomponent of the smooth surfaees. The widths of the funetions with p inerease as q inereases, but the detailed behaviour is dependent upon the exponent, 'Y •

a b

o o Q~ x

Fig. 7. Sehematie transverse seans for fixed q for (a) a perfeetly flat surfaee (b) a smooth surfaee Ce) a surfaee at a roughening transition and (d) a rough surfaee.

80

Finally, an important case of a rough· surface is when the left hand side of eqn (5.5) is ~ln r, because this behaviour occurs in the theory of surface roughening [8] and in the theory of liquid surfaces. The result for the scattering is singular as p -. 0, diverging as p- a with a = 2 - ~ q2/2, but with no delta function component. The schematic form of the scattering from the different types of rough surfaces is shown in fig. 7, and clearly a great deal of information can be obtained about surface roughness from detailed measurements of these profiles. Unfortunately in practice it is very difficult to distinguish between the different cases, and high resolution measurements must be made if the roughness is to be studied in detail.

6. STRUCTURE OF A NIOBIUM LA YER ON SAPPHIRE

As an indication of the type of data which can be obtained fairly readily about surface layers, in this section some recent experiments [9] on the structure of Nb layers grown epitaxiallyon sapphire will be described. The importance of these layers is that they are used as buffer layers for the epitaxial growth of many different metals and magnetic superlattices. The experiments were performed using Cu Kai radiation from a. rotating anode source with a Ge two bounce channel-cut monochromator. A Ge monochromator was used as the analyser. Initially the experiments studied the sapphire substrates which were prepared with nominally flat (1120) faces. Reflectivity measurements showed, however, a slight bending of the substrates so that scans were performed perpendicular to the reflectivity streak, and the reflectivity found by integrating under the peak in the scans. Secondly the substrate was only 1 cm x 1 cm, and at the smallest angles the beam footprint was larger than the sampie size_ After correction for this effect, the sapphire reflectivity was weIl described by the theoretical expressions with a surface roughness, (1, of 3.6 A. The (1120) reflection was then studied in detail and the truncation rods were found to vary as q-1.93 in good agreement with the expected q-2 behaviour. It was also found that the (1120) Bragg reflection was not in a parallel direction to the reflectivity, showing that the surface was off-cut from the (1120) plane by an amount which varied between 0.03° and 0.5°.

Studies were then made of the structure of Nb films deposited on the sapphire substrates. The Nb films are known to be oriented with (110) planes approximately parallel to (1120) sapphire planes. In fig. 8 we show the reflectivity from one of the Nb films and a fit using a two layer model for the film. The rapid oscillation arises from interference between the two sides of the Nb film and corresponds to a film thickness of 417 A. The longer period oscillation which destroys the structure near 0.25 A-I is due to a thin oxide film on the surface of the Nb of thickness 13 A, and increased slightly as the experiment progressed. Clearly this model gives a very good account of the reflectivity. In fig. 9 the scattering near the (1120) sapphire and (110) Nb Bragg reflections is shown. This scattering was measured using transverse scans through the rod of scattering and integrating to ensure that the integrated intensities were obtained correctly. With the high resolution tripie crystal arrangement it is very difficult to scan along the top of a scattering ridge especially when the surface and crystallographic planes may be misoriented by even small angles. Fig. 9 shows a sharp

81

0

-1 ....-.. ~ ~ -2 -rn Z -3 ~ ~ Z -4 -'-" tl.O • 0 -5 -:I

-6

-7 0.0 0.1 0.2 0.3 0.4 0.5

Q [A -1]

Fig. 8. The reflectivity of a Nb film on sapphire. The fall-off at low angles is due to the footprint of the beam becoming larger than the sample surface.

well defined sapphire peak and aseries of peaks arising from the Nb layer. The solid line shows a calculation of the scattering from the Nb layer assuming it was a regular block when the pattern is a conventional single slit diffraction pattern. Clearly the data is reasonably described by this simple model although there is also a weak systematic asymmetry in the measurements, the origin of which is currently uncertain.

82

o

-2

-4

-6

-8

-10

-12 0.92 0.96 1.00 1.04

1 [units of 27T/C] Figure 9. The longitudinal scattering near the (110) Nb Bragg reflection, on the scale for l, and the (1120) sapphire reflection.

The final figure, 10, shows a transverse scan through the largest Nb peak in fig. 9. The scattering is dominated by a sharp intense central peak which shows that the transverse mosaic width of the Nb film is less than 0.004°, and so is of very high quality and much more perfect than most metal single crystals. This peak is superposed on a broad background which has accurate1y a Lorentzian squared profile. Under the experimental conditions this implies that there are fluctuations in the position of the atoms about the mean lattice position with a correlation length of about 70 A. This is approximately the distance between the misfit dislocations for these films as determined by Lee et al [10]. In addition there are sharp weaker peaks showing that there is a nearly periodic oscillation about the mean layer position in the Nb films. The period is the distance between the steps necessary to account for the misorientation of the sapphire surface from the crystallographic planes.

These experiments show the power of high resolution X-ray scattering to elucidate the detailed structure of thin films. Other articles in this text will illustrate its use in considerably more complicated situations.

83

6 3 .------,---.....,

5

4 2

3 Nb 110

2 1

-0.002 0.000 0.002 ,............ 1 ~

h

E-t 0 ~

U)

Z -1 ~ 6 E-t Z 5 ~

'--" 4 0.0 0 3

.....:I 2

1

0

-1

-2 -0.01 0.00 0.01

h [units cf 27T/a]

Figure 10. The transverse scattering through the Nb (110) Bragg reflection, see also ref [11], and the sapphire (1120) reflection. Note the wings in the former case.

ACKNOWLEDGEMENTS

I am grateful to my collaborators in these experiments; S.R. Andrews, A. Gibaud, D.F.

84

McMorrow, C.A. Lucas and T.W. Ryan. Financial support was provided by the Science & Engineering Research Council.

REFERENCES

[1] A recent review is given by Robinson, I.K. and Tweet, D.J. (1992) Rep. Prog. in Physics 55, 599.

[2] Cowley, R.A. and Ryan, T.W. (1987) J. Phys. D. 20, 61.

[3] van Silfbout, R.G., van der Veen, J.F., Norris, C. and Macdonald, J.E. (1990) Faraday Discuss. 89, 169.

[4] Vlieg, E., Denier van der Gon, A.W., van der Veen, J.F., Macdonald, J.E. and Norris, C. (1988) Phys. Rev. Lett. 61,2241.

[5] Zachariasen, W.H. (1944) "Theory of X-Ray Diffraction in Crystals", Wiley, New York.

[6] Jackson, J.D. (1975) "Classical Electrodynamics", Wiley, New York.

[7] Berry, M.V. (1973) Phil. Trans. Roy. Soc. A276, 611. Andrews, S.R. and Cowley, R.A. (1985) J. Phys. C: Solid State Phys. 18,6427.

[8] Chui, S.T. and Weeks, J.D. (1976) Phys. Rev. B. 14, 4978.

[9] Gibaud, A., McMorrow, D.F. and Cowley, R.A. (1992) (to be published)

[10] Lee, C.H., Liang, K.S., Shieu, F.S., Sass, S.L. and Flynn, C.P. (1991) Mat. Res. BuH. 209, 679.

[11] Reimer, P.H., Zabel, H., Flynn, C.P. and Dua, J.A. (1992) (to be published)

X-RAY DIFFUSE SCATTERING AS A PROBE FOR THIN FILM AND INTERFACE STRUCTURE

SUNIL K. SINHA Corporate Research Exxon Research & Engineering Company Clinton Township, Route 22 East Annandale, New Jersey 08801 USA

ABSTRACT. The structure of thin films and interfaces can be probed by X-ray specular and off-specular (diffuse) scattering. As is weIl known, the former yields the average density profile across the film or interface. Diffuse scattering as treated here is the analogue for the interface of small angle scattering from bulk materials, but with the ability to probe much larger length-scales. We shall discuss how the diffuse scattering yields information regarding the detailed morphology of the interface roughness, the conformality of the roughness between successive interfaces, the morphology of the erosion or pit-structure at an interface, and various other kinds of defect structures that can exist in the film. We shall illustrate with results on several systems studied using synchrotron radiation at the National Synchrotron Light Source.

In actual electrochemical processes, such as corrosion or film growth, (irregular) fluctuations on fairly large length scales (lOA-104A) inevitably occur, and it is interesting to note that their statistical properties can also be probed with X-ray scatter­ing, particularly with the use of synchrotron radiation. In this paper, we shall discuss the information which can be obtained regarding the morphology of the interface structure from diffuse (i.e., off-specular) X-ray scattering from rough surfaces, thin films, and multilayers. The theory of specular reflectivity from single surfaces, films, and multilayers is rather standard by now [1-3] and will not be discussed he re in detail. We shall also not discuss truncation-rod scattering studied in Grazing Incidence Diffraction experiments, where one obtains information regarding the crystalline order, either of the surface itself (for studying surface reconstruction) or the crystalline ordering of adsorbed atomic layers. Instead, we are concerned here with larger meso­scopic length scale interface fluctuations, where the atomic struc­ture is averaged over, and the interface can be replaced by one between two uniform media of different refractive indices.

85

C. A. Melendres and A. Tadjeddine (eds.), Synchrotron Techniques in Inteifacial Electrochemistry 85-95. © 1994 Kluwer Academic Publishers.

86

As is well-known, the complex refractive index of a medium for X-rays may be written as

).2 n - 1 - ----z;- P , (1)

where ). is the X-ray wavelength and P is the scattering length den­

sity given by P - (e2/mc2) N f, N being the atomic density and f the (complex) atomic scattering factor, and the bar denotes an average. Consider an interface i between two media with scattering

length densities P; and p~, the +(-) signs referring to the upper

(lower) medium with respect to a z-direction chosen normal to the + -

average interface. Let 8P i - (Pi - Pi) and let zi(x,y) denote

the height of the interface above some (flat) reference x-y plane. We can write

zi(x,y) (2)

where the second term denotes the height fluctuations about some average interface plane.

In the Born Approximation, the scattering amplitude for the i th

interface for a wavevector transfer q is given by [4)

-iq Z. z l. e

-iq 6z i (x,y) -i(q x + q __ y) x II dxdy e z e x T,

(3)

and the scattering cross-section (per unit solid angle) for the whole system is simply given by

.. .. 2 S(q) - II b i (q) I .

i (4)

For a single interface, with Gaussian random height fluctuations, let us define

2 g(r) - <[6z(x,y) - 6z(x' ,y'») >

2 2 2 r - (x'-x) + (y'-y) .

(5)

87

For many kinds of rough surfaces, we can write

2 ger) - 2«6z) > - 2C(r) , (6)

where

h C(r) - <6z(x,y)6z(x' ,y'» - o2e -(r/e) (0< h < 1). (7)

2 2 o - «6z) > is the mean-square interface roughness, and h is

2h the so-ca11ed roughness exponent, since for r « e, ger) - r , defining a se1f-affine surface [5]. e is a cut-off 1ength for the roughness corre1ations. For such an interface,

2 2 2 1 A A 12 -q 0 q C(R)

S(q) = ~ A e z II dXdY e Z -i(q X + q Y)

e x y (8)

qz

where A is the total interface area, and (X,Y) are the relative separations (x'-x),(y'-y) in Eq.(5). Eq.(8) contains a pure1y

specu1ar component (proportional to 6(q )6(q » since x y

q2C(R) e z -+ 1

for 1arge R. Exp1icit express ions for the diffuse part scattering

2 2 IAAI2 -q 0 ~A z - 2 e

qz

q2C(R) -i(q X + q Y) x II dXdY [e z - 1] e x y

of the

(9)

have been discussed e1sewhere [4,6] and also been successfu11y used to fit scattering from a variety of rough surfaces [7,8]. Whi1e

analytical express ions for S(q) cannot be found for arbitrary q, the asymptotic (large qz ) form for Sdiff(q) as qx,qy -+ 0 has a simple

sca1ing form q-(2+2/h) [4]. Integration over one direction of q, z

(e.g., q ) with q -+ 0 (as wou1d be measured if the slits were 1eft x y wide open in one direction) wou1d change this dependence to

-(2+1/h) qz .

88

The Born Approximation resu1t given above may be improved by the use of the Distorted Wave Born Approximation (DWBA), which yie1ds the resu1t [4,6,9]

(10)

where T(a) is the Fresne1 transmission factor for the single inter­

face at grazing angle of incidence a. The factors IT(a)1 2 and 2 IT(ß)I are strong1y peaked when a or ß is equa1 to the critica1

angle of incidence resu1ting in the we11-known "Yoneda wings" observed in rocking curves about the specu1ar ref1ection from rough surfaces [10]. An examp1e of such a scan is shown in Figure 1 for a film of si1ver deposited on a silicon substrate.

Ag/Si in situ deposition 10-2 ,--------,----,-----.......,------r-------,------,

o

10-8 L-___ ---'-____ -'-___ --' ____ -'--____ L-__ .

-0.75 -0.50 -0.25 0 0.25 0.50 O.7~

[ -3 -1) qx xl0 A

Figure 1: Diffuse scattering scan in transverse (qx) direction at

q - o.lA- l for a si1ver film vapor deposited on a silicon substrate. z The sharp peak at q - 0 is the specular ref1ection, and the Yoneda x wings are c1ear1y visible on either side. (from C. Thompson, G. Pa1antzas, J. Krim, Y. P. Feng, and S. K. Sinha, to be published)

If we stay within the and (4) can easily be used from a system of multiple layers) to obtain

S(q) - L A i,j

89

Born Approximation, the result in Eqs.(3) to derive the expression for scattering interfaces (e.g., thin films or multi-

(11)

where A is the illuminated area, q~ is the mean-square roughness for

~

interface i, the sum is over all pairs of interfaces, and F .. (q) is ~J

given by

(12)

Here, cij(r) is given by

(13)

For i = j, it is given by an expression of the form of Eq.(7). For i t j, it is non-zero if there exists a degree of conformal rough­ness between the interfaces. Again, the purely specular component of the scattering can be subtracted off by subtracting 1 from

2 qzCij(R) .

e ~n Eq.(12). The result of having C .. (r) non-zero for i r j is to produce

~J

peaks and oscillations in the diffuse scattering as a function of qz which mimic the peaks in the specular reflectivity (from multi-

layers) and the oscillations in the specular reflectivity from thin films, as can be seen from Eq.(ll) (see also Ref.[B]). It is a surprising fact that most thin films and multilayers exhibit con­formal roughness as evidenced by structure in the diffuse scatter­ing as a function of qz. Figures 2(a) and 2(b) show a fit to the

specular and diffuse scattering along q (q nominally set equa1 to z x

zero, qy set close to zero by virtue of a small (0.05°) mis-set from

the specular condition for a 9-29 scan) for a GaAs/A1As multi1ayer, assuming the roughness in all the layers was complete1y conformal (i.e., perfect correlation from the substrate to the top of the multilayer) [lI]. As can be seen, a perfectly satisfactory account­ing for both the specular and the diffuse scattering can be made using Eq.(ll). In these fits, a random fluctuation (of - 1A)

90

HJ1

l(r'

j()-J

z;;. 'i ~ Hf* &

He?

Hl"*

1O-1t ll'

Figure 2(a): Specular reflectivity for a 77 bilayer GaAs/A1As multilayer prepared on a (100) single crystal GaAs substrate. The fit corresponds to a periodicity of l22.9A, a GaAs/A1As thickness ratio of 0.684, an rms interface roughness of 2.lA, and a thickness fluctuation of 1.07A. (from M. K. Sanyal et al., [11)

10"* L-'-_~--()..LA--"'---'--::ll'':,7-~-'''''''----:Ul

q! vr1}

Figure 2(b): The longitudinal diffuse scattering (q - q ~ 0) for x y the same multilayer fitted with a model of perfect conformal inter­face roughness, with a roughness cut-off length of 6400A and h = 0.4.

91

in the thickness of each layer was also included to account for the slight broadening (in the q -direction) of the peaks in both the z specular and the diffuse scattering. The GaAs/A1As multilayer experiment also illustrates another property of diffuse scattering from rough interfaces, namely anisotropy when the substrate is a single crystal with a slight miscut from a perfect facet. In this case, the miscut from a [100) facet resulted in steps running perpendicular to the direction of the miscut. The roughness is in fact due to these steps and is propagated from layer to layer in this rather perfectly grown multilayer. Figures 3(a) and 3(b) show measurements of the diffuse scattering along the q direction for

y fixed qz (i.e., transverse diffuse- scans) when the sampIe was

oriented so that the steps were respectively parallel and normal to qy. The fits correspond to the conformal roughness model discussed

above, where for each fit a single roughness exponent hand cut-off length e for the whole multilayer was employed as in Eq.(7). One notices that the diffuse scattering is highly anisotropic, reflected in different values for hand e along the two directions. The cut-off length e is much longer slong the steps than between them. (The roughness in the direction parallel to the steps corresponds to step meandering, and thus, the above result is in agreement with our expectations.)

Finally, let us consider a single interface where the roughness fluctuations are not Gaussian and the surface is not statistically self-affine. Specifically, let us consider_ a surface decorated with islands of various heights (or equivalently pits of various depths), sizes and separations, as illustrated schematically in Figure 4. The surfaces of these islands and of the base surface will also have roughness fluctuations which will lead to added complications which we will not consider here. To simplify the model, we assume that the heights of the islands are governed by a single distribution function F(z) and that the height of any island is uncorrelated with that of any other island or with its lateral size. A derivation of the diffuse (non-specular) seattering in the Born Approximation yields the result

(14)

where 6p is the difference in scattering density across the inter­face, ~ is the fractional coverage of the surface by the islands,

F(qz) is the one-dimensional Fourier trans form of F(z), f~(ql I) is

92

• Experimental -Fitted

. . -.

Figure 3(a): Transverse diffuse scattering for the GaAs/A1As multi­multilayer across the 5th order specular Bragg peak in the orienta­tion where the surface steps are parallel to the direction of q .

y The diffuse scattering is fitted with e - 6400A and h - 0.4. Note that the sharpness of the diffuse scattering makes it hard to distinguish from the specular peak.

Figure 3(b): Transverse diffuse scattering in the orientation where the surface steps are perpendicular to the direction of q. The

y diffuse scattering is The curves correspond peaks, as indicated.

now broad, and the specular peaks are evident. to scans across various order specular Bragg The fitted values are e - l200A and h - 0.64.

93

R·· 1J

z=O

Figure 4: Schematic of random pits distributed across an interface.

the two-dimensional form-factor of the projection on the (x,y) plane of the surface of island i, i.e., the 2D Fourier transform of its shape, and the < > sign indicates an ensemble average over all

islands. S2D(QI I) is the 2D Fourier transform of the pair correla­

tion function between the centers of mass of the islands in the (x,y) plane. Note that the purely specular reflectivity measures the· (x,y) averaged, density profile normal to the average interface and is given in the Born Approximation by

[1 - 2~ + ~F(q ) + ~F*(q )]. z z (15)

If the islands are all of the same height, F(z) = 6(z - ß), then -iqzß

f(qz) = e and both the specular reflectivity and the diffuse

scattering will show modulations as a function of qz' since the

islands essentially constitute a topmost layer of thickness ß, but of scattering density (~p) instead of p. According to Eqs.(14) and (15), the modulations in the diffuse scattering will be exactly out of phase with those in the specular, unlike the case of conformally rough surfaces where these modulations are in-phase. Figure 5 shows the specular and diffuse scattering (measured along q for q ,q ~ z x y 0) for a polymer film decorated with such islands of constant thick­ness. (The film corresponds to a polystyrene-PMMA block copolymer which has been annealed to produce a micro-phase-separated lamellar structure parallel to the silicon substrate. The last bilayer of such a structure cannot fill the whole surface area and is conse­quently left over as the islands on top of the film. The electron density contrast between the polystyrene and PMMA blocks is neglig­ible for the purposes of discussing the X-ray scattering.) Both

94

1E7

1E6

1E5

1E4 DIFFUSE

0.1 a 0.2 Z

0.3

Figure 5: Specular and longitudinal diffuse scattering as a func­tion of qz for a polystyrene/PMMA film decorated with islands on the

surface, as discussed in the text. The fitted curves are not shown for clarity since they are almost indistinguishable from the experi­mental curves. (from S. K. Satija, S. K. Sinha, T. P. RusselI, E. B. Sirota, and G. J. Hughes, unpublished)

these curves can be fitted extremely weIl with a model equivalent to Eqs.(14) and (15), but also incorporating conformal surface rough­ness from the top and bottom surfaces of the film. The rapid oscillations (Kiessig fringes) seen in the specular which correspond to the interference between the reflected beams from the top and bottom interfaces are mirrored in the diffuse scattering indicating a strong degree of conformal roughness. The slower modulations in both are due to the islands on top of the film. Note that these slower oscillations are out of phase with each other in the diffuse and specular components, but the rapid oscillations are in phase as expected.

Analyses of this kind can be used in experiments on diffuse scattering from surfaces which have been pittedor eroded electro­chemically or on which island deposition has occurred [12). One can in this way obtain global statistical information on the distribu­tion of such objects on the surface in a way which may be complemen­tary and perhaps more convenient to that obtained from various imaging microscopies. Such experiments are still relatively untried, but have the potential to yield very interesting results in the near future.

I wish to acknowledge the help and collaboration of many people on various aspects of the work described in this paper, including Y. P. Feng, H. Homma, K. G. Huang, G. J. Hughes, J. Krim,

95

C. Majkrzak, C. A. Me1endres, G. Pa1antzas, T. P. Russe11, M. K. Sanya1, S. K. Satija, E. B. Sirota, K. Siaradzki, and C. Thompson. The National Synchrotron Light Source at Brookhaven National Labora­tory is funded under Contract No. DE-AC02-76CH00016 by the Division of Materials Research, U.S. Department of Energy.

REFERENCES

[1] Parratt, L. G. (1954) Phys. Rev. 95, 359. [2] Nevot, L. and Croce P. (1980) Rev. Phys. App1. 15, 761. [3] Pershan, P. S. (1990) Faraday Discuss Chem. Soc. 89, 231;

A1s-Nie1sen, J. and Kjaer, K. (1989) in T. Riste and D. Sherrington (eds.), Phase Transitions in Soft Condensed Matter, Plenum Press, p. 145.

[4] Sinha, S. K., Sirota, E. B., Garoff, S., and Stan1ey, H. B. (1988) Phys. Rev. 38, 2297.

[5] Mande1brodt, B. B. (1982) The Fracta1 Geometry of Nature, Freeman, New York; Voss, R. F. (1985) in R. Pynn and A. Skje1torp (eds.), Sca1ing Phenomena in Disordered Systems (NATO ASI Series B133) , Plenum, New York, p. 1.

[6] Pynn, R. (1992) Phys. Rev. B 45, 602. [7] He, Y.-L., Yang, H.-N., Lu, T.-M., and Wang, G.-C. (1992) Phys.

Rev. Lett. 69, 3770. [8] Savage, D. E., Kleiner, J., Schimke, N., Phang Y-H., Jankowski,

T., Jacobs, J., Kariotos, R., and Laga11y, M. G. (1991) J. App1. Phys. 69, 1411.

[9] Vineyard, G. H. (1982) Phys. Rev. B 50, 4146; Dietrich, S., and Wagner, H. (1984) Z. Phys. B56, 207.

[10] Yoneda, Y. (1963) Phys. Rev. 131, 2010. [11] Sanya1, M. K., Sinha, S. K., Gibaud, A., Satija, S. K.,

Majkrzak, C. F., and Homma, H. (1992) in H. Zabe1 and I. K. Robinson (eds.), Surface X-Ray and Neutron Scattering, Springer­Verlag, Ber1in, Heide1berg, p. 91; Sanya1, M. K., et a1. (1992) in K. Liang, M. P. Anderson, R. F. Bruinsma, and G. Sco1es, (eds.), Mat. Res. Symp. Proc. 237, p. 393.

[12] Huang, K. G., Wong, R., Sieradzki, K., and Sinha, S. K., to be pub1ished; Me1endres, C. A., Feng, Y. P., and Sinha, S. K., to be published.

SURFACE MORPHOLOGY CHARACTERIZATION WITH X·RAY SCA'ITERING TECHNIQUES

c. THOMPSON Department 0/ Physics 6 Metrotech Center Polytechnic University Brooklyn, New York, 11201 USA

ABSTRACT. This paper describes applications of the x-ray reflectivity technique to characterize the evolution ofthe surface morphology of thin metal fIlms as a function of thickness during growth. The experiments show the sensitivity of the technique to the texture of the surface and to changes in the in-plane arrangement of surface structures which constitute the fIlm roughness. In the specular geometry, the scattering vector probes correlations in the structure perpendicular to the substrate interface, and thereby the density profIle of the surface may be obtained. The extensions to the conventional x-ray reflectivity technique include probing in diffuse and off-specular geometries, for which the scattering vector probes in-plane correlations of the surface structure.

1. Introduction

The availability of synchrotron radiation, with its characteristics of extremely high intensities, its particular collimation or polarization properties, its multi-wavelength accessibility and novel energy ranges, and other differences from conventional photon sources, continues to drive technical and theoretical advances in scattering and spectroscopy techniques. An exciting area developing is the exploitation of these advances in synchrotron radiation surface-specific probe techniques to study solution/solid interfaces in-situ.

Modem synchrotron x-ray scattering techniques are powerful methods to probe interface and surface structure down to atomic length scales. The surface-specific structural information available through such techniques as x-ray reflectivity, grazing incidence scattering, and standing wave fluorescence, includes the morphology of the substrate surface [1,2], the presence and epitaxial relationships of adsorbed layers [3], density profiles of surfaces [4] and other structural characteristics of film growth.

This paper describes the x-ray reflectivity technique, and extensions to this technique, which are particularly weIl suited to examine and characterize surface morphology on the atomic and microscopic level. X-ray reflectivity is a scattering technique using monochromatic x-rays. The tight collimation of a synchrotron bearn, and its high intensity, are useful and necessary for these experiments. The

97

C. A. Melendres und A. Tadjeddine (eds.), Synchrotron Techniques in Inter/acial Electrochemistry 97-107. © 1994 Kluwer Academic Publishers.

98

experiments and theory described are at a vacuum/metal interface during deposition, however, the concepts and the technique may extend to other deposition processes. The emphasis of the paper is to introduce the experimental technique, and its characteristics, to an audience who may not be familiar with the advances in this area.

The extension of these techniques to studies of the electrochemical interface will offer an excellent opportunity to understand electrochemical deposition and the film characteristics at a fundamental structurallevel. For example, in-situ measurements of the development of microgrowths between macrosteps and the evolution of microstep (less than 3nm) densities and rearrangements as functions of growth or dissolution parameters could be realized with the technique of x-ray reflectivity as a probe of surface morphology. As a sensitive technique to measure density profiles, the presence or absence of very thin layers which change under electrochemical processing conditions, and their thicknesses could be measured in-situ.

2. Suface Morphology : the SeIf-Amne Description

How are rough surfaces described? Processes at a surface may change or modify that surface. This could be a process of film deposition, or dissolution and corrosion. The description of the morphology of the surface, that is the arrangement of the mass on the surface, may depend on the length scales which we believe relevant and important for subsequent processing or end-product usage. However, a good description of a surface will help us understand the morphology over many length scales and how it evolved with time. For example, a surface may be mirror-like, that is smooth on optical length scales, yet exhibit very different properties under conditions which probe, or are influenced by, sub-opticallength scales.

There exists much theoretical work attempting to find underlying 'universal' descriptions of the surface structures which develop under non-equilibrium conditions [5,6]. It is hoped that many varieties of non-equilibrium growth processes may ultimately be described by similar surface morphology functions of time, thickness, and lateral aild transverse length scales, although the physical mechanisms and driving forces which rearrange or add atoms are different. If this is the case for a varied set of deposition processes, members of that set are said to be in the same 'universality' dass. Then the theoretical work may concentrate on the equations or simulations which simulate the dass behavior.

In discussions of roughness at a surface, we are interested in a description of the variation of the local height h(x,y), at different lateral positions on the surface. One may define an average surface height, < h >, for a single-valued gaussian rough surface h(x,y) characterized by a root-mean-square (rms) interfacial width (1 = <[h(x,y) - <h> ]2> Yl. This width (1 may be a function of the laterallength scale L over which the average is taken.

A variety of roughening processes produce a surface or interface which can be

99

represented by self-affme fractal scaling. A self-affine fractal surface is distinguished from a true fractal because the self-similarity relations are different between the lateral and the transverse dimensions. Globally, the surface is two-dimensional, however on a local scale, the surface can be characterized by a fractal dimension D=3-H.

For a self-affine fractal, the interfacial width, a(L) increases with the sampled lateral length L as In(a(L» oe Hln(L). The roughness exponent H take a value betweeil zero and one and characterizes the texture. Figure 1 shows examples of self­affme profiles with different H. The a saturates at large L, which implies a characteristic lateral or horizontal correlation length. As a surface growing rougher with time, the saturated interfacial width follows a In(a(time» oe ßln(time) law. For a growing film, it is assumed that the thickness of the film is proportional to the time, thus, In(a(h» oe ßln(h) where h is the total thickness of the film.

To be in the same universality class, a set of roughening processes will be described by the sanie numerical value for the scaling exponents H and ß. However, because of the ubiquity of surfaces which can be characterized as self-affine, these exponents are useful as characterization paranleters for a surface and its evolution. In the following seetions, the basic theory of the x-ray measurements are presented, and the manner in which these measurements can provide experimental values for the scaling exponents is also presented. The examples provided here show that the self-affine geometry description is useful.

o 200 400 x

Figure 1. Self-affine profiles for different values of H. The seIf-affine profiles all have the rms width of a = 1.1 [2].

100

3. X.Ray Reflectivity Techniques

The discussion of the reflectivity teclmique is separated into two complementary geometries which both use monochromatic x-rays. They may be done on the same high-resolution diffractometer. In the specular geometry, the scattering vector probes correlations in the structure perpendicular to the substrate surface. In the diffuse and off-specular geometries, information from in-plane correlations is also examined.

For the specular reflection geometry, the detector measures the intensity profile of the beam scattered from the surface at equal angles to the incident beam angle, i.e. the specular condition. The ratio of the specularly scattered intensity to incident intensity, R, is measured as a function of angle, i.e., scattering vector q = 4".sin(29detector/2) / l.

In the second geometry, the off-specular or diffuse scattering geometry, the detector measures the intensity profile offset fom the specular condition, generally by keeping the detector fixed and rocking the sampie, or by performing a scan similar to the specular reflectivity scan but with a slight constant offset in the sampie alignment. The total scattering vector q = 4". / lsin« Qi + Qr) /2) = 4". / lsin(29detector/2) where Qi and Qr are the angles with respect to the surface for the incident and exit beam. In this case, qz is denoted as the component of the total scattering vector which is perpendicular to the surface. With this definition, in the specular condition, ~ = q, however in the non-specular condition, this equality will not hold.

X-ray reflectivity is a low angle reflection technique. Because the x-ray index of refraction is less than 1 in materials, at small glancing angles below a critical angle, typically 0.2 to 0.5 degrees depending on the density of the material, the interface totally reflects the x-ray beam . At angles above the critical angle, the intensity falls off and the shape of the scattering curve is governed by the density profile of the sampie perpendicular to the surface. The experimental difficulties in general arise from the high resolution requirements of the experiments, which necessitates tight beam collimation and precise and reproducible sampie cell and detector manipulation. In addition, the substrate should be microscopically smooth.

Depending on the structural details in the density profile, the reflectivity technique can deliver information on structures from length scales of several tenths to several hundreds of nanometers. Hard x-rays, e.g. molybdenum Ka of .07nm (17keV), have an absorption length of 9mm in water, that is the intensity of a beam of 17keV photons is attenuated by 1/e in approximately 9mm of water. Solutions with higher electron density would have absorptions greater than this. This combination of properties makes x-ray reflectivity a likely candidate to examine and elucidate the in-situ microstructure of the electrochemical interface.

3.1 SPECULAR REFLECfIVITY.

The reflectivity R(q), where the scattering vector

101

q=47l/1sin«ai+ar)/2)=47l/1sin(29detector/2), is related to the index of refraction profile perpendicular to the interface. The angles ai and a r are the glancing angles the incident and reflected beam make with respect to the surface of the sampie. It may be seen that the index of refraction for x-rays is proportional to the total electron density, and therefore that changes in the reflectivity may be correlated to changes in the composition and electron density profile of the material. This density profile arises from density gradients at the interface, e.g., due to interdiffusion, or an effective density gradient at the interface due to steps, islands, kinks, and other sources of roughness, which when averaged over their distribution in-plane, give an effective interfacial density profiles.

The relevant length scales over which the reflectivity probes the interfacial density profile are related to the wavelength of the x-rays, (1 I::! O.lnm ) and the scattering geometry. Details in the profile with length scales of several tenths of nanometers to tens of nanometers can affect the reflectivity curve.

Changes in film thickness, in vertical extent of film roughness, and in the distribution of the structural elements constituting the film roughness which change the functional form of the average density profile distribution, are seen with the specular reflectivity technique. Changes in film thickness sensitively affect oscillation frequencies in the curve, changes in the vertical extent of interfaceial roughness affect the intensity at any point, and changes in the function which descibes the averaged density profile affect the entire shape of the reflectivity curve and the amplitudes of any oscillations. Note that the deviation of index of refraction from one for x-rays is small (6 -10-6). Thus the optical path length of a layer, nd, varies Httle from the layer thickness d, and therefore the oscillation spacing at the high q ranges of the reflectivity profile is very representative of the film thickness.

The index of refraction for x-rays in matter is slightly less than one, and is denoted by

n = 1 - 6 - iß (1)

The quantities 6 and ß are of the order of 10-6. The quantity ß is related to the linear absorption parameter p. by

ß = 1p./41f (2)

When the wavelength is far from the absorption edge, the quantity 6 may be given by

(3)

where p is the scattering electron density. Because the index of refraction is less than

102

one, there exists a critical angle, a.c~j(20), typically less than 0.5 degrees for vacuum solid interfaces, for which the x-ray beam undergoes total external reflection from the surface and the reflected beam intensity is equal to the incident beam intensity. The position of the critical angle is extremely sensitive to the density of the film. For a perfectly sharp interfacial density profile, (a step function in the electron densitites) between two regions, denoted by the subscripts 0 and 1, the specular reflectivity may be given by the Fresnel reflectivity equation.

(4)

The specular reflectivity depends on the effective density profile perpendicular to the surface and thus fits to the specular reflectivity will give us the interfacial width of the top surface, (J.

103

102

101

100

10-1

10-2 0

S. 10-3

10-4

10-5

10-6

10-7

10-8

10-9 0

~ ~.>, \': -3: 30.0

~\Y\ b 20 .•

\~~~ ,.,~) ,-\L"~~::::::::::··

__ .. -......... ~_ ........ . .... :: .. (a) -...... ....-... ..... . (b)

..... ~-~~-........... . ......... :::: ... (c)

'" " (d)

... (e)

0.5

Figure 2. Specular reflectivity measurements at room temperature for progressively thicker silver films grown at room temperature on a silicon substrate. (a) 9.8nm (b) 18nm (c) 36.7nm (d) 72.8nm (e) 150.2nm. The inset depicts a log-log plot of the interfacial width (J versus film thickness, h. The slope of this plot gives the exponent ß = 0.26±.05 [7,8].

103

3.1.1 Specular Reflectivity Measurement. The x-ray measurements [7] for Figure 2 were taken at the National Synchrotron Light Source on the x-ray scattering beam­line X-22C at a wavelength of 0.15377nm. Silver films were thermally evaporated onto a 10mm wide silicon substrate at room temperature. A high vacuum sampie chamber designed for this purpose, with beryllium windows for the beam input and output, an ion pump, and in-situ thermal evaporation unit, was mounted on the diffractometer. The deposition was monitored with a quartz microbalance, however, for the calculated film thicknesses, the results of fits to the reflectivity curves were used. At five thicknesses, ranging from 10nm to 150nm, the deposition was stopped in order to take the reflectivity measurements. The receiving slit geometry defined a detector scattering resolution of 0.012·. Due to a focussing mirror in the optics upstream of the two-crystal germanium monochromator, the incoming beam had a divergence of approximately 0.01 0 • The resolution transverse to the scattering plane was left open as 10

• The profiles have been corrected for the beam spillover at low angles, and the backgound scattering has been subtracted.

3.1.2 Specular Reflectivity: ß. The data were fitted usingan analysis based on multi-homogeneous stratified layers [4] where layer thicknesses, interfacial widths, and density are fit parameters. It is the interfacial width of the top interface, that between the evolving surface and the vacuum, which is connected with the o(t) discussed in the section on characterizing a rough surface. Because the bare silicon also contributed to an initial roughness of the surface, the top layer interfacial width plotted is related to the fit by 0(t)2 = Ofit(t)2 - os? An inset in Figure 2 plots the thickness of the layer versus the ° of the top surface. Examining this inset shows that for the silver evaporated onto a room temperature substrate, we have a ß of 0.26 ± 0.05 [7].

3.2 OFF-SPECULAR AND DIFFUSE X-RA Y REFLECTIVITY

In the off-specular and diffuse scattering geometry, the scattering vector also probes in-plane height-height correlations of the surface structure. Changes in the in-plane arrangement of surface structures give rise to changes in these diffuse scattering profiles. This could include re arrangements of surface microsteps or changes in the correlation of film surface roughness to the substrate. For the off-specular and diffuse reflectivity measurements, changes in the distribution of the structural elements constituting the film roughness affect the profiles, even if the functional form of the average density profile distribution in z remains the same.

The off-specular and diffuse scattering experiments generally require a. synchrotron source, due to the high resolution requirements of the profile determinations and the low count rates in the scattering volumes. Therefore, this class of experiments has a more recent history than the specular reflectivity technique. The development of basic theories, valid under the assumptions

104

considered of interest for film growth, is of current concern in the literature [1]. The diffuse intensity (assuming a single interface) may be given by

(5)

where S(qt)diff [1] is the structure factor with the specular component subtracted out and is related to the height-height correlation function C(X,Y) = <h(x,y),h(O,O».

Since the height-height correlation function characterizes the surface topology, the equations indicate that the off-specular and diffuse scattering experiments probe the nature of the roughness.

During deposition, a change in the specular reflectivity denotes a change in the average roughness of the surface or interface. The off-specular (rocking curve) profile changes signal morphological changes in the distribution of the roughness. This would include, for example, a transition between layer by layer growth to clustering and island growth, or a change in the surface distribution of microsteps and microgrowths. The x-ray scattering techniques are sensitive to changes in the growth mechanisms in the early stages, that is, changes which occur even over the deposition of only several monolayers of material.

3.2.1 Ojf-Specular Reflectivity Measurements. The experimental set-up is identical to that described in section 3.1.1. However, instead of keeping the incident angle and reflected angle equal, the sampie is continually off-set from the specular condition by 0.05 0 as the incident and exit angles increase during the scan. This type of scan is also used for the background subtraction from the specular reflectivity, since the purpose of the background subtraction is to ascertain the amount of scattering which is contributing to the specular component of the reflectivity. The background will include the diffuse component, and other sources of background. Due to the long path lengths and geometry, other sources of parasitic background are minimized. Note that in a sampie cell containing solution, small angle scattering from the solution may become another source of background, which must be eliminated from the diffuse scattering curves in order to elucidate substrate information in the manner described in section 3.2.2.

3.2.2 Ojf-Specular Reflectivity : H. By performing the integrals for the diffuse structure factor at conditions very close to specular, it is found [1] that the asymptotic limit (high W. diffuse cross-section for a self-affine surface has the form da / dn = qz-( + 2/H). Since the transverse scattering resolution is wide open, and converting the cross section to 1/10 units, and noting that the background has been corrected for beam spillover, the experimentally observed power law [7] which connects the background to the exponent H is

(6)

105

In Figure 3, the corresponding background curves are shown for the room temperature silver on silicon deposition. Straight lines are observed for the third, fourth, and fifth coverages. The value H = 0.63 ± .05 is obtained from fits for the high q regions. To compare, the lines with the slope -(3+ 1/(0.63» are shown for all the backgrounds.

100 [ 0 0 0 0

0

10-1 0 0 0 0 0

0 0 0 0

0 0 0

10-2 0 0 0 0

0 0

10-3 0 0 0

10-4 0 0

S 10-5

(a) 10-6 (b)

10-7 (e)

10-8 (d)

10-9 e)

10-1

Qz (A-1)

Figure 3. The off-specular measurements for the room temperature silver films. Linear fits to the log-log plots correspond to an exponent H = 0.63 for the single interface model. Solid lines with slopes corresponding to H = 0.63 are superimposed on the data sets for comparison [7,8].

4. Discussion

For the room temperature silver on silicon deposits, we found that the self-affine description of the roughness was consistent with the scattering from the surfaces. The scaling exponents for this experiment were H = 0.63 ± .05 and ß = 0.26 ± .05 . There is controversy in the theoretical literature as to the expected value of these exponents for this deposition. As a characterization tool, we can compare these exponents to

106

those measured experimentallyon another system. In this case, the silver was evaporated onto a silicon substrate which was held at 80 0 K. Tbese experiments were repeated in the same chamber and beam-line, although with a germanium exit analyzer providing the detector scattering resolution element. Tbe room temperature data was identical to the previous set. After measuring the reflectivity and diffuse reflectivity for 7 thicknesses [8] , the final thickness of approximately 1000Angstroms was allowed to warm to room temperature. In Figure 4a, note the striking difference in the reflectivity as fragile surface details collapse to create a more uniform film with a smoother surface interfacial width. We have analyzed the diffuse scattering in Figure 4b. Tbe H has changed between the cold and warmed sampIe as seen by the differing slopes. Tbe slope for the cold sampie is 6.1, giving an H of 0.32. After warming, the slope is 5.9, giving and H of 0.34. Tbe increase of H, referring back to Figure 1, implies a 'smoother' surface, again consistent with a collapse of the upper surface of the film and increased diffusion.

These experiments can be straightforwardly extended to cases of solution/ solid interfaces, and to the electrochemical interface. It would be exciting to use these techniques to study the evolution of a surface in electrochemical processing, for which we have such precise control of the driving forces for the surface re arrangement.

o S (4a) (4b)

10-12<-------------' 1 0-t 2 <----'----!.->-................ --"-....>.-............... ,...

o 0.5 10-2 10-1

Figure 4a. The specular reflectivity measurements at low temperature for a 100nm silver film deposited on a silicon substrate at 80 0 K and after the sampie has been warmed to room temperature. Tbe curves have been shifted for c1arity.

Figure 4b. Tbe off-specular reflectivity measurements for the same film under the same conditions. Tbe changing slope denotes the changing of H in the surface topography.

107

s. Aeknowledgments

These experiments were performed on x-ray scattering beam-line X22-C at the National Synchrotron light Source at the Brookhaven National Laboratory, USA. The authorgratefully acknowledges the continuing collaboration on the deposition experiments with J. Krim and S. Sinha, and the assistance of Y.P. Feng, G. Palasantzas, R Chiarello, and V. Panella.

6. References

[1] Sinha, S.K, Sirota, E.B., Garoff, Y. and Stanley, H.B. (1988) 'X-ray and Neutron Scattering from Rough Surfaces', Phys. Rev. B 38 2297.

[2] Chiarello, R, Panella, V., Krim, J. and Thompson, C. (1991) 'X-ray Reflectivity and Adsorption Isotherm Study of Fractal Scaling in Vapor-Deposited Films', Phys.Rev.Lett. 3408.

[3] Samant, M.G., Toney, M.F., Borges, G.L., Blum, L. and Melroy, O.R (1988) 'Grazing Incidence X-ray Diffraction of Lead Monolayers at a Silver (111) and Gold (111) Electrode/Electrolyte Interface', 1. Phys. Chern. 92220.

[4] Toney, M.F. and Thompson, C. (1990) 'X-ray Reflectivity on Perfluoropolyether Polymer Molecules on Amorphous Carbon',1. Chern. Phys. 923781.

[5] Indeku, J.O. (1993) 'Dynamics of Fractal Surfaces', to be published in Order­Chaos-Fraetals, Leuven University Press.

[6] For a review of current theoretical status (1991) F. Family and T. Vicsek (eds.), 'Dynamics of Fractal Surfaces', World Scientific, Singapore.

[7] Thompson, c., Palasantzas, G., Feng, Y.P., Sinha, S.K. and Krim, J. (1993) 'X-ray Reflectivity Study of the Growth Kinetics of Vapor-Deposited Silver Films', in preparation.

[8] Thompson, C. (1993) 'Characterization of Surface Morphology with X-ray Reflectivity', Bulletin 0/ the APS, 38 772.

STUDIES OF ELECTRODES BY IN SITU X-RAY SCATTERING

Michael F. Toney

IBM Research Division Almaden Research Center 650 Harry Road San .Tose, CA 95120

ABSTRACT: The atomic structure at solid-c1ectrolyte interphases can be determined directly with X-ray scattering, and this is hccoming a powerful method of characterizing such 'buried' interfaces. In this paper, I give an overview of X-ray scattering with an emphasis on problems of interest in intcrfacial e!ectrochemistry and corrosion. After describing the fundamentals of X-ray scattering, I discuss the application of this tech­nique to several diverse systems. These representative examples are taken from work in our laboratory amI inc1ude structure determination of the following systems: underpo­tentially deposited monolayers of Bi on Ag(lll); the double layer near a Ag(lll) electrode; and anodic Ti02 films on the (1120), (1010), and (0001) faces of titanium. j'inally, I describe future prospects for X-ray scattering studies in interfacial electrochemical science and technology.

1. Introduction The arrangement of atoms at solid-c\ectrolyte interphases is of fundamental im­

portance, since this atomic structure strongly arrects the chemical and physical proper­ties of the interphase. Despite this importance, our knowledge of atomic structure at electrochemical interphases is rather pOOl', due to the difficulty of in situ structure de­termination. However, in situ studies of such 'buried' interfaces are achievable with X-ray scattering, and substantial progress has recently been made in determining the atomic structure at solid-electrolyte interfaces and in thin anodic oxides (1-13).

Ilard X-rays (with energies of ~I() keV) are weil suited for in situ measurements of interphasial structure. X-rays have a large penetration depth in aqueous solution, and since X-rays interact wcakly with matter, the intcnsity data can be interpreted kinematically, which grcatly simplifics data analysis compared to eleetron diITraction. l'urthermore, X-ray wavelengths are comparable to atomic dimensions. Thus, X-ray seattering provides information on atomic-scale structure. Although the method is most useful when the atomic order extends over a long range (i.e., ;(:30A), X-ray scattering can provide important information e\'en if long-range order is absent, as we shall see bclow for measurements of the double layer structure.

The drawback ofthe weak interaction between X rays and matter is that the signals have low intensity. This is particlllarly true for surfaces and interphasial regions, since these contain only a sm all number of atoms. IIowever, this difficulty can be overcome with the use of the intense radiation from synchrotron sources (14), which have opened a new avenue of research into strllcture measurements at surfaces and interfaces.

109

C. A. Melendres and A. Tadjeddine (eds.), Synchrotron Tec/miques in Interfacial Electrochemistry 109-125. © 1994 Kluwer Academic Publishers.

110

The purpose of this paper is to provide abrief introduction to X-ray scattering and to describe hüw this technique can be used (in situ) to understand atomic structure in interphasial clectrochemistry. I begin with a brief review of the fundamental aspects of X-ray scattering and emphasize the information that can be obtained with this tech­nique. This is followed by representativc examples where X-ray scattering has been used for structure determination at solid-electrolyte interfaces and thin anodic oxides. I end with a short eliscussion of the future prospeets of X-ray scattering in eleetroehemieal science and technology.

2. Aspects of X-ray Scattering X-ray scattering has long been recognized as the most powerful technique for

structure determination of three-dimensional (30) matter, and in the past deeade, much progress has been made in the applieation of X-ray scattering to the study of surfaees, very thin films, and two-dimensional (2\)) adsorbed layers (7, 8, 15, 16). In this Sec­tion, I will discuss the basic principles of X-ray scattering in büth two and three di­mensions anel briefiy describe some or the experimental aspects of X-ray scattering with particuJar emphasis on electrochcmical systems.

2.1. Basic Principles

Figure 1 shows a typical X-ray scattcring experiment. The incident and diffracted wave vectors are k and k', respectivcly, and beeause of energy conservation, they have the same magnitude (k = k' = 2n/A, where A is the X-ray wavclength). The seattering angle 20 is the angle between k ami k', and the scattering vector is

Q=k' -k; (I)

it has a magnitude Q = (4n/A) sin O. In a X-ray scattering experiment, one measures the diITraction pattern; this is a map of the scattered intensity I(Q) as a function of Q, or equivalently, Q anel the sampIe orientation, which is given by the angles X and q, (see Fig. 1). For most thin layer systems, the kinematic approximation is valid, and

I(Q) = le IIp(r)eiQ.rd\r, (2)

where per) is the electron density of the material and le is the scattering intensity of a single clectron. This simple Fourier-transform relation between intensity and clectfOn density is a great advantage or X-my scattering compared to electron eliITraction. where a complicated dynamical treatment is ncedcd to calculate intensities. Note that since materials with low atOlnic number elements tend to have small electron elensities, surrace X-ray diffraction will be rather insellsitive to these.

For a 30 crystal1ine material, thc dirrraction pattern consists orßragg points at the positions of the reciprocal lattice vectors (17-19). Oiffraction peaks are observed when Q is equal to one of the reciprocal lattke vectors and the peaks are identified by their indices (hkf). For a 20 solid, however, the scattering consists of ßragg rods (8, 15, 16); these are continuous streaks or rods of diffracted intensity that run along a direction perpendicular to the 2D solid, which we take as the z direction. Oiffraetion is

111

Q

Figure I. Schematic representation 0(" a X-ray scattcring experiment. The incident alld dirrracted wave vectors are k anti k', respectively, and the scattering vector is Q~k'-k. The sarnple orientation is givell by the ;I/.imuthal angle 1) and the polar angle x. The ineidence angle is IX ancl the exit angle is {I. Taken from Ref. (13) with permission.

Figurc 2. (a) Sehematie rcciprocal space illustration oftruncation rods (for a bec (001) surfaee). The Bragg peaks are indicated hy thc solid points and the origin is shown by the plus. Thc cross-hatching shows regions where the truncation rod intensity is large. (h) The intensity orthe (1IQz) rod rClr a [Jat surfacc (solid line) and for a surrace with a commellsurate adsorbed laycr (dashcd linc). Takcn from Rcf. (7) with pcrmission.

112

observed when the in-plane eompollent of Q is equal to a 2D reciproeallattice veetor, whieh is labeled by its two indices (hk).

Measurements ofthe integrated scattering intensity permit the determination ofthe positions of atoms within a unit eell (17, 18). The integrated X-ray intensity from a 2D solid is (15, 16)

(3)

Ilere Qz is the component of Q perpendicular to the 20 solid, and Fhk(Qz) is the strue­tu re factor for the (hk) Uragg rod amI is the rourier transform of the positions of the atoms in one unit cell. Each atom is weighted by its form faetor, which is equal to its ~tomic number Z for small 28, but deercases as 28 inereases. The Debye-Waller faetor (e- 2M) accounts for the reduetion in intensity due to disorder in the crystal and Co contains geometrie factors. An expression similar to Eq. 3 holds for 3D materials, but lhk(Qz) is replaced by lhkl (17, 18).

By determining the positions of several diffraction peaks, one ean construct, in re­ciprocal space, the difTraetion pattern o[ the material under investigation. From this, one can determine the lattice and unit eell dimensions, and ifthere is only one atom per unit eell, the atomie strueture is essentially solved. rar materials with more than one atom per unit cell, lhk(Qz) (or J"kl for .'\1) solids) is measured for different diIfraetion peaks, and the atomic positions within the unit cell are determined from a erystallo­graphie analysis of the structure faclors (15, 16, 18). For thin films or aclsorbed laycrs on substrates, a eomparison of thc dini"action patterns of thc film and substrate shows the epitaxial ar orientational rc1ationship hetween the two. Altcrnatively, if the atomic structure of the material under investigation is believed known, then by eomparing the measured difTraetion pattern (the peak positions and intensities) with patterns from standards, the crystalline phase of the material ami its atomic structure can be identified. These aspects of structure determination are iIIustrated in the following Section.

The presence of an interface in a crystal creates X-ray seattering that has some si­milarity to that from a 2D solid. i\\though the scattcring is intcnsc ncar the Uragg points, it has long tails that extcnd pcrpcndicular to the intcrface (7, 8, 15, 16). This rod-like scattcring is referred to as a truneation rod and is schcmatieally illustrated in Fig. 2. In mcasurements of truncation rods, the component of the scattering veetor parallel to thc interfacc (Qn) is held constant and the ~cattered intcnsity is measured at different Qz aJong the rod (8, 15, 16). horn this intcnsity, one can determine Il~'k(Qz)12 for the truneation rod. Bcf'orc discussing what one can leam from truneation rod intensities, wc distinguish betwcen 'specular' and 'non-speeuJar' truneation rods; in the lattcr ease, Qx is non-zero, while in the former, Qx= 0 (i.e., (hk) are both zero).

When a eommensurate layer is adsorhcd on a surfacc, the scattcring from this layer will interfere with the non-speeular trullcation rod scattering from the substrate, and the resulting intensity will he different [rom that of the bare substrate. This is illustrated by the dashed line in Fig. 2b. Measllrements of this intensity permit adetermination of the adsorhed-Iayer registry (positions of tl1e adatoms relative to the substrate) and any ehanges in thc positions of thc suhstrate atoms relative to their bulk positions (15, 16, 20).

113

If we now consider an incommellsurate layer on a surface, this layer will only slightly affect the non-specular truncation rod intensities (see Ref. (3) for details), al­though it will strongly alTect the intensity ofthe specular truncation rod. Measurements of the latter can be used to deduce, for example, the spacing of the layer above the substrate, and in the case of solid-clectrolyte interfaces, aspects of the atomie structure of the electrolyte above the electrode. This will be illustrated below for a Ag(lll) elec­trode.

2.2. Experimental Aspects

The reader is referred to the literature for a description of the experimental aspects of X-ray scattering as applied to surfaces, adsorbed layers, and thin films at solid-vac­lIum (15, 16) and solid-liquid interfaces (7, 8). Here I only brief1y describe the impor­tant points and provide information that is necessary to understand the examples that follow. An essential part 6f in situ experiments are suitable electrochemical ceHs (see Rer:~. (1, 2, 7, 8». The key to our cell is a thin, flexible polymer window that contains the electrolyte above the electrode. The electrochemistry is done with the window dis­tended so a thick (~l mm) layer of electrolyte covers the electrode. The electrolyte is then partially withdrawn and the X-ray data are measured through the thin (;:S20tLm) layer that remains on the electrode. This thin-layer geometry minimizes the X-ray ab­sorption from the electrolyte and the background X-ray scattering from it.

Since surfaces and thin films are composed of a small number of diffracting atoms, X-ray scattering of thin layers requires intense X-ray sources. The advent of synchro­tron radiation, with its high brightness, has provided such a source and made measure­ments of surface Iayers feasible. The X-ray data described in this paper were obtained at the National Synchrotron Light Source (NSLS) beam line X20 (3, 4, 21). Typically, the X-ray energy was about 10 keV (..l~ 1.24 A) and was chosen as high as possible to minimize absorption by the electrolyte (8). The X-ray beam was focused onto the sampie with a spot size of approximately lxi mm2, ami the incident flux was approxi­matcly 10 li/sec. The dilfracted beam was analyzed with I milliradian Soller slits. This choice was a compromise: poorer angular resolution would have resulted in an exces­sivcly Iarge background (due to dilThse X-ray scattering from the polymer window and electrolyte), whilc better resolution wOllld have reduced the count rates too much.

3. Applications of X-ray Scattering to Studies of Electrodes In this Section I describe X-ray scattcring studies of monoatomic layers of Bi on

Ag(III), ofwater layers at Ag(lll), amI of anodically grown Ti02 on Ti. The goal of this Section is to iIlustrate the type of structural information that can be obtained with X-ray scattering; more detailed accounts are fOlllld elsewhere (4, 6, 7, 13).

3.1. Underpotential Depositioll of Bi 011 Ag (I Il)

The c1ectrochemical deposition of" metal layers onto a foreign metal substrate fre­quently occurs in distinct stages with the initial formation of one (or more) layers at clectrode potentials positive of the Nernst potential for bulk deposition (22, 23). This process is termed underpotential deposition (lJPD), and these initial deposits have been speculated to form ordered layers (23) - an idea proyen by our X-ray measurements

114

"......

'" E

+10

o

~ -10 "1 ~

-20

-30

(b)

o 100 200 V(mV)

Figurc 3. Cyclic voltammogram for the UPD of Bi on Ag( 111) in 0.1 M IICI04 containing 2.5rnM Bi20 J and 0.35mM NaC!. The poten­tials were measured relative to the Ag/AgCI (3M KCI) rcfcrence c1cctrodc in the diITraction cell. Taken from Ref. (4) with permission.

(3, 4, 7, S). The UPD layers are deposited hy ramping the electrode potential in a ne­gative direction from an initial potential that is positive enough that no metal is ad­sorbed. Figure 3 shows the current nowing to the Ag c1ectrode during such a potential ramp for Bi on Ag(lll) (4, 24). The peak with negative current (at about llOmV) re­sults from deposition of Bi, and our X-ray data show that this peak corresponds to the formation of a mono)ayer of Bi. The positive-currcnt peak is duc to stripping of the monolayer.

Figure 4 shows the in situ diITraction pattern for UPD Bi/Ag(lll) (4). This pattern consists of three symmetry-equivalent domains, amI for simplicity, the pattern that would be observed frmn one domain is shown in Fig. 4a. Notice that the Bi(20) peak occurs at the same position as a Ag truncation rod, which shows that the Bi monolayer is commensurate with the Ag substrate in one direction. We measured the integrated intensities of the peaks in Fig. 4 so that surface crystallography could be used to dcter­mine the positions of the atoms within the Bi unit Gell.

From the diITraction pattern amI the integrated intensities, we calculated the atomic structure of Bi/Ag( 111) using the methods outlined in Section 11. This structure is shown in FiguTe 5. The Bi monolayer has a rectangular lattice and is uniaxially com­mensurate with the Ag substrate along the Ag[211] direction (the brackets refer to a direction in real space) (4). There are two Bi adatoms per rectangular unit cell, but the observation of the Hi(30), (10) amI (12) diITraction peaks with nonzero intensity means that the monolayer does not have the ccntercd rectangular lattice. Rather, the adatom near the center of the rectangle is displaced from the per feet centered position along the Bi[IO] direction, and from an analysis of the integrated scattering intensity, this dis­placement is found to be 0.36 A. Since the (03) peak is not observed (sec Fig. 4), there is 110 dis placement along the Bi[OI] dircction. Our analysis of the integrated scattering intensities also shows that the disorder in the monolayer is anisotropie (e.g., M in Bq. (3) depends on direction): the root-Illeall-square displacement amplitude in the com­mensurate direction is 0.12 A, while in the incommensuratc direction, it is 0.23 A. This reneets the tendency of the substrate 10 'lock' the adatoms in position along the eom­mensurate direction.

115

08 80(21)

0 0 (20)

0 0 0 0

(30) (31) (32) (33) 0

(Rl) (R) 0 0 0 0 0

(20) (23) ~ • (02)cg

0 0 0 0 0 0 0 (10) (11) (12) (13)

0 0

0 0 • 0 X

(00) (01) (02) (03) 0 0

og 80 (a) (b)

Figure 4. In si tu surface X-ray diffractian pattern far Bi/Ag(lII). (a) One quadrant af thc pattern far a single domain. The size or the open circ1es indicates the structure [ac­tor [ar each e1iffraction peak, while the apen squares indicate peaks we did not measure. The intensity of the Bi(03) peak (indicated by the cross) was tao weak to mcasure. The Ag truncation rods are shown by rilled circles; note that the Bi(20) peak overlaps a truncation rod. (b) The observeel diffraction pattern, which is the incohcrent superpo­sition of the thrce domains present Oll the sur[ace.

j-b--j Ag[OlT] Bi[01 ]

Figure 5. Structure of UPD Bil Ag( 111). The open circ1es represent the surface atoms o[ the Ag suhstrate ant! the shaded circlcs reprcsent the Bi aelatoms. The relative sizes of these circ1es corresponel to the near neighbor spacing of bulk Ag (2.89A) anel hulk Bi (3.07A). The Bi monolayer is commcnsurate in the a direction, (a = 5.005A = ß times Ag nearcst-neighhor distance). hut is incammensurate in the h direction.

116

Since the Bi monoJayer is commensurate with the substrate in one direction, it is important to determine the registry of the monoJayer. To do this, we have measured the intensities of the Ag truncation rods. I\s discussed in Sec. 11, the Bi scattering interferes with the I\g truncation-rod scattering (since the Bi is commensurate with the I\g), and this can be used to determine the registry. Our measurements indicate that the Bi[IO] rows lie along the rows of three-fold hollow sites on the surface (see Fig. 5). This is reasonable, since these hollow sites are likcIy the minimum energy sites. (We have shown this for TI/Ag(III) (3».

The potential dependence of the structure of UPD Bi/ Ag(lll) was also investi­gated, and we find that the monolayer cornpresses uniaxially along the incommensurate direction. The ineommensurate lattiee constant h depends on the electrode potential, while the eommensurate lattiee eonstant a remains loeked to the Ag lattiee over the potential range where the monolayer is stable. Thus, the eornpression preserves the uniaxial cornmensurate structure. From the dependenee of the lattice constants 011 cIeetrade potential, one ean determine the 21) isothermal eompressibility of the mono­layer (4, 5, 8), and we find K2J)= 0.8 A2/ev. This is about the same for the hexagonal tlPD monolayers ofTI and Pb on Au(lll) and Ag(lll) am! is in reasonable agreement with K2J) for a 2D ffee electron gas (0.2 A2/eV) (4-6).

In addition to our study of UI'D Bi on Ag( 111), we have also investigated UPD Iayers of Tl am! Pb on I\g( 111) and 1\ u( 111) (1-7). This series of experiments has given insight into the structure determining illteractions for these heavy metals (TI, Pb, Bi) on these smooth (111) surfaees (Ag amI I\u). We deduce that the adatom-adatom inter­action is the predominant atomic interaction that tletermines the monolayer structure, and that the adatorn-substrate interaction only wcakly inOuences this structure. Fur­thennore, the solvent-adatom interaction does not influence the structure, and the structure is not signifieantly aflceted h~' interactions betwcen the adatoms and any an­ions adsorbcd on the UPD monolayer.

'1'0 summarize, our X-ray scattering experiments of these UPD systems have pro­vided detailed information Oll the atomic strllcture of these systems, have enabled a de­termination of the potential c1ependence of this structure, alld have provided insight into the structure tletermining interactions.

3.2. Water Layers Af{iacellt 10 Ag( 111)

The atornic strllcture ofwater layers adjaeent to electrode surfaces (i.e., the double layer) is of tremendous importance in elcctrochemistry and of fundamental scientific in­teI·est. This structure has been the subject of extensive theoretical work (25-33), but to date, there have beeil almost no experimental tests of these theories. To provide direct experimental data on this structure, we have recently investigated the behavior of the double layer at a Ag(lll) cIectrodc. Data have heen obtainecl in O.IM anti O.002M Na F, but here I only discuss the 0.1 M Na!' data, since these have been most completc\y analyzed. In these experiments, measurernents of the specular anti non-specular trun­cation rods were conducted at two electrade potentials: one O.5V positive of the po­tential of zero charge (pze), am! the other O.25V negative of the pzc. The goals were to determine the potential-induced changes of the Ag( 111) surface anel of the double layer strueture perpendicular to the substrate.

117

The structure 01' the c1ectrode surfäce was determined from the non-specular trun­cation-rod intensities. These are shown by the stars and circles in Figure 6 (+ 0.5 and -0.25 V of the pzc, respectively), ami the observation that the profiles are quite similar shows that the surface does not change substantially with potential (e.g., no recon­struction or surface roughening occurs). IIowever, quantitative analysis 01' these data (Iines in I'ig. 6) shows that the spacing betweell the topmost and second Ag( 111) planes does depend on electrode potential. At the positive potential, this interatomic spacing is contracted relative to the bulk spacing by about 0.03 A, while at negative potentials, the spacing is the same as in the bulk. This 'substrate electrostriction' likely affects the differential capacitance and has not been included in current models 01' electrochemical interfaces.

One possible explanation 01' the substrate electrostriction' is in terms 01' the po­tential dependence 01' the electron dcnsity near the c1cctrode surface. At positive po­tentials the surface charge is positive ami the near-surface electron density contracts 01'

is pu lied in from the top Ag(lll) plane relative to that for an uncharged surface. This generatcs a force that tends to puH thc topmost Ag(111) plane toward the bulk metal (e.g., contract the top-sec()!ld plane spacing), which is consistent with our observations. In contrast, for a negative surface charge (negative potentials), the near-surface electron density extends furtber from the electrode surface compared to an uncbarged surface. Now the force genera ted by the near-wrfaee electron density tends to pulI the topmost Ag(lll) plane away from the bulk metal (expansion). Tbc reason we observe a bulk top-plane spacing at -0.25 V 01' the p7.C (rather than an expansion) is that the top-plane spacings for uncharged surfaces in VaCU1l1ll are orten contracted eompared to bulk, and so, a negative surface charge is necessary to overcome this tendency. This explanation provides a qualitative elescription of our observations, which will hopefulIy stimulate a more quantitative, theoretical description 01' this errect.

The structure ofwater perpendiculnr to our Ag(lll) electrodc was elctermined from thc specular truncation rod data, which are shown in Figure 7 for + 0.5 V (stars) and -0.25 V (circlcs) 01' the pze. The dashed lines show the expected intcnsities for a bare Ag( 111) surface (e.g., JlO contributions !I'om the double layer). These curves are slightly different at the two potentials, because 01' the diflcrent top-second Ag(lll) plane spac­ings eliscussed above. That these profiles li.lil to describe the data anel that the mcasured intensities change substantially with potential shows that the double layer structure significantly influences the specular rod intensities. This structure can be quantifieel with an analysis o{" these data. From the best fits (solid lines in Fig. 7), we finel tImt at -0.25 V 01' the pzc: i) the water is ordered in layers extending about 2 or 3 layers {"rom the electrode; ii) thc layer-Iayer spacing hetween these is ~2.5 A; iii) the average elistance hetween the top Ag( 111) plane and the first (inner) water layer is about 3.7 A; anel iv) the coverage of the inner layer is abollt I water molecule per Ag atom. Analysis 01' the data at 0.5 V positive orthe p7.C indieates i) thc ordering in the water layers has increascd and hefe extends abaut 4 layers from the eJcetrode; ii) thc layer-Iayer spacing is again ~2.5 A; iii) the average distance between the top Ag(lll) plane and the inner laycr has decreaseel to about 2.7 A; allel iv) the inner layer coverage has increased to about 2 water molecules per Ag atom.

Point (i) in these observations wggests that the extent of water laycring ne ar the Ag( 111) eJectrode is potential dependcnt. This is quite reasonable. The electric fielel at

118

o 5

Figure 6. Corrected intensities (proportional to structure factor squared) for (a) the (10) rod and (b) the (0 J) rod of Ag( II I ) at + 0.5 V (stars) and -0.25 V (open circles) of the pzc. These are the integrated intensities corrected for sampie area, resolution function, and Ag atomic form factar. The dashed and solid Iines show the best fits to thc da ta at O.S V and -0.25 V of the pzc, respectivcly, where the surface roughness and the spac­ing between the top and second Ag(lll) planes were allowcd to vary. The abscissa is in reeiprocal lattice units (rlu), where I rlu = 0.88 A- t.

" .2-u III ... -"'0 00 ud

141 (rlu) 6,

rigure 7. Corrected intensities of the specular rod for Ag(l I I) at + 0.5 V (stars) and -0.25 V (circles) of the pzc. These are the intcgrated intensities corrected for sampie area, resolution function, and Ag atomic form factar. The dashed lines are the corrected intcnsities for bare Ag( 11 t), using the surface roughness and top-second spacing ob­tained from the fits in Fig. 6, but with no double laycr. The solid lines are thc best fits with the contribution from the double layer.

119

the electrode surface likely plays an important role in produeing the layering, and thus, potentials further from the pzc (e.g., +0.5 V) will tend to create more order or layering than potentials eloser to the pzc (e.g., -0.25V). Similarly, the spacing between water layers of ~2.5 A (point (ii» is also quite reasonable, since this spacing is comparable to that expected for packing ofhard sphere water molecules near a wall and with the results from computer simulations (27-29). To explain point (iii) in these observations, we note that as the electrode charge va ries from negative to positive, it is expected that there will be an orientational change in the inner-Iayer water molecules from oxygen-up to oxygen-down, sinee these orientations provide the most favorable electrostatie inter­actions for the oppositely charged surfaces (25, 26). The observed potential-dependent changes in the spacing between the inner layer and the top Ag(lll) plane (3.7 to 2.7 A) are consistent with such an orientation change, although the metal-water spacings are somewhat larger than might be expccted.

A most surprising result of these experiments (points (iv» is that the inner layer coverages are enormous - 1-2 water molecules per Ag atom. Note that one would ex­pect, based on the bulk density of water, about 0.7 molecules per Ag atom. Thus, our data show that water moleculesin the inner layer are compressed by a factor of about 2! This is unanticipated and promises to profoundly change our picture of the structure of metaljwater interfaces. We can, however, offer one qualitative explanation. At a charged eIectrode, the strong electric field orients the inner-Iayer water molecules through eIectrostatic interaction with the dipoles, since this lowers the e!ectrostatic en­ergy. Thus, water molecules in the bulk electrolyte fee! an attractive force pulling them toward the surface, where their free energy is lower. Consequently, there will be a build-up of water molecules in the inner layer. The magnitude of the resulting density increase will depend on the fiekl (among other variables), and is presently under inves­tigation. It is interesting to note that the compression we observe is rather similar to that observed near solvated ions where the e1ectric fields are comparable (34).

3.3. A1lodic Oxidatio1l 0/ Ti Si1lgle Crystals

Anodic oxide films have been widcly studied for many years beeause of their de­sirable electrical, optical, and mechanicnl properties and thcir excellent corrosion resist­ance. Despite this, surprisingly little is known about the phase, strueture, epitaxy, and strain of the film. To provide insight into some of these issues, we have used X-ray scattering to study the anodic oxide formed on the low index faces of Ti single crystals (e.g., the (1120), (1010), and (000 I) fi,cc~). These measurements werc made cx situ as a first step toward in situ measurements (7, 13, 35).

The anodization was carricd out by ramping the potential at 0.1 mV/sec from open circuit in O.t N 112S04 to 6-10 V (vs. SeE). Anodization of Ti results in full oxidation to Ti02, whieh has three crystalline modirications: rutile, anatase, and brookite, in order of dccreasing thermodynamic stahility at room temperature. On the Ti(1120) face, we ohserve rutile and anatase; while in striking contrast, rutile is absent on Ti(IOIO) and Ti(OOOI), and we only observe anatase. In what folIows, I summarize our results for Ti(1120) (see Ref. (13, 35) for details) and briefly diseuss anodization of Ti(IOIO) and Ti(OOOI), where data analysis is in progress.

120

3.3.1. Anodizatioll of Ti( 1120)

Figure S shows one quadrant of the surface difTraction pattern (X = 0). The rutile and anatase are indicated by solid ami dashed arcs, respectivc!y, and the angular widths of these arcs are equal to the mosaic spreads fJ.qJ. For rutile fJ.q>"""S.5°, and perpendicular to the surface, the mosaic spread is ~X"" ISO. J n contrast, the scattering [rom anatase is observed up to X """ 60°, and the azimuthai peak positions and widths depend on X (see hclow). We es ti mate that rutile accounts for 55± 15% of the observed oxide, with the remainder due to anatase. This is ohtained hy comparing the integrateel intensities of the R(IlO) and A(IOI) peaks (rutile and anatase are abbreviated by R anel A, respec­tivcly).

3.3.2. Rutile Structllre Oll Ti( 1120)

Rutile grows prcfcrentially with its c-axis normal to the Ti(1 (20) surface. It is al­most epitaxial, with its average a- and h-axes in the Ti [TI 00] and [0001] elirections, where the misfits are 10% and 2%, respcctively. Along the surface normal, the misfit is only 0.3%. Bccause these misEts are not too large, we bclieve that the rutile grows at the oxide-metal interface, since this matches as closely as possiblc the rutile unit cell to the Ti lattice. From the observcd orientational rclationship, we have suggested a possible, microscopic growth modcl (sce Rcfs. (7, 13, 35».

From the radial widths of thc rutile difTraction peaks, we estimate that the rutile is crystallographically eoherent over only a short range, about 40 A (13). In compar­ison, the average domain size of thc Ti suhstrate is over 400 A. The radial peak posi­tions indicatc that the rutile is only slightly expaneleel by 0.08±0.06°j" from its bulk lattiee parameter. Thcre is also cvidcllce of inhomogeneous strain, since the radial width of the R(330) is 30% larger than that of R( 110). Thc inhomogeneous strain is eharac­teri7:ed (17) by variations in the d-spacing of the rutile (110) planes of < (fJ.d/d)2 > 1/2 ~ 1.2%.

3.3.3. Allatase Structllre Oll Ti( t 120)

As is seen in Fig. S, the orient!! tional rclationship between anatase and the Ti( 1120) suhstrate is more eomplieiltcd than for rutile. '1'0 investigate this, we have de­tcrmined the peak positions in (p and the rnosaic spreads for several anatase peaks at varying polar angles x. (I f reciprocal space were a glohe with the surfaee normal, Ti(ll 20), deEning the pole, then q) and X correspond to longitude and latitude, respectivc1y.) Figure 9 shows these positions and widths for the A( 10 I) amI A(200) peaks. Accounting for this texture pattern requires two dilTcrcllt preferred orientations for the anatase, amI for the dominant orientation, we find that the surface plane corresponds to the A(013) plane (D). We ean gain insight into the growth process amI whether the anatase grows on the Ti substrate or on rutile, hy cOlllparing the epitaxy ofthe 1\(013) planes with both R(OOI) and Ti(1120). While neither shows a particularly good match, the growth of anatase on the Ti substrate appcars more prohahle, because of a doser match in Ti-Ti distances and orientations that are fount! for the A(O 13) and Ti( 1120) planes. Thus, we bclieve that the anatase forms at the Ti/oxide interface, whieh is consistent with the measured transport numhers (36).

1> = 9rJ' I

Ti(0002)

(0000)

80

~O)

TiOl/Ti( 1120)

~30) •

-""" A(204) ",

'\A(200) \

\

\A(004)

Ti(1100)

,

: \A(211)

IA(200) • -1>=rJ' Ti(Z200)

Anatase Texture Peaks / Ti(1120)

121

Figurc R. Onc quadrant of thc in­plane diITraetion pattern for anodi­eally oxidized Ti(1120). Thc Ti in-plane reeiproeal lattiee is shown hy solid points. Thc rutile and anatase seattering peaks are indi­eated hy solid al1(l dashed ares, re­spectively. The azimuthai angle 1) is defined to hc 0" along Ti[TIOOJ and 90" along Ti[OOOI]. Thc other threc quac!rants ean he reproc!ueec! hy renecting the pattern ahout cjJ = 00 anc! cjJ = 90". Taken from Rcr. (D) with permission.

(a) Anatase(1 01)

60 ~ ~ Ii) (I) (I)

C, 40 >----< -.~ (I)

~ Figure 9. Peak positions in >-1 1) from the seans at eon-20

stant X (e.g., Fig. R) for anatasc on Ti(l120). Thc

0 symhol size is proportional -90 -60 -30 0 30 60 90 to peak intensity, and the

$ (degrees) horizontal bars rcfcr to the

80 widths tJ.,p of thc profiles.

(b) Anatase(200) Takcn from Ref. (13) with permission.

60 Ci) Ql (I)

~r ~~ c, 40 (I)

~ 1-------1 >-----< t---o---{ 1-------i ,... 20 .-;---<

l--~---1 ........

0 -90 -60 -30 0 30 60 90

$ (degrees)

122

The preferred orientation is an average orientation, and there exists eonsiderable disorder about this average. Moreover, the disorder is anisotropie, leading to streaks of intensity in the texture plot (Fig. 9), rather than to isotropie broadening. The sim­plest deseription of the observed texture has a fixed axis with a high degree of rotational freedom about that axis. This is similar to fiber texture (19), but the fiber axis is not perpendicular to the film and the rotatiOTlal freedom about the fiber axis is not complete. To deseribe the data in Fig. 9, the fixed axis must be near A[133], whieh oeeurs at X = 18° and 1J = 35°. Traees of the A(lOI) and A(200) peaks are shown by the lines in Fig. 9 for a ±20° rotation ofthe average anatase strueture about this axis. The observed peak widths ö.1J in Fig. 9 are broader than predicted by this simple uniaxial orientation, and this ean be aeeounted for if the ftxed axis has an isotropie mosaie spread of ~ 12°.

The agreement in Fig. 9 is excellent, apart from a few small peaks found near 1J = O. To explain these we must invoke a second orientation where the A[010] prefers to grow along the Ti[OOOI] axis, with its A[IOO] axis canted 20±1O° out of the plane. The curve traced out by this orientational distribution is shown as a dashed line in Fig. 9. lt is interesting that despite the poor orientational order of the anatase, the erystal­lographic coherence length, 80-100 A, is larger than in the rutile (~40A). In addition, the anatase lattiee is expanded (by O.24±O.03%) relative to bulk anatase.

The causes of the dominant orientation and the disorder about the preferred axis are unclear, but this eomplieated texture suggests that the strain relief meehanism for anatase is different from that in rutile. We bc1ieve this differenee is related to the larger misftt between anatase and Ti(1I20) lattice than between rutile and Ti(1120) (13).

3.3.4. Anodization of Ti(IOIO) and Ti(OOOI)

Recall that for Ti(IOIO) and Ti(OOOI) only anatase is present. For growth on Ti(IO 10), the A[IIO] axis lies in the surl'ace plane, parallel to the Ti[121O] axis, but there is eomparatively littlc orientational prefercnce about this axis. This type of orientation is similar to both fiber texture (19) and to anatase on Ti(1l20), exeept that here the pre­ferred axis is in the surf'ace plane, rather than along a more eomplicated direetion (ana­tase on Ti(l120» or along the surfaee normal (fiber texture). The cause of this unusual texture is unclear. Finally, for Ti(lOIO) the average anatase strain is about 0.5% ex­panded, twice that for Ti(1120).

On the basal (0001) face the texturing is weak but reflects the hexagonal symmetry of thc underlying meta I substrate. In addition to our ex silu results, the (0001) face of Ti was also examined in situ, under potential control imrnediately after growth. The oxide was found to be essentially the ~ame as that on the emcrsed (0001) face, implying timt emersion docs not altcr the oxide ~tructure in any appreciable way.

In summary, our X-ray scattering experiments of anodic Ti02 have provided de­tailed, quantitative information on the atomic strueture of these teehnologieally impor­tant oxides. They have shown that the crystallographic orientation of the Ti substrate ean have a surprisingly profound eflect on the phase, structure, epitaxy, and strain of the oxide film. They have also provided a detailed growth model for the anodization ofTi(l120). What remains is to relate the oxide structure to the oxide properties.

123

4. Future Prospects In this paper, I have given an overview of X-ray scattering and described how this can be used to determine atomic structure in interfacial electrochemistry and corrosion. Recently, there has been rapid growth in research utitizing X-ray scattcring for in situ structure determination in this area, ami researchers have investigated surface recon­structions (9), metal oxidation (J 3, 37), UPD layers (3-8), and the double layer (38). In the future, I expect research in these areas to continue with an increasing emphasis on more complicated systems (e.g., UPD CujAu(lll), surface reconstruction on Ag(llO)). In addition, there will likely be an increase in the number o[ experiments on the struc­ture of the double layer - a poorly lInderstood, but crucially important, aspect of inter­phasial electrochemistry.

To date, most X-ray scattering studies have concentrated on static structure, but the future will likely sec experiments to investigatc the kinetics of structural changes. We have already conducted preliminary X-ray ref1ectivity experiments to investigate the evolution of interface morphology d",.illg the c1ectrodeposition ofNiFe (39). These were done on a time scale of the order of minutcs, but experiments on a much faster time scale (~ millisecond) are of more intercst and are possible. Such experiments might in­c1ude phase transitions (i.e., monotayer formation), etectrodeposition, and the formation of thin layers. The Advaneed Photo 11 Source (under construction at Argonne National Laboratory) will be a great boost to sut:h time-rcsolved experiments, since this source will have a brightness about 1000 times that of eurrent sources.

Acknowledgments The experiments described in this paper are the result of fruitful collaborations

between physicists and chemists, and they wOllld not have been possible without help from my colleagues: Mike Armstrong, Gary Borges, Joe Gordon, Jason I-Ioward, Chris McMilIan, Owen Melroy, Jocelyn Richer, Mahesh Samant, Bill Smyrl, Larry Sorensen, Gina Whitney, Dave Wiesler, and Dennis Yee. Much ofthis work was perfonned at the National Synchrotron Light Source (NSLS), which is supported by the lJ.S. Depart­ment of Energy.

References I. M.G. Samant, M.F. Toney, G.t. Borges, L. Blum, and O.R. Metroy, Surf Sei.

193, L29-L36 (1988). 2. M.G. Samant, M.F. Toney, G.L. Borges, L. B1um, and O.R. Melroy, J. Phys.

Chem. 92, 220-225 (1988). 3. M.F. Toney, J.G. Gordon, L.S. Kau, G. Borges, O.R. Melroy, M.G. Samant, D.G.

Wiesler, D. Yee, and L.B. Sorensen, Phys. Rev. B 42,5594-5603 (1990). 4. M.F. Toney, .l.G. <Jordon, M.Ci. Samant, G.t. Horges, D.G. Wiester, D. Yee, and

L.B. Sorensen, Langmuir 7, 796 (1991). 5. O.R. Melroy, M.F. Toney, G.t. Borges, M.G. Samant, J.B. Kortright, P.N. Ross,

and L. ßlum, Phys. Rev. B 3R, 10962 (1988). 6. M.F. Toney, .l.G. Gordon, and O.R. Mclroy, SPIE Proc. 1550, 140 (1991).

124

7. M.F. Toney, in The Application of Surface Analysis Methods to Environmental/ Materials Interactions, edited by D.R. Baer, C.R. Clayton, and G.D. Davis (The Electrochemical Society, Pennington, 1991), p. 200.

8. M.F. Toney and O.R. Melroy, in Hectrochemical Interfaces: Modern Techniquesfor In-Situ Interface Characterization, cdited by II.D. Abruna (VCH Verlag Chemieal, ßeTlin, 1991), pp. 57-129.

9. ß.M. Oeko,.T. Wang, A.T. Davenport, and II.S. Isaaes, Phys. Rev. Leu. 65, 1466 (1990).

10 . .T. Wang, ß.M. Oeko, A..J. Davenport, and II.S. Isaaes, Phys. Rev. B 46, 10321 (1992).

11. H. You, C.A Melendres, Z. Nagy, V.A Maroni, W. Yun, and R.M. Yoneo, P1ws. Rev. B 45, 11288 (1992).

12. I.M. Tidswell, N.M. Markovie, C.A. Lueas, ami P.N. Ross, submitted to Phys. Rev. B.,1993.

13. D.G. WiesleT, M.F. Toney, M.G. Samant, O.R. Melroy, C.S. MeMillan, and W.B. SmyrJ, Surf Sei 268,57 (1992).

14. E.E. Koch, llandbook on Synchrotmn Radiation Vol. J, (North-Hotland, Amster­dam, 1983).

15. I.K. Robinson, in Handbook on Synchrotron Radiation, edited by G.S. ßrown and D.E. Moneton (North Holland, Amstcrdam, 1991), Vol. 3, pp. 221-266.

16. R. Feidenhans'l, Surf Sei. Reports 10, 105-188 (1989). 17. L.H. Sehwartz and .T.B. Cohen, D([(ractionfrom Materials (Springer-Verlag, Berlin,

1987). 18. B.B. Warren, X-Ray Diffraction (Addison-Wcslcy, Rcading, 1969). 19. B.D. Cullity, Elements of X-RqJ) Di((raclion (Addison-Wcsley, Reading, 1978). 20 . .T.S. Pedcrscn, R. Feidenhans'l, M. Niclscn, K. Kjacr, F. Grey, and R.L. Johnson,

Surf Sei. 189/190, 1047 (1987). 21. M.l'. Toney, .T.G. Gordon, M.G. Samant, G.t. Borges, O.R. Melroy, D. Yee, and

L.B. Sorenscn, Phys. Rev. B 45,9162-9374 (1992). 22. D.M. Kolb, in Advances in Eleclrochemislry and Efectrochemical Engineering, edited

by H. Gerisehcr and C.W. Tobia~ (Wiley, Ncw York, 1978), Vol. 11, p. 125. 23. D.M. Kolb,.I. Vac. Sei. Technol. A 4, 1294 (1986). 24 .. L W. Sehultze and K.R. Brcnske, .l. I~·lectroanal. Chem. 137, 331 (1982). 25. J.O'M Boekris, B.E. Conway, ami E. Yeager, Comprehensive Treatise 0/

Electrochemistry Vol. 1, (Plenum Press, New YOTk, 1980). 26. W.R. l'aweett, S. Levine, R.M. dcNobriga, ami A.C. MeDonald,.I. Electroanal.

Chem. Ill, 163 (1980). 27. c.Y. Lee, J.A. MeCammon, and R..J. Rossky, .I. Chem Phys. 80, 4448 (1984). 28. E. Spohr, .I. Phys. Chem. 93, 6171 (1989). 29. G.N. Patey and G.M. Torrie, Chemica Scrip ta 29A, 39 (1989). 30. L. ßlum, in Advances in Chemical P/tysics, Vol. 78, edited by I. Prigogine and S.A.

Riee (Wiley, New York, 1990), p. 171. 31. W, Sehmickler and D. Henderson, I'rog. Su~r Sei. 22, 323 (1986). 32 .. LW. Halley and D. Priee, Phys. I?ev. B 35,9095 (1987). 33 .. LN. Glosli and M.R. Philpott,.1. Cl!em. Phys. 96, 6962 (1992). 34. ß.E. Conway, Chem. Soc. I?ev. (I?I~V. Chem Soc.) 2,<)3 (1992).

125

35. D.G. Wiesler, M.F. Toney, O.R. Mclroy, C.S. McMiIlan, and W.B. Smyrl, in X-ray Methods in Corrosion and lntelfacial Electrochemistry, editcd by A.l. Daven­port and l.G. Gordon (The E1ectrochcmical Society, Pennington, 1992), p. 89.

36. N. Khalil and l.S.L. Leach, Electrochim. Acta 31, 1279 (1986). 37. c.A. Melendres, H. You, V.A. Maroni, Z. Nagy, and W. Yun,.I. Electroanal.

Chem. 297, 549 (1991). 38. M.J. ßedzyk, G.M. ßomrnarito, M. Caffrcy, and T.L. Penner, Science 248,52

(1990). 39. M.l. Armstrong, G.M. Whitncy, amI M.F. Toney, in X-ray Methods in Corrosion

and lnterfacial Electrochemistry, cditcd by A . .I. Davcnport and .I.G. Gordon (The Elcctrochcmica1 Socicty, Pcnnington, 1992), p. 62.

SURFACE STRUCTURE AT THE Au(l1t) ELECTRODE

Abstract

B. M. OCKOt and JIA WANGtt Brookhaven National Laboratory Department 0/ Physicst Department 0/ Applied Sciencett Upton, New York 11973

This chapter summarizes Surface X-ray Scattering (SXS) studies carried out at the Au(ll1) surface under electrochemical conditions. In-plane diffraction measurements have been been carried out to accurately deterrnine the surface structures for bismuth and iodine adsorbate layers and for the reconstructed surface. Complementary x-ray specular reftectivity studies have been carried out to deterrnine the surface normal structure. These studies are sensitive to the structure of the electrode and the solution layers.

1. Introduction

The electrode/electrolyte interface differs from the ultra high vacuum (UHV) interface in that the charge of the electrode surface can be controlled by adjusting the applied potential. At the electrochemical interface, electric fields as high as 107 V / cm are acces­sible, and the induced surface charge can exceed 0.2 e/atom. An understanding of the atomic scale structure of electrode surfaces is important in many fields of science and technology. However, until recently, structural investigations of electrode surfaces relied primarily on indirect methods (e.g., cyclic voltammetry (CV) and ex situ techniques such as low-energy-electron-diffraction (LEED)). During the past several years, surface x-ray scattering (SXS) [1-3], scanning tunneling microscopy (STM) [4-7], and atomic force mi­croscopy (AFM) [8,9] techniques have been applied to single crystal electrode surfaces to study the electrodeposition of metals, the reconstruction of metal surfaces, the monolayer deposition of anion adlayers, the oxidation of metals, and the surface normal structure in the double layer region.

SXS is an extension of bulk x-ray scattering [10] which incorporates the semi-infinite nature of real crystals. The broken translational symmetry of the interface gives rise to weak scattering which can be distinguished from the more intense three dimensional bulk Bragg peaks. For the reader unfamiliar with surface x-ray scattering, there are several excellent reviews [11]. In addition to these more general reviews, several reviews pertaining to the electrochemical environment are recommended [12].

The structure of surfaces can be investigated with SXS within the surface plane and along the surface normal by controlling the direction of the scattering vector Q, wh ich is the difference between the incident and scattered x-ray wave vectors. Typically, information within the surface plane is obtained by orienting Q almost entirely within the surface plane.

127

C. A. Melemires andA. Tadjeddine (eds.), Synchrotron Techniques in lnteifacial Electrochemistry 127-155. © 1994 Kluwer Academic Publishers.

128

This corresponds to the grazing incidence angle geometry where the grazing incidence angle is typically a few degrees. On the other hand, when Q is aligned entirely along the surface normal direction (e.g., no in-plane component) information is obtained about the surface normal structure. This geometry is referred to as specular reflectivity since the angle between the surface plane and incident wave vector is equal to the angle between the surface plane and the scattered wave vector. Additional structural information can be obtained by determining the scattered intensity distribution along the surface normal direction at a fixed, finite in-plane wave vectors corresponding to either the bulk crystal structure or to the overlayer structure.

This chapter is intended to review SXS electrochemical studies carried out at the Au(l11) surface. In section 2, a brief review of the electrochemical and x-ray scatter­ing techniques is provided. In the following section the structures within the surface plane, obtained from grazing incident angle diffraction measurements, are presented and in the last section the surface normal electrode and solution phase structures, obtained from x-ray reflectivity measurements, are presented. At the Au(111) surface both hexago­nal and rectangular structures have been reported in the literature, however, this chapter is limited to uniaxial-incommensurate-rectangular (UIR) (p X v'3) structures. Here we present results for the reconstructed Au surface and for Bi, and I adlayers on the ideally terminated surface. Due to the high resolution of SXS, we are able to show that the iodine and bismuth adlayers are continuously compressible over a range of potential and that they show no ''lock-in'' transitions at commensurate lattice positions.

The x-ray specular reflectivity technique is presented in section 3.2 and results are shown for a variety of solutions. Here we explain the relationship between the surface normal structure and specular reflectivity with particular attention to simple density profiles. Analysis of the reflectivity profiles suggests that the first layer of water exhibits enhanced positional order. For an iodine adlayer, the iodine coverage, deduced from the reflectivity analysis, agrees with the value obtained from the in-plane diffraction study.

2. Experimental Techniques

2.1. SURFACE PREPARATION AND ELECTROCHEMICAL CONDITIONS

A gold disk electrode (10mm diameter by 2mm) was spark cut and aligned along the nom­inal [111] direction [13]. The disk was aligned within 0.1· ofthe (111) axis, mechanically polished, and then electropolished [15]. Finally, the crystal was sputtered using an argon beam at 5 X 10-5 torr at 800·C using a defocused beam at 1 keV and 2 X 10-6 A for several hours. The sampie was transferred through air to an electrochemical x-ray scattering cell constructed from Kel-F as shown in Fig.l.

A 61' polypropylene window sealed the cell with a thin capillary electrolyte film between the crystal face and the polypropylene film. An outer chamber was flushed with N2 gas to prevent oxygen from diffusing through the polypropylene membrane. The applied potential was referenced to a AgjAgCI(3M KCI) electrode connected to the cell through a micro glass frit. In order to reduce the possibility of chloride contamination from the reference electrode, a second frit was added and the path separating the two frits was filled with N aF electrolyte. Counter electrodes were either gold or platinum wires.

Gold Crystal

Figure 1. X-ray electrochemical cello

Polypropylene Window

129

After flushing the cell with N2 gas, the deoxygenated electrolyte was injected into the cell using a syringe. The cell was filled with enough solution to expand the polypropylene window leaving a thick electrolyte layer (several mm) between the face and the window. The potential control was then turned on and cyclic voltammograms were carried out in this geometry to check the electrochemical conditions. Before carrying out the x-ray scattering measurements the cell was deflated leaving a thin electrolyte layer which we estimate from the small angle reflectivity measurements to be between 10 and 20 J-L thick. In this thin electrolyte layer geometry, the effects of bulk impurities are greatly reduced relative to the thick electrolyte geometry.

2.2. X-RAY SCATTERING

The x-ray scattering measurements were carried out with focused, monochromatic syn­chrotron radiation at beam lines X22B, X22C, and X25 at the National Synchrotron Light Source (NSLS) at Brookhaven National Laboratory. In the four circle geometry, the sampie was oriented through its Euler angles 0, X, and <P by a spectrometer under

computer control [11,16,17). The magnitude of the scattering wave vector is Ik~ - ~I = (41r / >.)sin(20 /2), where k~ and k~ correspond respectively to the incident and scattered wave vectors and where 20 is the angle between these two vectors (10) . Diffraction mea­surements were carried out by measuring the scattering intensity along paths in reciprocal space in the "w = 0 geometry" (18).

To conveniently describe the scattering wave vector in terms of its components in and out of the surface plane, we employ a hexagonal coordinate system [13,19). In this coordinate system, an arbitrary position in hexagonal reciprocal space (see Fig. 2) is represented by the vector (H, K, 0) within the surface plane and (0, 0, L) along the surface normal

130

direction. The reciprocal space lattice constants are a* = b* = ;;Ta = 2.52 A -1, and c* = -!!Js = 0.89 A where a = 2.885 A is the gold nearest-neighbor spacing. The relationship between the cubic vector, (h, k, l)cubic, and the hexagonal vector (H, K, L) is given by the transformations h = -4H/3 - 2K/3 + L/3, k = 2H/3 - 2K/3 + L/3, and I = 2H/3 + 4K/3 + L/3. For example, (1, 1, 1)cubic = (0,0,3), (0, 0, 2)cubic = (0, 1,2), and (0,2, 2)cubic = (1,0,4).

(0) 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 ° A 0 0 0 0 0 0 000 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Lo ° 0 0 0 0 0 0 0 0 000 0 0 0 0 0 0 00 0 °

-1..-V

(b)

00000 0 0 0 0 0 000 0 0 0 0 0 0 0 000

00000 0 0 0 0 0 0 000 0 0 0 0 0 0 0 0 0

<1,1> ~ q,

• (1.1)

• • (1.0) (0.1)

• • (0.0) (1.1)

• • (0.1) (1.0)

• (1.1)

Figure2. a) Real space atomic structure for the ideally terminated Au(111) surface. b) In-plane reciprocal space pattern for the Au(l11) surface. The six lowest order reflections are arranged in a hexagonal pattern.

131

3. Res ult s

3.1. IN-PLANE DIFFRACTION STUDIES

Despite the underlying hexagonal symmetry ofthe Au(ll1) surfaee, the top layer of atoms often forms a uniaxial ineommensurate reet angular (UIR) phase. This phase, is denoted (p X v'3) sinee the repeat distanees, in units of the nearest-neighbor distanee (a), are p and v'3. In the following three subseetions, we present detailed information on the in-plane strueture and phase behavior of the (p X v'3) UIR phases for the reeonstrueted surface (no specifieally adsorbed species), for adsorbed bismuth adlayers, and for adsorbed iodine adlayers.

There are three equivalent rectangular surface domains which may form on the under­lying hexagonal substrate and all three contribute to the observed diffraetion patterns. For simplicity, only the domain for which the ineommensurate direction is along the (110) direction will be considered. The in-plane measurements were carried out in the graz­ing incident angle geometry with a grazing incident angle, a, equal to 1.250 • This angle eorresponds to L = 0.2 at all wavelengths. In this seetion, three dimensional wavevector (H, K, 0.2) will be referred to as (H,K).

At a grazing angle of 1.250 the incident x-rays illuminate a region of the erystal 0.5 mm wide across the entire crystal faee (10 mm). The speetrometer resolution, in reciprocal space, was primarily determined by the angular aeeeptance of the scattered radiation and the mosaie of the Au(111) erystal. Within the seattering plane the spectrometer resolution was determined by an array of equally spaeed parallel plates (Soller slits) wh ich provide a 28 resolution of 0.10 half-width-half-maximum (HWHM). The seattered intensity was measured with a scintillator detector on the 28 arm following the soller slits.

3.1.1. Au(l11) SURFACE RECONSTRUCTION

The possibility that gold surfaces might reeonstruet under eleetrochemieal eonditions was suggested on the basis of the hysteresis in capacity-potential curves [20]. Ex situ LEED studies have shown that the Au(111) surface, after emersion from an electrochemical eell, forms a (23 X v'3) phase in the negative potential regime [21]. In situ STM studies [5] in HClO4 solutions have confirmed the existenee of the (p X v'3) reconstruetion within the negative potential regime. The real spaee strueture of the reconstrueted Au(ll1) surface is shown in Fig.3a.

In a surface diffraction measurement, the (px v'3) reconstruction gives rise to additional in-plane reflections beyond the underlying (1 X 1) reflections. These are arranged in a hexagonal pattern (see Fig. 3b) around the integer reflections [13,14]. Here we limit our diseussion to diffraction along the qr axis, see Fig.3b, which eonneets the reeonstruetion peak at (fi / v'3, 1 + fI / v'3)to the (0,1) peak. In Fig.4a, the measured scattering intensity is shown at aseries of deereasing potentials between 0.1 and -0.8 V in a 0.01 M N aCI solution. At potentials of 0.10 V and above, the scattering is eentered at qr = 0, which indieates that the surfaee is not reeonstrueted. As the potential is redueed below 0.05 V, a seeond peak emerges, whieh signals the reeonstructed phase. The reconstruction peak position, (fI/v'3, 1 + fI/v'3) , moves outward corresponding to increased compression as the potential is decreased.

132

(0)

1\

I~ V 3

<1,1 >

(b)

5 7 9

..... (i.O)

a 0 & ...

o a

(0.1)

11 1 3 1 5 1 7 1 9 21 23

& • &

(1.1) q a 0 r

)

(0.1)

• • (0.0) (1.1)

a 0 ..... o 0

(1.0) a 0

& • &

(1.1)

Figure 3. a) Real space structure ofthe reconstructed (px v'3) Au(111) surface are shown as filled circles. Surface atoms in the left and right hand sides of the figure are in undistorted hexagonal sites (ABC stacking sequence). In the center of the figure the surface atoms are in faulted sites (ABA stacking sequence) [19]. For 24 surface atoms in place of 23 underlying surface atoms along the (11) direction, the compression is 24/23-1 = 4.4% and 0 = (J372)/23 = 0.038. b) Reciprocal space pattern for reconstructed Au(111) surface. In addition to the hexagonal pattern from the underlying Au(I11) layers (filled circles), the reconstructed surface exhibits reflection arranged in a hexagonal pattern (open symbols) surrounding the filled circles. Diffraction data was acquired by carrying out scans along the qr direction in reciprocal space.

In order to extract the unit cell size p (referred to as the stripe separation) the scattering profiles have been described as the sum of two Lorentzians [13,14)( centered at zero and 6). Fits to the Lorentzian form (solid lines in Fig.4a) describe most of the essential features of the profiles. The dependence of the fitted stripe separation, p = v'3/(26), is

133

shown in Fig.4b. Although the transition is reversible, the measured stripe separation depends on the sweep direction, as indicated in the figure. After the initial signs of the surface reconstruction at 0.05 V (corresponding to the emergence of the reflection at (o/va,l + o/va», there is furt her compression as the potential is decreased, as shown by the inverted triangles in Fig.4b. The maximum observed compression corresponds to p = 23 (0 = 0.038). This length is identical to that found in vacuum studies. This observation suggests that the Au(111) surface has the same underlying surface motifunder electrochemical and vacuum conditions.

The reconstruction formation/lifting transition can be directly related to the induced surface charge, u(E), independent of the anion species. The induced surface charge density has been obtained by integrating the capacitance curve from the PZC to E, that is, u(E) = f:zcC(E')dE'. This relation is correct in weakly adsorbing electrolytes and nonadsorbing electrolytes, such as in a 0.1 M N aF electrolyte. At a surface charge density of R:l 0.07 e/atom (excess electrons), in all solutions, the reconstructed (p X va) starts to form [13,14]. In NaF and NaCI electrolytes the reconstruction starts to lift at u = +0.07 e/atom. This lifting transition occurs at u = +0.05 e/atom in N aBT solutions. The appearance of the reconstructed surface at a common value of u, in all the three electrolytes, suggests a simple phase transition mechanism based on the excess surface charge.

The idea that excess surface charge might induce surface atoms to form a structure different than the underlying bulk layers has been discussed in calculations by Fu and Ho using a local-density-functional (LDF) model [22]. For Ag(llO) they predicted that the missing-row structure - absent at the elean metal surface - is induced by alkali adsorbates with an induced surface charge of R:l 0.05 e/atom. Within the context of the LDF model, adsorbates in vacuum mimic an applied electric field.

3.1.2. BISMUTH ADLAYERS ON THE Au(lll) ELECTRODE

During the electrodeposition process metal ions may be adsorbed and reduced at potentials positive of the bulk deposition potential. The metal adlayers which form under these conditions, referred to as underpotential deposition (UPD), are most often monolayer phases, however, in several instances bilayer phases may also form. The UPD potential range can extend over a one volt potential window and over this range there may be several different structures. These structures may exhibit either long range or short range positional order and in some instances island deposits may form.

The first in situ in-plane diffraction studies of a UPD structure were carried out by the IBM Almaden group on epitaxial Ag(lll) films [1]. These studies showed for the first time that UPD lead layers exhibit an incommensurate, hexagonally elose packed structure (hcp) and that the lead-lead spacing could be modified by the applied potential. Chen and Gewirth had previously utilized AFM to show that two distinct UPD Bi adlattice structures exist on the Au(l11) electrode surface [9]. As the potential is decreased, a commensurate (2 X 2) adlattice first appears which exhibits a 25% coverage. At more negative potentials (but positive of the bulk electrodeposited bismuth phase) a second bismuth adlattice appears which is rectangular (see Fig.5a).

134

1

~0.8 c: 1-1--"'''-::l

.ci 0.6 L­o -.... ~0.4 '00 c: Cl> 1:: 0.2 I!!-_eeee"'-

(0) 6=0.038

o L-~ ____ ~~ ____ ~ ____ -L ____ ~ ______ ~ ____ ~~

526 +=­~ o a.. Cl>

(f)

Cl> 23 a.. 'C

V5

-0.01 o

\7 neg. 6. pos.

0.01 0.02 0.03 0.04 0.05 q (units of 0*)

r

I

'V (b) sweep sweep D.

'V D. 'V

'V D. 'V

'V 'V

'V 'V D. D.

~ ~;q;n~: 'V 'V 'V 'V D. 1) D.D.D.D.D.D.D.AD.D.

-

-

20~~~~~---'~~-L-~~~'~~~~~~

-1.0 -0.5 0 0.5 EAg/AgCI (volts)

Figure4. (a) X-ray scattering profiles, a10ng the qr axis, for the reconstructed Au(1l1) surface (see Fig. 3b) in 0.01 M N aCI solution at aseries of potentials chosen from scans between 0.1 to -0.8V. The solid lines are fits to a Lorentzian line shape. (b) The stripe separation obtained from fitting the data to a sum of two Lorentzians. The triangles and inverted triangles correspond to positive and negative potential sweeps, respectively.

135

(a)

• 6. (b) (1,1) F

• 6. 6. 6. • (1,0) c 0 E (0,1)

• • 6. • (0,0) A B (1,1)

• • (0.1) (1,0)

• (1,1)

Figure5. a) Real space atomic structure for the(p x V3) bismuth adlayer on the Au(111) surface. b) In-plane reciprocal space pattern for the Au(111) surface with a (p x V3) bismuth adlayer. The underlying gold layers yield the integer reflections (filled circles) whereas the bismuth adlayer yields the reflections labeled B-F. Reflection A is not observed.

We have carried out a SXS study of Bi on Au(l11) in order to compare our results with the previous AFM results [9] and the results for Bi on Ag(l11) reported by Toney et al., [2]. Along the incommensurate direction, Le., (ll), the first diffraction peak occurs at (215,215) which is indicated by B in Fig.5b. Diffraction at this position implies that there are rows of atoms which are separated by a distance ap = a/(2t5) as shown in Fig.5a. The fact that there is no diffracted intensity at (0,0) implies that the unit cell is centered along the (ll) direction. The existence of diffraction at position A in Fig.5b implies that the unit cell is not weH centered along the (I1) direction. A full description of the (p X V3) diffraction pattern has been reported by ehen and coworkers [23].

136

""'":-0

E 700 0 0 600 N

@ 500 "'0 c: 8 400 Q)

~300 Ul ...... § 200 0

~100 ~ 'e;; 0 c:

Q) ...... c:

0.65

~ 0.64 N

11 <:Z:) ......... ~ 0.63 e ~

ü 0.62

0.60

o 0.17 V o 0.09 V t:;. 0.01 V

(0)

0.61 0.62 0.63 0.64 0.65 0.66 (X,X) (units of 0*)

0.61 L..J._ ............ _L...---"-_.L..---'-_...I-----J..._~ o 0.05 0.10 0.15 0.20

EAg/ AgCl (volts)

Figure6. (a) X-ray scattering profiles for the bismuth (p x v'3) adlayer on the Au(I11) surface along the (ll) direction at three potentials after background subtraction. (b) The bismuth coverage, () = 26 = p-l, obtained by measuring the peak position (26,26) versus the applied potential. The triangles and inverted triangles correspond to the positive and negative potential sweeps, respectively.

137

The atomic coverage of Bi can be measured very accurately from the diffraction peaks shown in Fig.6a since the coverage (J, relative to the underlying gold density, is equal to 28 [23]. Diffraction data was acquired along the (11) direction every 20mV. On the Ag/AgCI potential scale bulk deposition occurs very elose to zero volt. In Fig.6b, the bismuth coverage is shown in the positive potential sweep (triangles) and in the negative potential sweep (inverted triangles). When the (p x v'3) phase first appears at 0.19V, (J = 0.612. As the potential is lowered the coverage continuously increases in a linear manner with decreasing potential. The line shown in Fig.6b is a linear fit to the data with an electrocompressibility d(J/dE = -0.20S/V. This slope is about 30% larger at the Au(111) surface than for the Ag(111) surface [2]. At 0.01 V, just prior to bulk deposition, (J = 0.646 which corresponds to a Bi lattice constant of a/(28) = 4.4SA. The lattice constants of gold and silver only differ by 1/4% and not surprisingly the Bi adlayer achieves the same lattice constant on both the Ag and Au (111) surfaces before bulk deposition appears [2].

3.1.3. IODINE ADLAYERS ON THE Au(111) ELECTRODE

Recently, in situ STM studies have been extended to adsorbed iodine adlayers at the Au(l11) surface. A centered-rectangular (5 X v'3) and a rotated-hexagonal (7 X 7) R 21.S· structure has been reported by Gao and Weaver [7] whereas a (v'3 X v'3) R 30· and (3 X 3) has been published by Tao and Lindsey [6]. Here, we show that adsorbed iodine on the Au(111) surface forms an electrocompressive UIR (p x v'3) phase. The key results of our SXS studies of electrodeposited iodine on the Au(l11) surface are presented below. The complete study will be published elsewhere [25].

The present study was carried out in 0.01 M potassium iodide(K I) solutions. Within the (p X v'3) rectangular phase the diffraction pattern [25] is given by the basis vectors (8, 8) and (t, ~), where the iodine coverage (J = 28 = l/p. Since half of the spots are absent in the pattern shown in Fig.7b (i.e., complete cancellation of the scattering amplitudes) the real space unit cell is centered with two iodine atoms (see Fig.7a). Along the (11) direction the first diffraction reflection is at (28, 28) where 28 ~ 0.39. We note that the (5 X v'3) structure (four atoms per unit cell) is identical to the (p X v'3) structure (two atoms per unit cell) when p = 2.5 (28 = 2/5), and the commensurate (v'3 x v'3) R 30· structure is identical to the (p X v'3) structure when p = 3 (28 = 1/3). At an potentials the (v'3 x v'3) R 30· structure, expected on the basis of previous studies [21], was never observed.

Typical x-ray scattering profiles of the incommensurate-rectangular iodine phase through the peak at (28,28) are shown in Fig. Sa. The sharp diffraction peak in the profiles originates from rows of iodine atoms which are separated by pa/2 as shown in Fig.7a. As the potential is increased these diffraction peaks shift to larger 8 in a uniform manner. The coverage, (J, relative to the underlying Au(l11) layers is equal to 28. In Fig. Sb the coverage is shown for both the positive and negative sweep directions. There is a slight shift in (J obtained from positive and negative sweep directions by about 0.1%. This suggests that there is a fundamental difference between adding atoms (positive sweep) relative to removing atoms (negative sweep).

138

1\

I~ V

(0)

<1,1 >

(b)

-. -(1,0)

-. (0.1)

• (1.1)

.-(0,1)

• • (0,0) (1,1)

.-(1.0)

• (1.1)

Figure 1. a) Real space atomic structure for the(px v'3) iodine adlayer on the Au(111) surface. b) In-plane reciprocal space pattern for the Au(111) surface with a centered (p x v'3) iodine adlayer. The underlying gold layers yields the integer reflections (filIed circles) whereas the iodine adlayer yields the reflections shown by the short line segments.

At a critical coverage of 0.409 the (px v'3) phase transforms to a rotated-hexagonal phase in which the adsorbed iodine atoms no longer remain in the hollow gold rows [25]. This transition can be understood in terms of the competition between the I-I and the I-Au interactions. Within the (p X v'3) phase the nearest iodine atoms (four) are in neighboring

rows where the I-I distance, Dl-l = a/ -!i 2 + yt2, varies from 4.62 to 4.321 as 26 varies from 0.366 to 0.409. The next nearest iodine atoms (two) are always displaced by two gold rows, i.e, av'3 = 4.9981. The transition to the rotated-hexagonal phase appears at a potential when D1 - 1 = 4.321 js very dose to the Van der WaaJs diameter of 4.301. At this bond length the energy required to pack the adatoms doser can no longer offset

E 2000 o o C"-I

@ 1500 -0 c: o ~ 1000

"'" 23 c: 6 500 u -...-

0.42

0.41 ---.. ~ 0.40

~0.39 Q)

gO.38 L-Q)

> 0.37 0 u

0.36

0.35

o -0.15 V o -0.05 V 6. +0.05 V • +0.15 V

0.37 0.38

(b)

-0.3 -0.2

139

0.39 0.40 0.41 0.42 (X,X) (units of a*)

6. positive sweep 'V negative sweep

-0.1 0 0.1 0.2 EAg/AgCI (volts)

Figure8. (a) X-ray scattering profiles for the iodine (p x v'3) adlayer on the Au(ll1) surface along the (ll) direction at four potentials after background subtraction. (b) The iodine coverage, () = 28 = p-l, obtained by measuring the peak position (28,28) versus the applied potential. The triangles and inverted triangles correspond to the positive and negative potential sweeps, respectively.

140

the adsorbate-substrate energy gained by maintaining the I atoms in the hollow gold rows and the transformation to the rotated-hexagonal phase proceeds.

From the STM studies it is difficult to ascertain whether the incommensurate structures "lock in" to the commensurate structures [6,7). In the present SXS measurement a "lock in" transition would be manifested by a region of potential in which there is no change in the unit cello However, as demonstrated by Fig.8b there is no indication of a change in the monotonic increase in 26 with potential at the commensurate (5 X va) structure (26 = 0.40). A linear fit to the coverage versus potential plot is shown in Fig.8b as the solid line. The electrocompressibility dO / dE = 0.094/ V which is about a factor of two smaller in magnitude than for Bi on the Au(111) surface.

For the bismuth and iodine adlayers we have also compared the areal slope (dA/dE). The area per adsorbed atom A = a2 "!-O-1 and dA/dE = 7.21Ä2(-dO/dE)/ < 0 >2.

For the bismuth and iodine adlayers IdA/dEI equals 3.75Ä2/eV and 4.47 Ä2/eV, respectively. This finding shows that there is better agreement for IdA/dEI than for dOldE when comparing bismuth and iodine adlayers. At the present time, no model or theory adequately describes the observed electrocompressibility of electrode­posited adlayer phases.

3.2. X-RAY REFLECTIVITY

Specular x-ray reflectivity has been successfully applied to many different types of inter­faces including the solid/vacuum [19,26), solid/vapor, liquid/vapor [27), and solid/liquid interfaces [3,13,14,28). Reflectivity is a powerful probe of interfacial structure since it can be related to the average electron density profile along the surface normal axis in a straight forward manner. As previously demonstrated, at an electrochemical interface the reflectivity depends on the structure of the underlying atomic layers and of surface adsorbates [13). Quantitative information on the layer spacings, densities, and root-mean­square (r.m.s.) dis placement a.mplitudes can be obtained by fitting the reflectivity profiles to simple real space models. In Fig. 9 the atomic structure of the ideally terminated Au(l11) layer is shown and the corresponding reciprocal space pattern. Before describing the measured reflectivity spectra from the Au(l11) surface (section 3.2.3), the relationship between the surface normal electron density and reflectivity is shown (section 3.2.1). The experimental aspects of obtaining reliable reflectivity spectrums are presented in section 3.2.2.

3.2.1. SIMPLE REFLECTIVITY MODELS

First we consider the theoretical specular reflectivity for an ideally terminated Au(l11) surface. Within the kinematical approximation (also known as the Born approximation), the absolute specular reflectivity can be calculated from a sum over atomic layers. As previously shown [19,13), for the ideally terminated Au(l11) surface the reflectivity is given by

(1)

{cl {oon

•

1 (006)

a ! ---:,,(015)

(003)

• (01 2)

----t---- (010)

141

Figure9. a) Surfaee normal atomie strueture for the ideally terminated Au(1l1) surface. k; and kJ are the incident and outgoing wavevectors, respeetively, and in b) Qz is the momentum transferred . c) The eorresponding reeiproeal spaee pattern. The are eorresponds to a e roeking eurve path in reeiproeal spaee.

where r Q is the Thomson radius of the electron, FAu (Q z) is the atomic form factor of gold, and O"DB = 0.OS5Ä is the r.m.s. displacement amplitude (Debye-Waller term) [29]. The sum over atomic layers is given by

s(Qz) = feiQznd (2) n:O

where d = 4.0SÄ/v'3 for gold. Carrying out the geometrie sum over the atomic layers gives the relationship

1 I L 1-1 Is(L)1 = 2" sin(7r3 ) (3)

where L = Qz/c·. We point out that Is(L)1 is a measure of the number of gold layers whieh add coherently. When L is midway between two Bragg peaks the geometrie sum Is(L)1 is ~ and the scattered intensity corresponds to the equivalent scattering from half a monolayer of gold. Bragg peaks emerge from Eqs.1- 3 when L = 3n (where n is an integer) sinee the sum diverges when the scattering from all the atomic layers are in phase. The kinematical approximation is unphysical in the vicinity of Bragg peaks sinee the refleetivity can be greater than unity. In order to eorreet the model, the dynamical model of x·ray scattering, which takes into account the decrease in the incident electric field as a function of the penetration depth, must be utilized [10]. However, the kinematic approximation provides a good approximation to the dynarnical model when the reflectivity is less than 0.1% of the incident beam. An important feature of both models is the (L - 3n)-2 dependence of the reflectivity in the tails of the Bragg peaks.

142

In Fig.IOa we show the calculated R(L) for the ideally terminated Au(Ill) surface and the corresponding real space layer profile. The principal features are the Bragg peaks (divergence), the weak scattering between neighboring Bragg peaks, and the nearly 100% reflectivity at small L. For reconstructed or otherwise non-ideally terminated surfaces, the scattering amplitudes from the atomic layers elose to the surface are no longer equivalent to the bulk scattering amplitudes. We now consider the possibility that the top gold layer is different from all of the ideally terminated underlying gold layers [26].

In the single-Iayer model, we express the the sum over atomic layers as

s(Qz) = Ple~Q;(0'1B - O'i)eiQzdl + EeiQznd (4) n=l

where the top layer density (pd, the top layer Au-Au interlayer spacing (d - dl ), and the top layer layer r.m.s. displacement amplitude 0'1 may differ from their ideal valU!!S of 1, 2.355 A, and 0.085 A, respectively. Here we consider separately the effects of each parameter. Finally, we note that for real surfaces all three effects are important.

First, we consider how changes of the interlayer expansion or contraction of the top gold layer modify the reflectivity profiles. We consider the examples in which PI, and 0'1 are equal to their bulk values and d = 0.1 d (a 10% expansion). In Fig.IOb we show the density and reflectivity profiles for this model as solid lines and the ideal profiles are shown as dotted lines. The most striking feature of the reflectivity profile for the expanded top layer model, shown by the solid line, is the asymmetry in the vicinity of the Bragg peaks. Here the reflectivity is enhanced to the left of the Bragg peak and reduced on the right side. Note that the asymmetry increases as the order of the Bragg peak increases. Furthermore, the asymmetry increases as the top layer expands and the asymmetry reverses if the top layer contracts.

Secondly, we consider how the r.m.s. displacement amplitude of the top gold layer modi fies the reflectivity profiles. Here, as an example, we present the case where 0'1 is increased from its bulk value of 0.085 A [29] to 0.40 A and where PI, and d1 are equal to their bulk values. This effectively redistributes the gold atoms such that there is a greater range of scattered phase factors from the top gold layer relative to the underlying gold layers. In Fig.IOc we show the density and reflectivity profiles for this model as solid lines and the ideal profiles are shown as dotted lines. The enhanced r.m.s. displacement model reflectivity curve is always less than the ideally terminated curve and this reduction is most significant at intermediate wavevectors between Bragg peaks and increases with L. Within the context of Eq. 4, an enhanced r.m.s. displacement amplitude decreases the magnitude of the scattering amplitude of the top layer relative to the underlying layers

and this effect increases with L. For sufficiently large L the Gaussian factor, e-Q;0'~/2, approaches zero and the scattering amplitude from the top layer no longer contributes to the sumo At this point, the reflectivity approaches the ideally terminated reflectivity profiles (not shown in the figure).

Finally, we consider how the density of the top gold layer modifies the reflectivity profiles. Here, as an example, we present the case where the top layer density, PI is decreased from its bulk value of unity to ~ and where 0'1, and d1 are equal to their bulk values. Physically, this half-density occurs when there is a missing-row reconstructed

c o L.. ..... <..> Cl>

W

1.2 (a) 1.0

0.8

0.6

0.4

0.2 ::::::: O~UUUU

1.2 (b) 1.0

0.8

0.6

0.4

O·~~:lJ

1.2 (c) 1.0

0.8

0.6

0.4

0.2

0L-~-L~~~~~~

1.2 1.0

0.8

0.6

0.4

(d)

O·~LJI~ u U 0

~~~~~~~~1~ ideal

10-2

10~

10-6

10-8

L...._~~~_~~~_L..J 10-10

~~~~~~~~1~

10-2

10~

10-6

10-8

L...~~~~---,-_~~........J 10-10

10°

10-2

10~

10-6

10-8

L..._-,--~--,_-,--~--,_....J 10-10

.,.....---.-~-r--.-~~-..-::J 100

10-2

10~

10-6

10-8

10-10

-4 -2 0 2 4 6 8 10 0 1 2 3 4 5 6 7

z (A) L (units of c*)

143

Figure 10. Real space density profiles (left) and calculated reflectivity spectra (right) for the Au(lll) surface; (a) ideally terminated, (b - d) top layer modified, (b) expanded by 10%, (c) enhanced r.m.s. displacement amplitude, (d) 50% density. The ideally terminated profiles are shown as dotted curves.

144

phase. In Fig. IOd we show the density and refiectivity profiles for this model as solid lines and the ideal profiles are shown as dotted lines. The decreased top layer density model refiectivity curve (solid line) is always less than the ideal curve (dotted) and the refiectivity appears to go to zero at positions between Bragg peaks. Within the context of Eq. 4, at these half-order positions the magnitudes of the scattering amplitude from the top layer and from the sum of all the underlying layers are both ! and these two amplitudes are exactly out of phase. Thus, this complete cancellation gives rise to zero refiectivity at these positions.

In the proceeding discussion, we have separately considered the effects of a top layer expansion, of an enhanced top layer r.m.s. displacement amplitude, and of a reduced top layer density. These three effects occur simultaneously at real interfaces and there is the added complication that not all of the underlying layers are ideally terminated gold layers. By carrying out least-squares fitting analysis of the profiles using a model which incorporates several surface layers which are not bulk like, many of these deviations can be extracted. Finally, we point out that at the electrochemical interface other species may form an adsorbed layer at the electrode surface whose scattering amplitudes interfere with the amplitude from the underlying gold layers. These species may include a metal monolayer (e.g., Pb), an anion layer (e.g., I), or possibly even a layer of water.

3.2.2. EXPERIMENTAL

In a refiectivity experiment, the scattered intensity is measured along the direction normal to the surface. These measurements are complicated by the fact that the the refiected intensity must be measured over a large dynamic range (e.g., Qz) and this requires greater experimental care than in an ordinary diffraction measurement. For instance, the incident x-ray beam must not "spill-over" the edges of the crystal, the imperfect sampIe mosaic must be accounted for, and the diffuse scattering must be properly subtracted [11,17,26]. In order to avoid "spill-over" and to maximize the incident fiux, the defining slits (before the sampIe) are adjusted to illuminate a region of less than 5mm in length. The x-ray intensity on the sampIe is monitored by measuring the scattered intensity after the slits.

The resolution, in reciprocal space, is determined by the size of the incident beam, the size of the collected beam, and the sampIe mosaic [26,19,13,14]. In the present mea­surement, the detector acceptance of the scattered x-rays is controlled by a slit (2 mm by 10mm) located 750mm from the sam pIe on the spectrometer 28 arm. Since the inci­dent beam was always less than 1 mm by 1 mm, the detector slit is the dominant factor determining the resolution.

The resolution volume is a three dimensional ellipsoid with three characteristic widths [10,11,27,17,26]. Two of these widths /:;.Q2 and /:;.Q3 are in the scattering plane of the spectrometer, defined by the rotation axis 2(), and the third width /:;.Q1 is normal to the scattering plane. Within the scattering plane the resolution is determined mainly by the angular acceptance of the detector slit. The /:;,28 full-width-half-maximum (FWHM) is given by the ratio of the slit size (:::::2 mm) to the distance between the slit and the sampIe (:::::750mm). Typically, /:;.28 (FWHM) = 2/750 rads which is about 0.15°. The resolution function defined by these slits has a "box-like" profile where the width, /:;.Q3 (HWHM), equals ~cos(8)/:;'() (HWHM) and where /:;.() (HWHM) = ~/:;'(2()) (FWHM).

145

At ~ = 1.24 Ä (10 kev), fj.Q2 ~ 0.007 Ä -1. By construction, the in-plane transverse res­olution, fj.Q2 (HWHM) ~ fj.Qa sin (J (HWHM). Thus, fj.Q2 is always smaller than fj.Qa. Normal to the scattering plane, the resolution was typically determined by lOmm wide detector slits located on the detector arm. This yields a width, fj.Qh which is about 0.08 Ä -1 (HWHM). We note that the diffuse background intensity which often obscures the reflectivity is proportional to the detector slit area. In order to reduce the background scattered intensity, this 10mm wide slit width is often reduced to 5mm.

Typical (J rocking curves, along the specular direction at fixed 2(J values, are shown in Fig.11. The curves were taken in 0.01 M KI at -0.15 V. The values of L given in the figure correspond to the values at the center of the rocking curves. These diffraction profiles exhibit a sharp peak when the specular condition, (J = H2(J), is satisfied. Note that the rocking curves are broadest at small L. At large L the widths are elose to the spectrometer resolution.

L=0.3 2.9~.06 10

........... 2.8 ..ci 3.15 ....

0 '-"'

3.24 ~ 2.5 'in

1 c::: ß 0.1 3.6 .E

~ 0.01

0 5 10 15 20

theta (degrees)

Figure 11. Typical rocking curves, along the specular direction, in 0.01 M K I at -0.15 V.

We have fit ted the rocking curves, shown in Fig.ll, to a Lorentzian profile with a width 6(J, convoluted with the resolution function, in order to extract the intrinsic peak width. In Fig.12a, the fit ted width is shown versus L. At small L, 6(J = 0.15° and decreases

146

to Iess than 0.010 at the Iargest values of L. The effect of the resolution function is minimal at small L, and becomes very significant at large L. elose to the Bragg peaks the underlying layers contribute significantly to the diffracted intensity and the rocking curves are narrowest.

0.20 (0) S 0 1500 0 00

0 1250 0.15 0

.-.... 0 1000 d>

"'""" Q) "'0 0.10 750 'iC> ......... ......... <:b "0 500

0.05 250

0 (h) 0

0 1 2 3 4- 5 6 7 0 1 2 3 4- 5 6 7

L (units of c*)

Figurel2. (a) Lorentzian rocking curve width, MJ, and (b) coherence length , versus L for 0.01 M KI at -0.15 V.

In a diffraction measurement, the coherence length ( is equal to the inverse diffraction width (e.g., ~q-l). The coherence length, ( = (LC·CO)-l, calculated from cO (see Fig.12a), is shown in Fig.12b. Far from the Bragg peaks, the coherence length corresponds to the effective surface length over which reflected x-rays are coherent. The calculated coherence length varies with Land ranges from about 500 to 1500 A. Midway between the Bragg peaks the contribution to the reflectivity from the underlying layers is diminished and this leads us to conclude that the surface coherence length is around 500 A. This length is slightly larger than the width calculated from in-plane diffraction measurements.

To account for the spatial variations of the sam pie mosaic and the effect of the coherence Iength on the widths of the rocking curve, it is necessary to integrate the scattered intensity over a range of 0 at constant 20 as described above. The scan range must be broad enough

147

to calculate the integrated intensity after background subtraction. Typical 8 seans are shown in Fig.11. Sinee the out-of-plane resolution, !::!..Ql ~ 0.08 A (HWHM), is at least an order of magnitude broader than the transverse width from the Au surfaee, (Lc·o8), calculated from the roeking curves, the spectrometer automatically integrates over the out-of-plane direction. Under these conditions, we have previously shown [26] that the absolute reflectivity is given by

R(L) f d81.(8) (5) ! f d(28)lo(28)

where 1.(8) is the scattered intensity after background subtraction and 10 (28) is the direct beam intensity. It is often difficult to measure 10 (28) accurately since the fuU incident beam saturates the detector. For praetical purposes, we choose the reflectivity normalization by insisting that the reflectivity agrees with the idealized reflectivity profile elose to Bragg peaks. At these positions the reflectivity is almost completely insensitive to the details of the surface normal structure since many gold layers are contributing in the Bum over the atomic layers. Previously, it has been shown that both normalization techniques yield identical results [19]. In order to integrate the rocking curves, the curves were fitted to a Lorentzian lineshape with a sloping, adjustable background.

3.2.3. RESULTS OF REFLECTIVITY STUDIES

Before presenting the experimental reflectivity profiles, the potential dependence is pre­sented at fixed L. These results demonstrate that anion effects must be incorporated into the specular reflectivity models [13]. In Fig. 13, the potential dependence of the specular reflectivity is shown at (0, 0, 2.2). These curves were obtained in 0.01 M solution of N aF, NaCl, and J( Br, at a slew rate of 1 mV/sec. Scans at (0, 0, 3/2), exactly half-way between the (0, 0, 0) and (0, 0, 3) Bragg peaks, exhibit very similar potential dependence as at (0, 0, 2.2). At the most negative potentials the surface is reconstructed and the absolute reflectivity equals 1.56 ± 0.05 X 10-6 in all three solutions. The decrease in the reflectivity at (0, 0, 2.2), in the positive sweep, results from the transition from the reconstructed phase to the (1 X 1) phase and from the adsorption of anions at the gold interface. This intensity decrease is larger in the J( Br electrolyte than in the N aCl electrolyte, since bromine atoms have more electrons contributing to the scattering amplitude than chlorine atoms. In N aCI solution, the capacitance peak extends beyond 0.3 V which implies that the adsorbed chloride layer is not saturated up to 0.3 V. Correspondingly, the reflectivity at (0, 0, 2.2) continues to decrease as the potential is extended beyond 0.3 V as more chloride is adsorbed. Because the adsorption-desorption of anions at the gold surface is reversible and occurs rapidly upon changing the potential, the reflectivity versus potential curves obtained from both sweep directions overlap at the potentials above 0.3 V. In order to account for the effects of solution species at the gold interface and potential induced changes in the structure of the gold layers the reflectivity model given by Eq.4 must be extended.

To accommodate the adsorbed solution layer [13], the specular reflectivity is expressed as

(6)

148

2

---- KBr N N q2 0 '-" ..... 0

~ .> 1 +l 0 Q)

-.= Q)

a::: NoCI CD 0

2

NoF oÜ-~~~~~~ __ ~~~~ -0.8 -0.4 0 0.4 0.8

[Ag/AgCl (volts)

Figure 13. Potential dependence of the reflectivity at (0,0,2.2) on an absolute scale in 0.01 M solutions of KBr, NaGl, and NaF in the positive and negative sweep directions. The potential independent reflectivity at low potentials corresponds to the reconstructed phase. The decrease in reflectivity in the positive going sweep corresponds to the transition to the (1 x 1) state and anion adsorption.

The first term in Eq.7 corresponds to the scattering amplitude from a surface ion/water monolayer with a molecular charge Zo (fixed in the analysis), an atomic layer density po

149

relative to a Au(l11) atomic layer, and a root-mean-square (r.m.s.) atomic displacement amplitude, 0"0. Here the magnitude of the scattering amplitude from each gold layer equals the gold form factor, FAu which equals 74 at small Q •. The deviation from 79 is because not all of the electrons are considered free electrons at 8 kev [10]. Correspondingly, the magnitude of the scattering amplitude for an adsorbed monolayer (Po = 1) is Zo which equals 10, 17, 35, and 53 for water, fluorine, chlorine, bromide, and iodine, respectively.

The absorption losses from the polypropylene window and the thin electrolyte layer have been accounted for by assuming a slab of adsorbing material of constant thickness. This produces the factor e-Q···/Q·, where Q ah is related to the thickness, density, and mass absorption cross sections of these layers. In the subsequent analysis, the absorption correction given in Eq. 6, Q ab., is fixed at a value of 0.2c· for the N aF, N aCI, and K I data and at a value of O.4c· for the KBr data. Reflectivity within 0.05c· of the Bragg peaks and below O.4c· has been excluded from the fitting procedure. In the first case, the kinematic approximation does not apply near Bragg peaks and in the latter case it is difficult to control the footprint of the incident beam on the sampie at small angles.

In order to understand the effects of adsorJ:1ed species at the AU(l11) surface, it is useful to directly compare the electrochemical results with the calculated profiles from an ideally terminated surface and from the vacuum (23 X V3) surface [19]. In Fig.14a, the ideally terminated curve (dotted) does not exhibit the asymmetry around the Bragg peak and cannot describe all of the features of the data. The asymmetry around the Bragg peaks is most apparent in Fig. 14c in which the reflectivities have been normalized to the ideally terininated reflectivity profile. Since both the vacuum and electrochemical data support the same (23 X V3) motif, we believe that the scattering from the gold layers should be the same for both interfaces. The reconstructed vacuum model (0"1 = 0.15A, PI = 1.045 and (1 = -dt/d = 3.3%) shown in Fig.14a and 14c as a dashed line, exhibits the same asymmetry around the Bragg peaks. However, the model fails to describe the data between L = 0.5 and 2.5. This discrepancy is most apparent when the reflectivity is normalized to the reflectivity for the ideally terminated interface (Fig.14c). An improved description of the specular reflectivity, in the reconstructed potential range for all three electrolytes, is obtained if we incorporate a single water layer. In the analysis we have fixed the molecular charge, Zo = 10 (water). The parameters for the top gold layer were set to values obtained in vacuum (PI = 1.045, and (1 = 3.3%); and we have allowed Po, do, and 0"0 to vary. All three data sets are very weIl described by a model with Po = 1.0 ± 0.2, 0"0 = 0.60 ± 0.15A [30], and a gold-water layer separationof 2.9 ± 0.3A given by the solid lines in Fig.14a and 14c. A density of po = 1.0 for water only represents 13% of the electron density for the underlying gold layers. This water density is larger than the expected layer density (0.75) calculated from the water volume (30A3 per water molecule) raised to the 2/3 power. We note that it is difficult to distinguish between a water layer and submonolayers of adsorbed ions since both contribute to the electron densities.

The present model of the-specular reflectivity (Eq.7), including the effects of a water monolayer, also describes the reflectivity from the AU(111) surface in the (1 X 1) phase at positive potentials in N aF. Within the context of this model, the best fit is represented by a gold-water layer spacing of 2.9 A an r.m.s displacement amplitude of the water layer, 0"0 = 0.64A [30] and a water layer density of 1.0 ± 0.2. There is no apparent relaxation

150

1,,0 IV

10-1

10-2

f= i:: 10-3 (.) w ~ 10-4 w 0:::

0::: 10-5

:5 ::::> (.) 10-6 w CL Cf)

10-7

10-8

10-9

2.5

Cl w 2.0 N 1.5 :J « ~ 1.0 0::: 0 z: 0.5

0

I

(0) ~ 1.2 (b) • Ci) c 1.0 ...... ideal Q) 0.8 Cl ----- (23xIT) c 0.6 -t H20 layer 0 0.4 ~

-<-' 0.2 <.) ;. ----------(])

0 ............ w

-5 -4 -3 -2 -1 0 2

z (A)

• NaF (OVJ

~ ~~CI (~8.6~f ...... ideal ----- (23xJ.3) - (23x13) + H20 layer

(c)

0

• NaF (OV) o KBr ( -O.5V) l:>. NoCI (-O.6VJ

...... ideal ----- (23xJ3) - (23xJ3) t H20 loyer

2 3 4 5

(O,O,l) (units of c *) 6

3 4

7

Figure 14. (a) Absolute specular x-ray reflectivity from the Au(111) surface in 0.01 M solutions

of 1< Er, N ael and N aF in the reconstructed phase. Calculated reflectivity profiles are shown

for the ideally terminated surface (dotted line), the (23 x va) reconstructed surface (dashed line)

and for the (23 x va) reconstructed surface with a water layer (solid line). (b) Corresponding

real space density models. The density bump at about 3 A corresponds to the water layer. For

graphical purposes, some ofthe curves have been shifted upward. (e) X-ray reflectivity normalized

to the reflectivity from the ideally terminated surface.

151

of the top gold-gold interlayer spacing and the r.m.s displacement amplitude of the top gold layer, 0"1, is 0.12Ä in the (1 X 1) potential regime. Extending the model to inelude a second water layer does not improve the quality of the fit and suggests that the first layer density is elose to unity. This density is somewhat higher than the expected density of 0.75.

The introduction of a boundary condition, such as a metal wall, modifies the bulk liquid structure in the vicinity of an interface [31-34]. In addition to this steric effect, the surface electric fields may reorient the dipole moments at the interface. Far from criticality, it has been predicted that the ordering induced by the wall should decay away with a characteristic length scale of the molecular size [35]. The structure of water near an electrode surface has been studied using analytic theory, and Monte Carlo and Molecular Dynamics simulation. For an updated bibliography of these studies see Ref. 34. The present results strongly suggests that ordered water near the gold interface modi fies the x­ray refelctivity profiles. Detailed analysis provides information on the water layer density, the water-metallayer spacing and the width of the water layer (e.g. the r.m.s displacement amplitude). These results appear to be consistent with some of simulation results. At present, we can not detect the small change in the water-metallayer spacing which should occur with potential that are associated with the reorientation of the water dipoles. Also, the present data does not allow us to speculate on the nature of the water layers beyond the first layer. Studies which utilize electrodes with smaller Z will lead to a larger change in the reflectivity in the presence of ordered water layers. Preliminary studies by Toney and coworkers indicate that the top layer of water may have a density which is twice that of ordinary water at the Ag(111) surface [36]. This large water density is not supported by our data.

The specular reflectivity profiles in the (1 X 1) potential region in I( Er (0.5 V) and N aCI (0.6 V) have been fitted with the same single layer model. Here, Zo is fixed at 35 and 17 for the bromide and chloride anions, respectively. In the analysis, we allowed the three additional parameters describing the adsorbed anion layer (Po, do, and 0"0) and the parameters describing the top gold layer (0"1 and dt) to vary in the least-squares-fitting procedure. All of the remaining parameters have been constrained at their bulk values. The fit ted model is described by an anion layer with r.m.s dis placement amplitudes (0"0)

0.47 Ä (brOlnide) and 0.49Ä (chloride) [30]. The fitted atomic layer densities (Po) are 0.50 (bromide) and 0.67 (chloride). In both electrolytes the r.m.s displacement amplitudes of the top gold layer, 0"1 = 0.15Ä is slightly sm aller than the reconstructed value. The gold-bromide and gold-chloride layer spacings are 2.4 ± 0.3 Ä. This spacing is larger than the gold-gold layer spacing. Finally, we note that there is no apparent relaxation of the gold-gold layer spacing for the top gold layer (within 0.5%).

For iodine adlayers, the coverage obtained from the specular reflectivity analysis can be directly compared with the coverage calculated from the in-plane diffraction [25]. This comparison provides a stringent test of the reflectivity analysis. In Fig. 15 we show the specular reflectivity in 0.01M I(] at -0.15V (elosed circIes) in the (p x..j3) phase.As previously demonstrated, this deviation results from the destructive interference of specif­ically adsorbed anion layers. For iodine adlayers, Zo = 53, this effect is more significant than with Cl or Er adlayers. The best fit at -0.15 V gives po = 0.38 which was very

152

elose to the value of 8 obtained from the in-plane diffraction (8 = 0.37 at -0.15 V). In the subsequent analysis, Po was set at 0.37 and we allowed 0'0, 0'10 do, and d1 to vary in the analysis. The best fit was obtained with 0'0 = 0.26 Ä, 0'1 = 0.14Ä, with a top most interlayer gold spacing which is 1% greater than the bulk spacing and with a Au-I inter­layer spacing equal to 2.3 Ä. The A u-I interlayer spacing can be estimated in terms of the covalent bonding radii of gold (1.44Ä) and iodine (1.33Ä). If we assume that all of the iodine atoms are in "A top," "bridge," or "hollow" sites, then the interlayer spacings are 2.77, 2.36, 2.21 Ä, respectively. However, in the (p X v'3) phase the iodine atoms are at intermediate positions between "bridge" and "hollow" sites and we estimate an interlayer spacing of 2.28 Ä which is the average of these two. This estimated Au-I interlayer spac­ing is very elose to the measured value of 2.3 Ä determined from the reflectivity analysis. Furthermore, this distance supports the assumption of covalent bonding between the Au and I.

o

0.01 M KI -Fits -----Ideal Au( 111)

• -0.15 V

2 345 L (units of c*)

6 7

Figure 15. Absolute specular x-ray reflectivity from the Au(l11) surface in 0.01 M K I at -0.15 V in the (p x vIä) adlayer phase.

In summary, specular reflectivity in the reconstructed phase suggests that there is a well-defined monolayer of water at the interface. In the (1 X 1) phase, the adsorption of

153

anions has a drastic effect on the reHectivity and detailed analysis provides information on the coverage and position of this anion layer. Nonspecular reHectivity studies have confirmed that the phase transition from the (23 X v'3) phase at low potential is to an ideally terminated (1 x 1) phase at higher potential [13).

These studies demonstrate that specular reHectivity can provide detailed information about the surface normal structure at the Au(l11) electrode. We note that the gold surface normal structures under electrochemical control [3,13,37) and ultra high vacuum condi­tions [19,26) are in good agreemnt for both the (111) and (001) surfaces. An important aspect of the x-ray reHectivity technique in electrochemistry is the sensitivity to adsorbed solution species and to the underlying electrode layers. This structural information can not be directly obtained from other techniques. Therefore, x-ray reHectivity is likely to playa important role in the quest to determine the structure of electrode surfaces.

4. Acknowledgement

This work is supported by the Division of Materials Sciences, U .S. Department of Energy, under Contract No. DE-AC02-76CH00016. The Au(1l1) reconstruction study was carried out in collaboration with Hugh Isaacs and Alison Davenport (BNL/DAS). The bismuth study was carried out in collaboration with Chun-Chen, Keith Kepler, and Andy Gewirth (Univ. of ill.). The iodine study was carried out in collaboration with Gavin Watson (BNL/Physics).

5. References

1. M. G. Samant, M. F. Toney, G. L. Borges, L. Blum, and O. R. Melroy, J. Phys. ehern. 92, 220 (1988); O. R. Melroy, M. F. Toney, G. 1. Borges, M. G. Samant, J. B. Kortright, P. N. Ross, and 1. Blum, Phys. Rev. B 38 10962 (1988).

2. M. F. Toney, J. G. Gordon, M. G. Samant, G. 1. Borges, D. G. Wiesler, D. Yee, and L. B. Sorensen, Langmuir 7796 (1991).

3. B. M. Ocko, J. Wang, A. Davenport, and H. Isaacs, Phys. Rev. Lett. 65, 1466 (1990); B. M. Ocko and J. Wang, in Proc. 0/ the Workshop on Structuml Effects in Electrocatalysis and Oxygen Electrochemistry, edited by D. Scherson, D. Tryk, M. Daroux, and X. Xing, Electrochern. Soc.,Pennington, 1992, p. 147.

4. O. M. Magnussen, J. Hotlos, R. J. Nicols, R. J. Behrn, and D. M. Kolb, Phys. Rev. Lett. 64, 2929 (1990); T. Hachiya, H. Honbo, and K. Itaya, J. Electroanal. ehern. 315,275, (1991) ; M. P. Green and K. J. Hanson, SurE. Sei., 259, L743 (1991); S.-L. Yau, X. Gao, S.-C. Chang, B. C. Schardt, and M. Weaver, J. Am. ehern. Soc., 113, 6049 (1991).

5. X. Gao, A. Harnelin, and M. J. Weaver, J. ehern. Phys., 95, 6993 (1991).

6. N. J. Tao and S. M. Lindsay, J. Phys. Chem., 96, 5213 (1992).

7. X. Gao, and M. J. Weaver, J. Am. ehem. Soc., 114, 8544 (1992).

8. S. Manne, P. K. Hansrna, J. Massie, V. B. Elings, A. A. Gewirth, Seien ce, 25, 183 (1991). .

154

9. C. H. Chen, S. M. Vesecky, and A. A. Gewirth, J. Am. Chern. Soc., 114,451 (1992).

10. B. E. Warren, X-Ray Diffraction, Addison-Wesley (1969).

11. R. Feidenhans'l, SurE. Sei. Reports 10, 105 (1989); I. K. Robinson and D. J. Tweet, Rep. Prog. Phys. 55, 599 (1992).

12. M. F. Toney and O. R. Melroy, in Electrochernical Interfaces: Modern Techniques for In-Situ Interface Chamcterization, edited by H. D. Abruna, VCH Verlag Chernical, Berlin, 1991, p. 57; M. F. Toney, J. G. Gordon, and O. R. Melroy, SPIE Proc. 1550, 140 (1991).

13. J. Wang, B. M. Ocko, A. J. Davenport, and H. S. Isaacs, Phys. Rev. B, 46, 10321 (1992).

14. J. Wang, B. M. Ocko, A. J. Davenport, and H. S. J. Wang, A. J. Davenport, H. S. Isaacs, and B. M. Ocko, Seience, 255, 1416 (1992); J. Wang, B. M. Ocko, A. J. Davenport, H. S. Isaacs, in X-Ray Methods in COr7'Osion and Interfacial Electrochern­istry, edited by A. J. Davenport and J. G. Gordon, Electrochern. Soc., Pennington, 1992, p.34; B. M. Ocko, A. Gibaud, J. Wang, J. Vac. Sei. Technol. A, 10,3019 (1992).

15. J. L. Whitton and J. A. Davies, J. Electrochern. Soc., 111, 1347 (1964).

16. W. R. Busing and H. A. Levy, Acta Cryst., 22, 454 (1967).

17. I. K. Robinson, Phys. Rev. B, 33, 3830 (1986).

18. S.G.J. Mochrie, J. Appl. Cryst., 21, 1-3 (1988). 19. A. R. Sandy, S. G. J. Mochrie, D. M. Zehner, K. G. Huang, and D. Gibbs, Phys.

Rev. B, 43, 4667 (1991); K. G. Huang, D. Gibbs, D. M. Zehner, A. R. Sandy, and S. G. J. Mochrie, Phys. Rev. Lett., 65, 3317 (1990).

20. A. Hamelin, J. Electroanal. Chern., 142, 229 (1992); J. P. Bellier and A. Hamelin, C.R. Acad. Sei. (Paris) 280, 1489 (1975).

21. M. S. Zei, G. Lehmpfuhl, and D. M. Kolb, SurE. Sei., 221, 23 (1989).

22. C. L. Fu and K. M. Ho, Phys. Rev. Lett., 63, 1617 (1989). 23. C. H. Chen, K. D. Kepler, A. A. Gewirth, B. M. Ocko, and J. Wang, J. Phys. Chern.,

97, 7290 (1993). 24. B. G. Bravo, S. L. Michelhaugh, M. P. Soriaga, I. Villegas, D. W. Suggs, and J. L.

Stickney, J. Phys. Chern., 95, 5245 (1991); R. 1. McCarley and A. J. Bard, J. Phys. Chern., 95, 9618 (1991).

25. J. Wang, G. M. Watson, and B. M. Ocko, (accepted at Physica B).

26. D. Gibbs, B. M. Ocko, D. M. Zehner, and S. G. J. Mochrie, Phys. Rev. B, 38, 7303.(1988); D. Gibbs, B. M. Ocko, D. M. Zehner, and S. G. J. Mochrie, Phys. Rev. B, 42,7330 (1990); B. M. Ocko, D. Gibbs, K. G. Huang, D. M. Zehner, and S. G. J. Mochrie, Phys. Rev. B, 44, 6429 (1991).

27. A. Braslau, P. S. Pershan, G. Swislow, B. M. Ocko, and J. AIs-Nielsen, Phys. Rev. A, 38, 2457 (1988).

155

28. B. M. Ocko, Phys. Rev. Lett., 64, 2160 (1990).

29. S. P. Witherow, T. H. Barret, and R. J. Culbertson, SurE. Sei., 161,584 (1985).

30. The r.m.s displacement amplitudes given are larger by a factor ~ va than those given in ref. [13]. The present values have been modified because of a programming error which does not modify the fitted profiles ..

31. P. A. Thiel and T. E. Madey, Surf. Sei. Rep. 7,211 (1987).

32. Microseopie Models of Electrode-Electrolyte Interfaces, edited by J. W. HaIley and L. Blum, Electrochern. Soc., Pennington, 1993.

33. Condensed Matter Aspects of Electrochemistry, edited by M. P. Tosi and A. A. Ko-rnyshev, World Scientific, 1991.

34. J. D. Porter and A. S. Zinn, J. Phys.· ehern. 97, 1190 (1993).

35. B. J. Widom J. Stat. Phys. 19,563 (1978).

36. Michael Toney, private communications ..

37. I. M. Tidswell, N. M. Markovic, C. A. Lucas, and P. N. Ross, Phys. Rev. B,47, 16542 (1993).

In situ X-Ray Diffraction Studies of the Electrodeposition of Pb Monolayers on Au(lOO) Single Crystals

K.M. Robinsonl and W.E. O'Gradyl

lGEO-Centers, Inc. Ft. Washington, Md 20744

2Code 6170 Chemistry Division Naval Research Laboratory Washington, DC 20375

ABSTRACT. In situ high intensity synchrotron x-ray radiation has enhanced the study of the electrochemical solid-liquid interface. Recent work on the interfacial structure of Au, Ag and Cu single crystal electrodes has shown the need to merge known electrochemical techniques with x-ray diffraction experiments. In this report, an electrochemical based methodology for use with in situ x-ray diffraction from the electrode/electrolyte interface is applied to the electrodeposition of Pb on Au(100). The cell in this experiment provides control of the electrode potential, as determined by the cyclic-voltammogram, in conjunction with the x-ray studies. Growth of the Pb monolayer begins with diffusion of Pb+2 ions to the surface at potentials below the PZC for the Au(lOO)-(5x20) surface. Electrodeposition causes a lifting of the (5x20) reconstruction with the Pb atoms depositing in a c(2x2) structure. Domain sizes are small due to constriction by small Au islands which are formed when the (5x20) is lifted in the presence of a weakly adsorbing anion. Irreversible surface defects are observed below -032V (SCE). Potentials at which Pb on Pb deposition occurs result in a roughened surface and alloy formation.

1. Introduction

The interfacial structure of Au single crystal electrodes in contact with various electrolytes is fundamentally important to the understanding of electrochemical systems. It has been shown by many techniques that the low index faces of Au reconstruct under potential control [1-9]. In the presence of Pb +2 ions, there is also electrodeposition at potentials anodic (more positive) than bulk Pb deposition [8,10-

157

C. A. MelendresandA. Tadjeddine (eds.), Synchrotron Techniques in Interfacial Electrochemistry 157-169. © 1994 Kluwer Academic Publishers.

158

17]. The cyclic-voltammograms, CY's, of the Pb electrodeposition are very crystaHographically dependent and reflect a large amount of restructuring in the interface and of the surface. The Pb deposition on Au(l00) has been studied by LEED [8,18,19], STM [14-16] and by electrochemical methods [10-12,17]. The results of these studies show that Pb on Au(l00) is a complex system which exhibits restructuring and alloy formation. UHY studies of Pb on Au(l00) show a number of structures brought about by normal and lateral growth of alloys at submonolayer coverage [19]. The CVs of the Pb deposition on Au(l00) contain several transitions with narrow potential widths as weH as a broad transition which has been interpreted as Pb blocking of adsorption sites [12]. Because of these narrow transitions, any study of tbis system must be carried out with uniform potential control over the entire electrode surface. The ceH used in these experiments provides in situ x-ray diffraction capability as weH as good potential control [9].

There have been several systems studied with in situ x-ray scattering techniques which concern fundamental electrochemical systems such as Au, Pt and Cu oxidation [23-26] and Au(l00) surface reconstruction [1], as weH as electrodeposition on Au and Ag (111) faces [13,27,28]. Surface x-ray diffraction is a weH developed technique [21,22]. Details concerning the reconstruction, roughness and faceting of the surface can be determined by measuring the intensity variations along the crystal truncation rods, CfR's [21,22,29,30] and the surface or non-integer scattering rods. The fuH width at half maximum (FWHM) of the CfR's provide information on the surface domain size and the introduction of surface defects. The scattering vector is defined as Q = qz + qll' where ~ = c*l, and q~ = a*h + b*k. The vectors a*, b* and e* are the reciprocallattice vectors for the bulk crystal and h, k, I are the Miller indices. For CfR's, h and k have integer values while the vertical index, I, is not constrained to integer values due to the termination of the bulk crystal. For an abruptly terminated flat surface, the intensity of the rods change as I/li ~ 2, where /i qz is the distance between the bulk Bragg reflections. If the surface termination changes from a flat surface to a rough surface the intensity of the rod decreases more sharply.

Changes in the surface charge or atomic density and spacing, caused by surface reconstructions, also cause changes in the intensity of the CfR's [1,21,30]. By modelling the sum of the interferences from successive layers of the crystalline surface, a reflectivity, R(~), profile based on single layer deviations from an ideal termination can be constructed:

GO :2 am2

R(qz) =IT(qz)2F(qz) e-W(qz) 12 1E Pmeqz""2 eiqzc(m-€.,) 12 (1) m=o

[1,30]. The first part of equation 1 contains the Fresnel transmission coefficient, T(~), for low angle of incidence [21], atomic form factor F(~) and a bulk Debye­Waller factor W (=V2~2<U2». The data reported here are Lorentzian, area and

159

polarization corrected. Correcting the data, which removes the 1/Clz2 term [1], before analysis causes a reduction in the intensity of the (0,0) rod at low I which accounts for the observed differences in the reflectivity profiles reported here compared with those in reference 1. The second part of equation 1 contains the crystalline layer interference terms. The layer density, Pm' is relative to the bulk layer density. The layer fluctuation parameter, a m' acts as a layer dependent Debye-Waller factor. The layer displacement parameter, € m' allows the individual layers to deviate from the average multilayer spacing, c.

2. Transmission Electrochemical Cell

A detailed description of the transmission geometry ce11 has been published [9]. Briefly, the geometry of the ce11 uses the traditional thin layer cell concept, with the counter electrode directly opposite the working electrode, and the reference electrode tip situated in between. However, instead of a thin layer of solution squeezed between the electrodes, a larger drop of solution is suspended from a funnel shaped reservoir between the two electrodes. The larger drop provides clearance for the incident and exit x-ray beams, figure 1. Multiple electrolyte solutions can be introduced to the crystal surface in the ce11 to allow tests with

N 2 I niet ----,

,------------ Reference Electrode

Counter ElectrOde

,---------- Electrolyte Inlet

Water Saturated F i I ter Paper

r-'-'-----'--,~=*====I------ Resevo ir

: '? i ~ j 'f 1 Kapton Window

i __ ~ ... j ~. j f ~ t· i _____ :

Working ElecLrode

Figure 1 Transmission geometry cell for in situ x-ray diffraction. The entire cell is enclosed in a He bag.

160

varying concentrations. The cylindrical shape of the cell provides 36OOrotational symmetry and a 00-900 2a range for the specular reflection measurements.

The crystals used in this cell are 2.5mm in diameter by 5mm long and are held in a co11ette which seals about the base of the crystal. The co11ette mounts directly to the four-circle diffractometer. A groove in the lower half of the ce11 is filled with water saturated filter paper to provide a constant 100% humidity inside the cello This prevents the electrolyte drop from evaporating during the long exposure times required for co11ecting the diffraction data. This technique has kept a 2.5mm diameter cylindrical drop stable for aperiod of two days even while undergoing the multiple rotations of the four-circ1e diffractometer.

There are several advantages to this type of cell compared with the more often used reflection cell, in which a thin layer of electrolyte is trapped between a membrane and the electrode surface. The most important advantage is the ability to perform the electrochemistry in a manner consistent with previously published results. Figure 2 shows the CV for the Au(100) in 0.01M HCI04 in the cell with the x-rays on. This CV was co11ected on the first sweep and is consistent with published CV's [2,5]. In addition to the surface reconstruction worle, the design of the cell has been beneficial in the direct observation of the formation of AuCIOx complexes at the electrode/electrolyte interface [31].

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 V (vs. SeE)

Figure 2 Au(100) cyclicvoltammogram recorded in the transmission ce11 0.01M HCI04, 25 mV /sec.

161

The disadvantage of this geometry is in the large amount of absorption of the x-ray beam by the electrolyte. This generates a large background noise in the x-ray data. The noise can be reduced by operating at higher ineident x-ray energies, 10-11 ke V. There is also a change in the area correction involved in the integration of the data. At low angles, < 15° from the surface, an additional term, sin (2a )/2, must be included in the area correction. This accounts for the fact that the ineident beam completely covers the surface resulting in a loss of intensity in the diffraction data near the surface, I < 0.25. Figure 3 shows in situ x-ray data taken at OAV (SCE) in O.OlM HCI04 and the results of a calculation using equation 1 incorporating the area and Lorentzian corrections. The large asymmetry in the (0,0) data reported in reference 1 is greatly reduced when the area correction is applied.

5

4

1

1.25 1.5 1.752.0 2.252.5

(O,O,L)

Figure 3 In situ x-ray diffraction data and the calculated Au(100)-(5x20). The data are area and Lorentz corrected, which accounts for the lower intensity at low L than is reported elsewhere [1].

3. Experimental

The Au(l00) crystals were cut from a Au boule, bulk mosaic <0.05° (1,1,1), and

162

polished down to 1 micron alumina powder. The crystals were further electropolished in a 1:1:2 mixture of glycerol:ethanol:HCI(conc.). The crystals were then flame annealed according to standard procedures [30] and flushed with distilled water. This procedure has been shown to produce atomically smooth surfaces with a (lx1) surface structure [6,7,9,30]. The crystal was removed from the cell, repolished and the procedures outlined above were repeated to insure reproducibility. Only upon emersion at O.4V (SCE) was an incomplete (5x20) reconstruction observed [9,30]. Fresh O.lM HCI04 and l.OmM Pb(CI04)2 electrolyte, made from Ultrex grade HCI04, Aldrich Pb(CI04)2 • 3H20 and 18Mn-cm distilled water, was deoxygenated with 99.9995% pure N2 and pumped into the reservoir. A small drop, 2.5mm dia., was brought into contact with the electrode surface with the potential held at 0.4 V (SCE). A AuJ AuOx reference electrode was used in the measurements, however the data are reported versus an SCE reference for easy comparison with the literature.

The CV obtained following the above preparation technique is shown in figure 4. The letters indicate potential regions at which the x-ray diffraction data were measured. The result of holding the potential in a particular region will be discussed

2

.-. 1 CO

:2.. -- 0

-1

-2

2

(01 :2..

--0

-1

-2

A)

~C+-B A~ B)

-004 -0.2 0.0 0.2 004 V (vs. SeE)

Figure 4 Au(l00) cyclic voltammograms recorded in the transmission cell, A) first CV O.lM HCI04 + 1 mM Pb(CI04)2 25 mV Jsec, B) same CV after extended time at potentials negative of -0.32V (SCE).

163

later. This CV is identical to those reported in the literature [8,12] for Pb on Au(l00). The CV was recorded on the first cycle and is not a steady-state CV. Multiple cycles were avoided due to the potential alloy formation previously observed by STM [14,15]. The CV clearly supports the high purity achieved in the preparation procedure and that the cell geometry results in an excellent potential and current distribution.

The x-ray diffraction data was collected on the NRL X-23B Materials Analysis Beamline at the National Synchrotron light Source. The energywas fIXed at 9.5keV. This beamline is equipped with a fixed exit monochromator and x-ray optics for focusing a 1mm x 1mm spot at the center of the diffractometer.

4. Results and Discussion

Prior studies of the electrodeposition of Pb on Au(100) showed little more than a surface disorder in the [0,0,1] direction [32]. LEED results predict a c(2x2) structure for a coverage of 0.5 of a monolayer [8,18,19]. Figure 1b shows the expected diffraction pattern for the c(2x2) pattern using the fcc coordinate system and the positions at which data were collected. Because of the similarities of the atomic form factor of Au and Pb, the c(2x2) structure is expected to be very difficult to distinguish between the surface (1,0) position and the (1,1» CfR [33]. The intensity relationship,

(2)

developed for the c(2x2) at the (1,1,0) position [33], predicts a parabolic intensity variation for increasing coverage; however, comparisons of the (0,0,1), (1,1,1) and (2,0,1) CfR's allow adetermination of the Pb structure.

The integrated intensity for the (0,0,1) CfR's for the three potential regions of figure 4 are shown in figure 5. There are considerable differences in the fits associated with each of the CTR's. Table I gives the results from a fit of equation 1 to the (0,0» data. Bach potential region will be discussed separately.

Table I. Fit parameters of equation 1 to the (0,0» data in figure 3.

Potential 0"1 PI EI 0"2 P2 E 2 <U1/2>2

(SCE)

+0.40 - 0.0 0.2 0.60 -0.03 0.1 1.20 0.22 0.08

0.0 - -0.32 0.1 0.50 -0.10 0.1 1.0 0.05 0.08

-0.32 - -0.40 0.2 1.4 0.15 0.1 1.1 -0.60 0.08

164

4

3.5

3

~2.5 'w c ~ 2 c 0)

.3 1.5

1

0.5

o o 0.5 1

.. ~ .. ;:

1.5 2

(O,O,L)

. . • I I I

V

I I

I I

I I

I

2.5 3

Figure 5 (0,0» CfR and fits of equation 1 forthe regions A (. ) B (a, ) and C (., --) of Figure 2.

4.1 Potential Range +O.4V - O.OV (SCE)

Tbe (0,0» profile was best fitted with a model which depicts the (5x20) reconstruction and approximately 0.6 (lx1) coverage of the Pb+2 ions. Tbere is no significant eurrent measured in the CV to suggest that the Pb+2 has undergone electron transfer. In addition, the Pb +2 layer has a large fluctuation term, Table I, which suggests that the Pb +2 is not strongly interacting with the surface which would not be possible ü the Pb +2 had undergone electron transfer and was deposited on the Au surface. This suggests a diffusion of the Pb +2 ions to the surface as the potential is stepped below the pzc, 0.25V (SCE), for Au(100)-(5x20) [3,4]. Observations of the (1,1» and (2,0» CfR's depict a smooth surface termination.

4.2 Potential Range O.OV - -0.32V (SCE)

Tbe (0,0» profile was best fitted with a (lx1) surface with a total surface coverage of Au and electrodeposited Pb of 0.5 (Table I). As the Pb is deposited, the (5x20)

165

reconstruction is lifted and forms Au-Pb surface mosaics on top of the Au (lxI) structure. The behavior of the surface structure in this potential region is consistent with only 25% total coverage of electrodeposited Pb. When the 20% additional Au atoms from the reconstruction are included, this accounts, within error, for the surface fit density of 0.5. The (2,0,/) shows an increase in the surface disarder at low values of 1; however at higher values of I, the crR becomes nearly ideal in profile, consistent with an ideal bulk termination. Rocking curves taken near the critical angle, Figure 6, provide more insight into the surface disorder. Near the surface, the (200) planes are displaced ± 0.450 from the bulk (200) planes. There is no bulk roughening as observed by the (1,1,/) crR. An exact Au-Pb exchange would require a Au/Pb ratio in a single layer of 25/1 to account for a 0.450 distortion, assuming a ratio ofthe radii ofrPb/rAu = 1.2. The amount ofPb electrodeposited gives a surface ratio of Au/Pb = 4.

700

500

300

100

-2

o

+

o

+0/-

o 0

+ + o 0

o 0 o 0

o 0°+

0 0 0

o + o 0

00 0 0 0 +1-+ +

t + + + +~-fIt

-1 o cjJ

1 2

Figure 6 Rocking curves far the (0) (2, 0, 0.1); (+) (2, 0, 0.15) diffraction spots measured at -O.3V (SeE).

There are two possible positions for the excess Pb: 1) AuPb alloys on the surface or 2) a structure that slightly distorts the (lxI) surface. In a c(2x2) configuration, a Pb atom would have to be pressed only 0.02Ä (based on hard sphere calculations) into the (1x1) Au structure to cause the 0.450 distortion. There is no evidence, via alloy diffraction peaks, that there is any alloy formation at this potential, nar is there

166

any irreversibility in the CV that can be associated with alloy formation. Secondly, there is a slight decrease in the integrated intensity at the (1,1,0) surface diffraction consistent with the intensity predicted by equation 2. This suggests that there is a c(2x2) structure. The lack of any (1,0) or (0,1) scattering, associated with the c(2x2), can be accaunted for by the fact that there are Au islands, from the lifted (5x20), occupying 20% of the available surface. It has been shown that these Au atoms form small islands randomly situated on the (lx1) surface [34]. These islands would prevent the growth of large domains of c(2x2) structures and therefore would give weak in plane diffraction. The dynamics of how the Pb is deposited should also play a role in the development of large domain c(2x2) structures. H the lifting of the reconstruction is the trigger that allows electrodeposition to occur, one would expect larger domains ofboth (lx1) Au and c(2x2) Pb. However it would appear that some Pb deposition occurs first which causes a lifting of the (5x20), similar to the preoxidation which causes the lifting of the (5x20) in the cathodic direction. This localized deposition would cause an intermixing of Au and Pb atoms resulting in very small domains of Au(lx1) and Pb c(2x2). Spacings similar to the two Au-Pb stable alloys, AuPb2 and Au2Pb, may form more stable clusters, however no diffraction data similar to the spacings calculated from the alloy structures were observed in the in­plane diffraction.

4.3 Potential Region -0.32V - -0.40V (SCE)

When the potential is stepped below -0.32V(SCE), the observed distortions in the CV become irreversible, figure 4. The (O,OJ) profile is best fitted with a AuPb layer, possibly hexagonal, due to the high density from the fit, on top of a (lx1) surface, also with a slightly higher density, which could possibly be due to Au-Pb interdiffusion. Adense hexagonal overlayer has been observed with LEED for coverages greater than 0.60 [18,35]. There is no in-plane diffraction observed which would allow adetermination of the surface structure. A visual improvement to the fit would allow slight distortions to the third layer, however with the present error associated with the integrated intensity, there is no statistical improvement. Interestingly the (l,lJ) CfR remains unchanged, which means the permanent distortions involve only the first two to three layers. Below the critical angle, the (2,0J) contains many structural features. More data is necessary to fit these complex fluctuations.

4.4 Potential Region -O.4V - -O.5V (SCE)

At potentials cathodic to the irreversible Pb deposition, Pb deposits on Pb and the surface structure changes rapidly. Figure 7 shows the (O,OJ) rod in this region. The only fit to this data gives a highly disordered surface. The surface is so disordered that the (l,lJ) and (2,OJ) rods are not distinguishable from the background. Extra peaks have been observed in the data which are powder rings of the AuPb2 alloy.

6

p5 CI) c ~ 4 c 0)

o 3 --.J

2

1 1 1.5 2

(O,O,L) 2.5 3

Figure 7 The (0, 0, l) CfR for -O.4V to -0.5V (SCE). The new peaks are powder rings for the AuPb2 alloy.

167

An extended period, 15 - 30 minutes, at this potential increased the size of the alloy peaks, which remained after cycling the potential back to +0.4V(SCE). At this point the crystal had to be removed from the cell, etched in a HN03:HCI acid mix, repolished with alumina powder and annealed to regain a smooth, flat single crystal surface.

5. Conclusions

Three distinct stages of Pb electrodeposition on Au(lOO) have been studied by in situ x-ray diffraction. Initially, Pb+2 ions diffuseto the surface as the potential is stepped below the pzc for the Au(100)-(5x20) surface. The (5x20) reconstruction is lifted as electrodeposition occurs and the Pb and Au atoms form a c(2x2) surface on a Au(lOO)-(1xl) surface. The Pb distorts the (lxI) surface planes by ± 0.45°. An irreversible deposition peak at -0.32V(SCE) causes a permanent distortion, and probable Au-Pb atom exchange, after which the Au and Pb form a high density surface, composed of two to·three atomic layers, on the (lxI) surface. AuPb2 alloy forms, probably from the Au-Pb exchange sites, at potentials negative to the

168

reversible Pb deposition potential. Pb deposition produces an extremely disordered surface. These conclusions are in agreement with ex situ lEED studies on Au(100) and STM observations of AuPb alloys on Au(111).

6. Acknowledgements

The authors would like to acknowledge the Research Associate Program of the National Research Council. Additional support was provided by the Office of Naval Research. NSLS is supported by the Department of Energy, Division of Materials Sciences and Division of Chemical Seiences under contract 11 DE-AC02-76HOOO16.

7. References

[1] B.M. Ocko, J.Wang, A Davenport and H.lssacs, Phys.Rev.Letts., 65 (1990) 1466.

[2] D.M. Kolb, J. Schneider, Electrochim. Acta, 31 (1986) 929. [3] M.S. Zei, G. Lehmpfuhl, D.M. Kolb, Surface Seience, 221 (1989) 23. [4] A Hamelin, MJ. Sottomayor, F. Silva, S. Chang, MJ. Weaver, J.

Electroanal.Chem., 295 (1990) 291. [5] S. Strabc, RR Adzic. A Hamelin, J.Electroanal.Chem., 249 (1988) 291. [6] X. Gao, A Hamelin, MJ. Weaver, Phys.Rev.Letts., 67 (1991) 618.

[7] X. Gao, A Hamelin, MJ. Weaver, Phys.Rev.B, 44 (1991) 10983. [8] P. Hagens, PhD. Dissertation, Case Western Reserve University, January

1980. [9] K.M. Robinson, W.E. O'Grady, Rev. Sei. Insts., 64 (1993)1061. [10] RR Adzic, E. Yeager, B.D. Cahan, J.Electrochem.Soc., 121 (1974) 474. [11] J.W. Schultze, D. Diclertmann, Electrochim. Acta, 22 (1976) 489. [12] K. Engelsmann, W. J. Lorenz, E. Schmidt,J. Electroanal.Chem., 114 (1980)

1. [13] M.G. Samant, M.F. Toney, G.L Borges, L Blum, O.R Melroy, J.Phys.Chem.,

92 (1988) 220. [14] M.P. Green, K.J. Hanson, R Carr, I. lindau, J. Electochem. Soc, 137 (1990)

3493. [15] M.P. Green, KJ. Hanson, Surf. Sei. Letts., 259 (1991) L743. [16] NJ. Tao, J. Pan, Y.li, P.1. Oden, J.A DeRose, S.M.lindsay, Surf. Sei. Letts.,

271 (1992) L338. [17] D. A Koos, G.L Richmond, J.Phys. Chem., 96 (1992) 3770. [18] J.P. Biberian, G.E. Rhead, J.Phys.F: Metal Phys., 3 (1973) 675. [19] E. Bauer, Appl. Surf. Sei., 11/12 (1982) 479.

169

[20] For an overview of in situ techniques see: H.D. Abruna, cEd., Electrochemical Interfaces, VCH Publishers, Inc., New York, 1991.

[21] R. Feidenhans'l, Surface Science Reports 10 (1989) 105. [22] I.K. Robinson, in Handbook on Synchrotron Radiation, G.S. Brown, D.E.

Moncton eds., 1991 Elsevier Science, Amsterdam, p.221. [23] K.M. Robinson, W.E. O'Grady, preprint. [24] M. Fleischman, B.W. Mao, J.electroanal.Chem., 229 (1987) 125. [25] C.A Melendres, H. You, V.A Maroni, Z. Nagy, W. Yun, J.Electroanal.Chem.,

297 (1991) 549. [26] G.M. Bommarito, D. Acevedo, H.D. Abruna, J. Phys.Chem., 96 (1992) 3416. [27] M.F. Toney, l.G,. Gordon, M.G. Samant, G.L Borges, O.R. Melroy, L Kau,

D.G. Wiesler, Phys.Rev.B., 42 (1990) 5594. [28] M.F. Toney, l.G. Gordon, M.G. Samant, G.L Borges, D.G. Wiesler, D. Yee,

LB. Sorensen, Langmuir, 7 (1991) 796. [29] I.K. Robinson, Phys. Rev. B. 33 (1986) 3830. [30] K.M. Robinson, I.K. Robinson, W.E. O'Grady, Surface Science, 262 (1992)

387. [31] K.M. Robinson, W.E.O'Grady, To be Published. [32] B. Ocko, private communications. [33] P.W. Stephens, P J. Eng, T. Tse, Surface X-Ray and Neutron Scattering. H.

Zabel, I.K. Robinson, eds., Springer-Verlag, Berlin, (1992) p. 79-82. [34] N. Batina, D.M. Kolb, RJ. Nichols, Langmuir, 8 (1992) 2572. [35] l. Perdereau, l.P. Biberian, G.E. Rhead, J.Phys.F: Metal Phys., 4 (1974) 798.

OXIDATION OF Mo(OOI) SURFACES

I. K. ROBINSON Department of Physics University of minois Urbana, IL 61801, USA

ABSTRACf. X-ray diffraction using synchrotron radiation is now sufficiently sensitive to investigate surface structure in a fairly routine way. Among its advantages is the straightforward kinematic interpretation of the atomic arrangement at the level of O.OlA, where chemically relevant details of bonding become apparent The oxygen-induced reconstruction of Mo(OOl) has recently been studied and sheds light on the mechanism of oxidation.

1. Introduction

A major attraction of the use of photon-based techniques in studies of the electrochemical interface is that photons (optical or hard X-rays) can traverse a macroscopic quantity of liquid en route to the interface. Most other available experimental techniques with surface sensitivity attain that sensitivity by using a non-penetrating probe, electrons for example. These cannot penetrate the liquid of the ideal sampie geometry, and require tricks to be used, such as emersion, with all the accompanying caveats.

Photon-based methods are fundamentally different. They use the special symmetry of the interface for their specificity. Second-harmonic and sum-frequency generation, for example, use the non-inversion symmetry of the third-order dielectric susceptibility to obtain a unique signal from the interface. Surface X-ray diffraction similarly uses the broken translational symmetry of the interface for the same purpose. The diffraction from a 3D crystal is confined to sharp spots, while that of a truly amorphous or liquid ('OD') material is diffused everywhere. A flat interface between a crystal and a liquid has diffraction that has special properties: it is diffuse in the direction perpendicular to the interface, but sharp in the direction parallel to the interface. This makes possible the measurement of the integrated intensity distribution, needed for quantitative structural analysis. This is both straightforward and accurate, even though much greater quantities of crystal and liquid are present at the same time: the isotropically diffuse signal is subtracted as background, while the sharp 3D Bragg peaks can be avoided altogether.

Surface X-ray diffraction is however limited by the same constraints that gives its advantage. In order to establish c1early identifiable diffraction it is necessary for the electrode to be a single crystal and for the electrolyte to be an ideal liquid with a uniform scattering function. Moreover, for the diffraction pattern to be interpreted with certainty, it must be c10se to ideally 2D and unambiguously distinguishable from OD and 3D. This limits us to interfaces that are at most a few atomic layers thick. Some problems of interest in electrochemistry are not so readily idealised. Corrosion of polycrystalline or composite materials with

171

C. A. Melendres anti A. Tadjeddine (eds.), Synchrotron Techniques in lnteifacial Electrochemistry 171-180. © 1994 Kluwer Academic Publishers.

172

complex grain structures would not be easily undertaken. Similarly, not much information could be obtained about thick passivation layers that do not have specific ordered atomic structures.

Before the advent of scanning tunnelling microscopy (STM) and surface X-ray diffraction, few techniques were available to examine the electrochemical interface at the atomic level. It is already apparent that a variety of ordered structures has been found to exist under controlled, reversible preparation conditions. Unlike in the field of surface science where most of the structures were known before the X-ray techniques were developed, the existence of structures at the electrochemical interface is largely unexplored, notably by the present author. For these reasons this paper is about a surface structure in vacuum and not about an electrochemical system. It should serve as a tutorial for the kind of results to be expected from electrochemical interfaces (subject to the restrietions outlined about) that should be uncovered in the next few years.

2. Oxygen on Mo(OOl)

The wide range of oxides known for Mo demonstrate a dramatic variation of coordination and valency in their bulk crystal structures [1]. Given that much less is known about the surface oxides, a good place to start is the Mo(OOl)/O adsorbate system. This is an ideal model for understanding the role of precursors in the oxidation of Mo. Despite the fact that there have been a number of thorough LEED studies [2-10] over the years, not much more than the symmetry is known of the actual structure of these phases. The sequence of symmetries observed is completely different at different temperatures: room temperature (RT) adsorption of 0 on Mo(OOl) gives a consecutive series of ordered structures, c(2x2), followed by 6x2, then 6x1, then 3x1 and finally lxI [6,8,9]. On the other hand, high temperature reaction with 0 (above l000K) gives the series, c(4x4), 2x1, J5xJ5, 2xl, c(2x2), and eventually produces facetted structures [2-10]. The formation of these facetted states has been investigated and found to result in oxide layers [5,7,10] that start at l000L (1 Langmuir = 1O~ Torr sec) and saturate at 5000L [10]. Thermal desorption from the oxide state yields the high temperature series of surface structures in reverse order [6-10]. The likely reason for the variation with temperature is that the Mo atoms are mobile at 1000K so can form structures that involve diffusion in which the stoichiometry can change. At room temperature the number ofMo atoms in the structure must be conserved; most probably the RT structures are arrays of adsorbed 0 atoms with packing density corresponding to the coverage. For this reason it is the high temperature structures that are more interesting from the point of view of studying oxidation.

The oxygen coverages corresponding to each state have been determined by careful quantitative Auger spectroscopy measurements [8-10]. Of the various high-temperature phases, the J5xJ5, which occurs at an 0 coverage of 0.8 monolayers (ML), appears to have the widest stability range. It is also the highest coverage chemisorbed phase that is well­ordered, in the sense of having a sharp LEED pattern. All structures with higher coverages are somewhat diffuse in LEED, a fact that has been attributed to the onset of O-induced facetting [5-10] of the surface, followed by three dimensional oxidation. Our x-ray diffraction

173

study [11] focusses on the structure of the ,fSx,fS phase in order to gain some insight about this transition from the chemisorption to facetting regimes.

3. Experimental Method

The Mo(OOl) single crystal sam pIe was oriented and polished commercially prior to use [12]. It was found to have a slight miscut of 1.14±0.1 0 along a direction 180 away from [010] and a mosaic spread of <0.020 , as determined by X-ray diffraction. It was introduced into an Ultra-High Vacuum (UHV) chamber with a base pressure of 1xIO·IO mbar. LEED and Auger spectroscopy were available [13]. Oxygen was dosed into the chamber with a leak valve. The sam pIe was cleaned by oxidising it at l000K in an oxygen pressure of 1xl0·7 mbar for S min and then flashing to 2200K by electron bombardment from the back [14]. More than 20 cycles of cleaning by oxidation and flashing were performed. Finally, the ,fSx,f5 structure was prepared by stopping the cycling procedure with a flash only to lS00K that resulted in a sharp uniform ,f5x,fS LEED pattern.

k FobsO Fca1c 3

Mo(OOl)jO o

2

1

• h

3

Figure 1. Measured in-plane structure factors for one of the twin .fSx.fS domains; the other domain is omittedfor clarity, but can be generated by retlection in one of the principal axes. The radü of the unshaded half-circles are proportional to the observed amplitude at either extreme of the error bar; the shaded half-circle is the calculated value for the model described below. The grid indicates the size of the bulk reclprocallattice. The circles are positioned in the correct reciprocal space position, similar to the way they appear in the LEED pattern.

174

We made our X-ray diffraction measurements at beamline X16A, which is operated by AT&T Bell Laboratories at the National Synchrotron Ught Source (NSLS), Brookhaven National Laboratory. We used the Fuoss-Robinson design 5-circle diffractometer with its direct coupling to the UHV chamber [13]. Bending magnet radiation was collected and focussed with a toroidal Pt-coated mirror and a wavelength of 1.78Ä was selected with a double-crystal Si(111) monochromator. Tbe Mo crystal was aligned by means of the bulk [101] and [011] reflections. Crystallographic measurements of observed structure factors were made by integration of c...>-scans, followed by background subtraction and correction for Lorentz factor, polarization factor and active sam pie area [16]. We also used the appropriate out-of-plane area correction for the 5-circle mode [11].

A large in-plane dataset was collected at 1=0.1 and a grazing incidence angle of 0.4°. We used the method of Vlieg et al [16] for the 5-circle setting calculation. Tbe diffraction pattern was indexed conventionally for body-centered cubic, using the convention that Miller index 1 is perpendicular to the surface plane. Tbe diffraction from the ../5 x../5 superstructure of the reconstruction had in-plane indices hk which were multiples of 1/5, while 1 was a continuous variable. Tbe plane group symmetry of the ../5x../5 superstructure is p4, since it has no mirror planes. On a substrate with p4m symmetry two possible orientations of the superstructure can arise, called "twin domains". No significant difference between the intensities of the two twin domains was detected, however. Tbe eight p4m symmetry equivalents, four from each twin domain, of each reflection were therefore measured and averaged together. Tbe variation among them was used as an error estimate. Figure 1 shows the total of 29 independent fractional-order and 4 integer-order reflections that were collected altogether. 44 hours later some reflections were rechecked and found to have become somewhat larger in intensity, but still within the 10% average error estimate. Tbis measurement period was assumed to be the effective sam pie lifetime at 10.10 Torr.

A second preparation of the sam pie was then used to measure the I-dependence of four reflections, this time with four symmetry equivalents each [11]. Here grazing exit conditions [16] were used with an exit angle of 2°. Tbe two datasets merged with a scale factor of 1.5 that was found to match up the 1=0.1 points.

4. Results and Analysis

Tbe traditional way to start off a crystallographic analysis is the calculate the Patterson or pair correlation function. Tbis has been widely used in surface crystallographic analysis as weil [15]. Tbe Patterson, which is the Fourier transform of the observed structure factors, is a map of the structure in real-space. It is not however a map of the atomic density, but of the correlations in the atomic density. Instead of the atom positions, it shows the interatomic vectors, which are then used to guess a limited number of possible atomic models, which can be tested directly with the data. Tbe 2-dimensional Patterson function for our fractional-order in-plane da ta is shown in figure 2(a). Tbe origin has been suppressed from the plot, as have the peaks that would appear at bulk interatomic vectors, because of the omission of the integer-order data.

175

Figure 2(a) Contour map of the Patterson function of the in-plane fractional-order data. The origin peak has been suppressed. The indicated peak identifies the fundamental interatomic vector in the structure. Only positive contours are shown with equal spacing. Figure 2(b) Two atomic configurations consistent with the position of the observed Patterson peak.

The Patterson has a single, dear positive peak that identifies the principal interatomic vector in the structure. This is presumably a Mo-Mo vector, since 0 is a relatively weak scatterer of x-rays. Figure 2(b) shows two guesses of possible spatial arrangements of atoms consistent with the Patterson. A structure factor calculation was then carried out assuming p4 plane group symmetry as explained above. U sing the models of figure 2(b) for starting parameters and allowing two coordinates plus a thermal vibration parameter to vary until they reached optimal values, the calculation gave the X2 values of 16 and 37 indicated in figure 2. Clearly the 4-atom structure is the correct choice.

The secondary peaks of the Patterson and a Fourier difference map [15] were used to identify potential oxygen sites. The only site that supported refinement is the one shown in figure 3 that resulted in a X2 value of 3.98 upon optimisation of all parameters. The agreement between calculated and observed data is illustrated in figure 1. This model gives a structure with 4 Mo and 4 0 atoms in a unit cell with 5 times the bulk 1 xl area, hence a model oxygen coverage of 4/5 or O.8ML This agrees with the best determination of 0 coverage by Auger spectroscopy [8]: this gave a value of O.66-0.8ML with the brightest

176

LEED pattern at 0.8ML Other studies claim 0.72-0.86ML [9] and 0.45-O.83ML [10].

Next, the registry of the structure with respect to the bulk crystal below was determined using the integer-order data in the structure factor ca1culation. The integer-orders inc1ude the amplitude and phase information of the terminated bulk, so give strong interference with the surface structure factor components [17]. One parameter was adjusted here, the "surface fraction", which represents the fraction of the surface that is reconstructed, assuming the rest to be disordered or simple lxI, which had a value of 93±2% here.

Mo(OOl )/0

lxl Uni Cell

Figure 3. The Mo(OOl)/O ';5x';5 structure. One top-layer Mo site out of five per unit cell (box) is vacant Four O-atorns line the corner of the vacancy that is created. These are trivalently bonded to two Mo's in the surface plane and one in the layer below.

We attempted to look for the presence of second-Iayer displacements. These are widely present in surface structures, and are a particularly important component of the clean reconstructed Mo(OOl) and W(OOl) surfaces [18]. While this is in principle possible with in­plane data [19], this important question is best answered with the out-of-plane data [11]. A relatively wide range of perpendicular momentum transfer was covered, but only a very gentle downward trend was observed for all the structure factors considered [11]. This informs us that we have a single layer structure, a fact that is confrrmed by simultaneous refinement of all data with a 3D model, leading to a final X2 value of 1.62 and gave the parameters in table 1. At the 20 level, where 0 = O.OO4Ä, the second-layer Mo displacements are seen to be insignificant.

177

TABLE 1. Final structural parameters for the Mo(001)10 JSxJS structure. The x- and y­coordinates of the atomic positions are given as fractions of the JSxJS unit cell, which has an edge length of 7.02A. Bach atom appears four times in the unit cell at positions obtained by 4-fold rotation about the origin. MOl and M<>z denote flfSt and second layer atoms respectively; there is a fifth atom in the second layer at (0, 0). The parameter errors obtained by least-squares fitting are given in parentheses.

Atom

o

x-coordinate

0.2673(4)

0.1722(27)

0.2006(7)

y-coordinate

0.1030(4)

0.3880(25)

0.3988(6)

Debye-Waller

3.36(23)Ä2

0.10(57)Ä2

The table gives us the lengths of two indeyendent Mo-O bond lengths in the surface plane. These values, 2.12±0.02Ä and 2.12±0.02A, fall in the middle of the range of values seen for bulk Mo oxides [1]. The overall layout in figure 3 shows that there is a third bond from the o to the layer below. Unfortunate1y our teehnique is very insensitive to this length: we ean merely eonclude that the third Mo-O bond is 1.6±1Ä. The ehoiee of a 3-fold site would suggest three equal bonds (of2.12Ä), and this is eonsistent with our error estimate. The Mo­Mo bondlength is 2.851 ±0.005Ä, whieh is 4.6% longer than the bulk bondlength (2.725Ä), but 9.4% shorter than the separation of these atoms in an unreeonstrueted 1 xl surface (3.147Ä). It is interesting to note that the W-Wbondlength on the clean W(OOl) surfaee [18] (2.84Ä) is 3.8% longer than its bulk and that the Mo-Mo bondlength in the clean Mo(OOl) surface [18] (2.85Ä) is also 4.5% longer.

5. Discussion

Square clusters in a close-paeked array is a general summary deseription of the Mo(OOl )/0 strueture. Eaeh cluster eontains a eore of four Mo's and four O's decorating the edges. In some ways the Mo(OOl )/0 structure is similar to the clean W(OOl) and Mo(OOl) surfaees: they all have inereased eoordination for the surface atoms, either from clusters or the tendency to form ehains. The M040 4 cluster is apparently the elementary building block of the ,fSx,f5 strueture. The closest nonoverlapping paeking of these is the ,fSx,fs arrangement as figure 4 shows. Isolated clusters would align themselves with the substrate mirror planes (as shown in figure 4), but when they paek together into the ,f5x,f5 cell whieh has no mirror symmetry, they are free to rotate, and ehoose to do so by S.5°. The rotation is in the direetion favouring eontaet between the Mo of one cluster and the neighbouring cluster's 0, thus improving the paeking by partial bond formation. The rotation leads to a strueture with no mirror symmetry, reminiseent of the p4g strueture of N/Ni(l 00) [20], but different beeause of the Mo vacancy. Even though the two in-plane Mo-O bonds are within error of the same length, the 0 site is still asymmetrie beeause the bond angles are unequal.

178

The packing of the M040 4 clusters is not plane-filling and leaves one vacant site per ';5x';5 cell. This vacancy exposes a third layer substrate atom (shaded in figure 3). The temperature of formation of this structure must therefore be sufficient for lateral diffusion of vacancies to step edges to occur, so that an ordered 2D crystal can form. The ';5x';S phase is not seen at room temperature, also suggesting Mo diffusion is important in its formation. Along the oxidation pathway referred to in the introduction, the vacancy formation, exposing the third layer, may be a critical step towards oxygen-induced facetting of Mo(001). The decoration of the vacancy edges with 0 atoms lying in the plane of the top layer is analogous to the lining of missing rows in Cu(100)/0 [21] and Cu(110)/0 [22], and may playa similar role in the oxidation pathway of cop per.

Figure 4. Possible packing arrangements of MOP4 clusters on an unreconstrueted Mo(OOl) plane. If the plane is filled with clusters with relative positions A and B the ";5 x";S structure (figure 3) is generated, which has the highest packing density, but lower p4 symmetry. Relative positions Band C would lead to a p4m symmetrie 'c4x4' strueture with lower density. A 3x3 structure of sti11lower density is obtained by repetition of the AC paeking.

In the case of the clean ';2x';2 reconstructed W(001) surface, the building block is a zig-zag chain of bonded W's. When chains pack together the interaction is mediated by a significant second layer displacement with 25% of the magnitude of the top layer [18]; there is no direct contact between the chains. Clean Mo(001) also has second-Iayer displacements, but the magnitudes are sm aller [18]. For Mo (001 )/0 the situation isreversed: instead of the second layer displacement there is a direct attractive interaction between the Mo of one cluster and the 0 of the next (see figure 4), as manifested in the S.So rotation.

The description of the Mo(001 )/0 ';Sx';S surface in terms of M040 4 clusters in figure 4 allows other plausible structures to be genera ted by changing the packing. The figure suggests arrangements that lead to a p4m symmetrie c4x4 structure with lower density than

179

the ./Sx./S and a 3x3 strueture of stilliower density. Since a c4x4 is observed in LEED at lower 0 eoverage than the ./Sx./S [8-10], it is tempting to speeulate that it too is composed of M040 4 clusters in the manner indieated. From this model the expeeted coverage is 0.5, while the observed ranges are 0.2-0.6S (best at 0.4) [8], 0.2-0.5 [9] and 0.3-0.4 [10]. While this can be considered good agreement, the lower coverages could correspond to disorder in the form of further missing clusters.

6. Conclusion

We have identified the atomie strueture of the Mo(OOl )/0 ./Sx./S surface to be a cluster­vaeancy model. This is a new type of adsorbate-indueed reconstruetion in whieh the top metallayer is pinehed together into clusters. The covalently bonded M040 4 clusters overlap each other suffieiently to give rise to a 5.5° rotation. The 0 binding site is an unusual asymmetrie 3-fold hollow site at the edge of each cluster. An alternative, lower density paeking of clusters is a predietion of the structure ofthe lower coverage c4x4 phase, that will be investigated in a future experiment.

7. Acknowledgement

I am grateful to Detlef Smilgies and Peter Eng far collaboration and to AT&T Bell Laboratories for use of beamline X16A NSLS is supparted by the U.S. Department of Energy under grant DE-AC012-76CHOOO16. Support also came from the University of Illinois Materials Research Laboratory under grant DEFG02-91ER4S439.

8. References

[1] AF. WeHs, "Struetural Inorganic Chemistry", Oxford U niversity Press (1985); R. W.G. Wyekoff,"Crystal Struetures", Vol. 2 (John Wiley & Sons, New York, 1964).

[2] K. Hayek, H.B. Farnsworth, and R.L Park, Surf. Sei. 10 429 (1968).

[3] H.K.A Kan and S. Feuerstein, J. Chem. Phys. 503818 (1%9).

[4] G.W. Dooley m and T.W. Haas, J; Chem Phys. 52461 (1970).

[S] D. Tabar and J.M. Wilson, J. Crystal Growth 9 60 (1971).

[6] R. Riwan, C. Guillot, and J. Paigne, Surf. Sei. 47 183 (197S).

[7] H.M. Kennett and AB. Lee, Surf. Sci. 48 606 (1975); ibid. 624

180

[8] E. Bauer and H. Poppa, Surfaee Seience 88 31 (1979).

[9] E.I. Ko and RJ. Madix, Surf. Sei. 109221 (1981).

[10] C. Zhang, MA van Hove and GA Somorjai, Surf. Sci. 149326 (1985).

[11] IX Robinson, D.M. Smilgies and P.J. Eng, J. Phys. Condensed Matter 4 5845 (1992)

[12] E. Hulpke and D.M. Smilgies, Phys. Rev. B42 1260 (1991).

[13] P.H. Fuoss and I.K Robinson, Nuclear Instruments and Methods 222 171 (1984)

[14] RJ. Musket et aL, AppL Surf. Sei. 10 143 (1982).

[15] LK Robinson in Handbook on Synchrotron Radiation, voL IH, ed. D.E. Moneton and G.S. Brown (Elsevier, North-Holland, 1990).

[16] E. Vlieg, J.F. van der Veen, J.E. Macdonald and M. Miller, J. AppL Cryst. 20 330 (1987).

[17] R Feidenhans'l, M. Nielsen, F. Grey, RL Johnson and LK Robinson, Surface Seience 186 499 (1987).

[18] M.S. Altman, P.J. Estrup and IX Robinson, Physical Review B 38 5211 (1988); D.M. Smilgies, P.J. Eng and LK Robinson, Phys. Rev. Lett. 70 1291 (1993)

[19] LK Robinson, J. Bohr, R Feidenhans'l, M. Nielsen, F. Grey, RL Johnson, Surface Science 217 U35 (1989).

[20] L Wenzel, D. Arvantis, W. Daum, H.H. Rotermund, J. Stöhr, K Babersehke and H. !bach, Phys. Rev. B 36 7689 (1987).

[21] H.C. Zeng, RA McFarlane and KAR Mitchell, Surf. Sei. 208 L7 (1989); M. Wuttig, R Franchy and H. Ibach, Surf. Sei 213 103 (1989); LK Robinson, E. Vlieg and S. Ferrer, Physieal Review B 42 6954 (1990).

[22] R Feidenhans'l, F. Grey, RL Johnson, S.G.J. Moehrie, J. Bohr and M. Nielsen, Phys. Rev. B 41 5420 (1990).

EXTENDED X-RA Y ABSORYTION FINE STRUCTURE: PHYSICAL PRINCIPLES AND DATA-ANALYSIS.

D.C. KONINGSBERGER Debye Institute Utrecht University P.O. Box 80083 3508 TB Utrecht The Netherlands

ABSTRACT. In this paper the physical principles of EXAFS spectroscopy will be given. The method­ology of EXAFS data-analysis is discussed with a special emphasis on how to detect a low Z scatterer in an EXAFS spectrum of a particular sampIe, which is dominated by high Z scatterers. This problem is typical for strUctural investigations on platin um e1ectrodes in fuel cells, where Pt atoms on the surface of the plat­inum meta! particles interact with substrate atoms of the electrolyte.

1. Introduction

Extended X-ray Absorption Eine Structure (EXAFS) spectroscopy is a powerfuI technique to characterize aII forms of matter irrespective of their degree of crysta!linity. EXAFS spectrosco­py probes the Iocal structure of a material [1]. Traditionally, diffraction techniques (XRD, neu­tron diffraction, LEED) are being used for most of the structuraJ investigations and reliable structures can be determined for materials that exhibit a long-range structural order. To study the local structure of highly disordered solids, amorphous materials, liquids and small particles (diameter smaller than 40 Ä) EXAFS has to be used. EXAFS can be applied fOT structural in­vestigations in coordination chemistry, catalysis, bio-inorganic chemistry, biology, surface phys­ics and chemistry [2]. One of the major advantages of EXAFS is its atomic selectivity which enables the investigation of the Iocal structure of each different constituent of a sampie. A very important property of EXAFS spectroscopy is that experiments can be performed in-siru. Tbe recent availability of high-brightness synchrotron radiation sources has resulted in a prosperous development of EXAFS spectroscopy.

Tbis chapter fumishes abrief discussion of the concepts of EXAFS spectroscopy foIIowed by a more detailed discussion of the methods used for analyzing EXAFS data. Tbe EXAFS data­analysis will be illustrated with a study of the structure of very smaII (diameter< 8 Ä) platinum meta! particles dispersed on a high surface area zeohte support [3]. Tbis example was chosen since it mimics the problems encountered in the IocaI structure determination of platin um par­tieles dispersed on high surface area carbon. Tbis kind of material is being applied as electrode in fuel cells.

181

C. A. Melendres andA. Tadjeddine (eds.), Synchrotron Techniques in Interfacial Electrochemistry 181-198. © 1994 Kluwer Academic Publishers.

182

2. Principles of EXAFS

When a monochromatic X-ray beam passes through a medium, its intensity is attenuated expo­nentially according to the relation

(1)

where 10 is the number of the incident photons, I.is the number of the transmitted photons, J.I. is the linear x-ray absorption coefficient, x the thickness of the sampie and E is the energy of the X-ray photons. Figure 1 shows the X-ray absorption spectrum of a 25 J.I. thick rhodium foiL The

1.2 ,------------,

c o aO.8 ... o ." ~ 0.6

>0-E 0.4 X

0.2

o~~----~----~----~ 23 23.5 24 24.5

Energy (keV)

Figure 1. X-ray absorption spectrum of a rhodium foil, measured at RT.

initial monotonically decreasing part of the spec­trum is due to the interaction of X-rays with the outer shell electrons. When the X-ray energy hv is equal to or slightly larger than the binding en­ergy Eo the bound electron is emitted by the pho­toelectric process and a sharp increase in the absorption spectrum is observed. This is called the absorption edge and its position is character­istic for a particular type of absorber atom (for the rhodium K-edge, Eo = 23219 eV). For pho­ton energies larger than the binding energy, the X-ray absorption coefficient shows aseries of oscillations which may extend up to 2000 eV above an absorption edge. The oscillations from about 40 eV above the absorption edge are called EXAFS and they contain information about the local structure around the absorbing atom. The origin of the fme structure is outlined below.

In the photoelectric process, the energy of an X-ray photon is completely absorbed by a core electron, whieh in turn is promoted to an unoccupied valenee level or to the continuum. The ki­netie energy Ek of the emitted photoeleetron is

Et=hv-Eo• (2)

The ejeeted photoelectron can be desrcribed as an outgoing spherical wave having a wave­length Ä. defmed by the Broglie relation:

Ä. = 27t1k, (3)

where k is the photoelectron wave vector given by

(4)

183

Here is m the electron mass and h is Planck's constant. The probability of photoelectric absorp­tion for a K-shell electron is given by [4]

(5)

where I"'i) is the initial state wave function of the bound electron.l"'j) is the fmal state wave

Figure 2. Outgoing photoelectron wave scattered by surrounding atoms.

In phase

function of the ejected photoelectron and p(E,) is the density of the a1lowed states at the final state energy Er E is the electric field polarization vector of the X-ray pho­tons. r is the position vector of the scatterer with respect to the absorber atom. e is the electronic charge. co is the angular frequency of the photon and c is the speed of light. The x-ray absorption coefficient J.1 is proportional to the transition probability P. For an iso­lated atom, J.1 decreases smoothly at energies above the edge energy and does not show fme structure. This is the case for a mono­atomic gas such as krypton. However. when the absorbing atom is sUITounded by other atoms. as in a molecular gas or any con­densed phase. the outgoing photoelectron wave is backscattered by the surrounding atoms. as shown in Figure 2. Hence the final state wave function I",,) has two contribu-

)

Out of phase

Figure 3. Constructive and destructive interference of outgoing and backscattered wave.

184

tions, 1'I")OU'IOing and 1'I")backscaaercd· Tbe backscattered part of the wave interferes with the outgo­ing part creating an interference pattern. As the photon energy changes, the energy and thus the wavelength of the ejected photoelectron changes, causing constructive (Figure 3 left) or destruc­tive (Figure 3 right) interference at the absorber atom. Tbis interference modulates the fmal state wave function and thus the transition probability (see equation 5) of reaching this final state. Since the absorption coefficient is proportional to the transition probability, J.I. is modulated by the interference pattern as weIl. Tbe constructive interference as shown in Figure 3 leads to a maximum in the absorption coefficient, while the destructive interference results in a minimum in the absorption coefficient.

Tbe absorption coefficient above the absorption edge is defined as

J.I.(k) = J.l.o(k){ I + X(k)} (6)

where ~(k) represents the smooth varying (atomic) background and X(k) represents the oscilla­tory part of the total absorption coefficient J.I.(k). Since the oscillations are based upon the inter­ference between the outgoing and scattered wave and since this interference is detennined by the local atomic configuration, the fine structure X(k) contains information about the local structure around the ionized atoms. In general, the amplitude and the phase of the fine structure contain information about the coordination number, disorder and bond distances.

An accurate theoretical description of EXAFS spectra includes curved wave effects and an en­ergy dependent self-energy. Tbe curved wave EXAFS formula [ref 5-7] can be reduced to the planowave EXAFS formula (equation 7 and 8) without 1055 of accuracy, if the model function used in the data-analysis procedure is based upon the use of a reference absorber backscatterer coordination (obtained experimentally or theoretically ) having approximately the same distance as the coordination in the unknown material

Sh<l/s

X(k) = L Aj{k)sin [2kRj + <Jlj(k)] (7) pI

with Rj the coordination distance and cpj(k) the phase function determined by both the absorber and backscatterer. Tbe amplitude Aik) can be expressed as

S~e-2Rl). -2a'Jcl Aj(k) = • NjFj(k) • e }

kR 2 J

(8)

damping scattering disorder power

Tbe first group of terms describes the damping of the electron wave: the exponential term e·2J/jA

accounts for the fmite lifetime of the excited state and So2(k) is an amplitude reduction factor which accounts for the photo-electron energy loss due to many body effects and shake-up/shake­off processes in the absorber atom (not all the energy of the impinging photon is transferred to the photo-electron). Tbe second group of terms is characteristic for the scattering power experi­enced by the reflected electron wave: Fik) is the back-scattering amplitude of the neighboring atoms and Nj is the average coordination number in the jth shell. Tbe third term contains the Debye-Waller factor Oj which represents the root mean square fluctuation in Rj caused by ther­mal motion of the atoms and the structural disorder present in the material.

185

0.5

CD '1:::J

" a :E 0.4

-10 '''. e. .. E \ = ... ', :E C[ 0.3

\ (1)

\,. c:n

CD c -.: \ (1)

! 0.2 \ tU ' ... s:. a. tU \ ()

" (1)

" ......... oll:

-15 .... ~ 0.1 .. ;. .'., IXI

0 5 10 15 5 10 15

k (1/A) k (1/A)

Figure 4. a) Phase shift (cp(k» and b) Backscattering amplitude (F{k)) of Pt- Pt (solid fine) and Pt- 0 (dotted fine). Data were obtained from first shell I:SXAFS data of Pt foil and NII:zPt(OH) •.

Expression (7) and (8) are being used to analyze EXAFS data. The final goal is to detennine from the experimental data the coordination distance Rr the coordination number ~ and the Debye-Waller factor C1j for each individual coordination shell. EXAFS can distinguish between different types of neighboring atoms. As an example the phase shift of the Pt-Pt and Pt-O absorber-backscanerer pair and the backscanering amplitude of 0 and Pt are ploned in Figure 4a and b, respectively. It can be seen that both the Pt-Pt and the Pt-O phase are a non-linear func­tion of k. The backscanering amplitude of oxygen (typical low Z scanerer) is decaying rapidly with increasing values of k, whereas the amplitude of platinum (typical high Z scanerer) is sig­nificant at high values of k.

3. Data Analysis

3.1 EX1RACTION OF THE EXAFS FUNCTION

The first step in the analyses ofaXAFS spectrum is to separate the XAFS oscillations from the experimentally measured X-ray absorption coefficient. The necessary stcps for the extraction of the EXAFS function are: (i) pre-edge subtraction, (ii) determination of inner potential value, (iii) post-edge background subtraction, (iv) nonnalisation [8]. A standard procedure to remove the pre-edge absorption involves fitting the spectrum in the energy range sufficiently below the ab­sorption edge (typically - 200 to - 40 eV) with a polynomial. The fined polynomial is extrapo­lated past the edge and subtracted from the entire spectrum. This step removes the absorption due to all other electrons and isolates the contribution of a particular absorption edge from the total absorption. Tbe X-ray absorption spectrum of the rhodium K-edge with the extrapolated pre-edgc is shown in Figure 5a.

186

1.2

0.1 b

c: 0.6 0

ä 0.05 ~ 0.6

.lJ ce >- 0.4 CI!

:x 0 --l'o! -X 0.2 -0.05

0

-0.2 23 24 24 25

-0.1 5 10 15

Energy (keV) k (1/A)

40

c t/) 0.1 d LL

E 30 -0 -eil c: CI! ~ 20 -CI! ;:: ::J

.f 10

ce >< w

0.05 "Gi s:. tn

üi D ;: 'tI .!! 111 Ö ..(J.D5 .!!!

-0.1 2 4 6 6 10 5 10 15

R CA) k(l/A)

Figure 5. Data processing and first shell Fourier filtering of rhodium foil EXAFS data a) Pre-edge extrapolated X-ray absorption spectrum b) Normalized EXAFS spectrum c) Fourier transform d) Isolated EXAFS function for the first shell.

For EXAFS data analysis the smoothly varying post-edge background Ilo is removed by fit­tingthe post-edge region (typically starting at 20 eV above the edge) with a spline function and subtracting it from the data [9]. The EXAFS function is then normaIized by dividing the data through the edge jump J.lo at the absorption edge [8]. The normaIized EXAFS function is con­verted to k-space by using equation (4). In the initial stage of analysis. the value of Eo is chosen at the maximum derivative point on the edge (see further section 3.3). The normalized EXAFS spectrum ofthe rhodium foil is shown in Figure 5(b).

187

3.2 FOURIER FILTERING: SEPARATION OF COORDINATION SHELLS

As discussed above the EXAFS function shown in Figure 5(b) is a superposition of an unknown number of coordination shelIs. Fourier transformation of X.(k) yields peaks in r-space corre­sponding to individual coordination shells around the absorbing atom [8]:

(9)

These peaks are shifted to lower r-values from the real interatomic distances due to the influence of the phase shift function. The weight factor /(' can be used to compensate for the decay in am­plitude of the spectrum with increasing values of k or to emphasize a particular part of the EX­AFS spectrum. The Fourier transform of the rhodium foil EXAFS spectrum is shown in Figure 5(c).

The EXAFS function of a particular coordination shell can now be isolated by applying a win­dow in r-space and performing an inverse Fourier transform to k-space [8]:

(10)

This step produces a modulated sinusoidal function with amplitude A(k) and argument ct>j(k) = 2kRj + cpj(k) ,as given in equation (7). This means that A(k) and ct>(k) for a particular coordination shell can be obtained from an EXAFS experiment. Overlapping coordination shells have to be isolated by inverse Fourier transformation and a multiple shell non-linear least square fitting routine has to be applied. The structural parameters N, R and ilcr have to be re­fmed for every contribution (sheIl). The isolated EXAFS function for the first shell of rhodium foil is shown in Figure 5(d).

3.3 NON-LINEAR LEAST SQUARE MULTIPLE SHELL FITTING

To extract these coordination parameters from the EXAFS data the quantities cp(k), So2, e(·2M.)and F(k) has to be known. The phase CP(k) and backscattering amplitude F(k) of an absorber­backscatterer pair can be extracted by Fourier fUtering of an EXAFS spectrum from a reference compound in which the contributions are weIl separated, or calculated from first principles. To detennine F(k) one still has to know input values for So2 and e(·2M.). To circumvent this require­ment we assurne that not only cp(k) and F(k) but also So2 and e(·2M.) are transferable from one compound to another. The validity of this assumption has been shown for compounds with the same absorber-backscatterer pair [10,11] but even for absorbers or backscatterers which are neighbours in the periodic table [12,13]. This leads to the definition of a modified backscatter­ing amplitude F'(k) [3]:

(11)

This modified backscattering amplitude contains the Debye-Waller factor of the reference com­pound. Hence, the Debye-Waller factor (ilc:r) obtained from the data-analysis is relative to the

188

0.1

j 0.05

.- I

.:.: 0 1<

-0.05

-0.1

5

a

10 15 20 k (1/A)

75

E ... 0 -I/)

c 50 ~ ... CI) ;: ::s .f 25

b

2 345 678 R(A)

Figure 6 a) EXAFS spectra of rhodium foil (20 m thickness) measured at SSRL (solid line: liquid nitrogen temperature, dotted line: room temperature). b) Fourier transform 01 spectra shown in a. (solid fine, Ale 2.8 - 19.4 A-'; dotted line, Ale 2.6 - 15.8 A-').

Debye-Waller factor of the reference compound. The function to be minimized with a non­linear least square refinement becomes:

(12)

Here k' is the photo electron wave vector corrected for the difference in inner potential (6EJ be­tween the sampie and the reference compound:

(13)

As Mo is a fitting parameter, the values of k', and consequently F'(k') and <p(k') are subject to change during a refmement.

The reliability of the EXAFS analysis depends upon the transferability of the reference phase shift and amplitude functions [10-13]. Great care must be taken in selecting the reference com­pounds for providing the reference amplitude and phase functions. To obtain the highest accura­cy for the EXAFS coordination parameters it is imperative to measure the EXAFS spectra of reference compounds at liquid nitrogen temperature. In most cases the dynamic part of the De­bye Waller factor increases strongly in going from liquid nitrogen to room temperature. The amplitude of the EXAFS spectra measured at liquid nitrogen is higher with increasing k-values (compare in Figure 6a,solid with dotted line) which gives the oppDrtunity to apply a forward Fourier transform over a larger range in k-space. This leads to peaks in a Fourier transform which have a smaller width and higher amplitude (compare also in Figure 6b solid line with doned line). Iftheoretically calculated functions are used, they should be optimized by testing them on reference compounds [14].

189

The accuracy of the structural parameters obtained from EXAFS analysis depends upon many factors such as the quality of the data, the length of the data range, the amount of disorder, the choice of the reference compound and the complexity of the system being studied. Typical ac­curacies are 1 % for the determination of coordination distances, 10 to 20% for the coordination numbers and 10 to 20% for Debye-Waller factors.

0.08 10

a C E 0.04 E 5 ... ... 0 0 - in CI)

c C 111 ca i= 0 i= 0 ... ... u u

-.:: -.:: ::2 ::2

~ -0.04 0

-5 u.

-10 -0.08 1 2 3 4 1 2 3 4 R (A) R (A)

100

E 50 ... 0 I 1\

d

-ltJ C ca i= 0 ... u ;: ::2 0

-50 u.

-iI'i ~

~I I tu"-

~ ,

\I V

-1 L-____ ~~ __ ~~ __ ~ __ ~

1 234 -100

1 2 3 4 R (A) R (A)

Figure 7. Fourier transforms with &:: 3.1 - 18.6 k'. a) k'; b) k', Pt-Pt phase and arnpl. corr.; c) k"; d) k" , Pt-Pt phase and amplitude correction.

190

3.4. PHASE- AND AMPLITUDE CORRECTED FOURIER TRANSFORMS. USE OF IMAGINARY PART.

As pointed out in section 3.2 and shown in Figure 4 the phase factor <p.(k) is a non-linear func-• J

tion of k and the backscattenng amplitude F(k) is a function of k. This implies that the Fourier transformation of an EXAFS function does not lead to an optical transform (a transform of func­tion with constant phase and amplitude). GeneraIly, the peaks of the Fourier transform of an EXAFS function are asymmetric. For high Z elements the k-dependence of the phase and the backscattering amplitude may even lead to the appearance of multiple peaks in the Fourier trans­form of an EXAFS function describing a single absorber-backscatterer pair. In the past this has led to a lot of confusion in analyzing EXAFS data containing contributions of high Z elements [15]. A normal Fourier transform can be converted to an optical transform by removing the phase function and the backscattering amplitude. A phase- and amplitude-corrected (optical) Fourier transform can be obtained by transformation of the following function [15,16]:

(14)

Figure 7 shows anormal k' and J2 weighed Fourier transform of a single Pt-Pt EXAFS function calculated with N=7, R=2.77, ~<f=O.OO3 and t:.Eo=O. The phase shift and backscattering func­rion were obtained from EXAFS measurements on Pt foil (thickness 4J.l.) measured at liquid ni­trogen temperature. Since the Fourier transformation as defmed in equation (9) is complex it is possible to calculate both the magnitude (see Figure 7 solid line) and the imaginary part (dotted line) of the transform. It can be seen in Figure 7a that the magnitude of anormal k J weighed Fourier transform of a single Pt-Pt contribution is split in three peaks, whereas the imaginary part shows a complex behavior as a function of r. The applied weight factor /t influences the k­dependence of the amplitude of the /tX(k) function. The magnitude of anormal J2 weighed Fourier transform (Figure 7c) appears to be a more symmetric peak with a small sidelobe. Ap­plication of a phase- and amplitude-corrected Fourier transform leads in all cases (Figure 7b and d) to a single peak with a symmetrical imaginary part, having its maximum in the top of the magnitude at the right coordination distance.

The use of optical Fourier transforms can be of great help [15,17] in the identification of dif­ferent types of neighbors by application of different phase- andlor amplitude corrections. A Fourier transform, phase- and amplitude-corrected for an X-Y absorber-backscatterer pair must have a positive imaginary part peaking at the maximum of its magnitude if the EXAFS function indeed originates from an X-Y pair. Fourier transform peaks which are not symmetrical are a superposition of more than one contribution or have been analysed with the wrong absorber­backscatterer pair.

3.5 DETECTION OF LOW Z SCA TTERERS

It is tempting to apply a 12 or k' weighing to the EXAFS spectrum during the fitting in k-space to compensate for the decay in amplitude of the spectrum with k. Also Fourier transformation of a function which has an equalized amplitude results in less broadened peaks, which are easier to filter for inverse transformation. However, applying a 12 or k' emphasizes the high Z contribu­rions to the spectrum since high Z elements have more scattering power at high values of k than low Z elements, whereas low Z elements have the highest amplitude at low values of k (see

191

0.12 a

0.01 -~

N 0 .. -0.01

I' \'

·0.12 3 5 7 I 11 13

k (A' ')

0.08 b r I \

~ , \

0.08 E .. 0 ii c: 0.04 • .. ... .. ~ 0.02 ;; 0 I&-

0 0 2 3 4 5

R (A)

I c -. ~

E 4 ; ii c: • ;; .. 2 ~ ;; 0 I&-

0 0 2 3 4 5

R (A)

Figure 8. Simulated EXAFS spectra of Pt-Pt (solid line), Pt-O (dotted line) and Pt-Pt + Pt-O (dashed line). a) EXAFS functions, b) k' weighed Fourier transform (~ 3.5 - 13.0 k'), c) Ir' weighed Fourier transform (~ 3.5 - 13.0 k').

192

T ABLE 1. Parameters used in the calculation of the model spectra

Coordination N R (A) !J.er (N)

Pt-Pt

Pt-O

5.0

1.9

2.77

2.65

0003

0006

ßEu (eV)

o o

Figure 4 and section 2). Therefore. the use of a k?- or !C weighed EXAFS spectrum or Fourier transform makes the analysis much less sensitive for the presence of low Z contributions in the EXAFS data. This point has been overlooked in the past by some authors who tried to detect low Z scanerers in an EXAFS spectrum. which is composed both by low and high Z scanerers. The importance of k-weighing in the analysis is demonstrated in Figure 8. Pt-Pt and Pt-O EX­AFS model functions have been calculated with the coordination parameters as given in Table 1. These parameters are in the range of typical parameters found in the EXAFS analyses of data collected on small metal particles dispersed on high surface area supports. Pt-Pt and Pt-O phase shift functions and Pt and 0 backscanering amplitudes have been obtained from EXAFS data collected from Pt-foil and NllzPt(OH)6' which were measured at liquid nitrogen temperature. Figure 8a shows the individual Pt-Pt (solid line) and Pt-O (doned line) EXAFS spectra calcu­lated with the parameters of Table 1. The sum of the Pt-Pt and Pt-O EXAFS functions is also ploned (dashed line) and this function mimics the experimental data. The difference between the EXAFS spectrum with only a platinum contribution and the EXAFS spectrum with a plat­inum and an oxygen contribution is most pronounced below 6 A-I. In the !C weighed Fourier transform of the spectra (shown in Figure 8c) the difference between the spectrum with only a platinum contribution and the simulated experimental EXAFS spectrum (with a platinum and an oxygen contribution) is hardly detectable. The kl weighed Fourier transform (figure 8b) shows much larger differences. From this model study it is obvious that for a proper analysis of low Z contributions present in EXAFS spectra k1 weighed fits andlor Fourier transforms should be ap­plied.

3.6. CORRELA TION BETWEEN N AND !!.cr. USE OF BOTH k1 AND !C WEIGHED FOURI­ER TRANSFORMS

The determination of a unique set coordination parameters for a particular contribution to an EXAFS spectrum is often difficult due to the correlation between the value of the coordination number N and the Debye-Waller factor !!.cr. Kampers [17) showed that by simultaneous mini­mization of the difference between fit and data in a k1 and I2 weighed Fourier transform. a unique set of parameters can be found. This can be rationalized by examining the EXAFS equa­tion (8) and noting that the amplitude of the EXAFS spectrum depends on both N and ~cr. More particularly. different combinations of N and !!.cr will lead to the same peak amplitude of the Fourier transform of an EXAFS spectrum. However. this set of combinations depends on the weight factor, which has been used for the Fourier transform. The combination of N and ~cr that gives a good fit both in a k l and k' weighed Fourier transform offers a unique set of parame­ters. An example of this approach is given in figure 9. Figure lOb displays the I2 weighed Pt-Pt phase- and amplitude corrected Fourier transforms of three EXAFS spectra calculated with the combinations of N and ~cr of the !C curve in Figure 9. At the onset of the main peak differ­ences are presem. but the amplitude of the main peak of the transform of these spectra is equal.

~ .B E ~ z

15r---------------------~

c 10

i c ~ o o U

193

Figure 9. Combinations of coordination number (N) and Debye-Waller factor (.102) giving good frts in k' or !CI Fourier transform.

However, the k' weighed Pt-Pt phase- and amplitude corrected Fourier transfonns of the same three EXAFS spectra (figure 10a) are very different Only if Fourier transfonns witb both k' and ~ weighing show good agreement between model and experimental data a "good" combination of coordination number and Debye-Waller factor has" been selected. It is essential to use phase­and amplitude corrected Fourier transforms when applying this method, because otherwise the asymmetry of the peaks will obscure the results.

4 500 a b

E E400 ... .e3 0 -." CD C ~ 300 111 ~ F ... 2 ... -! cu

= '§ 200 0 0

I.L I.L 1 100

0 2.5 3 3.5 2 2.5 3 3.5

R (Ä) R (Ä)

Figure 10. Fourier transforms (6k: 3.1 - 18.6; Pt-Pt phase and amplitude corr.) of EXAFS model functions. Solid line: N=8.2,.t.a2= -0.0005 A-z; Dotted line: N= 10, !J.rr= 0 A-2; Dashed line: N=12.3, !J.a2= 0.0005 A-2. a) k' weighed, b) !CI weighed.

194

4. Application of EXAFS Spectroscopy: An example of the Structural Determination of the Metal-Support Interface of Highly Dispersed Supported Platinum Particles.

4.1 INTRODUCTION

A number of studies have appeared in the literature using EXAFS to investigate the structure of supported noble metal cataIysts. In the early studies, only the structural properties of the metaI particle were characterized [18]. Because the metal particIes were relatively large, the fraction of metaI atoms at the metal-support interface was small, and accurate information about the in­terface could not be obtained. Later studies indicated that the meta! particIe is supported directly on the oxide surface with reported metaI-oxygen distances between 1.92 and 2.07 A for platinum on alumina [20], platinum on silica [16], and rhodium on magnesia [21]. These distances ap­proximate the known metaI-oxygen bond lengths in metal oxide compounds. and are about equal to the sum of the covalent radii of the metaI and oxygen atoms. However, even in these later studies, the metaI particIes were stil11arger than about 15 atoms (coordination number about 6), limiting the accuracy with which the metaI-support distance could be determined.

In contrast, longer metaI-oxygen distances (ca. 2.7 A) have been reported for rhodium, iridi­um, and platinum on alumina, iridium on magnesia, palladium and platinum in zeolites, and rho­dium on titania [22]. For Rh/alumina [15]Jthe Rh-O contribution increased as the metaI particIe size decreased, indicating that this Rh-O contribution arises from the interface between the rho­dium and the oxide support. The lengthening of the metaI-oxygen distance has recently been ex­plained by combining the results of TPD and EXAFS studies on very small platinum particIes dispersed in LTI.. zeolite [3,23]. After reduction at low temperature (LTR: < 350 0c) hydrogen is present in the metaI-support interface, which in turn results in a long platinum oxygen distance. Reduction at high temperatures (H1R > 450 0c) removes the interfacial hydrogen leading to a direct contact of the interfacial platinum atoms with the support oxygen ions. In the following a short summary is given of the EXAFS data-analysis used to determine the local structure of the platinum atoms in the metaI-zeolite interface of a PtIH-LTI.. sampIe. Most of the EXAFS data­analysis concepts as discussed above are used and the same kind of approach can be used to ana­Iyze EXAFS data of smaII platinum particles dispersed on carbon [24].

60 0.15

a 40 b 0.1 E

~ 20 0.05 ..

c :x- c

~ 0 ';:( 0

~

~ .!!

-0.05 ~ ·20 "-

-0.1 .4()

-0.15 ~ 5 10 15 0 2 3 4 5

k(1/Ä) R (A)

Figure 11. a) EXAFS spectra of PtIH-L TI... reduced at 300°C (solid line) and 500°C (doned line)

b) Fourier transform [kl, !:;. k: 2.6 - 13.9 k ', Pt-Pt phase and arnplirude correctedl of spcctra in a)

4.2 EXAFS DATA-ANAL YSIS OF A PTIH-LTL CATALYST

Details of the sampie preparation and collec­tion of EXAFS data are given in reference [3]. The sampie was pressed into self-supporting wafers (calculated to have an absorbance of 2.5) and placed in a controlled-atmosphere cell [25]. The sampie was reduced in situ in flow­ing hydrogen (purified and dried) at 1 atm. Pt/H-LTL (Pt loading 1.2 wt%) was heated at 5 Klmin to the desired reduction temperature (L TR: 300 °C, HTR: 500 0c) and further re­duced for 1 hr. The sampies were cooled to room temperature under flowing hydrogen. The measurements were done with the sampie at liquid nitrogen temperature in the presence of hydrogen at atmospheric pressure.

The EXAFS data (average of 6 seans) for

E 0.5

~ c ca ~ 0 ~ll*Hrut*'~~~~ .. -! :s o I&. -0.5

2 3 R (A)

4 5 6

195

Figure 12. Nonnalized Fourier Transfonn of PrIH-L1L

(red 300"C) (solid line) and Pt-foil (dotted line) ~. &: 2.6-13.9 k' . Pt-Pt phase and amplitude corrected)

Pt/H-L TL reduced at 300 °C and 500 °C are shown in Figure Ila. At low k values the amplitude of the EXAFS function is higher after reduction at 500 °C than after reduction at 300 °C. In addition, the position of the nodes in the EXAFS function are located at different k values. Fourier transforms (l2 weighing, Pt-Pt phase- and amplitude corrected) of the EXAFS data for the Pt/H-LTL catalysts are shown in Figure llb. The Fourier transforms show an increase in the Pt-Pt contribution around 2.7 A. indicating a growth in the Pt particle during higher reduction temperature. Furthermore, for Pt/H-LTL reduced at 500 °C, an additional scatterer besides plat­inum is visible as a separate peak near 2.2 A in the Fourier transform. This peak results from an interference which is observable in the raw EXAFS spectrum (Figure Ila) by the presence of a bcating node at k=9-1 0 A·I .

Tbe normalized k l weighed Fourier transforms (Pt-Pt phase- and amplitude-corrected) of the EXAFS data for Pt/H-LTL reduced at 300 °C and for Pt foil are shown in Figure 12. Tbe Fourier transforms are normalized to facilitate comparison and k l weighed in order to emphasize the low-Z (oxygen) contributions. Differences are observed bctween the Pt/H-LTL catalyst and the Pt foil in both the magnitude and the imaginary part of the Fourier transforms betwcen 2.0-3.5 A. These differences indicate the presence of additional scattcrers bcsides Pt in the Pt/H­LTL catalyst. Multiple-shell fitting of the Fourier-filtered EXAFS spectrum (lCl. Ale: 2.6-13.9 k l , ~: 1.2-3.3 A) resulted in the identification of two significant contributions

TABEL 2. Structural parameters for Pt/H-LTL reduced at 300 ·C and 500 ·C

Coordination N R (A) I:l.cr (A2 x 10-3 ) l:l.Eu (eV)

Pt-Pt: Red300°C 4.1 =0.1 Red 500 °C 4.4 = 0.1

Pt-O: Red300°C 1.2 = 0.1 Red500°C 1.5 = 0.1

2.73 = 0.01 2.72 =0.01

2.65 =0.01 2.24 =0.01

3.4 = 0.1 4.4 = 0.1

7.0 = 1.1 -2.5 = 1.1

3.5 = 0.3 0.8=0.2

5.7 = 0.5 -13.9 = 0.1

196

0.4

~ 1:

n 0.2

!

~ ~ . -0.2

-0.4

4 6 8 10 k (1/A)

0.5 r.

~ - 0 x ~

-0.5

4 6 8 10 k (1/A)

a 0.4

-0.4

12 o

b 0.5

e .. .2 .. c ca ~ 0 .. GI

"§ 0

"" -0.5

12 0

2 R(A)

2 R(A)

3

3

4

d

4

Figure 13. Results of EXAFS analysis of PtIH-L TL reduced at a) 3OQ·C and b) SOO·C (solid fine: Fourier filtered data, dotted line: fit). Fourier transtorms [«2, ~k: 3.5 - 13.0 k'] of the spectra in a) and b) are shown in c) and d), respectively.

(Table 2): a Pt-Pt contribution at 2.73 A with a coordination number of 4.1 (ca. 6 Pt atoms/clus­ter) and a Pt-O contribution from the zeolite framework at 2.65 A and a coordination number of 1.2. The Pt-O contribution was statistically signifieant at the 88% confidenee level (F=2.51) [23]. A eomparison of the fitted spectrum with the Fourier-filtered raw data in both k- and r­space is shown in Figure 13a and 13e. Tbe Pt-O distanee of 2.65 A for this catalyst is in the same range as those previously observed for other supported metal catalysts redueed at low tem­peratures [22].

Tbe results of the multiple-shell fitting of the Fourier-filtered (12, nie: 2.8-13.9 kt, .6R.: 1.2-3.2 A) EXAFS data for the PtIH-LTL reduced at 500 oe are also given in Table 2. The Pt-Pt coordination number of 4.4 indicates a small inerease in the platinum particIes size to ap­proximately 9 atoms. Tbe Pt-O coordination number was 1.5. Fits in r-spaee are presented in Figure I3b and 13d. The Pt-O contribution at 2.65 A, which was present after reduction ar 300 oe, disappears after reduetion at 500 oe, with the coneurrent formation of a Pt-O contribution at 2.24A.

0.01

!§ .2 .. c .. ~ ... .!! l; 0 u..

-0.01

0.08 ,-----~----___,

E 0.04

I ~ Iii "l: ~ o u.. -0.04

2 3 4

197

Figure 14. Fourier transform [kl, k-range: 3.5-12.0 k l, Pt-Q phase corrected] of the EXAFS spectrum of PtIH-L TI..

minus the Pt-Pt contribution (solid line) and the Pt-O contribution calculated with the parameters ofTable 2

(dotted line). a) reduced at 300"C, b) reduced at 500"C.

Subtraction of the calculated Pt-Pt connibution from the raw EXAFS speetrum results in a difference spectrum containing only connibutions from the support. The Fourier transform (kl weighed. Pt-O phase-corrected) of this difference spectrum and the calculated Pt-O connibution is shown for PtIH-LTL reduced at 300 oe (Figure 14a) and at 500 oe (Figure 14b). The differ­ent Pt-O distances in the two catalysts can be clearly seen. Reduction at 300 oe results in a Pt-O distance of 2.65 A, while reduction at 500 oe results in a Pt-O distance of 2.24 A. The peak as­sociated with the longer Pt-O distance is significantly broader than that associated with the shorter Pt-O distance. The broader peak is an indication of a larger deviation about the mean distance, i.e., a larger Debye-Waller factor (ac:f). In addition to the difference in the Debye­Waller factor, the two platinum-oxygen connibutions also have different inner potential shifts (~) (see Table 2). These changes are indicative of a strengthening of the interaction between the platinum clusters and the oxide support during HTR.

4.3 eONCLUSIONS

The extremely small metal particles of 5 to 11 atoms present in these L TL zeolite catalysts al­lows for an accurate characterization of the structure of the metal-suppon interface by EXAFS. Low temperature reduction (LTR) leads to the formation of a long Pt-O distance (ca. 2.7 A). In contrast, high temperature reduction (HTR) leads to the formation of a shorter Pt-O distance (ca. 2.2 A). Combining the results of EXAFS and TPD experiments have led to the conclusion that the long Pt-O distance is due to the presence of spilled-over hydrogen in the interface between the platinum atoms and the oxide surface [3,23]. This interfacial hydrogen is lost during high temperature reduction, leaving the platinum atoms in direct contact with the oxide surface. This results in a stronger metal-suppon interaction, which affeets the chemical propenies of the pla­tinum clusters. reducing the chemisorptive capacity and influencing catalytic activity and selee­tivity for a variety of reactions.

198

The results given above may make dear that good quality EXAFS data combined with an ade­quate and careful data-analysis can give very accurate infonnation about the metal-support inter­face in supported metal particles. This example is also illustrative for the structura.l determination with EXAFS of electrolyte materials like platinurn electrodes in fueJ cells.

References

1. Sayers, D.E., Stern, E.A.. and Lytle, F.W., Phys. Rev. Lett. 27, 1204 (1971). 2. X-Ray Absorption: Principles, Applications, Techniques ofEXAFS, SEXAFS and XANES;

Koningsberger. D.C., and Prins, R .• Eds.; Wiley, New York, 1988. 3. Vaarkamp, M., PhD Thesis, Eindboven University ofTeehnology, Eindhoven, 1993. 4. Stern, E.A. in X-Ray Absorption: Principles, Applications, Techniques ofEXAFS,

SEXAFS and XANES; Koningsberger, D.C., and Prins, R., Eds.; Wiley, New York. 1988, pp 3-51.

5. Ashley, C.A., and Doniach, S., Phys. Rev. B 11, 1279 (1975). 6. Lee, P.A., and Pendry, I.B., Phys. Rev. B27, 95 (1975). 7. Stern, E.A., Sayers, D.E., and Lytle, F.W., Phys. Rev. BH, 48 (1975). 8. Sayers, D. E., and Bunker, B.A. in X-Ray Absorption: Principles, Applications, Techniques

ofEXAFS, SEXAFS and XANES; Koningsberger, D.C., and Prins, R., Eds.; Wiley, New York, 1988, pp 211-253.

9. Cook Jr. I.W., and Sayers, D.E., I. Appl. Phys. 52, 5024 (1981). 10. Citrin, P.H., Eisenberger, P .• and Kincaid, B.M., Phys. Rev. Lett. 22, 3551 (1976). 1l. Bunker, B.A., and Stern, E.A., Phys. Rev. B27, 1017 (1983). 12. Teo, B.K., and Lee, P.A., I. Am. Chern. Soc. 101,2815 (1979). 13. Lengeier, B., I. Phys. (paris) 47, 75 (1986). 14. van Dijk, M.P., van Veen, I.A.R, Bouwens, S.M.A.M., van Zon, F.B.M., and Koningsberg­

er D.C., in Proc. 2nd European Conl. on Progress in X-Ray Synchrotron Radiation Re­search, SIF (Bologna), 139-142 (1990).

15. van Zon, I.B.A.D., Koningsberger, D.C., van 't Blik, H.F.I., and Sayers, DE., I. Cbem. Phys. 82, 5742 (1985).

16. Lytle, F.W., Greegor, R.B., Marques, E.C., Sandstrom, D.R, Via, G.H., and Sinfelt, I.H., J. Catal. 95, 546 (1985).

17. Kampers, F.W.H., PhD thesis, Eindhoven University ofTechnology, Eindhoven, 1988. 18. Sinfelt, J.H., Via, G.H., and Lytle, F.W., I. Phys. Chern. 68, 2009 (1978). 19. Lytle, F.W., Wei, P.S.P., Gregor, RB., Via, G.H., and Sinfelt I.H., J. Chern. Phys. 70, 4849

(1979). 20. Lagarde, P., Murata, T., Vlaic, G., Freund, E., Dexpert, H., and Boumonville, J.B., lCatal.

84,333, (1983). 21. Emrich, R.J., Mansour, A.N., Sayers, D.E., McMillan, S.T. and Katzer, J.R., .J. Phys. Chern

89,4261 (1985). 22. Koningsberger, D.C. and Gates, B.C., Catal. Lett. 14,271 (1992). 23. Vaarkamp, M., Modiea, F.S., Miller, J.T.M., and Koningsberger, D.C., I.Catal. in press 24. 0' Grady, W.E., and Koningsberger, D.C., J.Electr. Chern.Soc., subrnitted. 25. Kampers, F.W.H., Maas, T.M.J., van Grondelle, J., Brinkgreve, P., and Koningsberger,

D.C., Rev. Sei. Instr. 60,2645 (1989).

THE USE OF X-RAY TECHNIQUES IN THE IN SITU STUDY OF CORROSION

Hugh. S. Isaacs Brookhaven National Laboratory Upton, NY 11973 U. S.A.

ABSTRACT. Applications of x-ray absorption and fluorescence techniques for in situ chemistry studies of passivity and localized corrosion have been described. X-ray absorption near edge structures showed that rapidly grown oxides on aluminum-chromium alloys incorporated chromium in the oxide. Repeated electrochemically induced changes between the chromium 3- and 6-valent state occurred without chromium dissolving. When the oxide was grown by small potential steps, the chromium dissolved when 6-valent chromate was formed. With iron-chromium alloys, dissolution of chromate took take place but 6-valent chromium was also incorporated in the oxide. The 6-valent state was reduced on exposing the iron alloy to air. X-ray fluorescence measurements, in conjunction with energy dispersive analysis, have been used to monitor iron, chromium and nickel concentrations in solution. The concentrations and concentration gradients during localized corrosion of stainless steel were used to determine salt solubility and relative diffusion rates.

1. INTRODUCTION

Synchrotron x-rays sources make it possible to observe chemical changes during in situ electrochemical measurements that could not be made using other available techniques. In general, in situ measurements, give real-time responses and enable direct unambiguous correlations with controllable variables. The removal and preparation of sampies for ex situ measurements can induce changes and introduce uncertainties. In addition there may be effects that develop during the measurement itself as many surface analytical techniques are carried out in and require exposure to vacuum.

There many possible applications of x-ray absorption techniques to corrosion studies [1]. Two techniques will be described in detail here. They were chosen as they bracketed extremes in corrosion rates. Here the approach will be to simplify the corrosion aspects for tbe x-ray specialist and detail experimental factors in tbe x-ray applications for tbe corrosion scientist. Further details of the theory of x-ray absorption may be found in other chapters of this proceedings and examples of corrosion are usually elose at hand.

199

C. A. Melendres andA. Tadjeddine (eds.), Synchrotron Techniques in lnteifacial Electrochemistry 199-214. © 1994 Kluwer Academic Publishers.

200

Corrosion processes have many faeets with each metal and alloy system showing its own quirks and surprises but also distinct similarities. Problems are usually engineered around rather than solved because of the complexities arising from the solution chemistry within the corroding environment, sensitivity to the multitude of impurities at the corroding interface arising from both the metal and the environment, and the heterogeneities in the metal such as inclusions, grain boundaries and atomic defects.

The corrosion resistance of metals and alloys of industrial importance is accounted for by the formation of a passivating oxide layer that kinetically limits the oxidation of the metal. Corrosion currents may be as low as a few nNcm2 which are equivalent to a penetration rate of less than a micrometer per year. The oxide layer is extremely thin having a thickness of only a few nanometers. The only metal that would not form an oxide or hydroxide layer when exposed to air-saturated water is gold, because of its thermodynamic stability. Even platinum is predicted to form a hydroxide in air-saturated water [2]. Thermodynamic stability, or cathodic protection, can be imposed on some metals by applying an external potential to hold the metal in a reducing regime, where the growth of oxide is prevented or the reduction of any prior oxide present may take place.

The thickness of passive oxides increases as the potential of the metal is increased at a rate of about 1-3 nmN [3]. The upper limit of potential is determined by the properties of the oxides. With metals such as aluminum and tantalum, hundreds of volts can be applied when the oxide is insoluble in the electrolyte. With other metals the potential may be limited by the electronic conductivity of the oxide or changes in composition. The oxides on iron and nickel are electronic conductors and oxygen evolution takes place when the potential is increased above a critical value. The applied current and rate of oxygen evolution then increase exponentially with potential.

Increasing the potential may also increase the oxidation state of the metal cation in the oxide and produee a more soluble species. The passive film then dissolves preventing further increases in passive layer thickness or even preventing any oxide from forming. Chromium is a notable example. The lower 3-valent chromium cation is highly insoluble and imparts an extremely protective oxide layer when used as an alloying element. However, at high potentials where the very soluble 6-valent chromium ion is formed, protection is lost and rapid dissolution of the chromium takes place.

For most corrosion studies oxide thicknesses of about 1-3 nm are of major interest. The chemistry of these extremely thin layers have been studied using a wide range of both in situ and er situ techniques. These have been reviewed by Landolt [4]. Questions of interpretation always arise with er situ methods as possible changes in the chemistry may develop on removal of the metal from its environment [5]. By far the greater number of studies have been carried out using ultra-high vacuum techniques which may lead to further changes [6,7].

When chlorides are present in solution, the passive oxide cover often looses its ability to prevent corrosion. Increasing the chloride concentration increases the probability of undermining the passive oxide. Corrosion initiates at only a few sites but onee started, the localized corrosion is rapid [8]. The rate of metal dissolution within the localized corroding site is up to 8 orders of magnitude more rapid than that for the passive surfaee. Continued

201

corrosion depends on maintaining the presence of high concentrations of the chloride corrosion products [9] at the localized corroding site. Tbe chloride concentration is a balance between an increase, due to dissolution of the metal forming dissolved metal chlorides, and a loss by diffusion. If the concentration of the metal chloride drops too low, the metal can repassivate and localized corrosion stops.

When diffusion is slow, the concentration of dissolved metal chloride increases and supersaturates the solution. Tbe metal chloride then precipitates and reduces the dissolution rate in a manner analogous to, but far less effective than the passive oxide. Tbe solubilities of the passive film and the salt layer mayaiso differ by over eight orders of magnitude and offer one possible explanation for the extreme differences in corrosion rates exhibited by passivating metals.

Examples of ceHs and methods using x-rays to analyze the corrosion products on passive surfaces and under simulated localized corrosion conditions will be discussed (in Section 3) to indicate the approach taken in investigating the chemistry of these corrosion processes.

2. X-ray Absorption and Fluorescence Techniques.

At incident monochromatic x-ray energies that excite inner sheH electrons to unoccupied levels, the absorption of photons increases rapidly and produces a step or edge in the absorption spectrum. After the inner shell is vacated, an electron from a higher shell drops into the inner sheH and is accompanied by emission of an x-ray with a characteristic energy. With energy increases up to a few tens of eV, successive empty energy levels are filled by the excited electron. Tbe position of the spectrum produced, or the x-ray absorption near edge structure (XANES), depends on the valency state of the excited element and its chemical bonding and incorporates multiple scattering effects associated with the ejected electron. Tbe spectrum, by comparison with known standards, can be used to characterize the state of the particular element, be it in solution, as a crystalline or amorphous solid.

Tbe XANES for Cr compounds in Figure 1A [10] clearly shows the changes for different valencies of the Cr atom. Tbe position of the edge for metallic Cr is at 5989 e V and the measurements were calibrated and plotted relative to this value. By convention, the edge position was taken to be the first maximum in the derivative of the metal edge spectrum. Increasing the valency of the Cr increases the binding energy and the edge is shifted to higher energies. Of particular interest is the pre-edge peak of 6-valent Cr. It arises from a 1s to 3d transition which is allowed when the coordination of the absorbing atom lacks inversion symmetry as for example in tetrahedral coordination [11]. Tbis peak clearly identifies the presence of 6-valent Cr.

A second example is shown for iron in Figure 1B [12]. Tbe shift in the edge is again seen as the valency varies from 0 to +3. Tbe edge for all the 3-valent compounds are close together, but the shapes of the spectra differ and the compounds show broad peaks. In general, the shape of the spectrum depends on the particular compound and varies considerably even though the valency remains the same. Tbe shape of the curves can be used as a "fingerprint" for the identification of a compound.

202

? -Cr METAL 'e: --- CrCI2 " -e _.- Cr203 2 ....... Cr02 z 0 _ .. - K2 Cr04 ;= Q. Ir

~ al cl

~ Ir , x

-20 20

ENERGY (eV)

40

1.6 .___-_r_--.--~--.-----r--..,

!z 1.4 w Ö 1.2 Li: u­w

8 0.8 z Q 0.6 tL a: 0.4

~< 0.2

--- Fe30. -y-FeOOH ·······u-Fep3 _.- y-Fe20 3

0_10 0 10 20 30 40 50

ENERGY (sV) RELATIVE TO Fe K EDGE (7112 sV)

(A) (B) Figure 1. A) XANES spectra of Cr metal and compounds relative to the Cr K edge at 5989 eV [10]. B) XANES spectra of Fe metal and compounds relative to the Fe K edge at 7112 eV [21].

At energies above the absorption edge the ejected electrons are weakly scattered and only single scattering processes are important. Interference between electron waves from the absorbing atom and backscattered waves from neighboring atoms produce an extended x-ray absorption fine structure (EXAFS) from about 50 eV to up to about 1 keV above the edge. With amorphous materials EXAFS signals beyond the nearest neighbors are not observed. With small quantities of material, as with thin films, the EXAFS signals are generally too weak to analyze_ However, efforts have been made to use EXAFS to determine the structure of passive oxide layers [13,14].

With incident monochromatic x-rays fluorescence process may be used to measure x-ray absorption_ The fluorescence is proportional to the absorption and the measurements are most useful in determining the edges of elements present in low concentrations. With polychromatic x-rays the fluorescence is used to determine the concentrations of particular elements. A broad polychromatic spectrum of the high intensity beam from the synchrotron is used and an energy dispersive detector is used to monitor the fluorescence [15,16]. Results of energy dispersive measurements are shown in Figure 2 for a saturated solution above the dissolving edge of a salt covered stainless steel foil dissolving in chloride electrolyte. The Ka peaks of the elements present are indicated.

51 Fe

4.5 NI -~ 4 Cr

~ Ar Cu 83.5 - As Br CI)

..2 3

2.5

5 10 15 20

Energy (keV)

Figure 2. Energy dispersive x-ray measurement of the solution at 50 pm above a salt-covered Type 304 stainless steel surface in a crevice dissolving in a chloride electrolyte showing the Ka peaks of the elements present [20].

203

10

woter E <.>

i= I Mo « 0.1 Ag "-

I .., Rb .., E

.01 Cu iron

:z « .., '" >- .001 « Cl<

I X

10 15 20 25 30

ENERGY ( k eV)

Figure 3. The mean free path for water, polystyrene and iron, as a function of x-ray energy, with the positions of the K shell binding energies for the elements shown [after 17].

In general, the absorption of x-rays increase with atomic number of the absorbing atom and decrease with the x-ray energy. Figure 3 shows the mean free path of x-rays (or the thickness that attenuates the beam by a factor of l/e) as a function of energy in a polyester plastic, water and iron [17]. The positions of the K-edge binding energy for various elements are also indicated. At low energies the penetration depths are small. At energies elose to the Cr edge, over 60% of the x-ray intensity is lost after passing through about 0.5 mm in water or the plastic and about 0.015 mm in iron. This limits the thicknesses of materials that can be used in constructing cells. However, if heavy elements are to be studied, e.g. Mo, the thicknesses can be increase by a factor of about 30. The absorption depth also indicates difficulties in the study of passive oxides on surfaces. The oxide being a few nanometers thick, is considerably smaller than the penetration in iron and would swamp attempts to measure the absorption by the oxide. In order to overcome this problem very thin layers of iron or iron alloys have been used where essentially all the metal is converted into oxide.

3. Experimental Configurations

A number of different experimental setups for x-ray absorption measurements are shown in Figure 4. Standards for XANES are measured using the configuration shown in Figure 4A The intensity 10 of the monochromatic x-ray beam is monitored on passing through an ionization detector, or proportional counter, and the intensity I after being absorbed by a thin layer of the powdered standard material held between adhesive tape, about .015 mm thick, is again measured. The absorption is given by the logarithm of the 10 /1 ratio as a function of the incoming energy. The results shown in Figures 1 and 2 were obtained in this manner and after subtracting background radiation extrapolated from below the edge, and for comparison, normalizing the edge step to unity. The edge step was obtained by extrapolating the absorption curve from weil above the edge.

In order to study the chemistry of the passive oxide the setup shown in Figure 4B was used [18]. The Mylar electrode window used was angled at 45° to both the incoming monochromatic x-ray beam and the detector which in turn was oriented at 90° with respect to the beam. The position is important on reducing the background signals. The detector used was composed of an array of 13 energy-dispersive solid state detectors (Canberra). When the edges of two element were scanned, a fraction of the 13 detectors was tuned to the Ka of each element. The Ka emission is at a lower energy than the binding energy. A large

204

Incid.nf x- ror biom

10 ..

SAIIPlE

(A)

ransmlft.d beam

I ..

(OWititet , 1t< lrOd. (p,.rnym)

FiguR 4. Schematic diagrams of x-ray measuring configurations. (A) Transmission measurements of standards.

..... \t • ., -' , y

(C)

(B) Electrochemical cell for in situ XANES measurements of thin layer alloy electrodes deposited on Mylar windows using incident monochromatic x-rays and a 13-element fluorescence detector. Tbe cell size was 70x50x25 mm [18]. (C) A cell for fluorescence measurements of concentrations of dissolved metals in solution within an artificial pit using polychromatic incident x-rays and an energy dispersive solid state detector [20].

background of the scattered incident x-rays having values elose to the binding energies are filtered out by tuning to the lower energy of the Ka emission.

Tbe advantage of the multiple detectors arises from the difficulties caused by the scattered radiation and the limitations of the detectors. Tbe detectors can only res pond to a maximum count rate independent of the incoming photon energy. Even though the energy discriminating detectors filters out the higher energy radiation, its presence gene rally determines the overall count rate. Tbis then limits the count rate due to the energies of interest. At higher count rates the "dead time" of the detector increases and the actual counting time or "live time" of the detector falls. Increasing the number of detectors increases the total count rate, and thereby decreases the counting times for a specific error margin.

Tbe solution used in the cell shown in Figure 4B was a pH 8.4 borate buffer, deaerated with nitrogen [17-19]. Tbe electrodes studied were thin metal or alloy layers sputtered onto a Ta or Nb substrate. Tbe Ta or Nb substrates were of the order of 10 nm thick and were sputter-deposited on a 6 I'm Mylar film. Tbe 10 nm of sputtered Nb or Ta deposit acted as current carrier. Tbe Mylar film was glued across a cutout in a plastic container with epoxy. The metallayer electrode was potentiostatically controlled relative to a mercurous sulfate reference electrode.

Figure 4C is the cell used for determining concentrations of elements using x-ray fluorescence [20]. In contrast to the chemistry studies using monochromatic beams, polychromatic x-rays are used because of the higher photon flux. Tbe beam size was collimated to a diameter of the order of 0.02 mm and scanned across a sampie to detect spacial variations in concentrations. An energy dispersive detector was again used but in these measurements the resolution was determined by that of the solid state detector. Tbe emissions detected relate

205

only to quantities of the element and not to its chemical state. Again, to reduce the background signal from scattered radiation the detector is placed at 90° to the incoming beam. A "pin-hole" was used to reduce the beam size down to its final size of 0.02 mm. The electrochemical cell was mounted on a stage that could be positioned in three dimensions with stepping motors and the area around where the beam entered the cell could be observed with the TV camera and microscope [15,16].

The cell was designed to measure concentration gradients in solution produced by dissolving stainless steel. The electrode was the cross section of a stainless steel foi! 0.0137 mm in thickness, 2 mm in width and about 25 mm long. The foi! was sandwiched between two pieces of 0.2 mm thick Mylar sheets with the foil protruding from one end. The other end was abraded after gluing to a solid square block of methyl methacrylate. This solid was attached to a hollow square cylinder with a window cut out on one side that was positioned over the foi!. The two square cylinders were attached with Teflon pressure sensitive tape. The solution was deaerated and stirred with helium and contained a platinum counter and a calomel reference electrode. The purpose of the solid square block was to produce a clearly defined interface between the gas-stirred bulk electrolyte and the solution in the cavity.

4. The Electrochemistry of Passive Oxides

X-ray absorption techniques have been used to elucidate the chemistry of passive oxide films. To demonstrate the methods, examples will be given based on studies on Fe-Cr and Al­Cr alloys. Alloying with Cr generally dramatically increases the corrosion resistance of many metals. At potentials where it forms a 3-valent compound it modifies and improves the properties of the oxide of the base metal, and being highly insoluble, may become the major component of the passive oxide. However, on increasing the potential the corrosion resistance due to the Cr may be lost or even further degraded, when under the highly oxidizing condition a extremely soluble chromate is formed with 6-valent Cr.

X-ray studies of the electrochemistry of Al-Cr alloys showed that the 6-valent Cr was retained in the oxide if the oxide was grown rapidly. The retained Cr was found to be electrochemically active, in that the valency could be repeatedly changed between the 3- and 6-valent states by changing the potential. Similar x-ray measurements also showed that 6-valent Cr was retained in the oxide on Fe-Cr alloy following chromate dissolution. However, on removing the electrolyte, the 6-valent Cr was reduced in contrast to what was found with the Al alloy.

4.1 Al-Cr ALLOYS

There has been recent interest in the corrosion resistance of Al alloys at concentration exceeding the limits of solubility [22,23]. These alloys are generally amorphous and can be formed by co-sputtering the alloying element with the aluminum.

Sputtered Al-Cr deposits on Mylar were polarized in borate solutions under different potential schedules using the cell shown in Figure 4B, and the XANES was monitored. Slow increases in potential led to dissolution of the alloy . Aseries of Cr edge energy scans are shown in Figure 5A [24]. The edge heights were not normalized and were a measure of the relative changes in the amounts of Cr present. The sampie was a 2 nm thick 12%Cr deposit.

206

Mter the first measurement at an open circuit potential of -1.08 Vmse (measured versus a mercurous sulfate reference electrode), the potential was then held constant during the 3 minutes required for each scan and then stepped 0.1 V starting at -0.7 Vmse. Changes in shape of the edges are compared with the standards in Figure lA The edge at open circuit was definitely metallic and remained so until a potential of about -0.3 Vmse was reached. At -0.2 Vmse, the edge position shifted to higher energies and the peak just above the edge increased, indicating a transition to a 3-valent Cr. The shape and magnitude of the edge remained the same until at 0.2 Vmse the magnitude decreased and continued to decrease with each subsequent scan up to 0.6 Vmse. Also associated with these spectra is the presence of a weak peak around 5992 eV. This peak arose because of the presence of 6-valent Cr as may be seen in Figure lA The presence of the higher concentrations of dissolving chromate was more readily seen during similar measurements on a pure Cr deposit or higher Cr contents in the alloy deposit. The presence of the chromate at the potentials were it would be expected to be the stable Cr species, increased currents from the potentiostat and the decrease in the edge height were all indicative of dissolution of the deposit resulting from the formation of soluble 6-valent Cr as chromate.

In contrast to the dissolution on slowly incrementing the potential, a single potential step to high potentials gave no significant dissolution of the Cr. In Figure 5B the edge for the

E (v MSE)

~ -1.08, oe

~ 0.0 2' 2 'c

------'c

:J 0.1 :J

-e

~ -e

..::, 0.2

~ c c

.2 0

e-

~ 0.3 ~

'-0 0.4 0

'" '" .D 0.5 .D « «

~ 0.6

5980 6000 6020 6040 5980 6000 6020 6040 Energy (eV) Energy (eV)

(A) (B) Figure S. (A) In situ XANES spectra of a AI-12Cr thin film electrode under potential contro!. The thin film electrode consisted of 2 nm AI-12Cr on 10 nm Ta sputtered on Mylar. The XANES measurements were taken at the potentials shown on the right of the figure. The potentials were held at these values for the 3 min to record each curve and then stepped to the next higher potential [24].

(B) In situ XANES spectra of a AI-12Cr thin film electrode under potential contro!. The thin film electrode consisted of 2 nm AI-12Cr on 10 nm Ta sputtered on Mylar. The solid Iines were XANES measurements of the alloy which took about 10 min to record. In (a) the alloy was first held for 12 min at open circuit, in (b) 7 min at 2.0 Vmse and in (c) 6 min at -1.5 Vmse. The dotted Iines show superimpose XANES standards: (a) Cr metal, (b) K2Cr04, and (c) Cr20 3 [18].

207

initial Open circuit exposure of an Al-t2 Cr deposit, ja shown by the noisy curve. a. The superimposed broken curve was for a Cr meta! standard and indicated Cr was present as the metal. Curve b, shows the edge after the sample was polarized by stepping the potential from open circuit to 20 Vmse, and compares it with the broken curve for a KzCrO. standard. The heights for the deposit were approximately the same as was seen in curve a, indicating Iittle lass of Cr. The pre-edge peak due to 6-valent Cr for the sample attains a height of about SO% that of the edge. The peak height for the standard is about equal to the edge height. Hence SO% of the Cr present was in the 6-valent state. 1be position of the edge was below that expected for the 6-valent state but distinctly above that for the meta! as seen on comparing with standards (Figure 1A). Hence, the remaining SO% was present in the 3-valent state. Curve c shows the edge after the sample was polarized to -1.5 Vmse. The corresponding curve for the CrzO] standard shows aß the er to have been reduced from the 6- to the 3-valent state. The Cr in the rapidly grown oxide remained electroactive. On repeated stepping of the potential between 2 and -1.5 Vmse similar changes in the er valency stales between 3 and 6 too place with no detectable lass of Cr.

The above results show marked differences in dissolution characteristics of the oxide on the aßoy depending on how it was grown. ether studies, on deposits 2.5 nm thick were made using XPS [10,24]. It was found that the air-formed oxide layers on the aßoys were predominantly Al oxide whicb tended to dissolve when the potential was slowly increased. The slow increase in potential to below where 6-valent Cr formed showed that Cr was further concentrated under the Al oxide. 1bis suggested that when tbe potential was slowly raised and reacbed a value at whicb the 6-valent Cr could form, only a thin Al oxide was present which was not suf6ciently protective to prevent the meta1lic Cr from dissolving.

4.2 Fe-Cr ALLOYS

Stainless steels owe their corrosion resistance to tbe presence of Cr. There is a marked improvement wben the concentration of Cr in Fe increases above about 13%. The passive film may, prior to analysis, alter on exposure to air or to vaeuum and electron radiation, when Auger spectroscopy or XPS is used. Thus the composition of the passive film on stainless steels determined by ex situ techniques may not accurately reflect the composition of the film in situ [4-7].

Tbe in situ bebavior of Fe-26Cr has also been investigated using a ceIl and sampie design similar to tbat descn"bed above for the Al-Cr studies. Figure 6, from BardweIl et al [19], sbowed tbere was agreement between the cyclic voltammetry for a sputtered and bulk electrode of the alloy in a pH 8.4 borate buffer solution conducted at 0.5 mV/so At the most negative potentials hydrogen is tbe dominant cathodie reaetion. The distinct hysteresis is associated with the reduetion of the oxide which takes place at tbe most negative potentials. The first positive cuerent peak on increasing tbe potential, was the formation of tbe passive oxide film. At higber potentials, the second positive euerent peak resulted from the dissolution of Cr as achromate. Above these potentials the marked increase in euerent was due to oxygen evolution. The negative euerent peak, at the higher potentials were associated with the reduction of chromate in the oxide and, at the lower potential, tbe negative euerent peak was due to reduction of the iron in tbe oxide. These peaks have previously been discussed in more detail [5].

208

\2r-----------------r1

POTENlIAL N ..... MSE)

FiguR 6. Cyclic voltammograms of Fe-26Cr bulk alloy and sputtered films in borate solution at a scan rate of 0.5 mV/So The thin film electrode consisted of 25 nm Fe-26Cr on 25 nm Ta sputtered on Mylar. The arrows indicate the potentials at which the XANES measurements in Figure 8 were collected [19].

In Figure 6 the first cycle for the sputtered film ditTered from the second and subsequent reproducible cycles. The subsequent cycles more closely resemble the bulk alloy. For the sputtered alloy the passive film formation was distinct on the first cycle but the peak due to Cr dissolution was suppressed. On subsequent cycles the passive film formation peak was suppressed and the chromate dissolution peak was defined. The cause of the various differences between the first and second cycle has attnbuted to an enriched surface layer of iron over the sputtered Fe-Cr alloy films, which protected the Cr. The first scan displays an uninhibited passive peak due to Fe oxidation, but a reduced peak for chromate formation. On subsequent cycles the Fe oxidation is reduced but the chromate formation is developed as shown by the enhanced dissolution peak.

XANES measurements of the Cr edge for the Fe-26Cr alloy, in Figure 7, were taken after stepping and holding the potential at the values indicated. Each edge measurement took 3 minutes. Here again the quantity of Cr can be deduced from distinct changes in the edge height and changes in position and shape of the spectra. The curves at -1.5 Vmse, followed open circuit exposure of the alloy on adding solution. Stepping the potential to regrow the passive film at -0.4 Vmse and to where chromate was expected to form, at 0.4 V, gave no major changes. Nor were changes seen on decreasing the potential to where the iran oxide was reduced on stepping to -1.5 V. Under these conditions it has been observed that Fe is lost to the solution and only a fraction of the iron in the oxide deposits as metal [21]. It was only after the Fe oxide was reduced and the protection afforded by the enriched Fe overlayer was lost, did Cr then dissolved on the second potential stepping from -0.4 10 0.4 V. The edge scans 8t these potentials showed a distinct drop in the edge height because of the dissolution of the Cr.

209

€NERGYI"V)

Figure 7. In situ XANES spectra of the Cr K edge for a Fe-26Cr thin film electrode under potential control. The thin film electrode consisted of 4 nm Fe-26Cr on 10 nm Ta sputtered on Mylar. The potential was held for 10 min at each value indicated while the XANES was collected [21].

Measurements to determine the behavior of the 6-valent Cr in the passive film are shown in Figure 8. They were carried out after a first cyc1ic voltammetry scan to remove the iron layer present after sputter deposition. The potentials at which the sequence of

Figure 8. In situ XANES spectra of the Cr k edge for a Fe-26Cr thin film electrode under potential control after a first potential step cyc1e. The thin film electrode consisted of 4 nm Fe-26Cr on 10 nm Ta sputtered on Mylar. The potential was held for 10 min at a given value while the XANES was collected. The vertical dotted line is at the pre-edge peak, 5993 eV, associated with 6-valent Cr. The insert shows a comparison between the summed spectra of (e) and (i) (both at 0.6 Vmse) and (h) and G) (both at -0.3 Vmse) [19].

210

measurements were taken are indicated by the arrows on the cyclic voltammetry curves in Figure 6. The edge scans (a) to (c) show that on stepping the potential Cr was not lost until the potential was increased to 004, (d) and 0.6 V, (e), i.e., above the current peak for chromate formation. The height of the main edge was reduced due to loss of the Cr into solution. A pre-edge peak (at 5993 eV) associated with 6-valent Cr can be discerned. Decreasing the potential to (1) and (g) produced no changes, and the pre-edge peak remained. On stepping below the peak to (h) the edge shape changed. There was on longer a clear indication of a pre-edge peak and also the height of maximum of the main edge increased. These two changes indicated a reduction of 6- to 3-valent Cr.

Changes in the edge shape on stepping the potential to (i) and G) showed that 6-valent was again formed and was then reduced to the 3-valent state without any losses of total Cr. The features associated with the 3- and 6-valent states are more clearly seen on summing the spectra taken at higher and lower potentials as shown in the inset in Figure 8. Hence, the passive film on the iron alloy was also electroactive and showed similar behavior to that observed with the oxide formed on the sputtered aluminum alloy.

The stability of the 6-valent Cr state when incorporated in the iron oxide was also investigated. A sampie with a passive oxide incorporating 6-valent Cr was washed with distilled water, and without drying, was exposed to air [19]. The near edge spectra are shown in Figure 9. The in situ measurement prior to removal was at 0.8 V and 6-valent Cr was present. At 1.5 h the 6-valent Cr was still present, but after 16.5 h it had disappeared. With dry sam pies reduction has been observed after 15 min [5]. Hence the 6-valent Cr was not stable and its presence cannot be maintained on removal of potential contral even in the presence of oxygen in the air. In reducing environments or in vacuum, the rate of reduction can therefore be expected to be more rapid.

POSITION OF Cr(VI) PRE-EDGE PEAK

~

~ :0 ~ z o f= a.. c:: o (J) CD <t

-. EXSITU.16.5HOURS

5970 5980 5990 6000 6010 6020 6030

ENERGY (eV)

Figure 9. In situ and ex situ XANES spectra of a 6 nm film of Fe-26Cr. (a) In situ potentiostated at 0.8 Vmse, and ex situ after (b)O.5, (c) 1.5 hand (d) 16.5 h, of air exposure. The vertical dotted line is at the pre-edge peak, 5993 eV associated with 6-valent Cr [19].

211

5. Localized Corrosion

In the presenee of chloride breakdown of passive films takes plaee and the dissolution of the metal is rapid. The composition and nature of the solution that develops within the corrosion area is important as it determine the subsequent behavior. If the coneentration continues to increase then a salt layer eventually precipitates on the metal surface and the rate of dissolution is controlled by diffusion gradients and the dissolution of the salt. If conditions prevail that lead to a drop in coneentration, the solution becomes less aggressive and the actively dissolving surfaee will repassivate.

In order to simulate the chemical processes in solution during localized corrosion one­dimensional artificial pits have been used [9]. X-ray fluorescenee measurements using white x-rays and an energy-dispersive spectrometer offer a powerful method for the in situ study of chemical and physical changes during electrochemical and corrosion processes. It enabled simultaneous counting of the x-ray fluoreseenee from all the elements in stainless steel at each position of the sampie and reduced the time required to obtain concentration gradients.

The electrochemical eell shown in Figure 4A was designed to simulate one-dimensional diffusion processes taking plaee in pitting or creviee corrosion. A 0.5 M HCI + 0.5 M NaCI solution was used with a saturated calomel referenee electrode (see). The steel was dissolved downwards at a constant potential of 0.6 Vsee leaving a rectangular shaped creviee [20].

Figure 2 shows an x-ray fluoreseenee spectrum of the solution inside the cavity immediately above the metal for real-time counting of 600 s. The counts under the Ka energy peak for Fe, Cr and Ni were integrated and used to determine their fluorescence intensities. The argon peak was from argon in the air. The souree of the As. and Br peaks could be due to impurities in the plastic, the Teflon window or the epoxy glue. These possibilities were not

10 ___ '···W~~.", __ ~. __ ~ __

f ,!iii 1 1 $~ " ........... _--~.

""I furI" jj) ...

<I., j)All "''''

~m diQ",.I" iW2 "' ...

A B

Q,$~ HoCI ~ Q,SW He)

P./onHol 1 &öl) "'V\" '$U)

d. d. ea"",t , 1),02 mA

0.$ I LS Dlsknc, (mm)

Figure 10. A) Variation of fluoreseenee intensities for Fe, Cr and Ni as a function of position above the cross section of a dissolving Type 304 stainless steel foil 0.014 mm thick. The depth of dissolution of the foil was 2.1 mm. The bulk solution was 0.5 M NaCl+O.5 M HCl [20].

B) Variation of coneentration of Fe, Cr, Ni and their summed total above the cross section of dissolving Type 304 stainless steel, derived from the results in Figure 10A [20].

212

pursued as they were only observed after long period of counting time (600 s) and were not seen when counting for 30 seconds real-time. This does, however, indicate a major application of x-ray fluorescence technique which is the quantitative analysis of trace impurities [15,16].

Figure 10A shows the intensities as a function of height above the stainless steeVsolution interface at a current density of 64 mNcm2 [20]. The iron shows an increased slope elose to the dissolving interface and again at the mouth of the artificial pit.

The relative intensities of the observed fluorescence is a complex function of the x-ray cross sections, the relative positions of the binding energies of the elements and their emission energies. Quantitative x-ray fluorescence analysis has been weIl established for many years [25]. One of two methods are usually used for quantitative analysis. The first is an empirical method exemplified by the multiple regression treatment of Lucas-Tooth and Pyne [26]. This method calculates the absorption and scattering effects of each element present in a sampIe from the fluorescence intensities measured. Standards are required which are similar to the sampie. The second method calculates the expected intensities based on a fundamental­parameters method [25-28]. This method derives the expected intensities from a given sampIe composition, geometry and three sets of basic information: the x-ray spectral distribution of the incoming beam, the mass absorption coefficients for each element in the sampIe, and their fluorescence yields. In this study, the fundamental-parameters method was employed to obtain calibration curves, and the reliability of this approach was investigated using known standards. A modified NRLXRF computer program [28] was used and referenced to the known concentration of the iron in the stainless steel in order to convert the measured normalized fluorescence intensities of Fe, Cr and Ni to solution concentrations.

The derived concentration variations are shown in Figure WB. The total concentration of metal ions is also given as a function of distance from the steel surface. The saturation value for the steel was 4.75 molar and the corresponding chloride concentration was 10.4 molar. This value is in agreement with the value of Hakkarainen [29] who observed that approximately 10 M chloride ion was required to give a saturated metal chloride solution on dissolving Type 304 stainless steel in HCI.

The relative Fe:Cr:Ni weight per cent ratios in the saturated solution at the metal interface were 67 : 22 : 11 compared to ratios of 68.7 : 18.2 : 13.1 in the metal. The differences in the ratios resulted from unequal mass transfer rates for the ions in solution. The mass transfer inc1udes an electromigration contribution in addition to diffusion, but as the contribution is not known, the results have been treated in terms of effective diffusion rates. The effective integral diffusion rates were calculated from the concentrations at the metaVsolution interface; the depth of the crevice and the current density. The values obtained were 6.0, 5.2 and 4.7 10"' cm2/s for the ions or chloride salts of nicke~ iron and chromium, respectively. These values are elose to published results for nickel oxide diffusion coefficients [30] and indicate that only a small electromigration contribution was present.

The concentration gradients in Figure WB were elearly non-linear with increased slopes at the dissolving metal interface and elose to the mouth of the crevice. This behavior shows that the differential diffusion coefficients have a maximum at intermediate

213

concentrations at abaut 2.5 M total metal ion. Once again this behavior is consistent with diffusion coefficients of many salts [31,30]. Nickel chloride solutions show a maximum diffusion coefficient but at a lower concentration of 1.5 M [30].

6. CONCLUSION

Two applications of high intensity x-ray methods have been presented as examples of corrosion studies. They demonstrate but a few uses of x-ray methods. There are many other approaches with advantages and possibilities that complement or may not be met, using other techniques. In particular, with electrochemical processes, including corrosion where the processes can easily be tumed on and off, there are many more in situ investigation that can elucidate the chemistry taking place.

7. ACKNOWLEDGEMENTS

The author wishes to thank Alison Davenport, J-H Cho, Jerry Frankei, Alex Schrott, Jennifer BardweIl and Barry MacDougall for their significant contributions to this work,. The work was performed under the auspices of the U.S. Department of Energy, Division of Materials Sciences, Office of Basic Energy Sciences under Contract No. DE-AC02-76CHOOO16.

8. REFERENCES

1. A Davenport and J. G. Gordon II, (ed.), "X-ray Methods in Corrosion and Interfacial Electrochemistry," VoI92-1, (The Electrochemical Society Inc., Pennington, NJ, 1992.

2. M. Pourbaix, Atlas o[ Electrochernical Equilibria in Aqueous Solutions, (Houston TX: National Association of Corrosion Engineers, 1974).

3. L. Young,Anodic Oxide Films, (New York: Academic Press, 1961). 4. D. Landolt, in Advances in Localized Corrosion, Edt. H. S. Isaacs, U. Bertocci, J.

Kruger and S. Smialowska, (Houston, TX: National Association of Corrosion Engineers, 1990), p. 25.

5. J. A BardweIl, G. I. Sproule, D. F. Mitchell, B. MacDougall and M. J. Graham, J. Chern. Soc. Faraday Trans., 87, 1011 (1991).

6. G. P. Halada, C. R. Clayton and D. H. Lindsley, Materials Sei. Eng., A 103, L5 (1988). 7. B. A DeAnglis, J. Electron Spec. Relat. Phenorn.,9, 81 (1976). 8. A J. Sedriks, Corrosion o[ Stainless Steels, John Wiley & Sons, New York, 1979. 9. U. Steinsmo and H. S. Isaacs, J. Electrochern. Soc., 140, (1993). 10. G. S. Frankel, A J. Davenport, H. S. Isaacs, A G. Schrott, C. V. Janes and M. A

Russak, J. Electrochern. Soc., 139, 1812 (1992). 11. B. McQuillan in Physics and Chernistry o[ Electrons and Ions in Condensed Matter, J.

Acrivos, N. Mott, A Yoffe, Edts., (D. Reidel Publ. Co., 1984), p. 135. 12. A J. Davenport, J. A Bardwell, H. S. Isaacs, and B. MacDougall, unpublished. 13. M. Kerkar, J. Robinson, and A J. Forty, Faraday Discuss. Chern Soc., 89,31 (1990). 14. G. G. Lang, D. A Fischer, J. Kruger, D. R. Black, D. K Tanaka, and G. A Danko,

Phys. Rev., 39, 10651 (1989). 15. Y. Wu, A C. Thompson, J. H. Underwood, R. D. Gaiuque, K Chapman, M. L. Rivers

214

and K. W. Jones, Nuc1. Inst. and Methods, A291, 146 (1990). 16. K. W. Jones, W. M. Kwiatek, B. M. Gordon, A L. Hanson, J. G. Pounds, M. L.

Rivers, S. R. Sutton. A C. Thompson, J. H. Underwood, R. D. Giauque and Y. Wu, in Advances in X-RayAnalysis, 31, Eds. C. S. Barrett, J. V. Gilfrich, R. Jenkins, J. C.

Russ, J. W. Richardson and P. K. Predecki, (New York: Plenum Publishing Co. 1988), P 59.

17. F. W. Lytle, In Situ Characterization o[ Electrochemical Processes, Publication No.NMAB 438-3, (Washington DC: National Academy Press, 1987), p. 63.

18. A J. Davenport, H. S. Isaacs, G. S. Frankei, A G. Schrott, C. V. Janes and M. A Russak, J. Electrochem. Soc., 138, 337 (1991).

19. J. A BardweIl, A J. Davenport, H. S. Isaacs, G. I. Sproule, B. MacDougall, and M. J. Graham, J. Electrochem. Soc., 139,371 (1992).

20. H. S. Isaacs, A J. Davenport, J-H. Cho, M. L. Rivers, and S. R. Sutton, Reference 1, p.354.

21. A J. Davenport, J. A BardweIl, H. S. Isaacs, B. MacDougall, G. S. Frankel and A G. Schrott Corrosion Sei., to be published.

22. G. S. Frankei, A G. Schrott, C. V. Jahnes, M. Mirzamaani and V. A Brusic, J. Electrochem. Soc., 136, 243 (1989).

23. W. C. Moshier, G. D. Davis, J. S. Aheam and H. F. Hough, J. Electrochem. Soc., 134, 2677 (1987).

24. G. S. Frankei, A J. Davenport, H. S. Isaacs, A G. Schrott, C. V. Janes and M. A Russak, submitted to J. Electrochem. Soc.

25. B. Dziunikowski, Energy Dispersive X-ray Fluorescence Analysis, (Amsterdam, Holland: Elsevier, 1989)

26. H. S. Lucas-Tooth and E. C. Pyne, "Advances in X-Ray Analysis", Vol. 7, (New York: Plenum Press, 1964), p. 523.

27. J. W. Criss and L. S. Birk, Anal. Chem., 20, 1080 (1968). 28. J. W. Criss, "NRLXRF, A FORTRAN program for X-ray Fluorescence Analysis",

Naval Research Laboratory, Washington, D.C. (1977). 29. T. Hakkarainen, "Corrosion Chemistry within Pits, Crevices and Cracks", Edt. A

Tumbull, (London, England: HMSO, 1987), p. 17. 30. R. H. Stokes, P. Phang, R. Mills, J. Solution Chem., 8, 489 (1979). 31. R. A Robinson, R. H. Stokes, "Electrolyte Solutions" (London, England: Butterworths

Scientific Publications, 1959), p. 290.

IN SITU X-RAY ABSORPTION SPECTROSCOPY INVESTIGATION OF UPO METAL MONOLAYERS

Abstract

A. TADJEDOINE L. E. I. -Co N. R. S. 1, Place A. Briand 92195 Meudon Cedex France

and

LURE - CNRS Bät. 2090- Universite Paris Sud 91405 Orsay France

In situ X-ray absorption spectroscopy (XAS) in fluorescence detection mode and at grazing incidence is a powerful technique for the investigation of the electronic structure and the local environment of thin layers at the electrolyte-electrode interfaces. We present the application of XAS to study the structure of UPO metal monolayers and submonolayers on gold single crystal substrates. The oxidation state of the adsorbed species and the partial charge transfer are obtained from XANES spectra; the structure of the adlayers is extracted from the EXAFS oscillations. Results on copper, nickel, and zinc electrodeposited onto gold single crystal electrode are presented and discussed.

215

C. A. Melemires andA. Tadjeddine (eds.), Synchrotron Techniques in Inteifacial Electrochemistry 215-245. © 1994 Kluwer Academic Publishers.

216

1. Introduction

Knowledge of atomic and molecular structural relatlonships at the sites of electrochemical activity is fundamental to understanding the reactivity of the electrode and its dependence on the nature and the coverage of the adsorbed species. Of special interest here is the investigation of the structural properties of metal monolayers electrodeposited onto a foreign substrate in the so-called underpotential deposition (UPD) region . UPD refers to the formation of a metal monolayer at potentials, E, posi ti ve from the reversible Nernst potential before bulk deposition can occur [1-5]. Investigated for a long time mainly in relation with analytical trace detection, UPD systems are now widely studied for their own interest both in fundamental and in many applied fields. It is a unique way to control and to change reversibly the metal coverage under thermodynamic equilibrium between 0 and 1 monolayer (ML). Since the coverage is controlled only by the electrode potential one can investigate in situ the electronic and the structural properties of the deposit under well-defined conditions .. Moreover the' predeposition of one or few ML is often the first step in metal electrodeposition and plays a röle on the further growth of the deposit. The deposit mayaIso change the reactivity and the 'electrocatalytic properties of the electrode. Correlation between the macroscopic behavior of the UPD system as deduced from thermodynamic and optical studies and i ts microscopic structure is a crucial step towards the understanding of the UPD process. In this respect electron spectroscopies under well-defined condi tions of emersion of the electrode out of the electrochemical cell and its transfer to UHV chamber have been successfully employed to study the structure and the composition of UPD modified single crystal surfaces [6-8]. However the need for emersion of the electrode from its electrochemical environment is an intrinsic limitation of ex si tu methods. As observed by Durand et al chemical reactions at constant electric charge may take place after emers ion so that the invest iga ted surface i s different from wha t i t was in the electrochemical cell [9]. This stimulated the development of structural in situ techniques based mainly on the use of X-ray radiation and atomic scale microscopies. This article presents the application of X-ray absorption spectroscopy [XAS] to in situ characterization of UPD layers.

In XAS experiment one measures the absorption coefficient near a particular absorption edge as a function of the photon energy. The magnitude of the absorption jump at the edge depends on the amount of the absorbing species. The absorption spectra display features related to the electronic state and the local environment of the absorber. The spectral features in the low electron kinetic energy range (up to 50 eV above the edge) are called X-ray absorption near edge structure (XANES) or Near Edge X-ray Absorption Fine Structures (NEXAFS) and

217

contain information on the oxidation state and the local symmetry of the absorber. Extended X-ray absorption fine structure (EXAFS) occurs in the region extending several hundred eV above the absorption edge. It is a modulation of the absorption coefficient due to elastic scattering of photoelectrons from neighboring atoms. EXAFS spectra contain local order structural information including local bond distances, coordination number and chemical identity of coordinating atoms. Furthermore absorption feature of different elements are well-separated in energy, and thus it is possible to probe the environment of a specific element. This unique propertie of XAS allows for the determination of the unoccupied density of states, of the bond lengths and of the local environment of a particular atom embedded in a matrix or deposited onto a foreign substrate [10,11]. Moreover, taking advantage of the polarization of the incident synchrotron X-ray beam, bonds whose vector lies in the plane of polarization contribute to the EXAFS oscillation whereas bonds whose vector is normal to the polarization do not. For well-oriented surfaces, the classical EXAFS relationship is still valid using an effective coordination number which accounts for the angle between the vector connecting the absorber and i ts neighbours and the electric vector of the incident beam. Hence, measurements at different polarizations give access to the orientation of the bond of the adsorbate and to the symmetry of the sites of adsorption [12]. Finally by using a fluorescence detection mode under grazing incidence one can make XAS surface sensitive and apply it in situ in an electrochemical cello Melroy et al used this technique for the first time to examine the structural characteristics of metal monolayers electrodeposited onto various substrates [13]. Later on several groups developed this technique to investigate the structural properties of UPD mono and submonolayers on single crystals as weIl as on polycrystalline substrates [14-16]. As recent books and review articles on this field are available [17-20] I will discuss mainly our experimental results and show how XAS can provide valuable information on the oxidation state of the adsorbate species, their local environment, their structure and its change with the substrate orientation and the adsorbate coverage as weIl as the detection of side reactions which could occur in the so-called underpotential region.

2. Experimental

XAS experiments were performed at LURE Orsay, using synchrotron radiation from the DCI storage ring running at 1. 85 GeV with an intensity of 250 mA and a lifetime of 45h. The X-ray beam was monochromatized by a Si(331) channel-cut single crystal with a resolution of 1 eV at the Cu K edge. The incident be am is collimated by slits and its intensity was measured by an ionization chamber. The XAS spectra were obtained in the fluorescence mode using a detector

218

especially designed at LURE which combines a plastic scintillator and a photomultiplier [21]. The plastic scintillator of 3 cm diameter is fixed at 2 mm above the working electrode so that more than 50 % of the total fluorescence flux is collected. The fluorescence X-rays pass through a filter, so chosen that it has a high absorption coefficient in the energy region of scattered photons, but a low absorption coefficient in the energy region of the fluorescent (K, L .. )

IX IX photons. For K-edge elements with atomic number Z = 23 through 43 the correct filter is the element defined by Z-l (20 ~ thick foil of Co at the Ni K edge, of Ni at the Cu K edge and of Cu at the Zn K edge).

2.1. Electrochemical cell

Measurements were performed in a spectroelectrochemical cell, designed to reproduce as much as possible the data obtained in a standard three-electrode cell and to allow the electrochemical response of the system to be checkedeasily at any time [14] .

The cell was made of Kel Fand had 3 electrodes. The working electrode was maintained in the horizontal plane by a small depression and the electrical contact was made at the back of the electrode; the system was coupled with a micro-translation screw so that the thickness of the electrolytic medium above the working electrode can be adjusted between 1 cm down to a few~. The reference electrode was a saturated calomel electrode (SCE) against which all the potentials were measured. Solutions were deoxygenated with argon in an auxiliary cell and added to (or removed from) the main electrochemical cell through Teflon ports. A kapton film 50 ~ thick was attached to the upper part of the electrochemical cell. 2 This kapton film was platinized on one side (except for a 2x1.5 cm area where the incident X-ray beam impinges on the working electrode) and used as the counter-electrode. The electrochemical cell was first purged with argon before introducing the deoxygenated electrolytic solution. A view of the cell is shown in Fig.1.

Electrolytes were prepared using high puri ty grade sal ts, acids and water. The concentration of the supporting electr~\r.te ranged from 0.1 is 0.5 M while the metal concentration was 10 M in order to minimize the contribution of the solution to the XAS signal. The procedure of metal monolayer deposition and polarization during the XAS measurements is defined from preliminary electrochemical experiments. For instance the copper monolayer was deposi ted in the potentiodynamic mode with a sweep rate of 20 mV/s and held at 0.10 V(vs. SCE) during all the XAS experiments. During the deposition, the electrolyte thickness was about 1 cm to decrease the diffusion contribution because of the lo.w copper concentration. After deposition, the gold electrode was translated up to the kapton window, leaving a thin electrolyte layer of the order of 10 Jlm. Wi th such thickness, the electrolyte absorption was quite low and no copper

Fig.l : Views of the electrochemical cell (top) and its arrangement for XAS experiments (bot tom)

220

species were observed in the XAS data when a bare gold electrode was used. Thus, the copper signal obtained after the electrochemical treatment came from the monolayer only.

2.2. Electrodes

Au(lll), Au(100) and Au(110) single crystal electrodes were disks about 8 to 12 mm diameter and 2 mm thick cut from single crystal rods by electroerosion. These were mechanically and electrochemically polished to a mirror like finish, chemically cleaned, flame treated just prior to use and transfered to the spectroelectrochemical cell with a drop of triply distilled water to prevent further contamination [22]. We have also studied monooriented polycrystalline electrodes of high purity. The surface crytallites were preferentially oriented perpendicular to the <100> direction, as shown in the X-ray diffraction spectrum of Fig.2.

50000

40000 ~ :::I C U ......, >-

30000. ; c: ., .... c:

20000 ~ L­I

><

o o N

100

,.. o o =t

(. 1\

I . 100

SO -'---'r--r---"-'

120 140

Fig.2 Diffraction spectrum of monooriented polycrystalline gold electrode : a : intensity of the diffracted X-ray beam as a function of scattering angle. b : enhancement of the y axis to show up the very small contribution of crystallites that are not oriented in the (100) direction.

221

2.3. Experimental procedure

The electrochemical cell with the fluorescence detector is mounted on a microcontrol system which allows a precise control of the z and e positions. All the experiments are performed with the E vector of the incident beam parallel to the gold surface (the angle e between the be am and the working electrode is less than 2°).

The first step is the control of the electrochemical behavior of the system using linear sweep vol tammetry. Fig.3 shows_4 the current-potential curve obtained for Au(100) in 0.5M Na2S04 + 10 M Cu

S04 (adjusted to pH 3 with H2S04 ) at a sweep rate of 20 mV. s -1,

between -0.05 and 0.6 V(vs SCE) [14]. The voltammogram is typical of the behavior of the Au(100) surface in sulfate solution [23,24].

r -,00 f1

P d /i

J ~[/V(SCE) ty--I

Fig.3 : Current-potential curve of Au(100) in 0.5M Na2s04+10-~ CuS04,

(pH 3) solution before (-) and after (- - -) polarization of the electrode at O.lV for 5h. The arrow indicates the applied potential during the XAS experiment.

222

A single copper desorp!:.~on peak around 0.3 V (vs SCE) with a charge of 395 ± 20 ~C cm (after correcting for double-Iayer charging) is obtained for the oxidation of the copper monolayer. This value is_~ agreement with that reported by Schultze et al [23] (375 ± 20 ~C cm_2) and the expected theoretical value for the system (400 ± 20 ~C cm ). We observe no modification of the curve after the XAS experiment (Fig.3), which is indicative of the absence of any oxygen diffusion into the cell through the kapton window. The XAS spectra are recorded at a potential of 0.1 V(vs.SCE) for which the copper monolayer is stable.

XANES data are collected in scans of 15 min duration wi th an energy step of 0.5 eV; 16 scans are added to obtain a good signal to noise ratio. For the EXAFS experiments (recorded over 600 to 800 eV), 16 scans of 20 min each and an energy step of 2 eV are averaged to obtain a good signal to noise ratio.

The analysis of the EXAFS oscillations involves a background substraction. The various neighbouring shells are sorted out by a Fourier transformation of the EXAFS signal : the peaks occur at values of R that differ by a phase shift from the real interatomic distance [25]. By an inverse Fourier transformation into k space, the EXAFS oscillations corresponding to only one ne ighbouring shell are obtained. Comparison of the phase and amplitude of model compounds with those of the sampie yields the structural parameters.

3. Results

3.1 Determination of the oxidation state of the adsorbate by in situ XANES

XANES contains information about the oxidation state and the local symmetry of the absorber [11,Z6-Z8]. The mean free path of low kinetic energy electrons is high. Higher-order terms of the correlation function of the atomic distribution become important in the XANES energy region and multiple scattering by neighbours must be invoked in order to explain the shape of the edge. A detailled review of re cent theoretical advances in XANES and its application in surface science can be found in ref. [29]. However, in XANES experiments, basic information on the local symmetry and the electronic structure of the absorber could be gained from a fingerprint approach usi2~

model compounds. Fig.4 shows the XA~ES spectra at the CuK edge of Cu (CuCI2 in aqueous solution), Cu (CuZO) and Cuo (copper foil),

referred in the energy scale to the first inflection point of the metallic copper edge (8979.8 eV) [14]. The energy position of the absorption edge is dependent of the effective charge density of the absorber and reflects the tendency of an electron-deficient atom to bind more tightly the remaining electrons. The edge separation of the

223

metal and the Cu+ oxide is small (1.1 eV) since both compounds have a 10 . 2+ .

3d conf1~uration. In contrast the Cu edge d~fers strongly from that of Cu due to its partially filled d-band (3d) and is shifted by 7.5 eV to higher energy. Because of the dipole selection rule and multiple scattering effects, the shape of the edge carries information on both the type and the symmetry of the ligands .

. ..c L.. -< , ::§

"""' c o ... .... ... ..

1.5

~ 0.5 ;j!

....... , , A'" , .. , ..

I .. .. I

I

10 20

[nergy (eV)

.. .. .. ..

30

-2 Fig.4: CuK edge XANES spectra of t~ree copper species (A) 5. 10 M aqueous CuC12 solution; (B) Cu20 (Cu ) compound; (C) metallic copper

foil.

The XANES features of metallic copper arise mainly from multiple scattering of the photoelectron by the different copper shells in an fee structure [30]. The Cu + oxide has a linear geometry wi th two oxygen at 0.185 nm; it displays aprepeak at 2.7 eV above the edge

that is assigned to a dipole allowed 1 s --7 3d104p transition with xy

the z axis liing a12~ the Cu - ligand direction. The octahedral configuration of Cu ions in aqueous solution with six water moleeules at 1.95 A results in a characteristic strong white line due to multiple scattering of the photoelectron between copper and oxygen neighboring atoms [31].

It is then possible to obtain the oxidation state and the local symmetry of the adsorbate by a simple comparison of its XANES features wi th the XANES spectra of known compounds measured under the same condi tions. This particular characteristic of XANES can be used for

224

the determination of the actual charge of adsorbed metal atoms and to obtain information on the partial charge transfer between the adsorbate and the substrate. This is shown below for Ni and Cu monolayers on gold (100) substrate.

3. 1. 1. Oxidation state of Ni monolayer on Au(100)

The deposition of Ni on gold in aqueous solutions requires the use of a complexing agent to bring the plating and dissolution reactions within a suitable potential range [32]. We have used potassium thiocyanate as complexing agent. Fig.5 shows _4the voltammograms of a Au(100) electrode in 0.1 KN03 + 0.1 KSCN + 10 Ni

(N03 )2 electrolyte, for various negative potential limits, at a scan

rate of 5.10-3 V s-l [33]. A first monolayer is deposited at -0.65 V during the negative scan before Ni bulk deposition occurs. The bulk and ML stripping peaks are well-defined in the positive scan. XANES spectra were recorded with the electrode potential held at -0.65 V.

d .. 4

i (JJA/CM2)

.c • • .. ~ ... .. . '. ~ :5 •• t

.. .. • 0 c ~

• 0

• o

• o

.2 ECV)

Fig.5 :Voltammograms obtained for Au(100) in 0.1M KN03 + O.lM KSCN + -4 dE -3 -1

10 M Ni (N03 )2 at dt = 5.10 Vs for various negative potential

limits Ec : a : -0.5, b : -0.6, c : -0.75, d : -0.9 V/SCE.

225

Fig.6 shows the XANES spectra of several nickel compounds taken under the same condi tions : nickel foil (a), NiO powder (b), NiCl2 solution (c) and the electrolyte used for MI... deposition (d). The origin of the energy scale is the first inflection point of the metal nickel edge (8333 eV).2We note significant difference in the shape of the edge of Nio and Ni +. A strong white line appears for the oxidized species due to the multiple scatterlng of the photoelectron between the nickel and ligand atoms. It is then possible to assign unambiguously the oxidation state of the nickel monolayer from the shape of the edge without any further assumption.

Fig.7 shows the XANES spectra at the Ni K edge of Ni monolayer electrodeposited on Au(100), the electrode be1ng held at E = - 0.65 V in (a), and at open circuit in (b) . The XANES spectra of Ni and NiO are shown in (c) and (d) respectively. The similarity between curves a and c indicates that the oxidation state of Ni in the monolayer under potential control is Nio; the adsorbed Ni is totally discharged and there 1s no partial charge transfer to the the gold substrate.

Another observation can be made from a comparison between curves a and b obtained for Ni monolayer with and without potential control. The appearence of a white line in curve breveals that the Ni monolayer is oxidized when the electrode is not under potential contro!.

Z 0 ... I-Q. ct: 0 U) 1Zl a:

2

1.8

1.6

1.~

1.2

.8

.6

.4

.2

-20 -10 0 10 20 30 10 S0 60 70 a0 90 E-Eo(eVl

Fig.6 : Ni K edge XANES spectra of nickel compounds measured by a step-by-step XAS _2'periments a : metallic foil; b_4 __ NiO compacted powder. c : 5.10 M aqueous NiCl2 solution; d : 10 ~ Ni (N03 )2 in

(O.lM KN03 + O.lM KSCN) solution.

226

d 1.8

1.6

1.1

Z 1.2 0 ~

.... a. !>: 0 .8 l/}

'" er: .6

.1

.2

-2111 -1111 111 1111 2111 3111 1111 5111 6111 7111 9111 9111 E-Eo(eVl

Fig.7 :Ni K edge XANES spectra under the same conditions as in Fig.3 : a : Ni monolayer electrochemically deposi ted on Au(100) and polarized at E = -0.65 V; b : same Ni monolayer without potential control; c : Ni metal, d : NiO.

3.1.2. Oxidation state of Cu monolayer on Au(100)

Fig.8 presents the XANES spectra at the Cu K edge of Cu metal (curve A), Cu20 (curve B), and the copper monolayer electrochemically

deposited onto Au(100) electrode at E = 0.1 V (curve C) [14]. The ability of XANES to differentiate divalent, monovalent and metallic copper allowed us to determine the oxidation state of the copper monolayer electrodeposi ted onto Au(100). Jhe energy edge position of the monolayer is very close to that of Cu (Fig.8bl, which means that the oxidation state of copper atom is close to Cu+ . Copper atoms are then not totally discharged in the monolayer, indicating the occurence of a large partial charge transfer between the adsorbate and the substrate.

3.1.3. Discussion

A partial charge transfer occurs during the metal deposition, due to the difference in electronegativities between the adsorbed metal atom and the metal substrate [2]. As a consequence a dipole layer is buil t up to ensure the electronic equilibrium between the substrate and the adsorbate. At the metal-vacuum interface the dipole moment is

227

T

"'"' 1.

21 CII

t ... .... c: /.

=> ': . 0.9 T , . .... B ,: .. t

oe '\ .... ....... c: I ' ...... e .... O.~ t .... ... .. e I CII (a) ....

l oe

e le 20 3e Energy/eV

1 .~ t cl

,-... CII b .... T .... c:

=> 1 \. . ....

·"f .. oe ....... I c: Ci e .... i a. I .... ... .-.. I .-c' CII " .... .-

oe -.. ----:

Energy/eV

o 1 Fig.8: a: Cu K edge XANES spectra of Cu (A). Cu (8) and Cu monolayer (C) electrochemically deposited onto Au(1.00) surface at E = O.lV/SCE (reference energy : E = 8979.8 eV);

o b : same as (a) but with expanded scale.

a linear function of the change of the substrate work function upon adsorption and decreases from 9 = 0 to 9 = 1. At the metal-electrolyte interface the situation is by far more complicated and depends on the structure of the double layer. on the ions adsorbed and on the potential distribution across the interface. The adsorbed species of

228

charge zad fell only a fraction g of the potential drop. Their

behavior is then described by the geometrie factor g and the partial charge transfer 0 = z-zad' Z being the charge of the metal ion in the

solution. As discussed by Schultze, g and 0 are correlated with the

electrosorption valency 71 that accounts for the charge consumed to adsorb the UPD ion at constant potential [34]. 71 can be determined from thermodynamic and kinetic measurements [34] or in "in si tu" quartz microbalance experiments [35]. However the determination of 0 needs the knowledge of gwhich cannot be measured directly and is just estimated from "reasonable" assumptions, based mainly on a comparison with the metal-vacuum interface. It is assumed that the metal adatoms must be totally discharged at full coverage and strongly bound to the substrate.

This behavior is actually observed in many metal adsorbates like Pb, Bi, Cd that are totally discharged on platinum and gold . XANES measurements show that the Ni atoms adsorbed on Au are completely discharged in the monolayer, under potential control. However the behavior of Cu adatoms is different since our XANES s~1ctra at the Cu K edge indicate clearly an oxidation state close to Cu . Since copper monolayer formation is a two electrons per atom process, charge transfer from copper to gold must take place. This resul t has been confirmed by in si tu XANES experiments performed by J. McBreen et al [36] at both the Cu K and the Pt LIII edges on Cu underpotentially

deposi ted on carbon-supported platinum. TheV resul ts indicate that the copper has an oxidation state close to Cu and show a reduction of the intensity of the white line of Pt XANES, consistent with partial filling of empty Pt d-band vacancies with charge transfered from Cu to Pt. The same bahavior has been also observed for+ 1Cu monolayer on Pt (100) substrate by Durand et al who found a Cu like near-edge structure of the copper K edge [16]. Finally in situ quartz microbalance study confirms the occurence of a charge transfer from Cu to the gold [35]. The occurence of such charge transfer in the case of Cu deposition onto gold substrate has been inferred in 1970 by Schultze [34] to explain the observed increase of the electrosorption valency 71, with the coverage e from 0 to 1 ML. However this explanation was ruled-out later because of its inconsistency with the behavior of metal adsorption at the metal-vacuum interface [2]. The behavior of 71 was then tentatively explained by a change of the geometrie factor g wi th the coverage. Indeed, ei ther g or 0 can be invoked in the change of 71 and, up to now, there was no means to separate between their contributions. Moreover interactions of the UPD layer with electrolyte ions has been found to playa rale [37-40]. It is therefore very important to use technique like XANES that provides information on the charge of the adatom.

229

3.2.Local structure of the adsorbed layer : EXAFS

The XAS spectra at the CuK edge of the copper mO~4:ayer on Au(lll), Au(100) and Au(110)in contact with O.lM Na2S04 + 10 M Cu 504

(pH 3) solution are shown in Fig.9. The electrode was polarized at 0.1 V(SCE) in each case.

o .

o 100 201l 31l1l

E(EV)

Fig.9 : In situ fluorescence X-ray absorption of a copper monolayer electrochemically deposited onto (a) Au(lll), (b) Au(100)and (c) Au(llO).

The analysis of such data consists of background substraction, normalization, conversion from the E-scale to the wavenumber k-scale and weighting by kn to get the EXAFS oscillations kn X(k). The various neigboring shells are sorted out by Fourier Transformation of the EXAF5 signal into the R-space as shown in Fig.l0 for a copper ML on Au(111),Au(100) and Au(110). The peaks occur at R values that differ by a phase shift from the real interatomic distances [25].

By inverse Fourier Transformation into k-space of each fil tered peak the EXAFS oscillations corresponding to only one neighbouring shell are obtained and fitted using the phase and the amplitude of model compounds to obtain the structural parameters [42]. An example of the fits is presented in Fig.ll for copper monolayer on Au (111). In some cases the peaks corresponding to two different shells are not resolved, resulting in the appearence of a broad structure in the EXAF5 signal. In this case one has to fit the inverse filtered Fourier Transform of the broad peak by using a two shell model [14].

230

Au 100

Fig.l0 : Fourier transforms of the Cu k3-weighted EXAFS spectra of a copper monolayer underpotentially deposited on Au( 110). Au(111) and Au(100).

0.01

0.02 (a)

~

~ 2 3 5 " 2 3 4 5 .. 0 , 0 ,

'" -:; -;:---;;:

.,

- 0.02 0.01

Fig.ll : Inverse Fourier transforms of the Cu K-edge EXAFS spectrum of a Cu monolayer on Au(lll) substrate; experimental curve (-) and theoretical fit ( ••• ) a : first peak- b: second peak

231

3.3.Effect of substrate orientation on the structure of the adsorbed copper monolayer (ML).

Fig.10 shows the k3-weighted Fourier Transforms (FT) of the EXAFS oscillations above the CuK edge for a copper ML underpotential deposited onto Au(111), Au(110) and Au(100) substrates [41]. They display very different shapes indicative of very different structures of the adsorbate which depend on the substrate surface orientation.

3.3.1. Structure of the Cu ML on Au(lll)

Two well-defined peaks are observed. The first peak is around 1.6 A and the second at 2.5 A (not corrected from the phase shift). In this system three different atoms (oxygen, copper and gold) could surround the copper atom. The first peak corresponds to scattering from oxygen atoms while the second one is due to scattering from copper and/or gold atoms. To obtain a quantitative estimate of the bond lengths, the nature of the neighbours and their numbers, the inverse filtered FT of each peak ~re fitted with. the phase and amplitude functions of Cu-O from a Cu + aqueous solution and metallic copper (Fig. 11). Replacing copper by a gold backscatterer leads to poor results. The parameters obtained are given in Table I and yield copper-oxygen and copper-copper distances of 1.95 ± 0.03 A and 2.89 ± 0.03 A, respectively with corresponding effective coordination numbers of 4.5 and 5.9. These results lead to a (lx1) epitaxial arrangement of the copper monolayer on an unreconstructed gold (111) substrate, the copper atoms sitting in a threefold hollow site [42], in agreement with the results of Melroy et al [43].

More important is the presence of oxygen atom backscatterers in the present experimental configuration where the electric field lies in the plane of incidence. This observation rules out the atop position of oxygen on the copper adlayer suggested in [43]. The value of the effective coordination number of 4.5 is consistent with a hexagonal structure of the oxygen atoms with an angle of 30· between the vector connecting the absorber and its oxygen neighbours and the substrate surface [42].

3.3.2. Structure of the Cu ML on Au(100)

The X-ray absorption spectrum from the copper monolayer on Au(100) shown in Fig.10 differs strongly from the spectrum of Cu ML on Au(111). Instead of two well-defined peaks of the latter system, we observe a first unresolved peak in the 1-2.5 A range followed by a second peak at 3.5 A. The fits of the inverse filtered Fourier transform of the first broad peak using a two shell model leads to the parameters given in Table I. The copper-oxygen distance of 1.97±0.03 A is equal to the distance obtalned on Au(l1l) while the effective

232

eoordination number of 3.90 is eonsistent with a loeation of the oxygen atoms in the eopper plane. The first eopper-eopper distanee of 2.66 ± 0.03 A lies between the gold-gold lattiee spaeing (2.88 A) and the eopper-eopper (2.54 A). This ean be aeeounted for by an epitaxial arrangement of the eopper adlayer, the eopper atom sitting in an atop positions on the reeonstrueted gold (100) surfaee [14].

The fits of the inverse Fourier transform eurve for the seeond peak reveal the presenee of eopper-eopper and eopper-gold seatterers atoms at 3.72 ± 0.03 A (Table I), the main eontribution eoming from the in-plane eopper atoms. With an atop site eonfiguration and a eopper distanee of 2.66 A, the seeond nearest eopper-eopper distanee should be equal to 3.75 A, in good agreement with the observed value. Thus the strueture of the copper monolayer underpotential deposited on Au(100) appears to be the following :

i) the first gold layer undergoes struetural rearrangement whieh results in a deerease of the gold-gold spaeing down to 2.66 A.

ii) thf eopper atoms are on atop sites with an oxidation state elose to Cu .

iii) the adlayer is stabilized by strong interaetions with oxygen eoming from water or sulfate ions.

TABLE I

EXAFS parameters of Cu monolayer on Au(100), Au(lll), Au(110) and Cu bilayer on Au(100)

R(A) N 102der -1 dE/eV nm

Monolayer Cu-O 1. 97 3.90 2 0.05 CU/Au(100) Cu-Cu 2.66 4.10 1

Cu-Cu 3.72 3.95 3 Cu-Au 3.70 2.05 1 0.08

Monolayer Cu-O 1. 95 4.50 1 0.12 Cu/Au (111 ) Cu-Cu 2.89 5.90 -1 0.12

Monolayer CU-O 1. 95 4.50 1 0.12 CU/Au(110) Cu-Cu 2.87 5.8 1.2 .25

Cu-Cu 3.58 6.2 1.6 .25

Bilayer Cu-Cu 2.59 6.20 0.06 0.03 Cu/Au (100) Cu-Cu 2.90 4.15 0.03

R Bond lengths N Coordination number der Debye-Waller faetor dE Energy variation

(aeeuraey : R = ± 0.03 A, N = ± 15%)

233

The oxygen-copper interactions exist only for submonolayer and monolayer coverages and disappear at higher coverage as we have observed in the case of the electrochemical deposition of two copper layers on Au(100) [14].

3.3.3. Structure of Cu ML on Au(110)

The quantitative analysis of the X-ray absorption spectrum of Cu monolayer on Au(110) leads to the following results (Table I) :

i) presence of oxygen backscatterers at the same distance (1.95 ± 0.03 A) with an effective coordination number of 4.5 and strong copper-oxygen interactions.

ii) appearance of two hexagonal structures of the copper adlayer with two Cu-Cu distances. The Cu-Cu value of 2.87 A equal to the gold-gold spacing corresponds to an epitaxial (lxl) phase. The second Cu-Cu distance of 3.58 A can be related to a c(5x5) phase identified on Au(lll) at submonolayer Cu coverage by EXAFS [42] and near field microscopies [44,45]. Such a local structure can be explainedby Cu deposition onto the (111) oriented facets of the (lx3) reconstructed Au(110) surface, which was recently identified by in situ X-ray reflectivity and diffraction studies [46].

3.4. Effect of adsorbate coverage on the structure of the adlayer Cu/Au(l11) [41,42].

Fig.12 shows the Fourier transforms of the Cu k3-weighted EXAFS spectra of copper underpotentially deposited on gold· (111) for various coverages. Curve a corresponds to a freshly deposited 0.3 ML, curve b to the same 0.3 ML 1 h after deposition, curve c to 0.6 ML recorded during the deposition scan, curve d to 1 ML and curve e to 0.6 ML recorded during the stripping scan. The different shapes of the spectra are indicative of the different structures of the adlayer, which depend on the coverage, the time, and on the direction of the potential scan. A quantitative analysis of the spectra following the procedure already described leads to the EXAFS parameters shown Table Ir.

i) At all coverages we observe a sca t ter ing from oxygen atoms with a Cu-O distance of 1. 95 A and an effective coordination number close to 4.5. This is consistent wi th a hexagonal structure of the oxygen atoms with an angle of 30· between the substrate surface and the vector connecting the absorber and its oxygen neighbors.

ii) The fits of the inverse filtered FT of the other peaks reveal in all cases the presence of Cu scatterers with a coordination number of 6, which is indicative of the appearance of a well-ordered hexagonal structure of the adsorbate. However the Cu-Cu distances strongly depend on the coverage, the time, and the direction of the

234

DISTANCE:(A)

Fig.12: Fourier transforms of the Cu k3-weighted EXAFS speetra for Cu adlayers on Au(l1l) : a, "freshley" deposited 0.3 ML; b, 0.3 MI after 1h; e, 0.6 ML during the deposition sean; d, 1 ML and 0.6 ML during the stripping sean.

potential sean, whieh leads to adetermination of the loeal strueture of the eopper adlayer at eaeh stage of deposition and stripping :

3.4.1. Deposition scan :

For a freshly deposited 0.3 ML (E = 0.22 V/SCE) the Cu-Cu distanee

of 4.88 A (elose to 2. 89~ A) leads to the (hJ:;') strueture identified previously in ex situ LEED [47] and in situ STM [44] investigations of the same system under the same eonditions. However XAS data show that this strueture is not stable in the eleetroehemieal environment. After 1 h of polarisation at E = 0.22 V the EXAFS speetra

yield two Cu-Cu distanees, one at 4.88 A due to the (h~) phase and one at 3.59 A expeeted for a densely paeked c(5x5) hexagonal structure. Thus an equilibrium between these two phases in the copper adlayer is reached in about 1 hand preserved during the 5 h of the EXAFS measurement.

After stepping the potential to 0.15 V, corresponding to 0.6 ML, only the c(5x5) structure is observed and when a full ML is adsorbed (E = + 0.08 V) the Cu-Cu distance of 2.89 A corresponds exactly to the

235

(lxI) epitaxial arrangement of the adlayer already discussed in this paper.

3.4.2. Stripping scan :

The structure of the adlayer appears to be sens i ti ve not only to the coverage but also to the direction of the potential scan as shown for 9 = 0.6 ML in Fig. 12. During the anodic stripping scan 3 Cu-Cu

distances corresponding to an equilibrium hetween(h~), c(5x5) and (lxI) structures are observed. This result could be explained by an inhomogeneous process of copper oxidation over the electrode surface. The stripping rate is higher in the area covered by the

(hxh) phase than in the area eovered by the c(5x5) phase and very weak in the surfaee region where the (lxI) phase is present.

TABLE 11

. -1 9(ML) R(A) Neff t.cr(nm ) t.E(eV)

0.3a * Cu-Q 1. 95 4.5 0.01 .12

Cu-Cu(1)+ 4.88 5.9 0.015 .13

0.3b Cu-Cu(2) 3.59 5.85 -0.01 .24 Cu-Cu(3) 4.87 5.8 -0.01 .25

0.6e Cu-Cu(2) 3.57 5.85 -0.02 .15

1. d Cu-Cu(3) 2.89 5.9 -0.01 .12

e Cu-Cu(3) 2.86 5.7 -0.025 . 1 0.6 Cu-Cu(4) 3.56 6.3 0.015 . 1

Cu-Cu(!) 4.96 5.8 0.02 .15

• The same Cu-Q EXAFS parameters have been obtained for all Cu coverages.

+ 1,2,3 relate- to the same labelled peaks in Figure 11. (bond lengths, R; effective coordination number, Neff ; Debye-

Waller factor, t.cr; energy variation, t.E) (accuracy:R:±0:03A, Neff :±10%).

236

In summary, in situ EXAFS shows that the local structure of adsorbed copper on gold single crystal electrodes depends strongly on the substrate surface orientation and on the adsorbate coverage. The structure of the full monolayer changes from an epitaxial (lx1) arrangement on unreconstructed Au(111) surface, (the copper atoms sitting in a threefold hollow site) to an atop epitaxial deposition on a reconstructed Au(100) surface. The identification of two hexagonal structures of the Cu monolayer on Au(110) ((lx1) and c(5x5) hexagonal phases) shows that the copper deposition oeeurs on a (lx3) reconstructed Au(110) surface, in agreement with X-ray reflection and diffraction studies [46]. Thus the difference of reaetivity and the eleetrochemical behavior of single erystal electrodes like Cu/Au(hkl) system must be related to the observed differenee of strueture. However more extensive theoretieal and experimental investigations are needed to establish the eorrelations between the structure and the reactivity of electrodes.

As far as the Cu-Cu distances are concerned XAS results are in good agreement with in situ 5TM [44] and AFM [45] data. These show that the struetures of adlayers are a true struetures of the electrochemical interface. Further information can be gained from the unique ability of XAS to determine the chemieal nature of the neighbouring atoms. Our investigation shows the occurrence of strong copper-oxygen interactions at monolayer and submonolayer coverages. Furthermore the Cu-oxygen interactions desappear for higher coverage, as we have shown for a bilayer of copper on Au(100) (Table I, [14]). Questions on the origin of oxygen atoms (solvents, ions) need further investigation to clarify their effect on the UPD process.

3.4.3. Structure of nickel monolayer on Au(100)

Fig. 13 shows the radial strueture functions of a Ni monolayer underpotentially deposited on Au(100) at E = - 0.65 V(SCE) (a), of Ni metal foil (b), of NiO compacted powder (c), of 0.2 M NiC12 solution

(d) and of the electrolyte solution used in this study (0.1 M KN03 +

0.1 M K5CN + 10-4 Ni(N03 )2 (e) obtained from EXAF5 spectra recorded

above the NiK edge on the same beam line under the same experimental conditions. Details on the structure of Ni compounds presented Fig. 13b, c and d can be found in the literature [48-50] and could be used to obtain some information on the monolayer. Quantitative analysis is underway and will be presented elsewhere.

The EXAFS spectrum of a Ni ML (Fig.13e) shows 3 peaks located at 1.8, 2.75 and 4.1 A (not corrected from the phase shifts). The first peak corresponds to scattering from oxygen or/and sulfur neighbours.

~ 1

e \ '2 ::I .3 T .cl ~ .21 Cl> "0 ;: ö. T E .1 t oS

I

a

distances Ä

b

distances Ä

distances A

'"' 1 e I '§ .3 T .cl I

1 distances A

I ] 1.6 I ~ I ;;i 1.2 T

--; 1 ] .at c.. ~ T

.4 T I

distances A

237

d

e

Fig.13: Fourier transforms of the Ni k3-weighted EXAFS spectra of Ni metal foil (a), NiO powder(b), NiCl2 O.2M solution (c), (KN03 O.1M +

KSCN O.1M + NHN03 )2 10-4 )solution (d) and Ni monolayer on Au(100)

electrode (e).

238

It peaks exaetly at the position of the first maximum of the eleetrolyte speetra (1.8 A) and differs markedly from the first peak of NiO (eurve e) and NiCl2 (eurve d) speetra (1.4 A) that evidently

eorresponds to seattering from oxygen atoms. This observation is eonsistent with a strueture of the Ni ML where the Ni adatom is bound to the 5 end of the eomplexing ions SCN-, most likely in a tetrahedral arrangement.

The seeond maximum peaks at 2.75 A before phase shift eorreetion. This distanee is greater than the Ni-Ni spaeing in nickel foil (eurve a) and should correspond to scattering from Ni neighbours in (1x1) epitaxial arrangement on unreeonstrueted Au(100) surfaee, the third distanee of 4 A before phase shift eorreetion is eonsistentwith this strueture.

3.5 Deteetion of side reaetion by XAS substrate [51]

UPD of zine on Au(100)

The eleetrodeposition of zine on a foreign substrate plays a role in many teehnologieal systems such as zine-si 1 ver oxide and zine-niekel oxide batteries [52]. However only few studies on the early stages of zine deposition are available [52,53 and referenees eited therein]. We have defined in preliminary work the eleetroehemieal eonditions of underpotential deposition of Zn on gold single crystals consistent with in situ XAS experiments. Fig.14 shows

typical voltammograms obtained bet~1en -0.75 and ~~45 V at 10mV s-l on Au(100) in eontaet with Na2S04 10 M + ZnS04 10 M solution (pH 8).

Curve a refers to a continuous potential sweep and eurve b is obtained after polarization of the eleetrode at -0.75 V for 15, 30 and 60 min. The deposition of a monolayer during the negative sean results in two negative peaks (1,11). Three positive peaks appear during the positive potential sean. l' and II' peak at the same potential as land II; peak III appears at very positive potential and eould be related to the oxidation of strongly bound zine atoms.

The voltammogram is slightly modified during the polarization at -0. 75 V and reaehes a steady state after 15 min. Thus the XAS experiments were performed using the following proeedure the eleetrode is polarized at a given potential for 30 min; then the electrolyte gap thickness is decreased down to ca 10 ~m to start the XANES and the EXAFS speetral aquisition. The XANES speetra at the ZnK edge of Zn underpotentially deposited onto Au(100) under these eonditions are presented in Fig.15. Speetrum (a) was reeorded on a bare gold eleetrode, speetrum (b) at E = -0.55 V, speetrum (e) at E = -0.7 V speetrum (d) at E = +0.6 V and speetrum (e) after polarization at E = +1.25V for 30 min.

239

E/V (SCE)

_.5

I

-1 Fig.14 : Voltammograms of Au(100) in eontaet with (Na2S04 10 M +

-4 ZnS04 10 M, pH 48) solution; a : eontinuous potential sean. b : after

polarization of the eleetrode at -75V for 1h.

,....., 1/1 ..... .8 .... c: =:

.Q ... . 6

'" ......, z: Cl .... - .4 .... CI< Cl

'" "" ... .2

" " E-Eo(eV)

Fig.15 : XANES speetra at the ZnK edge _~f zine eleetrodeposited on gold (100) surfaee in 0.1 M Na2S04 + 10 M ZnS04 , (pH 8) solution :

(a) bare eleetrode, (b) E = -0.55 V, (e) E - 0.7 V, (d) E =0.6 V and (E) after 30 min. of polarization at 1.25 V. (Zero energy = 9659 eV).

240

Analysis of these spectra leads to the following conclusions : i) The magnitude of the edge-jump is larger for Zn than for Cu

and Ni at the same coverage, as deduced from the oxidation charge of the adsorbate during the positive potential scan.

ii) In the case of copper and nickel deposition, the height of the absorption jump is controlled only by the electrode potential in the first monolayer region and does not change with time as long as the electrode is maintained at the same potential. On the contrary, in the case of zinc deposi tion, the magnitude of the edge- jump is no longer controlled by the potential; it increases continuously with the time of polarization, reflecting an increase of the amount of zinc remaining on the electrode. As we have noted the voltammogram and consequently the electode surface in contact with the solution are in a steady state.

Since the XAS signal is unambigously related to the amount of Zn electrodeposited on the initially zinc-free gold substrate (curve (a) shows no Zn signal on a bare electrode) these results reveal clearly the formation of a zinc-gold alloy during the underpotential deposition, which is in contradiction with previous conclusion deduced

lBl~--------------------------------------------------------,

-CI>

.~ c :::s

..c ... ~

a

.2 .4

Zn 68 depth profile

.S .B 1.2 1... 1. S 1. B 2

Fig.16 : Zinc depth profile in Au(100) electrode (a) bare electrode, (b) after polarization of the electrode at -0.75 V/SCE for 5h.

241

from electrochemical data that ruled-out alloying in the UPD region [53]. The fact that the electrochemical response is stable could be explained by a penetration of zinc atoms into the gold matrix in such a way that the surface of the substrate does not change during the polarization. Indeed the zinc-gold compound is very stable and can be partially removed after a strong polarization at very positive potential (curve 15 e).

The formation of zinc-gold compound has been confirmed by independant measurements using Secondary Ion Mass Spectoscopy (SIMS). Experiments have been performed on the same gold electrode in the following way : Half of the surface electrode was immersed in the electrolyte and polarized for 5 h at E = -0.7 V, the second half being kept out of the solution. Fig.16 shows the depth profil of Zn in the bare (a) and in the emersed (b) region of the electrode. These spectra show clearly the presence of zinc in the treated region and confirm the zinc-gold alloying process which was revealed by XAS measurements.

4. Conclusion

The use of X-ray absorption spectroscopy in the study of thin layers electrodeposited at the electrode-electrolyte interface appears indeed to be a powerful way of obtaining unique structural and electronic information on the adsorbate. The method is applicable in situ, is not destructive, and yields specific information on the oxidation state and the local structure of a given element. In the particular case of UPD layers adsorbed on gold single crystal substrate, XANES appears to be unique tool in probing partial charge transfer from the adsorbate to the substrate. EXAFS yields structural information on the local environment of the adsorbate and allows for the determination of' the structure of the adsorbed layer and its dependence on the substrate surface orientation, on the adsorbate coverage, on the electrolyte composition, etc. Hence XAS provide structural and chemical information at the microscopic scale that should contribute to the understanding and the control of many electrochemical processes.

The developement of new fluorescence detectors availabili ty of new intense synchroton beams should significantly the time for aquisi ti on of XAS spectra in future. This would improve the capability of XAS in interfacial properties, in particular the investigation of systems that can hardly maintained at the same state for many

and the decrease the near

probing reactive hours.

242

5. Acknowledgments

The studies with G.Tourillon Dr. D.Guay, Dr. Lahrichi.

6. Bibliography

described here were carried out in collaboration (LURE). l' d like to thank all my co-worker' specially M.Ladouceur, Dr. A.Boutry-Forveille and Mrs. A.

1. W.J. Lorenz , H.D.Hermann, N.Wultirich, F.Hilbert,J. Electrochem. Soc. 121, 1167 (1974).

2. D. M. Kolb, in "Advances in Electrochemistry and Electrochemical Eng.", H. Gerischer and C. W. Tobias Eds., Wiley, New-York 11, 125 1974.

3. K.Juttner, W.J.Lorenz, Z. Phys. Chem. 122, 163 (1980).

4. R. Adzic, in .. Advances in Electrochemistry and Electrochemical Eng.", H. Gerischer and C. W. Tobias Eds., Wiley, New-York 13, 159 (1985).

5. S.Szabo, Rev. Phys. Chem. 10, 207 (1991).

6. D.M.Kolb, G.Lehmpfuhl and S.Zei in "Spectroscopic and diffraction techniques in interfacial electrochemistry", G.Gutierrez and C. Melendres Ed., NATO ASI series 320, Kluwer Ac. Pub!., 361 (1990).

7. D.Aberdam, Ibid 383.

8. R.Koetz, Ibid 409.

9. D. Aberdam, S. Traore, R. Durand and R. Faure, Surf. Sci. 180, 319 (1987).

10. D.E.Sayers, F.W.Lytle and E.A.Stern, Phys. Rev. Lett. 27, 1204 (1971 ).

11. J. Stohr, in "X-ray absorption Principes, Applications, Techniques of EXAFS, SEXAFS and XANES" , R. Pr ins and D.Koningsberger Eds., Wiley, New York 443, (1986).

12. P.H.Citrin, J.E.Rowe and P.Eisenberger, Phys. Rev. B28, 2299 (1983).

243

13. L. Blurn, B. Borges, (1986) .

H.D.Abruna, J.H.White, M.J.Albarelli, J.G.Gordon, M.Sarnant and O.R.Melroy, J. Chern. Phys. 85, 6732

14. G.Tourillon, A.Tadjeddine and D.Guay, J. Electroanal. Chern. 289, 263 (1990).

15. T. E. Furtak, L. Wang, E. A. Crek, P. Sarnanta, T. M. Hayes, G. Kenda11, W.Li, G.Liang and C.M.Lo, Electrochern. Acta 36, 1869 (1991).

16. R. Durand, R. Faure, D. Aberdarn, C. Salern, G. Touri l1on, D. Guay and M.Ladouceur, Electrochern. Acta 37, 1977 (1992).

17. H.D.Abruna in "Electrochernical Interfaces: Modern Techniques for in situ Interface Characterization", H. D. Abruna Ed., V. C. H. Publ., New York 3, (1991).

18. "X-ray Methods in Corrosion and Interfacial Electrochernistry", A.Davenport and J.G.Gordon II Eds, Electrochern. Soc. Ed. Publ. Penington (1992).

19. H.D.Dewald Electroanalysis 3, 145 (1991).

20. A.Tadjeddine and G.Tourillon, Analusis 20, 309 (1992).

21. G. Touri11on, D. Guay, M. Lernonnier, F. Bardol and M. Badeyan, Nucl. Instrurn. Meth., A294, 382 (1990).

22. J.Clavilier, R.Faure, G.Guinet and R.Durand, J.Electroanal. Chern. 107, 205 (1980).

23. J.W.Schultze and D.Dickerrnann, Surf. Sci. 54, 489 (1976).

24. H. Angerstein-Kozlowska, B. E. Conway, A. Harne 11 in and Stoicoviciu, Electrochirn. Acta 31, 1051 (1986).

25. J.Stohr, R.Jeager and R.Brennan, Surf. Sci. 117, 503 (1982).

26. A.Bianconi, Appl. Surf. Sc. 6, 392 (1980).

27. P.H.Citrin, J. Phys. (Paris) 47-C8, 437 (1986).

28. D. Norrnan, J. Phys. (Paris) C19, 3273 (1986).

29. A. Bianconi and A. Marce11i in "Synchrotron Radiation Research : Advances in Surface and Interface Science, Vol. 1 : Techniques",

244

R.Z.Bachrach Ed., Plenum Press, New York 63, (1992).

30. G.N.Greaves, P.J.Durharn, P.Quinn and G.Diakun, Nature 294 139 (1981 ).

31. G.Tourillon, E.Dartyge, A.Fontaine and A.Jucha, Phys. Rev. Lett. 57, 603 (1986).

32. M.M.Nicholson, Anal. Chern.32, 1058 (1960).

33. A.Tadjeddine, A.Lahrichi and G.Tourillon, J. Electroanal. Chern. (1993), in press.

34. J.W.Schultze, Ber Hunsenges. Phys. Chern. 74, 705 (1970). J.W.Schultze and K.J.Vetter, J. Electroanal. Chern. 44, 63 (1973). J.W.Schultze and F.D.Koppitz, Electrochirn. Acta 21, 327 (1976). S.Bruckenstein and S.Swathirajan, Electrochirn. Acta 30, 851 (1985).

35. M.R.Deakin and O.Melroy, J. Electroanal. Chern., 239, 321 (1988).

36. J. Mc Breen, W. E. 0' Grady, G. Tourillon, E. Dartyge and A. Fontaine, J. Electroanal. Chern.307, 229 (1991).

37. G.Horanyi and G.Vertes, J. Electroanal. Chern.45, 295 (1973).

38. G.Horanyi, J. Electroanal. Chern. 45, 45 (1974).

39. M. S. Zei, G. Qiao, G. Lehmpful and D. M. Kolb, Her Busenges Phys. Chern. 91, 349 (1987).

40. C. H. Chen, S. M. Vesecky and A. A. Gewirth, J. Am. Soc. 114, 451 (1992) .

41. A.Tadjeddine and G.Tourillon and D.Guay, Electrochern. Acta, 36, 1859 (1990).

42. A.Tadjeddine, D.Guay, M.Ladouceur and G.Tourillon, Phys. Rev. Lett.66, 2235 (1991).

43. O.R.Melroy, M.G.Sarnant, G.L.Borges, J.G.Gordon, L.Blurn, J. H. White, M. J. Albarelli, M. McMillan and H.Abrufia, Langrnuir 4, 728 (1988).

44. O.M.Magnussen, J.Hotlos, R.J.Nichols, D.M.Kolb and R.J.Behm, Phys. Rev. Lett. 64, 2929 (1990).

45. S.Manne, P.K.Hansrna, J.Massie, W.B.Ellings and A.A.Gewirth,

245

Science, 252, 183 (1991).

46. B.M.Ocko, J.Wang, A.Davenport and M.Isaacs, Phys. Rev. Lett. 65, 1466 (1990).

47. Y.Nakai, M.S.2ei, D.M.Kolb and G.Lehmpfuhl, Ber. Bunsenges. Phys. Chern. 88, 340 (1984).

48. J.Mc Breen, W.E. O'Grady, G.Tourillon, E.Dartyge, A.Fontaine and K.I.Pandya, J. Phys. Chern. 93, 6308 (1989).

49. K. I. Pandya, R. W. Hoffrnan, I. McBren and W. E. 0' Grady, J. Electrochern. Soc. 137, 383 (1990).

50. M.F.Toney and J.McBreen, Electrochern. Soc. Interface 2, 22 (1993).

51. A.Tadjeddine, M.Ladouceur, A.Lahrichi, D.Guay and G.Tourillon, in "X-ray rnethods in corrosion and interfacial electrochernistry", A.Davenport and J.G.Gordon 11 Eds., Electrochern. Soc. Ed. Pubi., Pennington (USA), 159 (1992).

52. G.Adzic, J.McBreen and M.G.Chu, J. Electrochern. Soc., 128, 1691 (1981) .

53. A.R.Despic and M.G.Pavlovic, Electrochern. Acta, 27, 1539 (1982).

In Situ X-Ray Absorption Spectroscopy of Nickel Oxide Electrodes

William E. O'Grady' and Kumi I. Pandya2

'Naval Research Laboratory Code 6170 Washington, DC 20375

2Brookhaven National Laboratory Upton, NY 11973

ABSTRACT: In M!l! x-ray absorption spectroscopy (XAS) has been used to study the changes in the structure of nickel oxide electrodes in real time and in the fully charged and discharged states. The extended x-ray absorption fme structure (XAFS) data showed that the nickel oxide electrode undergoes irreversible structural changes in the initial charging step but then the electrode cycles between two states which are reversible. The XANES data from the real time dispersive XAS experiments also show the nickel oxides undergoing the same changes observed in the XAFS data. The analysis of the XAFS data is discussed in detail for these oxide materials.

1. Introduction

The development of high intensity synchrotron light sources has created a revolution in the study of the structure of electrodelelectrolyte interfaces. As exemplified in this workshop it has now become possible to examine the structure of electrodelelectrolyte interfaces with a whole new battery of techniques which are sensitive to the electronic, atomic and molecular structure of these interfaces.

One very practical application of tltese techniques is in the area of oxide electrode structures. Oxides play a very important role in electrochemical processes such as chlorine production where the dimensionally stable anode (DSA) composed of primarily ruthenium and titanium oxides is used. The nickel oxide electrode or nickel hydroxide electrode (NOE) is another important oxide material which finds extensive use in batteries, electrolyzers and fuel ceHs. Another very important area where oxides playa role is in the area of corrosion prevention and metal surface stabilization. In alt of these cases the in situ structures of the metal oxides, including both the electronic and geometrie structures, remain largely unknown. In this paper the nickel oxide electrode will be used as an example of how x-ray absorption spectroscopy can be used to determine the in situ structure. We will focus on the data analysis as this

247

C. A. Melentires anti A. Tadjeddine (eds.), Synchrotron Techniques in lnteifacial Electrochemistry 247-261. © 1994 Kluwer Academic Publishers.

248

is the most critical aspect of XAS. The nickel oxide electrode has been used extensively in batteries. In spite of

the NOE's long history the development has been highly empirical and the understanding of the electrode reactions remains an enigma. The amorphous nature of the hydroxide and its oxidized product are the primary reason for this situation. The nature of the XAS technique is that it only monitors the local structure, as opposed to the long range order, making it ideal for studying disordered oxide systems.

2. Experimental

The cell designed to carry out these in situ experiments is shown in figure 1. The details of the cell and electrode preparation have been discussed in detail (1,2) and will only be summarized here. The ß-Ni(OH)z electrodes were a composite of nickel and cobalt hydroxides, graphite powder, carbon fibers and a poly(vinylidene fluoride) plastic binder. The electrodes were circular wafers 25.4 mm in diameter and 0.25 mm thick with a nominal charge capacity of 2 mAh. The electrochemical cell consisted of the nickel oxide electrode and a circular Grafoil electrode separated by three sheets of filter paper saturated with S.4M KOR. The electrodes and filter paper were held in Teflon gaskets which in turn are held in between two acrylic plastic blocks with 0.75 mm acrylic plastic windows at the center of each block. The entire assembly was bolted together under a slight compression. The reference electrode capillary mouth was elose to the edge of the NOE. A zinc wire was used as a reference electrode.

Figure 1. Schematic of the in situ x-ray absorption spectroscopy cello

249

To obtain reproducible results the nickel oxide electrodes were charged by anodically oxidizing the electrodes at 2 mA for 16 hours. Three ceHs were prepared in this way and then two of them were discharged by cathodically reducing the electrodes at 6 mA until the ceH voltage reached 1.0V vs the zinc reference electrode. One of these cells was then recharged at 3 mA for 7 hours. The times of these experiments were staggered so that each charging or discharging operation was completed within one hour of carrying out the XAS experiments. In another series of experiments the initial charging of the ceH at 2 mA for 16 hours was followed by a slow sweep cyclic voltammogram (ImV/sec) over the potential range of 0.6 to 2.1 V vs the zinc reference. The charging characteristics of these ceHs (1) suggest that 97-99 % of the nickel hydroxide was oxidized on charging assuming a one electron oxidation process.

Two series of experiments were carried out, one using standard transmission XAS and another using dispersive XAS. The set up for a standard transmission XAS is shown in figure 2a. The x-rays from a synchrotron source pass through a monochromator, where they are Bragg reflected from two crystals producing a monochromatic beam of photons. The beam then passes through an I., detector, the sampie and into an I detector, to determine the amount of absorption in the sampie. A spectrum is generated by stepping the crystals of the monochromator to different angles (different energies) and then recording I., and I. This procedure is repeated until the entire energy range has been covered. Operation in this mode requires 15-30 minutes to record a complete spectrum. This amount of time for recording a spectrum is acceptable when the sampie is in the fully charged or discharged state but it is clearly a drawback if data at voltages in between the limits are of interest. To overcome this time limitation, facilities have been developed to carry out what is referred to as dispersive x-ray absorption spectroscopy (dispersive XANES and XAFS) experiments. The facility described here is at LURE in Orsay, France (2). The dispersive technique allows time resolved spectra to be recorded in times as short as 4 msec (3,4). The spectrometer is composed of the x-ray dispersive optics and the detection system shown schematically in figure 2b. The x-ray optics consist of a 23 cm long triangle shaped Si(lll) or Si(311) crystal and a mechanism to bend the crystal. The curvature of the crystalleads to a progressive change of the Bragg angle along the center line of the crystal and this results in a photon-energy reflecting angle correlation and ultimately to a pixel number-energy correlation. Further, the bent crystal also leads to a focusing of the dispersed polychromatic beam with a 350",m wide focai spot. The sampie to be investigated is positioned at this focai point. Upon passing through the sampie the beam continues to diverge until it reaches the position sensitive detector which is placed so that the beam is spread across the entire detector array. The Reticon photodiode array is composed of 1024 sensing elements (2500",m high and 25",m wide) and here the pixel number-photon energy correlation is established.

In these experiments the ceH is mounted on the line and is potentiostated at 1.88V vs Zn (the voltage it had foHowing charging) and then a cyclic voltammogram is initiated. The potential is cycled at ImV/sec down to 0.6V and then back to 2.1V.

250

'.

DAtA ~lSln(M SYSTEH

Figure 2a. Schematic of the experimental setup of a step-wise XAS

Figure 2b. Schematic of the experimental set up for dispersive XAS.

Sixty spectra were recorded at I-minute intervals with each spectrum being an average over 13 seconds. The first 50 spectra recorded the data for a complete voltammogram.

3. Data Analysis

The XAFS function X(k) is defined as, X(k) = IL-ILj ILo where IL and lLo are the x-ray absorption coefflcients of the absorbing atom in the material of interest and in the free state respectively. The difference WlLo depends upon the local structure of the absorbing atom and represents the XAFS. The division by ILo normalizes the XAFS oscillations to a per atom basis. The wavevector k of the ejected photoelectron is given by k = [2m/~(hp-E..-EJ] where m is the mass of the electron, pis the frequency of the x-ray photons, E.. is the binding energy of the electron and Eo is the correction to the binding energy caused by the atomic potentials.

The XAFS oscillations were separated from the absorption background using a cubic spline background removal technique (5) and were subsequently normalized to a per atom basis by dividing by the step height of the absorption edge. Since the reference compounds were normalized in the same way, the step normalization was satisfactory. The theoretical expression which relates the measured XAFS parameters

251

to the single scattering approximation is given by (6-8).

x(k:) = I:Aj(k:)Sin(2~ + cPj(k:» (1)

Here the sum extends over j neighboring shells. R; is the average interatomie distance from the absorber atom to the neighbor atoms in the jth shell. cbj(k:) is the total phase shift suffered by the electron in the outgoing and backscattering processes. ~(k:) is the amplitude funetion of the jth shell and is given by,

Nj

Aj(k:) = So2(k)F;(r,k) e-2(Rj-A)f)j-2D:J"2t2

kR2

(2)

where Nj is the average coordination number and Fj(k:) is the backscattering amplitude of the atoms in the jth shell. U;2 is a Debye-Waller term whieh accounts for the thermal vibration (assuming harmonie vibrations) and statie disorder (assuming Gaussian pair distribution) present in the material. So2(k:) is an amplitude reduetion faetor whieh takes into account the relaxation of the absorbing atom and multi-electron excitations (shake up/off) processes at the absorbing atom. A(k:) is the mean free path of the photoelectron and ä is a correction faetor (~= R1) to the mean free path since So2(k) and Fj(k) already account for most of the photoelectron energy losses in the first coordination shell. Taking the Fourier transform of the XAFS data yields the radial structure funetion which contains information about the individual coordination shells and is given by

(3)

The function x(k) is multiplied by a factor Je' to equalize the envelope of x(k) over the transformation range. The value of n is normally chosen from 1 to 3 depending upon the amplitude variation and the signal-to-noise ratio of the measured data. The radial structure function 9 n(r) contains aseries of peaks which reflect the local structure. In order to reduce the termination errors which. arise from the finite range of the Fourier transform, the values of k",;.. and k.n.x were chosen to coincide with nodes of the x(k) function.

If the coordination shells in the radial structure function are weH separated, a subsequent inverse transformation can be performed to ftlter out the XAFS function of a particular shell of interest. The limits for the inverse transform are chosen to coincide with the nodes in the imaginary part of the complex Fourier transform in order to reduce truncation errors. A fitting in k space is then performed using an iterative least square technique. The parameters obtained from the fitting are checked by comparing the Fourier transforms of the experimental XAFS with the fitted XAFS function_

252

When a k" weighted Fourier transformation is applied to an XAFS signal containing high Z scatterers, sidelobes are introduced in R-space on both sides of the peak (9,10). A small contribution to the sidelobes is caused by the fmite integration range which cannot begin at zero. The major contributions to the sidelobes are nonlinearity in the phase shift function and low frequency variations in the backscattering amplitude. These low frequency variations increase with increasing atomic numbers and are important for elements which have atomic numbers higher than titanium (Z=22). Sometimes, these sidelobes have amplitudes which may be as large as 30% to 40% of the amplitude of the main peak. When peaks due to other shells are proximate to the main peak, these sidelobes may significantly interfere with the signals of the other shells. With standard Fourier transform techniques, it is impossible to separate these peaks by an inverse transformation as described above. By increasing the power of k" in the Fourier transform, the intensity of the sidelobes may be reduced to a considerable extent. However, this amplifies the noise at high k values. The most important disadvantage in using higher n values is that the Fourier transform becomes less sensitive to the contributions of low Z scatterers (like oxygen) which have their scattering amplitude primarily at low k values.

A Fourier transforrn, which corrects for the phase shift function and the backscattering amplitude reduces to a transform of a sine-like function with very small sidelobes due to truncation effects. Sueh a corrected Fourier transformation is given by (9):

x(k)e-i.pj(k)e2ikr

---------------------- dk Fl"·,k)

Here cMk) and Fllr,k) are obtained from suitable referenee eompounds. This correction technique has several advantages: (a) since the phase shift is

removed, the interatomie distances may be obtained direct1y from the transformation peak positions, (b) nonlinearities in the phase shift and amplitude funetions are removed, henee, the peaks are narrower and more symmetrie. This minimizes the ambiguity during inverse transformation making a single shell more readily isolated.

Analyzing the imaginary part together with the magnitude of a corrected Fourier transform has important advantages: (a) If the imaginary part of a phase and amplitude eorrected Fourier transform is asymmetrie, it indicates that contributions from other types of atoms are present, i. e., the separation of the shell is not complete. (b) From the shape of the imaginary part, it is often possible to distinguish the types of atoms contributing to a given shell, because of the phase shift dependence of different atoms. (e) When the proper phase eorrection has been made, the imaginary part of the phase corrected transform will peak at the same point that the magnitude of the transform peaks. If they do not peak at the same value a eorrection in the inner potential (EJ is required.

253

4. Results and Discussion

Phase and amplitude functions for a specific absorber-scatterer pair are required for the analysis of XAFS data. These functions can be obtained from theoretical calculations or from the XAFS data of weil defined reference compounds. Ni, NiO and ß-Ni(OH)2 were used as reference compounds and the phase and amplitude functions obtained from the Ni and NiO were used to analyze the ß-Ni(OH1 sample and the ß-Ni(OH)2 was finally used as a reference for analyzing the structure of the in situ battery materials (l ,2).

In figure 3a the normalized XAFS data recorded at 77K for ß-Ni(OH)2 are shown. The corresponding radial structure function is shown in figure 3b and it exhibits three well-resolved peaks. The first peak corresponds to the first Ni-O shell, the second peak corresponds to the first Ni-Ni shell and the third peak corresponds to the third interplanar Ni-Ni shell. The positions of these three shells are shown in figure 4 where the unit cell of ß-Ni(OH1 is shown together with the structure of the basal plane which contains the atoms giving rise to the observed XAFS.

10

('J

~

" ~ 0 ~

-10

0 5 10 15 k $.-1

12.5 b

10.0

E=<7.5 Ii. D.O

2.5

0.0 0 2 4 6 8

RA

Figure 3. (a) Normalized XAFS spectrum of ß-Ni(OH1 measured at 77K; (b) Radial structure function for ß-Ni(OH)2 measured at 77K (-) and at 297K (---).

254

°1 . Ol2A LU

ONi ~OH

A B

Figure 4. Structure of ß-Ni(OHh (a) interplanar structure; (b) stacking along the c­axis; Ni-Ni shells marked 1,2,3.

The analysis of the 77K ß-Ni(OH)2 will be discussed in detail serves as a model for the other sampies analyzed. The amplitude and phase functions derived from the first Ni-O and Ni-Ni shells of nickel oxide were used to analyze the corresponding Ni­o and Ni-Ni shells of the ß-Ni(OHh. The amplitude and phase functions from the fourth shell of nickel foH were used to analyze the third coplanar Ni-Ni shell in ß­Ni(OH)2' The magnitude of the third coplanar shell is unusually large and is attributed to the shadowing by the first Ni-Ni shellieading to what is called the focusing effect (11). This effect arises because the first shell of Ni atoms act as a lens "focusing" the electron wave on the next interplanar Ni atoms in the third shell(see figure 4a). In order to analyze this shell a reference material with a similar focusing effect is required and this criteria is satisfied by the fourth shell in Ni foil.

The first Ni-Ni shell contribution was isolated by performing a Icl-weighted Ni­Ni phase-corrected Fourier transform followed by an inverse transform. A fitting of this Ni-Ni shell was carried out in K-space over the interval ak = 4-14k1 varying the parameters N, R" aa'- and aBo. Based on this fit an XAFS function was calculated, Fourier transformed and inverse transformed in precisely the same way as the ß­Ni(OH)2 XAFS spectrum to produce a new reference spectrum. This new reference spectrum was used in the successive fitting procedures because it contained the same transformation artifacts as the measured XAFS and lead to more rapid convergence in the fitting calculations. Following a rigorous scheme to check the possible correlation between R, aBo and N, aa'- (12) the best fit results are shown in Table I. Using these

255

pammeters an XAFS function was calculated and is compared to the experimental data in figure 5a. In figures 5b and 5c the imaginary and real parts of the Fourier Transform (Fr) of the experimental and calculated XAFS functions are compared.

From the experimental ß-Ni(OH)2 XAFS data the calculated Ni-Ni XAFS is subtracted. From this difference file the Ni-O XAFS contribution was isolated by applying a k1-weighted Fr. The same procedure used for fitting the Ni-Ni XAFS function was used for the Ni-O XAFS function. Fitting over the k-space interval dk = 3.6 - lOkl with a k1-weighting and using the amplitude functions obtained from the nickel oxide XAFS produced the results in Table 1. These values were used to calculate a Ni-O XAFS function which is compared to the experimental data in figure 5d. In figures 5e and 5f the imaginary and real parts of the Fr of the experimental and calculated XAFS functions are compared .

5 • '

0.3 I.S ft

'" '" ... ~ '" ~ 'l< 'l< 'l< 0 0

0

-5 -02 -1.5

0 5 k A-I 10 15 0 2 1.: 1-16 8 10 12 0 5 k .1.-1 10 15

b 0.2 e h

5 t: 0.1

,g }, f'I.

0 0

11 0.0

-I

-s -0.1

0 Z RA4 6 8 0 2 RA 4 6 8 0 :;: R J. 4

C h I.S

O.IS

5 1.0

~ 0.10

0.05 O.S

\.. 0.0 .-~ ~

0 0.00 0 2 RA4 8 0 2 RA 4 6 8 0 2 R J. 4

Figure 5. XAFS results for ß-Ni(OHh measured at 77K using NiO and Ni foil as the reference compounds. (a) Isolated experimental (-) and calculated (---) XAFS functions for the first Ni-Ni shell; (b and c) Imaginary parts and magnitudes of the phase corrected fourier transform of the XAFS data in Figure 5a; (d) Isolated experimental (-) and calculated (---) XAFS for the first Ni-O shell; (e and f) Imaginary parts and magnitudes of the phase corrected Fourier transform of the XAFS data in Figure 5d; (g) Isolated experimental (-) and calulated (---) XAFS for the coplanar third Ni-Ni shell; (h) Imaginary parts and magnitudes of the phase corrected Fourier transform of the XAFS data in figure 5g.

256

A further check of the quality of the Ni-Ni fit is carried out by subtracting the calculated Ni-O XAFS function from the experimental ß-Ni(OHh XAFS data. Then aphase and amplitude corrected Fr "is carried out on this difference file. If the Ni-Ni peak in the imaginary part of the Fr was symmetric then the fit parameters were considered satisfactory. However, if it was not symmetric the inverse Fr of the difference spectrum was reanalyzed until a symmetric peak was obtained.

As already discussed, the third peak in the radial structure function of ß­Ni(OH)z is quite unusual. The fourth shell of Ni foil was found to provide a suitable reference and it gave meaningful results as shown in Table I. The XAFS function calculated from these results is compared with the experimental data in figure 5g. The imaginary and real components of the Fr of the experimental and calculated XAFS are shown in figures 5h and 5i. The structural parameters obtained from the XAFS analysis of ß-Ni(O~ recorded at 77K and ß-Ni(OH)z recorded at 297K using nickel oxide (77K) and nickel foil (77K) are summarized in Table 1.

Table 1 Structural Parameters from XAFS Analysis Using NiO (77K) and Ni(77K) as References

Sampie TempK SheH R(A) N fl.rr(AZ) !lEJ.eV)

ß-Ni(OH)z 77 Ni-O 2.07 6.0 -0.0012 -1.89 Ni-Ni 3.13 6.2 0.0002 1.85 Ni-Ni 6.26 6.8 0.0018 -0.13

ß-Ni(OH)2 297 Ni-O 2.07 5.9 0.0004 -3.07 Ni-Ni 3.12 6.0 0.0026 -1.31 Ni-Ni 6.26 6.0 0.0054 -1.92

As a further check on the quality of the fitting, amplitude and phase functions were extracted from ß-Ni(OH)z (77K) for the Ni-O and Ni-Ni interactions. These functions were then used to analyze the ß-Ni(OH)2 (297K) XAFS; the results are summarized in Table 11. The values for the parameters here agree very weH with those found using nickel oxide and nickel metal as references.

Figure 6 shows the experimental XAFS data for the dry electrode and the fuHy charged electrode. The 12-weighed fl.k = 2.6 -14k1 Fr's for the dry electrode and the charged electrode are shown in figure 7. In both figures 6 and 7 an increase in the magnitude of the XAFS and the radial structure functions are observed when the ß­Ni(OH)2 is oxidized. This increase is in agreement with the mathematical description of the magnitude in equation 2. There the magnitude is shown proportional to lIR2 therefore if the atoms move eloser together the magnitude of the XAFS will increase as ohserved. The cyeled electrode XAFS spectra were analyzed using the Ni-O and Ni­Ni phase and amplitude functions extracted from the dry ß-Ni(OHh electrode XAFS recorded at 297K.

257

10

(') .!( .. 0 ~ ><

-10

10

(') 5 .!( .. ~ 0 ><

-5

-10

0 5 10 15 k X-I

Figure 6. Normalized XAFS spectra (a) dry electrode; (b) charged (oxidized) electrode.

8

6

4

2

8

6

4

2

2

b.CHARGED

4 6 8 RA

Figure 7. Radial structure functions. (a) dry electrode; (b) charged (oxidized) electrode.

258

4.1 Char~ed Electrode. The XAFS data for the electrode charged at 2mA for 6 hours are shown in figure 6b. The three peaks in the Ff (figure 7b) are sufficiently separated that they were isolated by simple inverse Ff's. In the case of the Ni-Ni peak a small shoulder on the high R side was observed in both the imaginary and real parts of the Fr as seen in Figure 8a and b. The isolated Ni-Ni XAFS shell was fitted in k-space over the interval dk = 4-13A-l with Ic2 and I(J weighting. It was not possible to achieve a satisfactory fit with meaningful parameters for this shell using only a single shell fit, so a two shell fit was attempted. Again using a scheme (12) of fixing some variables while allowing others 10 float led to a satisfac10ry fit in k-space and the results are summarized in Table 11. The quality of the fits was checked in R­space for both the imaginary and real parts as shown in figures 8a and 8b. Figures 8c and 8d show the relative contributions of the two Ni-Ni shells in k-space and R-space. It was not possible to fit the Ni-O contribution of the charged electrode with a single shell. Following a procedure similar to that applied to the Ni-Ni contribution above, a good fit with meaningful parameters was achieved by using two Ni-O shell contributions and the results are shown in Table 11. It was not possible 10 achieve a satisfactory fit for the third Ni-Ni shell using the amplitude and phase functions obtained from the dry ß-Ni(OH)2 electrode. Since both valence and core electrons are involved in the focusing effect it may be that it is not possible to transfer the amplitude and phase functions from a Ni(Il) compound to a Ni(1I1) compound.

I 10 f--

S

"" ;>:

• t 5 r­

~ 0 ';;;: E 0 ~ II~Wh~,~~_.........j

-Sf-- -s

0 2 RA 4 6 8 0 5 10 15 k X-'

10 - I A b - 10 d

8 - 8

F 6 c- -: F 6

- f:,. 4

f:,. 4 f--

2 l- - 2

4 0

6 8 0 0

0 2 2 4 6 8 RA RA

Figure 8. Comparision of the experimental and calculated data for the first Ni-Ni shell of charged electrodes. (a and b) are the respective imaginary parts and magnitudes of the Fourier transforms of the experimental (-) and calculated (---) data; (e and d) show a comparision of the experimental data (-) of the XAFS and Fourier transform. The calculated data show the major (---) and minor (""") contributions.

259

Table n Struetural Parameters from XAS Analysis Using Dry ß-Ni(OH)2 as a Referenee

Sampie Shell R(A) N .6.r2(A2) .6.Eo(eV)

Dry Ni(OH)2 * Ni-O 2.06 6.0 0.0017 -2.81 Ni-Ni 3.12 6.0 0.0027 -2.63 Ni-Ni(3) 6.21 5.5 0.0037 -0.94

Charged Ni-Ol 1.88 4.1 -0.0039 7.01 Ni-02 2.07 2.2 0.010 0.32 Ni-Nil 2.82 4.7 -0.0023 6.55 Ni-Ni2 3.13 1.0 0.0072 -0.09

Charged & Ni-O 2.05 5.7 0.0016 1.92 Diseharged Ni-Nil 2.82 1.0 0.0009 -7.92

Ni-Ni2 3.13 5.0 0.0007 0.08 Ni-Ni(3) 6.25 5.7 0.0029 -2.65

Recharged Ni-Ol 1.88 4.1 -0.0039 7.05 Ni-02 2.07 2.2 0.009 0.48 Ni-Nil 2.82 4.7 -0.002 6.63 Ni-Ni2 3.13 1.0 0.006 0.11

* Using ß-Ni(OH)2 (77K) as a referenee

4.2 Diseharged Electrode. The XAFS analysis of the data for the diseharged electrode gave results only slightly different from those found for the dry starting electrode. The Ni-O data eould be fitted with a single shell fit while the Ni-Ni data again required a two shell fit. An excellent fit was achieved for the third Ni-Ni shell. The struetural parameters obtained from these fits are shown in Table Ir. 4.3 Recharged Electrode. The XAFS data for the recharged material were almost identieal to those obtained for the material eharged onee. The struetural parameters are shown in Table n.

In summary the fit of the dry sampie data shows the Ni eoordinated to six equally distant oxygen atoms and six equally distant nickel atoms indicating a symmetrie octahedrally eoordinated nickel. When the cell is eharged two Ni-O and two Ni-Ni distanees are observed indicating that the nickel coordination shell has become distorted. When the sampie is discharged we see that there is only one Ni-O distance; however two Ni-Ni distanees are required to aehieve a satisfactory fit of the data. Hence, once the coordination sphere is distorted it does not return to that observed with the starting ß-Ni(OH)2' When the eell is recharged a second time it is seen that the same set of values are obtained in the fitting which suggest that the eoordination state of the Ni in the charged eell is that of a distorted octahedron.

260

Now we consider the real time dispersive XAFS results (13) to see what further insights tnay be obtained. In figure 9 a set of XANES data obtained during a cyclic voltammogram for ß-Ni(OH)2 is shown. Figure 9a shows the changes that occur when the potential is swept from 1.36 to 2.08V and the sampie is oxidized; figure 9b gives the corresponding results when the potential is swept from 1.88 to 1.0V and the sampie is reduced. The first thing to note from this data is that the inflection point of the edge shifts from 8343.8eV to 8345eV due to the transformation of Ne+ to NP+. Another very interesting point to note is that no NP+ is observed in the XANES spectra until a potential of 1.90V is reached. This is O.15V higher than the potential at which Ni2+ to Ni3+ oxidation should occur. We have no explanation at this point for this observation. In figure 10 an expanded version of the XANES data for an uncharged sampie, a charged sampie and a discharged sampie is shown. The ß­Ni(OH)2 has brucite type structure in which the Ni atoms are each surrounded by six oxygen atoms with an octahedral configuration (figure 4). In this configuration the final state of the metal components have only d-character, transitions from the ls level are therefore dipole forbidden because the octahedron has a center of inversion. Thus no pre-edge features are expected and only a strong white line due to multiple scattering in the Ni06 cluster are expected; this is observed in the uncharged sampie in figure lOc. Any distortion of the octahedral environment would remove the center of inversion (p-d hybridization) and this would lead to a weak pre-edge peak or shoulder. In the case of both the charged sampie (figure 1Oa) and the discharged sampie (figure lOb) a shoulder is observed at -5eV. This new shoulder and the overall decrease in intensity of the white line is evidence for the development of a distorted octahedron upon charging which is not reversible upon discharging. Hence, the XANES data are in excellent agreement with the XAFS results, showing that an irreversible structural change occurs when the ß-Ni(OH)2 is oxidized to NiOOH in a nickel battery.

1 t-I! .. -e .:! ~

t ~ ... <

4

3

l!

-20 -10 0 10 20

Energy (.VI

B

-20 -10 0 .10 20

Enargy (ev>

1.oV(a)

1.22V(c)

1.28V(c)

1AOV(c)

1.Ii8V(c)

1.84V(c)

1.70V(c)

1.88V(c)

Figure 9. XANES data from the dispersive experiments. (a) during the anodic (oxidation) scan; (b) during the cathodic (reduction) scan.

261

-I

ENERGY(eV)

Figure 10. Expanded XANES from the dispersive experiments. (a) fully charged electrode; (b) discharged electrode; (c) dry electrode.

s. Acknowledgements This work was supported by the Office Naval Research.

6. References

1. J. McBreen, W.E. O'Grady, K.I. Pandya, R.W. Hoffman and D.E. Sayers, Langmuir, ~, 428 (1987).

2. K.I. Pandya, R.W. Hoffman, J. McBreen and W.E. O'Grady, J. Electrochem. Soc., 137, 383 (1990).

3. E. Dartyge, L. Depautex, J.M. Dubuisson, A. Fontaine, A. Jucha, P. Leboucher and G. Tourillon, Nucl. Instrum. Methods A, 246,452 (1986).

4. J. McBreen, W.E. O'Grady, G. Tourillon, E.Dartyge, A. Fontaine and K.I. Pandya, J. Phys. Chem., 93, 6308 (1989).

5. J.W. Cook and D.E. Sayers, J. Appl. Phys., ~, 5204 (1981). 6. D.E. Sayers, E.A. Stern and F.W. Lytle, Phys. Rev. Lett~, 27, 1204 (1971). 7. E.A. Stern, Phys. Rev. B, W, 3027 (1974). 8. P.A. Lee and J.B. Pendry, Phys. Rev. B, 11, 2795 (1975). 9. J.B.A.D. Van Zon, D.C. Koningsberger, H.F.J. Van't Blik and D.E. Sayers,

J. Chem. Phys., 82, 5742 (1985). 10. E.C. Marques, D.R. Sandstrom, F.W. Lytle and R.B. Greegor, J. Chem.

Phys., 77, 1027 (1982). 11. B.K. Teo, in EXAFS: Basic Principles and Data Analysis, Springer-Verlag,

Berlin (1986). 12. K.I. Pandya, W.E. O'Grady, D.A. Corrigan, J. McBreen and R.W. Hoffman,

J. Phys. Chem., 94, 21 (1990). 13. J. McBreen, W.E. O'Grady, G. Tourillon, E. Dartyge, A. Fontaine and K.I.

Pandya, J. Phys. Chem., 93, 6308 (1989).

Tbe UPD of Copper on Pt(lOO). "In situ" EXAFS and "ex situ" structural LEED investigations

D. ABERDAM and Y. GAUTHIER Laboratoire de Cristallographie, CNRS BP I66X, F-38042 Grenoble cedex

R. DURAND and R. FAURE Centre de Recherche en Electrochimie Minerale-Genie des Procedes, ERA # 1212 Institut National Polytechnique, BP 75, F-38402 St Martin d'Heres cedex

ABSTRACT. The adsorption of copper on platinum, obtained by electrochemical under-potential deposition (upd), is a complex phenomenon. As observed by cyele voltammetry, the upd is not limited to a narrow, weIl defmed range of electrochemical potentials, but is spread out on a wide range of potentials. This is particularly true on the Pt(lOO) surface, where, in addition, a lack of reversibility of adsorption and desorption occurs, and a gradual change in the shape of the voltammogram takes place during potential cyeling.

In this paper, we describe the results of some investigations of this complex system by two very different surface science techniques, namely "in situ" X-ray Absorption Spectroscopy (XAS) and "ex situ" Low Energy Electron Diffraction (LEED). It is not known apriori whether the objects studied "in situ" by XAS and "ex situ" by LEED are the same, and in addition the experiments are rather difficult to perform. Furthermore, any analytical method, used elose to the limits of its possibilities, may be the source of reliability problems, in particular in relation with the extension of the data basis. We discuss these various aspects of our investigations, putting more emphasis on the LEED analysis, because the XAS study was recently published 1.

Introduction

Electrochemical studies of the underpotential deposition 2,3 of copper on Pt(100) were performed by Kolb 4, Scortichini and Reilley 5, Andricacos and Ross 6, and EI Omar 7. Various shapes of voltammograms, changes upon potential cyeling, unsolved questions relative to charge transfer prove that this is a complicated system. Electron spectroscopies are attractive to investigate the structure and composition of the adsorbate layer because they are very surface-sensitive (3-10 monolayers). They are qualified as "ex situ" techniques because of the prerequisite of electrode emersion. This emersion may be the source of uncontrolled chemical reactions and "in situ" techniques are desirable. Most of them make use of X-Rays, which are able to penetrate a thin layer of an aqueous electrolytic solution. However, and in spite of the efficiency of specially designed tricks to improve surface sensitivity, they are by far less surface-sensitive than electron spectroscopies. Consequently each

263

C. A. Melendres anti A. Tadjeddine (eds.), Synchrotron Techniques in Inteifacial Electrochemistry 263-280. © 1994 Kluwer Academic Publishers.

264

type of technique will give a limited but hopefully complementary answer to the question of the structure and composition of the interface.

In this paper we describe the results of "in situ" XAS and "ex situ" LEED investigations of the electrochemical interface CuA>t(I00), and discuss the results from both the knowledge gained on the interface, and the questions arising due to the experimental procedures and their reliability. The paper is organised in three main sections. In section 1 we recall the results ofthe XAS study, with the information necessary to understand the experimental conditions and the reasoning which lead to the results. In section 2 we describe the LEED investigation with some emphasis on methodology, in order to introduce the discussion of section 3 which deals with the differences between the results obtained by both techniques.

1. X-Ray Absorption Spectroscopy study of the upd of Cu on Pt(100).

1.1. EXPERIMENTAL SET UP.

The experiment was performed at LURE (Orsay, France), using the synchrotron radiation of the DCI storage ring. The detection of the EXAFS signal was made at the Cu K edge located at about 8980 eV, in the fluorescence mode, to select information from the actual interface. The platinum (100) surface was that of a single crystal 3 mm thick, 11 mm in diameter. Cleaning of the surface prior to copper upd was accomplished in the flame of a Hz-02 blow pipe, according to the recipe designed by Faure and described in reference 8.

The electrochemical cell, made of KEL-F®, had been designed by Tadjeddine for his previous studies of the upd of copper on gold 9,10. The electrolytic solution was 1M NaCI04, 10-3 M HCI04 to achieve a good conductivity with a moderate acidic pH, and 10-4 M Cu2+ to keep negligible the contribution of copper ions in solution to the fluorescence signal. The small number of atoms absorbing X-Ray photons required a long accumulation time (28 EXAFS spectra, 600 e V wide, were recorded by steps of 2 e V at a time rate of 4 s per step, i.e. more than 9 hours), not counting the recording of XANES spectra.

Voltammograms recorded either in the EXAFS electrochemical cell or in a more standard cell are not perfect. The kinetics of copper diffusion is slow, due to the low concentration of the solution in copper ions, so that the adsorption peak is spread out. The single crystal was placed in a PTFE ring to minimise the contribution of its lateral surfaces to the voltammogram. This increased the electric resistance elose to the crystal and induced a distortion of the voltammogram.

Altogether, this makes the electric charge measurement under the voltammetric wave uncertain. The result is about 420 ~C/cm2, corresponding to the adsorption of a copper atom per platinum, assuming a charge transfer of 2 electrons. After more than 9 hours of exposure to the X-Ray beam, the anodic dissolution showed that the copper layer was still present on the surface.

1.2. INFORMA 'flON FROM THE SHAPE OF THE THRESHOLD.

When compared to the shape of thresholds recorded 10 for metallic copper, Cu+ and Cu 2+, the shape of the copper threshold in our experiment is not typical of any of these oxidation states, but resembles more that for Cu+. This might be an indication that copper atoms are not all in the same oxidation state.

265

1.3. EXAFS DATAANALYSIS ANDRESULTS.

Standard analysis of the data consists in background subtraction, normalisation, conversion of the energy E-scale to wave-number k-scale, and weighting by k to get the k.X(k) function. A Fourier filtering of the various shell contributions to the EXAFS signal is performed to fit these contributions with calculated models. Atom scatterinr properties, necessary ingredients for the model calculations, were taken from McKale's tables 1. The energy range of the processed data was not very large, about 450 eV, because ofthe presence of a Bragg peak which shortened the usable EXAFS range. Consequently, the Fourier transform of the k-weighted, normalised, k.X(k) function were somewhat broadened.

Main peaks in the Fourier transform ofk.X(k), not corrected for atom phase-shifts, are located at 0.15 and 0.30 nm, (fig. 1). The latter is the superposition of unresolved contributions, with a shoulder located roughly at 0.37 nm. These unresolved contributions were Fourier filtered as a whole.

30

25 CI)

..:! :::::I 20

"C 0 ::!: 15 I-LL LL 10

5

2 3 4 5 6 7

r (A) Figure 1 : Fourier Transform ofk.X(k), not corrected for phase-shifts. Peak at 0.15 nm is assigned to oxygen neighbours. Peak at 0.30 nm is the unresolved contribution of first and second copper neighbours.

The first peak was assigned to the contribution of a first neighbour shell of oxygen atoms located at a distance of 0.197 nm from the absorbing copper atom. This distance agrees well with known copper-oxygen distances in various copper-oxygen compounds 12.

Fourier filtering of the second composite peak showed a sharp beat node. Consequently, this peak is built up from the contributions of two shens of neighbours at wen defined distances. These contributions were assigned to copper first and second neighbours, no platinum contribution being able to fit the data. (fig. 2). The first neighbour distance dlCu-Cu is found to be 0.309 nrn. This is rnuch larger than both the dPt-Pt frrst neighbour distance and the dcu-Cu first neighbour distance in copper oxides The second neighbour distance d2Cu-Cu (0.393 nrn) is equal to the lattice parameter of bulk platinurn.

266

No significand contribution of platinum was found, which may be an indication that the Cu-Pt first neighbour bond is perpendicular to the electric vector of the X-ray beam, that is, perpendicular to the surface. Cu atoms may thus be located on top of the platinum atoms.

-. 0.2 ~

I

oe:( ....... -. 0 ~ ....... X

-0.2 ~

4 6 k (A -1)

8 10

Figure 2. Solid line : Fourier filtered EXAFS contributions of Cu neighbours. Dotted line : calculated EXAFS with Cu at 0.309 and 0.393 nm.

The following model rationalizes these results. (i) A square array of Cu atoms sits on top of the Pt atoms of the unreconstructed Pt(lOO)

surface, with half-monolayer density (1 Cu / 2 Pt), thus forming an adlayer with a c(2x2) surface mesh periodicity. Oxygen atoms sit in the plane of copper atoms, at the centre of the squares.

(ii) An outer adlayer, identical to the previous (inner) one, sits at a distance normal to the surface of 0.135 nm. Parallel to the surface, the outer adlayer is shifted so that a copper sits above the oxygen of the inner adlayer. The total copper coverage in this model is about one copper per platinum, as expected from the UPD process.

The oxygen is located at the centre of the squares of copper atoms because (i) the Cu-O distance (0.197 nm) is half the in-plane Cu-Cu distance (0.393 nm), and (ii) the strong contribution of oxygen neighbours is possible only if they lie in (or dose to) the plane of copper atoms, the electric vector of the X-ray beam being parallel to the surface.

In principle, this geometry requires that the forward scattering of the electron wave by oxygen atoms on its path from the Cu absorber to the Cu second neighbour (and conversely) be taken into account. Doing this proved to produce little change in the least mean square fit between the calculated and the Fourier filtered EXAFS. The quality of the raw data is not good enough to justify the addition of the required additional parameters. The number of neighbours of a copper atom in the first (oxygen) co-ordination shell is not reliably determined. In the two further (copper) shells, it is roughly equal to 4, as expected from the model. The accuracy on the number of neighbours is poor, because these numbers are always difficult to determine, and, in addition, our data were not particularly good.

267

1.4 SUMMARY OF RESULTS.

We summarise the above results as follows : (i) Electrochemistry says : about one copper per platinum. (ii) Threshold says : oxidation state non zero, some oxygen must be present. (iii) EXAFS says : two adlayers of copper of half density in c(2x2) configuration, shifted

parallel and perpendicular to the surface; some oxygen is located at the centre of the squares, in the planes of the adlayers.

Weshali discuss the peculiarities of this modellater on.

2. LEED study of the upd of CuIPt(lOO).

2.1. EXPERIMENTAL SET UP AND PROCEDURES.

The "ex situ" experiments with surface science techniques requires the transfer of the electrode from the vacuum system to the electrochemical cell, and conversely, without contamination. Platinum adsorbing organic molecules easily, this transfer must take place without any contact with the atmosphere. For this purpose, a vacuum system has been designed, to which an electrochemical cell can be attached. The vacuum chamber is equipped with an Auger electron spectrometer (AES) of the CMA type, a quadrupole residual gas analyser, a standard four-grid LEED optics, an ion bombardment gun, a vacuum gauge, and a liquid nitrogen cooled gas trap. This sys~em has. already been ~escribed in earlier pa~ers, together with the procedures allowing "clean" ImmerSIOn and emersIOn of the electrode 1 -15. Let us only recall that the water and electrolytic solutions are vacuum degassed, and that the atmosphere over the cell in the electrochemical chamber attached to the vacuum system is either flowing high purity argon or flowing high purity nitrogen.

2.2 SAMPLE PREPARATION AND UPD OF COPPER.

The platinum crystal was grown by Faure 8. It was spherically shaped, with a diameter of about 3 mm. This sphere was then X-Ray oriented, cut along the diametral (100) crystallographic plane by electro-erosion, and mechanically polished. The Pt(I00) surface was prepared in the vacuum system according to the standards of surface science techniques. It was ion bombarded with argon. The vacuum chamber was filled with argon gas up to 1.3xl0-2 Pascal (10-4 torr) in the presence of a liquid nitrogen cooled gas trap. This is essential to avoid graphitic residuals from depositing on the platinum surface. The crystal was then annealed at about lOOooK for a few minutes and cooled rapidly to prevent the segregation of dissolved impurities.

The process was repeated until (i) a careful analysis of surface composition with the CMA shows that no impurity is detectable, and (ii) a sharp, contrasted LEED pattern exhibits the well known complex reconstruction of Pt(100).

The crystal was then transported under vacuum into the electrochemical chamber, which was then filled up to atmospheric pressure with high purity argon or nitrogen. It was then covered with a fresh droplet of 18 Mn-cm water and immersed in the electrolytic solution with the dipping technique. The electrolytic solution was 1M HCI04 (pH=O) and 0.5xlO-2 M Cu2+; the reference electrode used was a reversible hydrogen electrode (rhe). The initial voltammogram was very similar to that obtained by EI Omar 7. It is characterised by a sharp anodic dissolution

268

peak at a potential depending on the potential sweep rate, e.g. elose to 770 mV (rhe) at a sweep rate of 50 mV/so The corresponding adsorption peak was very broad and located at a much lower potential at the same sweep rate. The electric charge measured under the voltammogram, between the minimum (0.3 V (rhe)) and the maximum (0.8 V (rhe)) of the sweeping voltage was 836 ~Clcm2, that is very elose to that expected for copper monolayer coverage (1 Cu / Pt atom) iftwo electrons are transferred on copper adsorption. (fig. 3).

I As EI Omar showed, the voltammogram changes when the electrode is subjected to potential cyeling in the range 0.3 to 0.8 V (rhe). The adsorption peak moves to higher potentials and increases in height, the anodic dissolution peak moves to 10wer potentials and decreases in height, up to a point where they are of similar height. Their separation in potential is still of about 75 mV at 50 mV/s, and is small at slow sweep rate (5 mV/s), so that the adsorption­desorption is nearly reversible. The initial shape of the voltammogram is restored by a few sweeps at higher potentials, in the range of oxygen adsorption on platinum.

"""' N

S 600

1 400 0 1-1 U ·s '-' 200 Emersion oint

"Ei

~ 0 U

300 400 500 600 700 800

Potential (mV (rhe» Figure 3 Cyelic voltammogram for the upd of copper in 1M perchloric acid, 0.510-2 M Cu2+. Sweep rate: 50 mV/so Reversible hydrogen reference electrode. The electric charge under the voltammogrem corresponds approximately to the adsorption and the desorption of a full copper monolayer. Emersion at 500 mV (rhe). The copper coverage at this potential is about 2/3 of a monolayer.

The interpretation of this behaviour is not elear. Andricacos and Ross 6 published sharp and reversible voltammograms. Oxygen might be the eiue. To make our LEED study, we first chose to look at the surface which exhibits the voltammogram in its initial form, as fig. 3 shows. The electrode was maintained at 500 mV (rhe) for 2 minutes before emersion. The electric charge contained in either the anodic or the cathodic parts of the cyele, from 500 to 800 mV (rhe) is equal to 300 ~C/cm2, so that after correction for the double layer capacitance, the coverage is expected to be at most 280/420=0.67 monolayer of copper.

269

At 500 mV (rhe) some perchloric ion electrostatic adsorption is expected, so that the presence of some chlorine and oxygen is expected on the surface of the electrode after emersion, irrespective of the thickness of the film of electrolytic solution left on the electrode after blowing the liquid droplet with dry argon. Qualitatively, these expectations were fully confirmed by Auger spectroscopy. It was observed that under the exposure to the Auger spectrometer electron beam, the Auger peak amplitude of chlorine decreases and that of oxygen increases. After some hours of rest, these Auger amplitudes are back to their initial values. This may be tentatively explained by the destruction of some CI04 radicals by the electron beam, and afterwards, a restoration of the uniform distribution of these mobile species.

The processing of LEED images recorded by the video camera (see below) shows that, in addition to the LEED spot pattern, there is a very uniform, isotropic background of intensity, in contrast to the structured, anisotropic background usually observed on single crystaIs prepared in ultra-high vacuum. This is likely in relation with a (static or dynamic) disordered layer of molecules, and it is natural to make the CI04 species responsible for this uniform background. Of course this is an assumption, but this assumption has to be checked, and it will be the next step in our investigation of this interface. Note, however that this is independent of the possible contribution of oxygen to the diffraction pattern, and this is part of the task we are going to discuss.

2.3 LEED SETTING, ACQUISITION AND IMAGE PROCESSING.

2.3.1. LEED geometry and image recording. The standard LEED optics is supplied with an axial (horizontal) electron gun. The rotation axis of the sampie manipulator is also horizontal, normal to the electron gun, and goes through the centre of the spherical grid arrangement. The Pt(lOO) surface is approximately normal to the electron gun axis, and contains the rotation axis of the manipulator. If this geometry was perfectly set, rotating the manipulator would change the polar angle of incidence without changing the point where electrons hit the surface, nor the azimuthaI angle of incidence. Fortunately, the normal to the surface of the crystal, when horizontal, is tilted with respect to the gun axis by a few degrees. Rotating the manipulator by a few degrees thus not only changes the polar angle of incidence by a few degrees, but also changes a lot the azimuthai angle of incidence. One takes advantage of this situation in adjusting the polar angle in order to set the azimuth of the plane of incidence in a direction of high symmetry, and thus reduce the size of some of the matrices entering the LEED calculation. This is illustrated in fig. 4.

Prior to any accurate setting of the polar and azimuthal angles of incidence, it is necessary to compensate for magnetic fields which distort the electron paths. This is only partly achieved with the help of two pairs of Helmoltz coils suitably oriented. Then one chooses the azimuthal angle so that a high symmetry is achieved for the LEED pattern.

Images are recorded every 2 e V with the help of a video camera, stored digitaIly and processed off line with a program specifically developed in our group. An image at 0 e V energy is recorded at the end of recording process, and subtracted from each of the LEED image, eliminating spurious contributions of light coming from the cathode of the electron gun and reflected by the gun drift tube, the crystal holder, and other pieces of equipment. After recording the images, the electron beam current impinging on the crystal is recorded as a function of energy, to aIJow further normalisation of spot intensities to a constant beam current.

270

• Gun

• Figure 4. Two settings of the polar angle of incidence, such that the azimuthai angle of incidence is -450 and +450

2.3.2.Preparation o/I(V) data curves. The software designed to process images deals with background subtraction, lattice fitting to follow diffracted spots on their path when electron energy (wavelength) is changed, integrating spot intensity within windows of adjustable size, averaging intensities of beams equivalent by symmetry, normalisation to a constant beam current, etc. The aim of the game is to extract the maximum number of Ihk(V) curves, also called energy profiles, (hk) being the two dimensional Miller indices of the diffracted spots. Usually, a large enough data base is built with 7 to 10 energy profiles.

The accurate value of the polar angle of incidence is required for the LEED computation. This angle has been set to achieve the maximum possible symmetry for the diffraction pattern. But its value is unknown. It may be computed by measuring the angles between the parallel component of the wave vector and the direction of the various spots on a recorded image, and putting these data in a program published by Van Hove 16. To increase accuracy, one repeats the operation for a set of images at various energies.

2.4. COMPUTATION AND MINIMISATION SCHEMES.

2.4.1. Multr,le scattering calculation. The principles of LEED calculations were best reviewed by Pendry 7. On the computational ground, we used the program written by Moritz 18. LEED intensities are calculated with the help of the layer doubling method for interlayer multiple scattering, (the layering being parallel to the surface), and the symmetrized matrix method for intralayer multiple scattering. To avoid lack of convergence associated with small interlayer spacing, layers too elose to one another are considered as one composite layer in which multiple scattering is calculated in angular momentum space. The size of the matrices is reduced by taking advantage of symmetry. In a LEED calculation, (a run), one tries to find the atomic positions (and primarily interlayer spacings), intralayer shifts perpendicular to the surface (buckling), and occupation factors (concentrations), which lead to the energy profiles which compare best with experiment. A run may inelude up to 150 different models.

271

Figure 5 illustrates the layering of a crystal parallel to the surface, and the surface zone where parameters are varied. In the following, interlayer distances normal to the surface are denoted by dl, d2, d3. interatomic distance are denoted by dCu-Pt .... , concentration (occupation factors) by Cl. C2. C3·

Outer platinum layer

1d3 3-dimensional platinum

Cu

Cu,O Pt

Figure 5. Side view of the crystal showing the layer stacking. The surface region, above the 3-dimensional pure platinum substrate is that optimized in the present work. Left : bare platinum, with unlrnown d3 interlayer distance. Right : platinum covered with an outer layer of copper and an inner composite buckled layer containing copper and oxygen. Registries and concentrations not stated.

The real and imaginary parts V 0 and Vi of the inner potential, expressed in e V, are taken as energy (E) dependent, according to the following model 19 :

Vo = -3 - 88/ (E+12)1/2, Vi = 0.85 (E-Vo)1I3 Scattering properties are described by T-matrices built from atom phase shifts. The phase

shifts for copper and platinum were computed from a crystal potential of pure metal resulting from the superposition of atomic ~otentials. Relativistic spin averaged phase ·shifts were computed for platinum by Rundgren 0, 21. It is possible to simulate perfect chernical disorder in a geometrically wen ordered structure by using the average T-matrix approximation (ATA) 21. It consists in replacing the (different) atoms in equivalent crystallographic sites by an average scatterer the scattering amplitude of which is a weighted average of those of the species involved. The weight is equal to the occupation factor (concentration) of the site by the species. One of the species may be a vacancy, with zero scattering power.

2.4.2 Comparison 0/ computed and experimental curves. In LEED, one has to build a model of structure, compute the energy profiles, and compare them to the experimental ones. This is a difficult task, not only because of the multiple scattering calculations, but also because of the need to make quantitative and meaningful comparisons between theory and experiment. A model is accepted as good not when some of the theoretical curves fit best some of the measured ones, but when the whole set of theoretical curves fits best the whole set of experimental ones.

To evaluate the fitting of individual energy profiles and of the whole set of profiles, use is made of five metric distances, (MD) 22, 23. A model is considered as valid only when the five metric distances converge towards the same answer. In addition the LEED program computes

272

reliability factors designed by Pendry 24 and by Zanazzi and Jona 25 which do not have the mathematical properties of metric distances, but which can help in searching for good models .. In the following, MD are denoted by Dl, D2, Dy2, D4, Dy4.

It may happen that the surface structure is built from domains of two (or more) different structures, each domain being larger than the size of the coherence zone. Then the two different structures contribute in intensity, not in amplitude, to the observed energy profiles in proportion to the area they cover on the surface. Testing systematically such niixtures constitutes a huge computational task, and the practical way to do it consists in testing only those models which are elose to the best one found for the individual structures.

2.5 SUMMARY OF THE STARTING CONDmONS, STRUCTURES TESTED, AND RESULTS.

Let us first summarize the characteristics of the interface subjected to the LEED investigation, the assumptions made before starting, and the geometry chosen to record the data.

( i) The upd is taken elose to its initial state, according to EI Omar 7. ( ii) The copper coverage is not expected to exceed 70% of a monolayer. (iii) According to Auger measurements, only platinum, copper, oxygen and chlorine are

detected on the surface. A small carbon contamination occurs at some places. Chlorine and most of the oxygen are components of CI04 - ions assumed to be totally disordered ort the surface, and thus do not contribute to the diffraction.

(iv) Geometrically, the LEED pattern is mainly a p(lxl) pattern, with relatively weak c(2x2) extra spots, not easy to measure, because of proxiniity to the detection liniit and insufficient resolution of the system. The (hk) Miller indices are labelled with respect to the priniitive square surface net of Pt(100), that is a square net with unit meshes of 0.277 nm. We compute both integer and fractional order (hk) profiles, but consider only integer orders for optimisation with respect to the experiment.

( v) The set of images was recorded from 400 down to 50 eV of electron energy, at apolar angle of incidence such that the plane of incidence is parallel to the [11] direction (azimuth q,=45°). This set, containing ten non-equivalent diffracted beams, was best suited for data acquisition. The angle of incidence was found to be (3.1 ± 0.3)°. Different arrangements of metallic copper were first tried, with 100% occupation factors : Cu p(lx1), Cu c(2x2), in 4-fold (hollow), 2-fold (bridge), 1-fold (top) sites. The hollow site gave significantly better results than others, and in particular better than the top site which is the position resulting from the EXAFS analysis. Then, in view of the maximum copper coverage of 70% measured on the voltammogram, occupation factors less than 100% were tried, improving somewhat the metric distances obtained for 100% occupation factor.

Table 1 shows the Dl metric distances, the Pendry (RP), and Zanazzi-Jona (RZJ) reliability factors for some of the arrangements of metallic copper, with the corresponding occupation factors (Ci) and the interlayer distances di. An occupation factor less than 1 means that ATA is made for an atom and a vacancy. In other words, the cheniical disorder in the occupation of crystallographic sites at the atomic scale is described as a statistical average of scattering properties, and the surface is considered as homogeneous at the macroscopic scale.

Combination oftwo superposed copper layers, such that (Cu p(1xl)/Cu p(1xl), Cu c(2x2)/Cu c(2x2), (outer layer in 4-fold or 2-fold positions with respect to the inner layer), Cu c(2x2)/Cu p(1x1), did not result in much improvement, and were comparable when the

total occupation factor for copper was elose to that found in the previous step. Thi~ is onee again related to the ineomplete eoverage of the surfaee by eopper.

273

Table 1. Results of LEED calculation for homogeneous copper structures (ATA)

model Cl C2 DI RP RZT dl d2 d3 ,

Pt(Ix1) 1.0 0.248 0.580 0.250 2.06 Cu p(lxl) 1.0 0.223 0.586 0.165 1.78 2.01 Cu c(2x2) 1.0 0.265 0.468 0.162 1.85 1.98 Cuc(2x2) 0.7 0.206 0.444 0.151 1.84 2.02 Cu p(lxl)1Cu p(lxl) 1.0 1.0 0.263 0.691 0.154 1.64 1.78 1.96 Cu p(1xl)/Cu p(lxl) 0.12 0.47 0.230 0.460 1.65 1.85 1.96 Cu c(2x2)/Cu c(2x2) 0.25 0.75 0.225 0.414 0.137 1.60 1.85 1.96

At this stage, the agreement between theory and experiment is not very satisfactory, and it is difficult to discriminate between the models giving the lowest metric distances. The best combination appears to be a superposition of incomplete c(2x2)'s.

Models ineluding oxygen were then built. The number of such models is potentially very large. Because, in some cases, the sites assigned to oxygen (or to copper) are not equivalent in the geometry of the experiment, two independent calculations are required, followed by an intensity average of the two results. We tested basic models built to mimic the structure of bulk copper oxides CuO and CU20, and a number of models derived from these basic ones. The list of these model is not given because it will be too long and not of much interest. None of these combinations brought any improvement to the metric distances or r-factors obtained for metallic copper.

Since the best result was obtained for structures with metallic copper coverage less than one monolayer, the question arises whether the surface is homogeneous at the macroscopic scale, as assumed in the previous calculations, or inhomogeneous with islands of different structures larger than the coherence zone. To test this assumption, weighted mixtures in intensity of two simple basic structures were made, and the corresponding metric distances computed. The weight assigned to the basic structure is nothing but the relative area it is supposed to cover on the surface. It is regarded as an adjustable parameter. There is no longer any ATA approximation. One computes atomic structures without chemical disorder, assuming the sc ale of dis order is larger than the coherence zone. Average has now to be performed on intensities, not on amplitudes.

Each basic structure being computed for a set of parameters such as interlayer distances, buckling, etc, the number of models in a LEED calculation may be as large as 150. Combining these models by pairs would be endless. A guide to selecting the most interesting combinations is to choose models elose to those giving the best results for each basic structure independently. However, this is not a theorem!

Basic pairs of structures used for this weighted average were mainly pure platinum on one side and metallic copper arrangements such as Cu p(lxI), (Cu p(lxl) / Cu p(lxI)), Cu c(2x2), (Cu c(2x2) / Cu c(2x2)). Pure platinum has also been combined with CuO or CU20 like arrangements. The best result appears to be obtained with the simplest of these combinations, Pt p(lxl) and Cu p(lxl) / Pt (p(lxl) in equal proportions. In other words, islands of Cu p(1xl) cover about half of the crystal surface. In table 2 are listed the values of the MD obtained for the best combination. The other mixtures which gave good values for the MD's correspond also to a copper coverage elose to half of a monolayer.

274

(-( -t>

~'" :\ .:' .:.... 0""': \ ... -- ' . ~ ..

( 0

(2 2)

wo 150 200 250 300

(1 2)

(-1 t>

(-1 2)

( 0 -t) .

. Fy:J~ ( 1 -2)

taO 15D 200 250 300

Figure 6. LEED energy profiles for the ten beams used in the optimisation. Dashed Hnes : experiment. Fullline : calculation. Figures in parenthes.es are hk indices. The comparison is made for the weighted average in intensity shown in table 2. Pt(100) p(lxl) (50%). Cu p(lxl) (50 %)

275

Table 2. Metric distances and parameters for basic and optimally mixed structures.

metric distances (%) optimal parameters

model Dl D2 Dy2 D4 Dy4 dl d2 x

Pt p(lx1) 24.7 5.7 4.2 12.2 6.9 2.04 1.0 x Pt (1xl)+(1-x) Cu p(1xl) 16.8 3.7 2.7 7.6 5.0 1.79 2.00 0.5 Cu p(Ix1) 22.3 4.1 3.2 9.4 6.2 1.78 2.02 1.0

Examination of the data in table 2 shows a substantial improvement of al1 the MD's when domains are mixed. Moreover, inspection of the individual MD's response (not displayed) shows that all five MD's have converged very elose to a unique answer. This is an indication that in spite of its imperfeetions, this model is a fairly good description of the surface structure. The weak (cx2) we have observed is, however, not accounted for by the model. It is likely that the actual domain structure is not binary, as in the model, but that minor contributions (in relative area), ineluding c(2x2) are indeed present. In fig. 6 the experimental and theoretical energy profiles are displayed in the optimum case. Fig. 7 shows the variations of Dl with x, the relative surface occupation by the bare Pt p(1xl). Dl shows a very elear minimum elose to x=ü.5. The same is true for the other MD's.

24

22

20

18

o 0.2 0.4 0.6 0.8 x

Figure 7. Metric distance Dl (%) versus the relative area x occupied by bare Pt(loo) for the superposition ofLEED intensities ofPt(lOO) and Cu p(lx1) / Pt(1OO)

276

3. Discussion

3.1. COMPARISON OF RESULTS FROM XAS AND LEED.

The answer from LEED is very satisfactory in view of the starting point. The maximum expected copper coverage was expected to be about 70% of a monolayer. This indicates that either the surface is macroscopically homogeneous, with statistical occupation of crystallographic sites, (ATA), or macroscopically inhomogeneous, with microscopically homogeneous domains (Intensity averaging). LEED clearly favours the second possibility and a rather cmde model, with only two components, gives a very reasonable value for the MD's. Values of Dl range from 0.13 (for excellent agreement) up to 0.20 (in a still acceptable situation). The difference between the expected copper coverage of 0.7 monolayer and the LEED answer of 0.5 may have various origins. The structural model of the surface rnay be too cmde. LEED is mostly sensitive to long range order, and a fraction of copper may be in a disordered state and escapes detection. In addition the area probed by the electron beam is significantly smaller than the area involved in electric charge measurement and local fluctuations of density cannot be exc1uded. Secondly, a small fraction of the electrosorbed copper may have been lost by dissolution upon emersion of the electrode, if traces of molecular oxygen were present in the solution (in spite of its careful vacuum degassing).

The LEED answer seems unambiguous relative to the oxidation state of copper. It is metallic copper. The interlayer distances measured normal to the surface are typically metallic; the copper platinum distance is d2 = 0.179 nm, 1 % shorter than the bulk copper interlayer spacing d = 0.181 nm, which is almost equal to the copper platinum interlayer spacing deduced from the hard sphere model. The first platinum interlayer spacing is d3 = 0.200 nm, expanded by 2% with respect to the bulk value. This is consistent with the results generally obtained on fcc p(1xl) (100) meta! surfaces, where the observed relaxations are weak or zero.

These results are in contrast with the interpretation of XANES and EXAFS data (as described above), where the oxidation state of copper is somewhat similar to Cu+, and where the first neighbours of copper are oxygen atoms at about 0.200 nrn. In addition)LEED shows that copper sits in hollow sites of the platinum (100) surface, in a bulk-like stacking, whereas the absence of a c1ear contribution of platinum to the EXAFS suggested that copper sits on top of platinum atoms, because the electric vector of the polarised X-Ray beam lies c10sely parallel to the surface. The difference in the results of the LEED and EXAFS investigations requires some further discussion. Sampie preparation is not the same, analytical tools are not sensitive to the same properties, and the environment of the interface is very different. The question is to try to understand what is the dominant effect.

3.2. THE NATURE OF THE INTERFACES OBSERVED IN LEED AND IN EXAFS.

3.2.1. Substrate preparation. Vacuum versus blow-pipe. After the pioneering work of Clavilier et al. 8, 26, who discovered the drastic effect of oxygen adsorption desorption cyc1es on the voltarnrnograms of Pt(lll) surfaces prepared in the flame of a blow pipe, it has been confmned by LEED Auger that identical results were found on Pt(111) prepared in UHV 27. It is now well established that platinum single crystal surfaces may safely be prepared either in a blow pipe

277

with cooling in an H2-Ar atmosphere, or in UHV, with results of equally good quality. It seems quite unlikely that the origin of the problem is in the surface preparation.

3.2.2. Underpotential deposition. Surface structure in solution. As we have mentioned above, the voltammogram for copper upd on Pt(lOO) changes upon cycling, and the role of oxygen, hydrogen or anion adsorption in this change, (in relation with surface reconstruction), is not clear. A very interesting discussion of these problems was made by Clavilier et al. 28, who pointed out the potential role of the adsorption of small amounts of chloride anions in facilitating a conversion from the Pt(lOO)-p(1xl) to the pseudo hexagonal reconstructed surface structure in sulphuric solutions. Thus, one of the challenges is to ascertain that copper upd takes place on surfaces of identical structures. This has not been actually proved in the experiments reported here. However, it might also be that copper adsorption converts any initial surface structure to a p(lxl), and that changes in the voltammogram may be induced by a progressive incorporation of copper in the first layer of platinum (in relation with the presence of small amounts of adsorbed oxygen). It is not the case in the present investigation, where cycling was stopped before there was any significant change of the voltammogram. This is consistent with LEED, which states that a Cu-Pt rnixture in the top layer is very unlikely.

3.2.3."In situ" exposure versus emersion and exposure to vacuum .. Let us now assume that electrochemical conditions are identical in both experiments (same initial surface structure, same solution, same amount of cycling, etc.). The question then arises on the difference between the "in situ" versus "ex situ" observation.

EXAFS undoubtedly shows the presence of some oxygen (or of some other anion?), which gives the copper a non zero oxidation state, whereas LEED does not. However, oxygen is not absent from the surface investigated ex situ. It is present together with chlorine, as demonstrated by AES. It is natural to assume that both are constituents of Cloi ions which were electrostatically adsorbed at the emers ion potential. LEED results indicate no (or weak) interaction between copper and these ions. On the other hand, artifacts cannot be excluded from our EXAFS experiment because the cell was sealed with a kapton@ foil, which is permeable to molecular oxygen. Molecular oxygen might react with adsorbed metallic copper; however, this reaction seems difficult to explain without some loss of the potentiostatic control. Working with a very large electrode, and recording data in a very short time, Furtak et al. 29 also observed a contribution of oxygen to the EXAFS of copper. They proposed that this oxygen belongs to coadsorbed sulphate anions. These anions would be the cause of the presence of a residual charge on the copper atoms, thus making coverage estimates from electric charge measurements in error. Is the behaviour of copper in perchloric solution sirnilar? How may such a situation be altered by the emersion process? The number of question marks concerning this system is still rather large.

3.3. THE RELIABILITY OF ANALYTICAL TECHNIQUES.

In addition to the problem of oxygen and oxidation state of copper, the structural answers of EXAFS and LEED are different. EXAFS concerns primarily the local environment of a copper atom absorbing a photon, and LEED is mostly sensitive to long range order.

A good data base is the clue of a reliable analysis. This condition is rather well fulfilled in the case of the present LEED work. Due to good surface sensitivity and large number of energy profiles, signal to noise ratio is good enough, and the extension of data base in making a

278

comparison to calculation is rather large. It is much more questionable in the case of our EXAFS work. The sensitivity of the fluorescence detector was poor, the surface sensitivity also was poor due to the small number of absorbing copper atoms, and worse, the energy extension of the data was not very large, so that some broadening of the peaks in the Fourier transform occurred. Altogether, this made the signal processing quite sensitive to imperfections in background subtraction, to windowing, etc. As a result, confidence in the geometrical answer is reduced. However this answer is in agreement with analysis of the upd of copper on gold 9, 10, where similar problems with oxygen were encountered, and where copper was also found to sit on top of platinum atoms.

Conclusion

From the preceding discussion it appears that the copper on Pt(100) interface is far from being understood. It is likely that the differences observed in EXAFS and in LEED relative to the chemical state and to the geometrie al structure of copper adsorbate on Pt(lOO) have a common origin. In EXAFS, there is a complicated "two floors" superposition of half monolayers of copper in a c(2x2) periodicity with respect to platinum, with nearly coplanar oxygen involved, and a non zero oxidation state of copper. In the LEED context, this whole scaffolding collapses, metallic copper makes compact planar islands, and oxygen escapes the problem, but is still present in a mobile form such as perchloric ions. It might be that the electrostatic field in the electrochemical cell brings the anions so elose to the adsorbate metal that some interaction, with a partial charge transfer occurs, and that some sophisticated ordering is the consequence of these interactions. Out of the cell, in the absence of the electric field, anions or anionic radicals recover their freedom, loose their ordering, and thus contribute to the isotropie electron diffuse scattering. The adsorbate metal keeps the residual electronic charge which was missing, and takes the most favourable metallic configuration, according to the net sign of lateral interaction, repulsive or attractive. In the present case, LEED indicates that interactions might be attractive.

In order to convert this poetic description of the interface into plain knowledge, more work appears to be necessary in the following directions :

( i) Go further in the LEED analysis to check out any possible contribution of anion residuals. ( ii) Define more rigorous electrochemical procedures to make sure that surfaces subjected to

copper upd are identical. (iii) Try to eliminate any possible artifact involving oxygen. (iv) Perform more accurate EXAFS experiments with a better fluorescence detector, and in a

more extended range of energy.

Acknowledgements

The authors wish to thank gratefully R. Baudoing-Savois who has followed this work all along and helped a lot to its materialisation.

References

1. R. Durand, R.Faure, D. Aberdam, C. Salem, G. Tourillon, D. Guay, and M. Ladouceur: Electrochimica Acta 37 (1992) 1977-1982

2. D.M. Kolb, in "Advances in Electrochemistry and Electrochemical Engineering", H. Gerischer and C.w. Tobias (edit.), J. Wiley, N.Y., 11 (1978).

3. R.R. Adzic, in "Advances in Electrochemistry and electrochernical engineering" , H. Gerischer and C.W. Tobias (edit.,) J. Wiley, N.Y., 13 (1984).

4. D.M. Kolb, R. Koetz and K. Yamamoto, Surf. Sei. 87 (1979) 20.

5. C.L. Scortichini and C.N. Reilley; J. Electroanal. Chem. 139 (1982) 233.

6. P.C. Andricacos and P.N. Ross, J. Electroanal. Chem. 167 (1984) 301.

7. F. EI Omar, thesis, University ofGrenoble (France) September 19,1986, pp 128-131.

8. J. Clavilier, R. Faure, G. Guinet and R. Durand, J. Electroanal. Chem.107 (1980) 205.

9. G. Tourillon, D. Guay and A. Tadjeddine, J. Electroanal. Chem. 289 (1990) 263.

279

10. A Tadjeddine, D. Guay, M. Ladouceur and G. Tourillon, Phys. Rev. Lett. 66 (1991) 2235.

11. AG. McKale, B.W. Veal, A.P. Paulikas, S.K. Shan an G.S.Knapp, J. Am. Chem. Soc. 110 (1988) 3763.

12. G. Martens, P. Rabe, N. Schwentner and A Wemer, Phys. Rev. B17 (1978) 1481.

13. R. Durand, R. Faure, D. Aberdam and S. Traore, Electrochim. Acta 34 (1989) 1653.

14. D. Aberdam, S. Traore, R. Durand and R. Faure, Surf. Sei. 180 (1987) 319

15. D. Aberdam, R. Durand, R. Faure and F. EI Omar, Surf. Sei. 162 (1985) 782.

16. M.A Van Hove, W.H Weinberg, and C.M. Chan, "Low Energy Electron Diffraction" Springer-Verlag, Berlin (1986).

17. J.B. Pendry, Low Energy Electron Diffraction, Academic Press, London, (1974)

18. W. Moritz, J. Phys. C 17 (1984) 353.

19. R. Baudoing-Savois, Y. Gauthier and W. Moritz, Phys. Rev. B 44 (1991) 12977.

20. J. Rundgren, Private Communication.

21. Y. Gauthier, Y. Joly, R. Baudoing and J. Rundgren, Phys. Rev. B 31 (1985) 6216.

22. J. Philip and J. Rundgren, in "Determination ofSurface Structure by LEED". Eds. P.M. Marcus and F. Jona, (Plenum New York (1984)).

280

23. Y. Gauthier, R. Baudoing, M. Lundberg and 1. Rundgren, Phys. Rev. B 35 (1987) 7867

24. J. B. Pendry, J. Phys. C 13 (1980)937.

25. E. Zanazzi and F. Jona, SurJ. Sei. 62 (1977) 6l.

26. J. Ciavilier, J. Electroanal. Chem. 107 (1980) 21l.

27. D. Aberdam, R. Durand, R. Faure and F. EI Omar, SurJ. Sei. 171 (1986) 303.

28. J. Ciavilier, A. Rodes, K. EI Achi and M.A. Zamakhchari, J. Chim. Phys. 88 (1991) 129l.

29. T.E. Furtak, L. Wang, J. Pant, K. Pansewicz ant T.M. Hayes, in Proceedings ofthe symposium on X-Ray methods in Corrosion and Interfaeial Electrochemistry, A. Davenport and J.G. Gordon 11 (edit.), The EIectrochemical Soc. Pennington, USA. 92-1 (1992).

CUARACTERIZATION OF NEW SYSTEMS FOR TUE CATALYTIC ELECTROREDUCTION OF OXYGEN BY ELECTROCUEMISTRY AND X­RA Y ABSORPTION SPECTROSCOPY.

M.C. MARTINS ALVES, J. P. DODELET (*), D. GUAY, M. LADOUCEUR and G. TOURILLON LURE, Batiment 209 D, 91405, Orsay, France (*) INRS, Energie, c.P. 1020, Varenes, PQ J3XIS2, Canada 11.

ABSlRACT. Electrochemical and X-ray absorption techniques have been used to determine the influence of heat treatment in electrocatalytic activity for 02 reduction for two Co catalysts. The catalysts are cobalt phthalocyanine (catalyst 1) and polyacrylonitrile + cobalt acetate (catalyst 2) adsorbed on carbon black and heat treated at several temperatures. A maximum for the catalytic activity was obtained for PcCo at 850°C and for the PAN + Co catalyst at 950°C with subsequent decrease. The results obtained by XANES and EXAFS data clearly show that metallic cobalt aggregates with different size are synthesized in the range of increased activity. In the region of highest activity were observed the smallest cobalt clusters (20 A). For higher temperatures these cobalt aggregates became bigger (100-200 A), which corresponds to the decrease in the catalytic activity. TEM was used as a complementary technique and it confmns the influence of the annealing temperature in the size of the cobalt aggregates obtained. XANES measurements at the Co and N K edges confirm that CoN4 centers and nitrogen atoms are no longer detected after heat treatment in the region of increased activity.

1 . Introduction.

One of the main objectives in fuel cell research over the past 25 years has been the deve10pment of a viable system using an acid electrolyte. Platinum has traditionally been employed as the oxygen reduction catalyst in such systems on account of its ability to meet the three criteria of electrocatalytic activity, stability, and electronic conductivity (1).

The investigation of compounds that do not use precious metals is very important from practical and theoretical points of view. Jasinski (2) reported oxygen reduction electrocatalysis with Co phthalocyanine, adsorbed on carbon and nickel electrodes. After that many other macrocyclic organic N4-chelates have been investigated as catalysts for the cathodic reduction of oxygen. Such compounds constitute a promising class of catalysts because they are relatively inexpensive, and posses semiconductor properties and thermal stability. Figure I illustrate the structures of some N4-chelates studied for the O2 reduction.

These chelates became more interesting when it was demonstrated that the heat treatment of these materials adsorbed on high area carbon under inert atmosphere improved their stabilities and activities for 02 reduction (3-5).

Several authors have attempted to explain the origin of the high activity and stability and especially the nature of the active species for the oxygen reduction after the thermal treatment:

281

C. A. Melendres and A. Tadjeddine (eds.), Synchrotron Techniques in Intetjacial Electrochemistry 281-293. © 1994 Kluwer Academic Publishers.

282

Wiesener (6) proposed initially that during the heat treatment, a "special kind of carbon" is synthesized on the substrate. According to others (7-9) the heat treatment leads to a ligand modification which preserves the central N4-metal part. Gupta et al.(lO) demonstrated that the N4-metal centers are not essential to the electrocatalysis. They studied a system composed of a mixture of cobalt or iron salts and polyacrylonitrile adsorbed on carbon black and annealed up to 1000 oe. The catalytic activities of such compounds are identical to those of the corresponding transition metal-N4 macrocyc1es. In their opinion the active species is a modified carbon surface on which transition metal ions are adsorbed through interactions with residual nitrogen atoms derived from the heat-treated macrocyc1es. Other authors sustain that nitrogen is necessary for the electrocatalytic reaction (11) but the exact chemical nature of the active sites is unknown.

The formation of of a mixture of oxides (12) and elementary metal (7,8) during the heat treatment is also reported in the literature but these compounds are not considered as a possible origin of the catalytic activity.

In order to understand the processes occuring during the thermal treatment, we studied two systems by electrochemical and X-ray absorption techniques : (1) cobalt phthalocyanine (PcCo) and (2) polyacrylonitrile (PAN) + Co acetate adsorbed on carbon black and treated at different temperatures in the 200-1100 °C range.

A B

Figure 1 - SOIIIe macrocyclic N4 -cbelates structures M = transition Meta!, A - Meta! Porphyrin, B - Meta! Phtbalocyanine.

2. Experimental

2.1. CATALYSTPREPARATION

The catalysts were prepared from active carbon Vulcan XC-72 (surface area of 254 m2/g) obtained from CABOT with a very low content of metallic impurities. Catalyst 1 (PcCo + Vulcan XC-72) - Following the route described in (13), this catalyst was prepared by dissolution of the PcCo (Fluka Chemie) in 96 % H2S04 (Merck) and precipitated on the support by pouring into ice-water to yield PcCo/support ratio of 1/8 (w/w).

Catalyst 2 (PAN + Co + Vulcan XC-72) - This catalyst was prepared as previously described in (10). It was prepared by dissolving PAN (Aldrich) and Co (11) acetate (Aldrich) in warm DMF (Merck). A portion of Vulcan XC-72 was added to this solution. The DMF was removed by evaporation. The final concentration was 10 % PAN + 0.59 % Co + Vulcan XC-72.

283

Heat Treatment· The thermal treatment of the catalysts was performed in a horizontal quartz furnace for 2 hours under a continuous flow of argon and then allowed to cool down to room temperature while still under flowing argon.

Electrode Preparation . The working electrode was prepared by following the method proposed by van der Putten (14). The catalyst particles were attached to the disc of a vitreous carbon rotating electrode via incorporation in a polypyrrole film. In order to prepare the polypyrrole film, 2.5 mg of catalyst was added to a 5 mL solution composed of 0.1 M LiCI04-O.5M pyrrole and acetonitrile (Merck). The pyrrole was distilled just before use. The suspension was then put in contact with the working electrode and a potential of 0.6 V vs SCE was applied. The potential was switched off when acharge of 40 mC was passed. The deposited layer was washed with ethanol and then dried.

2.2. ELECTROCHEMICAL MEASUREMENTS

The electrochemical experiments were conducted in a standard three-compartment electrochemical cell filIed with H2S04 solution (pH == 0.5) saturated with oxygen. A platinum grid was used as counter electrode and a saturated calomel electrode (Tacussel) as reference. The rotating disk was the working electrode. The net oxygen reduction currents reported are the differences at - 0.150 V/SCE between currents at 25 rps and current responses of the stagnant electrode in 02-saturated conditions. The sweep rate was always 10 mV/so The voltammograms were obtained with a Princeton Applied Research Model 273 potentiostat.

2.3. X-RAY ABSORPTION EXPERIMENTS

The experiments at the Co K edge were performed at the DCI storage ring, LURE Orsay, running at an energy of 1.85 OeV and a current of 300 mA. The X rays were monochromatized with a Si 331 channel-cut single crystal with an energy resolution of 1 e V at the Cu K edge. The XANES and EXAFS spectra of the catalysts and of the reference compounds (Co metal foil, cobalt phthalocyanine and Co acetate) were recorded at room temperature in the fluorescence and transmission modes, respectively. The thickness of the sam pies in transmission mode was adjusted such that ~x (~ is the absorption coefficient and x the thickness) on the high-energy side of the absorption edge was 1. In the transmission mode the transmitted beam intensity (I) was determined by using an ionization chamber. The spectra recorded in the fluorescence mode were obtained using a detector specially designed at LURE which is based on a plastic scintillator and a photomultiplier tube (15).

The experiments at the N K edge were conducted in an ultrahigh vacuum (UHV) system (base pressure of about 10-10 Torr). They were carried out at the VUV Super-ACO storage ring, LURE, on the SACEMOR beam line using a high-energy TOM monochromator (resolution of 0.2 eV at the N K edge). For the XANES, the incident photon beam intensity (10) was monitored by collecting the total electron yield from an 85 % copper metal grid freshly coated with gold. The total electron yield (I) from the sampie was then normalized with respect to (10).

The analysis of the EXAFS data involves a background subtraction by means of a cubic spline function (16). The various neighboring shells were obtained by a Fourier transformation of the EXAFS signal. The various peaks were sorted out by a window separation. By an inverse Fourier transformation into k space, the EXAFS oscillations corresponding to only one neighboring shell were obtained (17). Comparison of the experimental phase and amplitude functions deduced from a model compound (cobalt meta!) with those of the sampie yields the structural parameters.

284

3. Results

3.1. ELECTROCHEMICAL CHARAcrERISTICS

Figure 2 presents the perfonnances at a potential of -0.150 mV /SCE of both catalyst materials for a wide range of temperatures. For the PAN + Co catalyst, the activity increases from 500°C up to 950°C where a maximum is obtained. The catalyst activity decreases for higher temperatures. For PcCo the activity is initially quite constant until 700 oe. A maximum is obtained around 850 oe. For temperatures higher than 950°C, the current for both catalysts is very similar.

200 -

-:( "- 150,. -

100 -

400 600 800 1000

TEMPERA TURE (OC)

Figure 2 - Variation of electrocatalytic current for 02 reduction in H2S04 (pH = 0.5) at -150 mV vs SCE, as function of the annealing temperature (.) PcCo on Vulcan XC-72 1/8 (w/w); (.) PAN 10%, Co 0.59 % + Vulcan XC-72. Catalyst loading = 6.4 mg/cm2.

3.2. X-RAY ABSORPTIONSPECTROSCOPY ATTHECoKEDGE.

3.2.I.XANES characteristics oi Catalyst 1. Figure 3 shows the Co K edge XANES for the catalyst 1 (PcCoNu1can XC-72) as a function of the annealing temperature. The XANES data for the pure compounds (cobalt phthalocyanine (a), cobalt metal (f) and the untreated sampIe (b) are given for comparison. The reference energy (0 eV) corresponds to the first inflection point of the metallic cobalt edge (7709 e V).

Cobalt phthalocyanine has a D4h symmetry where the metallic center atom is in a square-planar environment. The cobalt phthalocyanine spectrum (figure 3a) exhibits several transitions labeled 1,2,3,4 and 5; their energy values are listed in Table 1. The transition labeled 2, which is observed for all compounds in a square-planar environment is a fingerprint of the Co-N4 structure (18) and any modification of the coordination greatly affects this transition. The XANES spectra for cobalt metal in a hcp structure is given in the curve 3f. The features observed in the spectra arise from multiple scattering effects of the photoelectron by the different cobalt shells.

When the phthalocyanine is deposited onto the carbon support and upon heating up to 700 °C (figure 3b,c) the XANES data reveal that the Co-N4 structure is retained. For temperatures above 700 °C (curves d and e), significant changes are observed, especially in

285

the pre-edge region. In particular, transition 2 is no Ion ger observed which means that the square-planar configuration is destroyed. Moreover the XANES spectra become very similar to the one of cobalt metal. Thus the XANES characteristics clearly reveal the appearance of metallic cobalt particles.

4~----------------------------------~

3

~ OE ~

.ci

~ z 2 0 f= 0-cr: 0 (/) a:J «

o -+-----------20 o 20 40 60 80 100

ENERGY(eV)

Figure 3 - Co K edge XANES for PcCo on Vulcan XC-72 a) pure PcCo; b) PcCo on Vulcan XC-72; c) PcCo on Vulcan XC-72 treated at 700 °C; d) at 800 °C; e) at 1000 °C; f) Co meta!. The zero energy reference corresponds to 7700 eV.

TABLE 1: Energies and proposed assignments for features observed in the Co K XANES spectrum of cobalt phthalocyanine.

features energy (eV) assignment 1 7709 Is-ßd (p-d hybridization) 2 7716 Is~pz + ligand hole 3 7724 Is---74pz 4 7728 IS---74pxy + ligand hole 5 7735 Is---74pxy

286

3.2.2 EXAFS Characteristics of Catalyst 1. Figure 4 shows the k3 weighted Fourier transfonn of the Co K edge EXAFS for the catalyst 1 in the range of temperatures studied. The standard compounds are given for comparison.

The Fou~ier transform of cobalt phthalocyanine (figure 4, curve a) exhibits a peak located at 1.6 A (uncorrected for the phase shift) and is relative to the Co-N distance in the phthalocyanine structure. The Fourier 0 transform of Co metal (figure 4, curve f) exhibits three peaks located at 2.2, 4, and 4.7 A. The first peak is related to the Co-Co distance in the first neighbor shell and the others are related to more distant neighbors.

For the catalyst annealed to 700°C (figure 4 b,c) the peak located at 1.6 A due to the Co-N distance in the Co-N4 structure is still observed. For temperatures between 800 and 1000 °C, the intensity of this Poeak continuously decreases and three new peaks appear at approximately 2.2, 4 and 4.7 A. The comparison of these distances with those found in cobalt metal demonstrates the formation of metallic cobalt in good agreement with the XANES data.

0.6

:§'

:3 .D

~ 0.4 w a :;)

t: -' a. ::;; «

0.2

0.0

o 2 4 6 8 1 0

DISTANCE (A)

Figure 4 - Fourier Transform of the k 3 weighted EXAFS data for PcCo on Vulcan XC-72 a) pure PcCo; b) PcCo on Vulcan XC-72; c) PcCo on Vulcan XC-72 treated at 700 °C; d) at 800 °C; e) at 1000 °C; f) Co meta!. (Ak = 1.65 - 6.48 A-I).

3.2.3. XANES characteristics 0/ Catalyst 2. Figures 5 and 6 show the XANES spectra and the Fourier transfonn obtained at the Co K edge for catalyst 2 (PAN + CoNulcan XC-72) treated at different temperatures. The XANES data of pure compounds CoAc (a), Co metal (f) and untreated sample (b) are given for comparison. The Co acetate

287

(C02(CH3CH2COO)4·2H20) is characterized by a CO-CO distance of 2.65 A. Each Co is situated in a distorted octahedral environment. Four of the six apexes of the octahedron centered on the frrst Co atom are oxygens from two acetate groups. The Co-O distances is 1.97 A. Another oxygen from ~O with a Co-O distance of 2.2 Ä is on the f1fth apex while the other Co atom is on the sixth (19). The XANES spectrum of cobalt acetate (figure 5, curve a) exhibits mainly three transitions at 7709,7724 and 7765 eV respective1y (table 2). The pre-edge feature is due to a dipole forbidden transition to 3d states. The intense white line (feature 2) corresponds to the transition of the 1s e1ectron from the core level to 4p states. Transition 3 is the first EXAFS Co-O oscillation.

The XANES characteristics of the catalyst annealed up to 900°C reveal significant modifications. The intensities of the white line (feature 2) and the fmt EXAFS oscillation (feature 3) decrease, indicating modifications in the local order of the organic matrix. For temperatures above 900 °C, the transformations are similar to those already observed with catalyst 1, the spectra evolving to that of metallic cobalt

~ 3

"e: ::l

.e 5-z 0 i= 2-11. a: 0 CI) CD «

·20 o 20 40 60 80 100 ENERGY(eV)

Figure 5 - Co K edge XANES spectra of PAN + Co + Vulcan XC-72 sampies heated at various temperatmes a)pure Co acetate; b) PAN + Co + Vulcan XC-72 untreated sample; c) PAN + Co + VulcanXC-72 treated at 800 °C; d) 900 °C; e) 1160 OC; t) Co metal. The zero euergy reference corresponds to 7709 eV.

288

T ABLE 2. Energies and proposed assignments for features observed in the Co K XANES spectrum of cobalt acetate

features energy (eV) assignment 1 7709 ls~3d (p-d hybridization) 2 7724 ls~4p

3 7765 first EXAFS oscillation

3.2.4. EXAFS characteristics of Catalyst 2. The k 3 weigqted Fourier transform of Co acetate (figure 6 curve a) exhibits a peak located at 1.62 A (uncorrected from the phase shift) and is relative "to the Co-O distance. For annealing temperatures up to 900 °C the Co-O distance at 1.62 A is still present. Considering that the Co acetate inserted in the polymerie matrix is not stable at high temperatures, it could be transformed to an amorphous oxide. For the temperatures between 900 and 1160 °C, the distances observed at 2.2, 4 and 4.7 A are relevant to the Co-Co distances in metallic Co as it was demonstrated by the XANES analysis.

0.7

0.6

0.5

~ 'c :::l

.c 0.4 ~ [] t-u..

u.. LU Cl 0.3 :::l t-:::i a. :2 «

0.2

~ 0.1 @J

@] 0.0

0 2 4 6 8 10 DISTANCE (A)

Figure 6 - Fourier Transform of the k 3 weighted EXAFS data for PAN + Co + Vulcan XC-72 sampies heated at various temperatures a) pure Co acetate; b) PAN + Co + Vulcan XC-72 untreated sarnple; c) PAN + Co + Vulcan XC-72 treated at 800 °C; d) 900°C; e) 1160 °C; f) Co rnetal.(ßk = 1.77 - 7.05 A-l).

289

In order to obtain quantitative estimations of the bond lengths and the coordination numbers around the cobalt atoms, the inverse Fourier transforms of the first shell has been fitted. For catalysts 1 and 2 the backscattering amplitude and the phase shift were deduced from the EXAFS spectra of the metallic cobalt foil.

The fit parameters (bond lengths, coordination numbers and Debye-Waller factor) are shown in table 3. The results clearly show a continuous increase of the coordination number with increasing temperature while the Co-Co distances remaining constant at 2.46 A. The change in the coordination number from 8 to 12 suggests that in the lower temperature range (900 °C for PcCo and 950 °C for PAN + Co) very small metallic clusters are synthesized (size in the order of 20 A (20». When the temperature increases, these aggregates become bigger and for temperatures as high as 1000 °C for PcCo and 1160 °C for PAN + Co, 12 neighbors are obtained, which correspond to the metallic cobalt foil.

TABLE 3. Coordination number (N), bond lengths (R), and Debye-Waller factor (Llo-) obtained by fitting the first shell in the Fourier-filtered EXAFS spectra for catalyst 1 and 2 as a function of the annealing temperatures (T).

coordination T (0C) number N distance (A)

900 8.4 1000 12

Cometal 12

All the parameters were left floating during the fit Accuracy: R, ± 0.01 A; N, ± 15%.

coordination

2.46 2.46 2.46

T (0t) number N distance (A)

950 8.8 1000 10 1090 11.5 1160 12

Co metal 12

All the parameters were left floating during the fit. Accuracy: R, ± 0.01 A; N, ± 15%.

2.46 2.46 2.46 2.46 2.46

Debye-Waller R variation Llo-2

- 0.002 - 0.001

Debye-Waller R variation Llo-2

- 0.001 - 0.001 - 0.001 - 0.001

An examination of the PAN + Co catalyst by TEM was performed in the range of temperatures between 900 and 1090 °C. At 900 0 C the clusters are very small, the average size being in the range 15-20 A. At higher temperatures, the size increases being in the range 100-200 A. These results are in agreement with the results deduced from the EXAFS data. They confirm the influence of the heat treatmen~on the size of the clusters.

Our results clearly reveal that (i) the increase in the catalytic activity is directly connected to the synthesis of metallic cobalt aggregates and (ü) there is a direct correlation between the electrocatalytic activity and the cluster size of the metal. For both catalysts, the optimum activity is obtained when the smallest cobalt clusters are dispersed on the surface of the carbon black. The decrease in efficiency of the electrodes is associated with the increase in size of the cluster. Thus it seems that the origin of the catalytic effects is directly

290

connected to the synthesis of small metallic Co particles, the Co-N4 centers being destroyed.

3.3 X-RAY ABSORPTIONSPECTROSCOPY ATlHENKEDGE

In order to verify if nitrogen atoms remain after heat treatment even at very low concentration, we performed XANES experiments at the N K edge on eatalyst 1 and 2 treated at several temperatures (figures 7 and 8). The spectrum of cobalt phthalocyanine (figure 7, eurve a) is eomposed of six transitions labeled 1,2,3,4,5, and 6, (21) their energy values are listed in table 4. The XANES speetra of untreated PeCoNulcan XC-72 (figure 7, curve b) are similar to that for pure PcCo. For PeCoNulcan XC-72 annealed up to 600 °C; !iule modifieations are observed in the XANES, compared to pure PeCo. Dramatie ehanges are observed for temperatures higher than 600 °C; at 800 °C the edge jump deereases, indieating the loss of the major part of the maeroeycle of the phthaloeyanine; finally nitrogen atoms are no longer deteeted at 900 oe.

0.08

~ 0.00 <= :::I .ci 3.

~ Z 0.04 0 ;:: Cl.

6 a: 0 rJl co

0.02 «

1 2 3

·0.01 395 400 405 410 415 420 425

ENERGY (eV)

0.09

0.07

~ E

'b I :::I

0.05 .ci 3. z 0 ;:: 0.03 Cl. a: 0 rJl co « 0.01

·0.02 395 405 410 415 420 425

ENERGY (eV)

Figure 7 • N K edge XANES spectra obtained for catalyst 1 a) PcCo; b) PcCol Vulcan XC-72 untreated sampie; c) PcCol Vulcan XC-72 beat treated at 600 °C; d) at 800 °C; e) at 900 oe.

291

TABLE 4. Energies and proposed assignments for features observed in the N K XANES spectrum of cobalt phthalocyanine.

features energy (eV) assignment 1 399.8 7t* (eg) 2 401.3 7t* b 1 u (b2u) 3 403.8 7t* ego a2u 4 407.8 0* 5 409.8 0* 6 418.8 0*

0.8-r-------------------,

0.7

lii 0.6

1: :::l 0.5 .c ~ 0.4 z g 0.3 Cl. a: ~ 0.2 C!l « 0.1

400 405 410 415 420 ENERGY (eV)

425 430

O.05r--------------------,

1f § 0.03 .c ~ z o >= Cl. a: ~ 0.01 C!l «

-o.0~:-:95:----:4OO-::::----:40-::5:---4-1 0:---4-15--4-20:---4-25-----:430

ENERGY(eV)

Figure 8 - N K edge XANES spectra obtained for catalyst 2 a) PAN; b) PAN + Co + Vulcan XC-72 treated at 600 °C; c) at 950°C.

292

TABLE 5. Energies and proposed assignments for features observed in the N K XANES spectrum ofPAN.

features

2 3 5 6

energy (eV) 400.8

403.8 406.9 411 423

assignment

1t*(C=N) multielectronic I or focusing effect

0'* (C=N)

The XANES spectra of PAN (figure 8, curve a) exhibit the transitions labeled 1,2,3, 5 and 6 (table 5). The sharp feature 1 is assigned to transition of the Is electron of N to the 1t* (C=N) orbital. The broad features 2, 3 and 5 are due to multielectronic transitions I or focusing effect . The transition 6 corresponds to the 0'* (CN) shape resonance (22). For the PAN + Co catalyst, after annealing at 600 °C, the XANES spectra are strongly modified. Feature 1, the Is ~ 1t* (C=N) transition, decreases and the Is ~ 0'* (C=N) transition vanishes. A new resonance appears at 408 eV (4 feature), which is assigned to the transition of a Is electron to the 0'* (C=N) orbital. These modifications reveal that the major part of

the CN groups is removed from the PAN + Co catalyst and that some unsaturated C=N bonds are formed. Finally, no nitrogen is detected when the catalysts are heat treated up to 950°C.

4. Conclusions

The experimental techniques used (voltammetry, XANES, EXAFS, TEM) made possible the characterization of the active centers for 02 reduction in the heat-treated catalysts containing Co loaded on carbon black. The results clearly reveal that metallic cobalt clusters are formed upon thermal treatment. The average size of these aggregates increases with increasing annealing temperature. The highest activity for the 02. reduction is observed for small metallic cobalt clusters having a diameter of about 20 A Upon annealing at temperatures higher than 950 °C, a decrease in the activity of the catalyst corresponds to the appearance of quite large aggregates (100-200 Ä). XANES measurements at the Co and N K edges confirm that Co-N4 centers and nitrogen atoms are no longer detected after annealing at high temperature. Finally, these findings permit clarification of the origin of the electrocatalytic activity of such systems and malre possible their optimization for their use in fuel cells.

References

1. Appleby, AJ.and Faulkes, F.R. In Fuel Cell Handbook, Van Nostrand Reinhold: New York, 1989, Chapter 12.

2. Jasinski, R., Nature 1964,201, 1212. 3. Van Veen, J.AR., Colyn, H.A.and Van Baar, J.P., Electrochim. Acta 1988,

33(6), 801. 4. Radyushkina, K.A, J. Res. Inst. Catal., Hokkaido Univ. 1982,30 (3), 135. 5. Sawagushi, T., Itabashi, T., Matsue, T.and Uchida, J., J. Electroanal. Chem.

1990, 279, 219.

293

6. Wiesener, K., E1ectrochirn. Acta 1986,31 (8), 1073. 7. McBreen,1. In Extended Abstracts, The E1ectrochernieal Society: Pennington, NJ,

1987, Vol. 87-1, P 146. 8. Wingerden, B.V., Veen, J.ARand Mensch, C.T.J., 1. Chern. Soe., Faraday

Trans. I 1988,84 (1), 65. 9. Savy, M., Coowar, F.: Riga, J., Verbist, J.J., Bronoe1, G.and Besse, S., J. App1.

E1eetroehern. 1990,20,260. 10. Gupta, S., Tryk, D., Bae, 1., Aldred, W. and Yeager, E., J. Appl. Eleetrochern.

1989, 19, 19. 11. Franke, R, Ohms, D. and Wiesener, K., J. Electroanal. Chern. 1989,260,63. 12. Seherson, D.A, Gupta, S.L., Fierro, C.and Yeager, E.B., E1ectrochim. Acta

1983, 28(9), 1205. 13. Van Veen J.A.Rand Visser, c., E1ectroehim. Acta 1979,24,921. 14. Van der Putten, A, Visscher, W.and Barendreeh, E.B. E1ectroanal. Chern. 1985,

195,63. 15. Tourillon, G., Guay, D., Lernonnier, M., Barto1, F.and Badeyan, M., Nuel.

Instrum. Methods Phys. Res. 1990, A294, 382. 16. Lenge1er, B.and Einsenberger, P., Phys. Rev. B 1980,21,4507. 17. Stohr, 1. Jaeger, Rand Brennan, S. Surf. Sei. 1982, 21, 4507. 18. Briois, V., Cartier, c., Momenteau, M.: MaiHard, P., Zarernbovitch, 1., E.,

Fontaine, A, Touril1on, G., Thuery, P.and Verdaguer, M., J. Chim. Phys. 1989, 86,7/8, 623.

19. WeHs, AF .. In Struetural Inorganie Chemistry, Clarendon Press: London, 1986. 20. Greegor, RB. and Lytle, F.W., 1. of Catalysis 1980, 63,476. 21. Alt. H., Binder, H.and Sandstede, G. J. E1ectroanal. Chern. 1971,31, App. 19. 22. Tourillon, G., Guay, D., Fontaine, A, Garret, R, Williams and G.P. Faraday

Discuss. Chern. Soe. 1990, 89, 275.

IN SI'l.'U AHn EX SI'l.'U EXAMINATION OF PASSIVATING Cu20 LAYERS WITH EXAFS AHn REFLEXAFS

H.-H. STREHBLOW, P. BORTHEN and P. DRUSKA Institut für Physikalische Chemie und Elektrochemie Heinrich-Heine-Universität Düsseldorf 40225 Düsseldorf Germany

ABSTRACT. Passive layers on Cu and thick Cu20 films formed by cathodic reduction of a saturated cuO/- solution in 5 M KOH have been examined previously with electrochemical, photoelectro­chemical and surface analytical methods. These layers consist of cu2 0 with a CuO, Cu(OH):z overlayer at sufficiently positive potentials. They have semiconducting properties with a similar band gap as crystalline CU:zO with however large quantities of interband states. The spectroscopic results may explain quanti­tatively the electrochemical behavior of these surface layers on the basis of a simple semiconductor model. To decide whether these films are crystalline or amorphous EXAFS measurements were performed in transmission and reflection. The in situ trans­mission studies of thick cathodically deposited Cu2 0 films yield the short range order of crystalline CUzO. ReflEXAFS investi­gations have been successful only for thicker films. The angle of total reflection of about 0.36 degree requires extremely small incident angles and a sm00th and flat surface. Measurements with passivated Cu films vapor deposited on float glass showed that these films were still not smooth enough to prevent the contribution of the underlaying bulk Cu.

1. Introduction

Many reactive metals are protected against corrosion by passive layers. They are generally nonporous oxide or hydroxide films of some few nm thickness which cover the metal surface continuously. Any metal dissolution has to occur through this oxide barrier which reduces the corrosion rate by many orders of magnitude. A high electric field strength of some 10· V/m causes a slow ion transfer through the film leading to the formation and growth of the layer in the passive state. The composition of the oxide layer and of the electrolyte determine the rate of cation transfer at the oxide/ electrolyte interface [1,2].

These passive layers usually have a complex multilayer or at least a duplex structure even for the case of pure metals [3,4].

295

C. A. Melendres and A. Tadjeddine (eds.), Synchrotron Techniques in Intetfacial Electrochemistry 295-310. © 1994 Kluwer Academic Publishers.

2%

The chemical structure has been examined quantitatively with surface analytical methods [5,6J. Thus for several pure metals and binary alloys the composition and thickness of the total layer and the different sublayers have been determined by XPS and ISS studies [3,7-9J.

Similar to these general trends Cu shows a simple or duplex structure of the oxide film depending on the electrode potential [10,11]. This situation may be illustrated by the potentiodynamic polarization curve in alkaline solutions as presented in Fig. 1 for 0.1 M KOH. Anodic and cathodic peaks mark the potential range of the presence of the different species wi thin the surface layers. At AI a Cu20 layer is formed whereas at All CuO/CU(OH)2 layers form on top.

0.4

0.2

""!; -0.2 .< e :: -0.4

-0.6

-0.8

Cu O.1MKOH

-1.0 CuzO-t.

CI

-0.8 -0.6 -0.4 ·0.2 0.2 0.4 0.6 0.8

E(SHEl I V

Figure 1. Potentiodynamic pOlarization curve of Cu in 0.1 M KOH with the anodic and cathodic peaks and indication of the corres­ponding formation and reduction of the oxides, the formation of soluble species and the potential range of the existence of related surface layers.

This duplex structure has been confirmed by ISS depth profiles and XPS investigations [11]. At CII and CI the reverse reactions occur as for the anodic peaks All and AI. with the rotating ring disc electrode cu+ and Cu2+ ions are detected.

297

The composition of the passive layer at its electrolyte inter­face determines the kind of dissolving species. The charges under the reduction peaks permit the calculation of the amount of Cu (lI) - and Cu (I) -oxide. Fig. 2 gi ves an example for borax buffer pH 9.2.

The passive layer on Cu is a typical example for the situation of quasi dissolution equilibrium of the oxide layer with the adjacent electrolyte. Best passivity is observed in weakly alkaline solutions with a minimum of solubility at pH 9.2 [12J.

In strongly acidic electrolytes oxide layers on Cu are not stable and no passivity is observed. In strongly alkaline solutions they are still protecting however with an increased dissolution especially at potentials of the anodic peak All. The passive range with regard to the electrode potential and pH is in good agreement with the Pourbaix diagram which involves only thermodynamic but not kinetic aspects [12]. The situation is basically different for other passive metals like e.g. Fe, Cr, Ni, and their alloys.

(uO Cu(OH~

/0". Cu,O" ---=-\ _.-.~ ..

/0 • e_e o--O~

'§ 05 I o<"~

-0...-0- '-cuo /0 Cu(OHl, '-'

0/0 borax buffer pH= 9.2 o+-----~~~~~~~~---+

o 0,5 1.0

Ep I V (SHEl

E

~ 6

~ c

4 ~ ~

2 §

" u

Figure 2. Thickness of the Cu(I)- and Cu(II)oxide layer in dependence of the applied potential formed in borax buffer (pH 9.2) as deduced from the reduction charges under the cathodic peaks CI and CII [11]

In these cases oxide layers are even a corrosion barrier in strongly acidic electrolytes where their oxides are weIl soluble. For these situations the surface films are far from dissolution equilibrium. Their stability is a consequence of the extremely slow transfer rate of metal cations from the film surface to the electrolyte.

The anodic oxides grow to a limited thickness within the passive range (Fig. 2). Thicker Cu2 0 films may be grown by cathodic reduction of Cuo/-in strongly alkallne electrolytes. To get a continuous homogeneous layer independent of any nucleation phenomena and their kinetics, the cathodic growth should occur

298

on top of an already existing anodic film. An optimum procetlure involves the formation of the duplex layer on Cu in 0.1 M KOH at E = 0.6 V after cathodic reduction at E = -1.0 V and a subsequent partial reduction at E = -0.32 V in the range of the Cu(II) oxide reduction. Fina11y addidiona1 Cu,O is grown from a saturated solution of CUO,2- in 5 M KOH at E = - 0.235 V [13]. The amount of Cu (I) oxide may be determined by potentiodynamic reduction which occurs at appropriate potentials below peak CII (Fig.1). Fig. 3 gives a schematic diagram of the oxide formation.

The oxide film grows 1inearly with time and may reach some ~m wi thin some hours. The oxide growth is an electrochemical reaction at the phase boundary of the electrolyte and therefore depends on1y on the electrode kinetics and the electronic properties of the oxide layers. These thick Cu20 films behave similar to thinner anodic layers. They form only thin Cu(II)oxide films on top at sufficient1y positive potentials within the passive range which corresponds to the overlayers of the duplex anodic films [13].

01 M KOM I SM KOHlsat. (uOl- I 01 M KOM

E • ..0.31 V t1l1201'.' tfl,·-O.BSV '-vor ..• *-002VJS

Figur 3. Schematic potential diagram for the formation of Cu2 0 surface layers on Cu

Besides thickness, chemical structure and electrochemical corrosion behavior the electronic properties of these surface films are very important. Part of the overall electrochemical corrosion is the cathodic reduction of an appropriate redox system compensating for the anodic metal dissolution. At a passive metal surface this partial process requires the transfer of electrons or holes across the surface layer which in many cases behaves like a semi-conductor. Cathodic currents at Cu2 0 layers on Cu have been observed in the dark and under illumination especially in the presence of redox systems. Various

299

redox systems have been examined as e. g. dissol ved oxygen, [Fe(CN)6]'-/[Fe(CN)6)4-, H20 U As03 '-/ASO/-' [Co(NH3 )61'+ and even a thin Cu(II)oxide overlayer [13,14). The potential dependence of the photocurrent i ph yields by extrapolation of a ~ i ph 2-E plot" a flat band potential of E~ = - 0.25 V for both the thin anodic and the thick cathodically deposited Cu:zO films. The photocurrent spectrum agrees with the absorption spectrum of erystalline CU:zO of Brahms and Nikitine [15]. The appropriate evaluation yields a band gap of 3.0 eV for a direct and 2.3 eV for an indireet trans i tion for anodic Cu.O [14] and 2.4 eV to 2.6 eV for the direet and 2.0 eV to 2.3 eV for the indirect transition for the electrodeposited films depending on the oxide thickness [13]. Apparently the band gap is deereasing with inereasing oxide thickness. All the detailed electrochemical resul ts may be summarized in the following simple band structure model of passivating cU:zO layers (Fig.4). It behaves like a p-type semi­conductor with a high density of interband states which are responsible for the observed. dark currents. Conductivity llIeasure­ments for crystalline Cu.O by other authors suggest even a band for the states within the band gap elose to the Fermi level {16,17]. Electronic eonduetion may oeeur across these states by a hopping mechanism to the empty states of the redox system eorresponding to the observed dark eurrent.

.]

.J

~ 'u .. E,·

-5

-6

(u,O electrolyte

~ '" .1

-1

_1 ~ i!;

<

Figure 4. Shematic diagram for the band structure model of Cu20 surface layer in the energy seales relative to the vaeuum level and to the standard hydrogen eleetrode with indieation of the Co ( NH,) 6) .+/3+ redox system [18)

Photocurrent will involve the excitation of eleetrons to the conduction band and their transport to the oxide/electrolyte

300

interface within the field of the band bending. These electrons are trapped at surface states from where they are transferred to the electrolyte (Fig. 5). The existence of these states within the band gap is concluded from pronounced transients of the photocurrent which are explained by filling of these traps when light is switched on and their retarded evacuation to the redox system when it is switched off [18]. The large tailing of the photocurrent spectra at lang wavelengths may be seen also as a deviation from an ideal crystalline semiconductor.

Addi tional UPS measurements provide quanti tat i ve data about the work function and threshold energies and thus about the position of the Fermi level or the valence band edge relative to the vacuum level or the standard hydrogen electrode. Figure 4 gives the relative positions of the bands to each other and to the two energy scales. The consequence of the increasing electrode potential and the related band bending is shown in figure 5. Starting with the flat band potential at EPb = -0.25 V the valence band edge will cross the Fermi level at E = 0.55 V according to a potential increase of 0.8 V which equals the energy difference between the valence band edge and the Fermi level [19].

(u,o (ulU) Cu{Ul

~,o (u{nr

0,­E"09SV~-------_-----.. uy tormattOl'l

'"

Figure 5. Semiconductor model and phase formation for increasing electrode petential E of passive layers on Cu [19].

This situation will introduce a large quantity of positive holes i.e. Cu(II)ions which finally causes the separation of a

301

Cu(II)oxide overlayer. After this re arrangement within the passive layer and the potential distribution Le. the duplex film formation a further potential increase by 0.4 V is required to reach again a crossing of the Fermi level and the valence band edge. Further potential increase will then be located at the Cu(II)oxideje1ectro1yte interface with an increase of the transpassive dissolution of Cu(II)ions and oxygen evolution. All these spectroscopic data are in good agreement with the potentiodynamic p01arization curve as may be seen by comparision of Fig. 5 and Fig. 4 [19].

In conclusion, from the electrochemica1 and surface ana1ytica1 results, e1ectrochemica1ly formed Cu.O may be seen as a highly disordered p-type crystalline semiconductor with a 1arge quantity of interband states. Other authorf'; prefer the concept of an amorphous structure. For a clear answer diffraction methods shou1d be applied to thick 1ayers and EXAFS or ref1EXAFS to passive or thin oxide 1ayers.

2. EXAFS in Transmission

For in situ transmission studies a thin film e1ectrochemica1 ce11 has been used, simi1ar to that of McBreen et a1. (20). This cell is shown in Fig. 6.

Electrolyte Port

Plotinum Counter Electrode

Workin Electrode

X-Roy Window

Acrylic Block with Kopton Foil

PTFE Goskets

lic Block with Kopton Foil

Figure 6. Thin film electrochemica1 ce11 for in situ EXAFS transmission studies.

Two PTFE gaskets served as spacers to get an electro1yte film of 0.1-1 mm thickness. A Pt wire served as a counter e1ectrode and a smal1 AgjAgC1 e1ectrode was used as a reference electrode.

302

The electrolyte was 0.1 M Na,B.07 solution (pH 9.2). Wor·king electrodes were prepared as folIows. First, a 15 nm thick gold film was vacuum evaporated onto a 0.1 mm thick PTFE foil. Then, an approximately 5 nm thick copper film was deposited onto the gold covered PTFE foil. The copper film was passivated in 0.1 M NaOH and the Cu(II) part of the duplex film reduced to CUzO. Subsequently, a 5-10 ~m thick Cu,O layer was grown by reduction from a Cuo,z- solution in KOH as described previously [21].

A diffraction pattern of a 7 ~m thick Cu2 0 layer recorded with a Guinier camera is shown in Fig. 7. All peaks corresponding to the crystalline Cu,O (cuprite) are marked. The remaining peaks are due to the underlying material. It is evident that the CUzO layers grown according to the procedure described above are crystalline Cu,O [22].

2000

1600 Cup

1200 t -.; 0.

.!<. Cup Cup BOO

t Cu,O

400

27.2 36.1 45.0

e [deg]

Figure 7. Difraction pattern of a 7 ~m thick CUzO layer.

Stepwise reduction of these Cu,O layers was examined in si tu with EXAFS at the EXAFS II beamline of HASYLAB (DESY, Hamburg.) [21]. The reduction was performed for 10 min at each potential. Figure 8 shows the variation of the background substracted absorption spectra with increasingly negative electrode potentials at the Cu K edge. As references, spectra of a copper foil and crystalline Cu2 0 are also shown.

Fourier transforms of the related EXAFS functions are shown in figure 9. The reduction of Cu2 0 starts at -0.60 V, leading to the formation of pure Cu metal at -0.75 V. At potential va lues between -0.60 V and -0.75 V both species are simul taneously present on the working electrode leading to mixed absorption spectra. In Fig. 10 Fourier transforms of synthetically generated absorption spectra are shown. These spectra were obtained by addition of normalized Cu and Cu,O standard spectra with various ratios.

c: o :;:; a.. .... o CI)

...a o

303

-0.75

Cu

energy

Figure 8. Normalized absorption spectra of a CU"o layer recorded at various electrode potentials. Cu and Cu"O are references •

o 2 3 4 5 6 7 B 9 10

R / A Figure 9. Normalized magni­tudes of Fourier transforms corresponding to absorption spectra in Fig. 8. (Not phase-shift corrected)

o

. , . \,''''\.,'\ .. _" _ ~:.o!'t'!·!'~

0.25/0.75

1.00/0.00

2 3 4 5 6 7 6 9 10

R / A.

Figure 10. Magnitudes of Fourier transforms corresponding to absorption spectra obtained with mixed standard Cu and CUzO absorption spectra. The numbers are mixing ratios. 1.00/0.00 marks pure Cu, 0.00/1.00 marks pure Cu"O.

304

comparison of the measured absorption spectra and their Fourier transforms with the synthesized spectra yields molar Cu2 0/Cu ratios of about 0.25/0.75 and 0.45/0.55 for -0.60 V and -0.65 V, respec-tively. The shift of the absorption edge between Cu20 and Cu is small (8981.2 eV and 8980.3 eV, respeetively). However, the Fourier transforms show a prounouneed change, finally approaehing the situation of eompletely redueed oxide with the atomie distanees and eoordination numbers of the Cu bulk material.

The detailed evaluation of the EXAFS speetra of the freshly preeipitated Cu2 0 layers is summarized in table 1. The values obtained for the eoordination radii, the eoordination numbers and t.he Debye-Waller faetors are found to be elose to those of cuprite.

Table 1. EXAFS results for Cu(I) oxide preeipitated on PTFE/Au with the phase-shift uneorreeted distanee R-a, phase-shift eorreeted distanee R, the eoordination number N and the Debye-Waller faetor a for the eorresponding shell

First shell Cu - 0 R l nm Re<r nm <Tl nm NI

Seeand shell Cu - Cu

Rznm Rz-<:L nm Uz nm Nz

0.185 0.140 0.0330 2.0

0.303 0.280 0.0222

12.3

3. EXAFS measurements at grazing incidence

The eomplex refractive index in the x-ray range ean be written as n = 1 - 6 - iBo The linear absorption eoefficient IJ is correlated with B by IJ = 4~ß/A, A being the x-ray wavelength. The real part of n is always slightly less than 1. Therefore total reflection occurs at glancing angles 6 below 6 c ~ (26) '/2. For a given 6 and B for metal and oxide, the reflectivity and the penetration depth at oxidized, ideally smooth metal surfaees can easily be caleulated. Combining the Fresnel equations wi th a matrix formalism for stratified media these ealculations are straight forward [23,24]. An example of such calculations for the cu/CuO system at 8600 eV photon energy is shown in Fig. 11.

305

The optical constants used here are: Ooxlde = 1. 2X10-s , Boxide = 3.0x10-7 and 6'ntd = 2. 3x10-5 , ß •• tal = 4. Ox10-7 [25].

1.0

0.8

>-:<:::: 0.6 > .....-u 0.4 Q)

........ Q) 0.2 L

2.5

5.0

10.0 nm , , 1000.0 nm/, ,

0.0 \ nm \

o 1 2 3 4 5 6 7 8 9 10

angle (mrad)

Figure 11. Calculated Refectivities of CujCuO with various oxide layer thicknesses.

Apparently, thin oxide layers up to about 5 nm da not significantly change the reflectivity of the metal surface. The surface roughness has probably a much more pronounced influence on the reflectivity.

EXAFS experiments in the reflection mode may provide valuable information about the local order in thin oxides and passive layers. For such measurements, it is essential to have same idea of the x-ray intensity distribution within the surface layer of the oxidized metal. This information is given by the penetration depth curve as a function of e [23,26]. Calculated penetration depths for copper oxide and copper metal with the same 6' and B va lues as those used for the calculations of Fig. 11 are shown in Fig. 12.

"-""10 E c

-.....; 8

..c +-' 6 0.. Q)

-a 4-

~ +-' 2 Q) C Q) 0 0..

oxide metol I , I , I , ,

I , , , , ----

o 1 2 3 4- 5 6 7 8 9 10

angle (mrad)

Figure 12. Calculated penetration depths. eS and B as in Fig. 11.

306

If the copper oxide layer thickness is on1y 3 nm thick· the penetration depth exceeds this value for S > 3 mrad. For glancing angles above 3 mrad a significant contribution of copper bulk signals will be observed in the EXAFS spectrum. Below 3 mrad the EXAFS spectrum should mainly contain information from the thin surface oxide layer. However, due to the surface roughness the effective penetration depth will be larger than expected from the calculations for an ideally smooth surface. Thus the EXAFS spectrum may contain large contributions from the metal bulk. To minimize this effect, it is necessary to work at glancing angles weil below Sc.

An instrument for glancing angle EXAFS measurements is shown sChematically in Fig. 13a and in more detail in Fig. 13b.

b)

sl1t 2 I, slit 1

arm

Figure 13. a) Schematic drawing of the experimental set up b) Grazing incidence EXAFS instrument

The x-ray beam from a Si(311) double-crystal monochromator is collimated by a 30 ~m slit (slit 1). Both the incident (10 ) and the reflected (I,) beam are measured by ionization chambers. A third ionization chamber (I,) is used for measurements of standards to calibrate the energy scale. A second slit (slit 2) and the ionisation chambers for the reflected beam (1, and I 2 )

are mounted on an arm which is rota ted about the same axis as the specimen stage. The specimen stage can be rotated with a

307

resolution of about 0.05 mrad. All motors are controlled via a CAMAC crate by a PDP11 computer. The experiments were carried out at the beamline RÖMO II of the storage ring DORIS II (DESY, Germany) [27].

Measurements were performed on electrochemically and thermally oxidized copper films. The Cu films were vapor deposited onto unheated 3 x 5 cm2 float glass substrates. The thickness of the copper films was about 100 nm. The electrochemica1 preparation in 1 M borax buffer (pH 9.2) inc1uded 5 min oxidation at +0.60 v (SHE) and subsequent reduction of the grown Cu2 0/CuO duplex layer at -0.30 V for the next 5 min. As a result, the copper film was covered with a Cu20 layer of about 3 nm thickness [13,14]. For thermal oxidation, copper films were heated at 160' for 1 h in air.

The reflectivity spectra R recorded in the vicinity of the Cu K absorption edge at fixed g1ancing angles for the oxide covered copper films were converted to the absorption spectra by taking the (1-R)!(1+R) ratios [26]. After background substraction and normalization the EXAFS oscillations x(k) were extracted from the absorption spectra [28]. The Fourier transforms of X(k) for the electrochemically and thermally oxidized copper films are shown in Fig. 14 and Fig. 15, respectively.

" I' I, I' " , I " , , I ,

f \ I , , " , , , \

I \

" \ C 0 " 't,""", ,'\ U2 .. ...-......... - ........ '- ......

Cu

o 1 2 :5 4 5 6 7 8 9 10

R / A

Figure 14. Magnitudes of the transforms for electrochemically oxidized copper film.

...... (JJ c:: o '-I-

I "­CI.) .;:: ::l

G:

,_ .... -

" " , , I' I' I , , I I , I I , ,

mrad

Cu

o 1 2 :5 4 5 6 7 8 9 10

R / A

Figure 15. Normalized Fourier transforms for thermally oxidized copper film.

308

For eomparison, Fourier transforms for the eopper foil' and erystalline Cu2 0 are ineluded in both figures. The Fourier transforms are not phase-shift eorreeted.

Comparision of the Fourier transforms of the eleetroehemieally oxidized eopper films with those of the standards (Fig. 14) shows, that at both glancing angles (3.3 and 4.4 mrad) eontributions of the underlaying bulk copper dominate and not, as expeeted, those of the about 3 nm thiek eopper oxide film. The reason for this result is probably the roughness of the copper film surface. The Fourier trans form at e = 3.3 mrad in Fig. 14 may be obtained synthetically by mixing the normalized standard spectra of Cu and CUzo in a ratio 0.60: 0.40 wi th a subsequent Fourier transformation (Fig. 10). This comparision shows, that the bulk eopper contributes with at least 60% to the observed EXAFS spectrum.

Thermally oxidized Cu films have formed much thicker oxide films eompared to the eleetrochemically grown passive layers. The eontri-bution of this oxide film is clearly detected with the grazing incidence EXAFS. Normalized Fourier transfarms for 3.0 mrad and 5.5 mrad glaneing angles are shown in Fig. 15. The radii of the first (1.4 A, Cu-O) and the second (2.8 A, Cu-Cu) eoordination shells at 3 mrad are elose to that of the crystalline Cu20. However, the width of both peaks and the much smaller height of the 2.8 A peak compared to the CUzO standard indieate that the surface oxide has a significantly disordered strueture. The speetrum recorded at 5.5 mrad contains also, in contrast to that recordered at .3 mrad, some contributions from the bulk eopper.

4. Conclusions

In situ EXAFS examinations in transmission mode has been applied during electrochemieal reduction of cathodically formed CUzO films. The evaluation of the EXAFS data for freshly formed films yields bond distanees, coordination numbers and Debye-Waller factors for the first two eoordination shells whieh are very elose to those of the crystalline Cu20 standard. This result has been supported by diffraction studies of the CUzO covered eleetrodes. For a stepwise reduetion, the relative amounts of CuaO and metallic Cu on the eleetrode were determined by eomparision wi th synthetically eomposed EXAFS speetra of the pure standards. After eleetroehemical reduction at E = -0.75 V (SHE), the strueture of crystalline copper was found. EXAFS measurements at grazing incidenee were only succesful with a thicker thermally oxidized copper films. For these oxide layers the short range

309

structure of a disordered crystalline Cu20 was found. OXide layers on copper obtained by electrochemical passivation were hardly detected with grazing incidence EXAFS suggesting a strong influence of the surface roughness on the actual penetration depth for x-rays.

5. Aeknowledgements

The support of this work by the Bundesministerium für Forschung und Technologie (project 05 435FAB) is gratefully acknowledged. We also acknowledge the help of Mrs. C. Druska with the x-ray diffraction studies.

Referenees

1. K.J. Vetter, Electrochemical Kinetics, Academic Press, New York, 1967, page 733 ff.

2. H. Kaesche, Die Korrosion der Metalle, Springer Verlag, 1979, page 173 ff.

3. H.-H. Strehblow, Corrosion 91, NACE Proceedings 76, March 1991, Cincinnati, Ohio, USA

4. H.-H. Strehblow, Surf. Interf. Anal., 12 (1988) 363 5. P. Marcus and J. Grimal, Corr. Sei., 33 (1992) 805 6. S. Misehier, A. Vogel, H.J. Mathieu and D. Landalt,

Corr. Sci., 32 (1991) 925 7. H.-W. Hoppe and H.-H. Strehblow, Corr. Sci., 31 (1990) 167 8. A. Rossi, C. calinski, H.W. Hoppe and H.-H. Strehblow,

Surfe Interf. Anal., 18 (1992) 269 9. C. Calinski and H.-H. Strehblow,

J. Electrochem. Soc. 136 (1989) 1328 10. H.-H. Strehblow and H.D. Speckamnn,

Werkst. Korr. 35 (1984) 512 11. H.-H. Strehblow and B. Titze,

Elektrochirn. Acta 25 (1980) 839 12. M. Pourbaix, Atlas of Electrochemical Equilibria in Aqueous

Solutions, Pergarnon Press, London, 1966, page 384-392 13. U. Collisi and H.-H. Strehblow,

J. Electroanal. Chem., 284 (1990) 385 14. U. Collisi and H.-H. Strehblow,

J. Electroanal. Chern. 210 (1986) 213 15. S. Brahms and S. Nikitine,

Solid State Commun., 3 (1965) 209 16. C. Noguet, M. Tapiero and J.P. Zielinger,

Phys. Stat. Sol. (a), 24 (1974) 565

310

17. J. P. Zielinger, M. Tapiero and C. Noguet, Phys. Stat. Sol. 33 (1976) 155

18. U. collisi, PhD Thesis, Düsseldorf (1988) 19. H.-H. Strehblow, U. Collisi and P. Druska,

International Symposium on Control of Copper Alloys oxidation, Rouen France, 1992

20 J. McBreen, W.E. O'Grady, K.I. Pandya, R.W. Hoffmann and D.E. Sayers, Langmuir, 3 (187) 428

21. P. Druska and H.-H. Strehblow, J. Electroanal. Chem. 335 (1992) 55

22. Z.G. Pinsker, R.M. Imamov, soviet Physics-Crystallography, 9 (1964) 334

23. L.G. Parratt, Phys. Rev. 95 (1954) 359 24. R.M.A. Azzam and N.M. Bashara, Ellipsome~ry and Polarized

Light, North-Holland, Amsterdam, 1977 25. F. Stanglmeier, PhD Thesis, KFA Jülich, 1990 26. L. Bosio, R. cortes, A. Defrain and M. Froment,

J. Electroanal. Chem., 180 (1984) 265 27. R. Frahm, Rev. Sei. Instrum., 63(1992) 873 28. B.K. Teo, EXAFS: Basic Principles and Data Analysis,

Springer Verlag, 1986

IN SITU AND EX SITU SPECTROELECTROCHEMICAL AND X-RA Y ABSORPTION STUDIES ON RECHARGEABLE, CHEMICALLY-MODIFIED AND OTHER Mn01

MATERIALS

B.E. Conway, D. Qu and J. McBreen Chemistry Department, University of Ottawa, Ottawa, Canada and Brookhaven National Laboratory, Upton, N.Y., U.S.A.

ABSTRACT. A combined series of in situ and ex situ UV spectroelectrochemical and X-ray absorption studies have been made on Mn02, chemically-modified by small amounts of Bi(lIl), and comparatively on other Mn02 materials such as a blank (Bi-free) and )'-Mn02. These procedures are applied in order to follow the oxidation-states of Bi and of Mn during the course of discharge and rech arge of Mn02 as a battery cathode material, and the extents of rechargeability that can be achieved with such materials. The presence of Bi appears to provide a preferred "heterogeneous" discharge/recharge pathway involving a soluble Mn(lll) intermediate, over the alternative "electron-proton" hopping, solid-state mechanism.

From the XAS results, it is concluded that the presence of Bi although not affecting the O-coordination, does influence the Mn-Mn coordination, influencing the way the "MnOz" coordination octahedra are connected.

1. INTRODUCTION

As solid-state cathode materials for aqueous battery systems, oxides of transition and other metals are preferred electrochemical reagents, e.g. Pb02, NLO.OH, HgO, AgO, Mn02, etc. Cathode materials that are rechargeable in a practically realizable way, with good cycle-life and attractive energy and power densities, however, are relatively few, Pb02 and Ni.O.OH being the predominant cathode reactants of practical acid or base electrolyte systems, respectively. Mn02 has held apreeminent position as an excellent cathode material for primary batteries such as the Leclanche cell or the alkaline Mn02/Zn cello

The mechanisms of MnOz discharge and its limited recharge have been widely studied by many researchers for some decades. Among such mechanisms, Kozawa's cathodic reduction scheme seems to be widely adopted [1-7]. A two-step mechanism was postulated for reduction of )'-MnOz to Mn(OH)2 in alkaline solution. Conversion of Mn02 to Mn01.5 is the first step, in which reduction of MnOz takes place through a so-called "electron-proton mechanism". For the latter, it was assumed that the reduction of Mn02 toMn.O.OH takes pi ace homogenously in a single phase of the oxide electrode lauice; the second step is further reduction of Mn01.5 to Mn(OH)2' in which process it has been supposed that soluble Mn(I1I) species in the electrolyte arise as intermediates. The latter step has been further proved more visibly by Ruetschi [8] who used an ultrathin electrolytic cell to make a direct microscopic observation of the

311

C. A. Melendres and A. Tadjeddine (eds.). Synchrotron Techniques in Inteifacial Electrochemistry 311-334. © 1994 Kluwer Academic Publishers.

312

morphological change of Mn02 particles that is undergone du ring the discharge process, due to Mn(III) dissolution.

Various attempts have been made to achieve rechargeability of Mn02 [1-7; 9-13], especially in recent years by Kordesch et al. [14-16] based on restricted discharge of Mn02 to less than the "one-electron" stage of reduction, following which recharge to Mn02 is achievable but only over a limited cycle life. Such a system was explored many years earlier by RCA. However, there would be a major market for a multiply rechargeable Mn02 cathode coupled with, e.g., a Zn or a metal-hydride (e.g. NisLa) rechargeable anode, excluding Cd for environmental reasons or Fe for other reasons such as H2 gassing on recharge.

The chemical processes in Mn02 reduction and the species produced at various stages of discharge have been characterized in various papers by Yeager and Kozawa [1,6] and by Kozawa et al. [2-5; 7], while McBreen examined the cyclic voltammetry behavior of -y-Mn02

[9]. Recently, a major breakthrough was made by Wroblowa et al. at Ford Scientific Lab.,

who demonstrated [10-13] that Mn02, in a chemically-moditied (CM) form, with Bi(III) as dopant at levels of 0.2 to 2 mol % [10], could be discharged and recharged in alkaline solution up to ca. 2000 times. The characteristics of the discharge and recharge processes were determined by slow cyclic voltammetry and so me galvanostatic experiments. Although it was suggested that incorporation of the Bi ion into the Mn02/Mn(OH)z layer-lattice could have the effect of preventing irreversible lattice expansion and contraction effects associated with discharge and recharge, and hence promoting rechargeability, it is difficult to understand how a small mol % of an additive such as Bi(III) ion could influence a bulk 3-dimensional oxide or oxy/hydroxide structure. In fact, the effects of Bi are still detectable even down to "catalytic" levels of dopant.

Attempts to examine structural changes induced by Bi in the CM Mn02 by means of X-ray diffraction were unproductive owing, evidently, to the amorphous nature of this material. Therefore a different approach, by means of X-ray absorption spectroscopy, was sought.

In the present paper we describe so me preliminary X-ray absorption (XANES and EXAFS) experiments on Bi-modified Mn02 which have been carried out both ex situ and in situ (with respect to the electrochemical discharging and recharging cell) at the Brookhaven National Laboratory synchrotron facility. Complementary electrochemical and spectro-electrochemical analytical measurements have also been made in situ du ring discharge and recharge of chemically modified and regular -y-Mn02 • Both these kinds of measurements have been undertaken in order to attempt to provide some new insights into the origin(s) of the remarkable effects of Bi [10-13] in promoting rechargeability of Mn02 •

The X-ray absorption experiments can provide: a) information on the changes of oxidation-state of the reactive species during discharge and recharge and b) the coordination environment (from EXAFS) of the reacting species [21]. The UV -visible spectrophotometric measurements enable the production and consumption of soluble Mn(III) species that is produced on discharge or recharge, to be followed at the CM Mn02 in comparison with the behavior at unmodified -y-Mn02 (International Common sampie no. 2). Useful general reviews on EXAFS spectroelectrochemistry by Sharpe, Heineman and EIder and by Eisenberger and Kincaid are to be found in refs. [17] and [18], respectively.

313

2. EXPERIMENTAL

2.1 X-ray Absorption Experiments (XANES and EXAFS)

Ex situ X-ray absorption (XAS) measurements were made at both the Bi L3 edge (13419 eV) and at the Mn K-edge (6539 eV). In these experiments, washed and dried sampIes of the chemically modified Mn02 from cycled cells were mixed with boron nitride powder and pressed into pellets that were then held in a sampIe holder between two layers of Kapton tape. In the case of the Bi, ~-edge measurements were done at both 77K and 298K.

The ex situ measurements were done at both Beam line XIIA and X23A2 at the National Synchrotron Light Source (NSLS) in Brookhaven National Laboratory. The X-ray ring was operating at 2.58 GeV and at a ring current of 110-240 mA. All the ex situ measurements were carried out in the transmission mode. Edge calibration was achieved through use of aBi or Mn foil, and a third reference detector.

The in situ X-ray absorption measurements were carried out on Beam Line X23A2. XAS spectra were obtained at the Bi L3 edge, in the transmission mode, in a 3-electrode cell similar to that described previously [19]. Attempts to do corresponding in situ XAS experiments at the Mn K-edge in the transmission mode failed because of excessive absorption by the 9M KOR electrolyte and other components of the cel!. The experiments were finally carried out successfully in another cell that permitted EXAFS measurements to be made in the fluorescence mode.

Fluorescence spectra were recorded using a Lytle detector. The electrochemical cell will be described in a later publication [20]. The quality of the data obtained in the fluorescence mode was excellent. The cell also permitted calibration of the edge through use of a reference Mn sampIe using signal detection in the transmission mode. All the in situ experiments were carried out under potentiostatic control at selected potentials along the discharge/recharge curves (cf. Fig. lA). Spectra were usually recorded when the current, after a given adjustment of potential, had declined to a very small value. At some potentials Ce.g. -O.4V on discharge and -0.15V on recharge), the current persisted for a long time Coften for some hours). In these cases, several spectra were recorded including several edge spectra which were taken within a small energy interval (-50 to 50 eV) together with complete EXAFS spectra.

The methods used for conducting the EXAFS data analysis were as previously described in several publications [20-22].

2.2 In Situ Spectrophotometric Experiments

The special experimental cell for in situ spectrophotometric characterization of soluble Mn ion intennediates, formed du ring discharge and rech arge of Mn02 has been described elsewhere [19]. It is based on a flat-plate type of 3-electrode mini-cell sealed on to and above a spectrophotometer cuvette (l cm path length). The whole combined spectroelectrochemical cell could be located in the chamber of a UV visible spectrophotometer. The absorption spectrum of species formed in solution during discharge or rech arge could be continuously monitored at various extents of discharge or recharge capacity, or at corresponding electrode potentials du ring these processes. The spectroelectrochemical results were recorded and plotted "three­dimensionally" (see below) as optical absorption as a function of wave-length for various extents of discharge/recharge or corresponding potentials.

314

2.3 Mn02 Electrode Preparation

CM Mn02 was prepared according to the procedure of Wroblowa et al. [10-13] by joint precipitation of Mn(OH)2 and Bi(OH)3 from solutions of soluble salts of Mn and Bi by addition of aqueous KOH. The Mn (11) species were converted to Mn02 by bubbling O2. Electrodes of the resulting CM Mn02 were made into film structures by milling with Lonza graphite (1:14 Mn02:C or 1:1 Mn02:C) with an addition of a Teflon fibre suspension, and roll-pressing the resulting mixture as described elsewhere [19]. These electrodes were examined in a thin, plate­geometry cell provided with a counter and a reference electrode (Hg/HgO). All experiments were conducted in 9M aqueous KOH made up from the high-purity BDH "Aristar" material.

2.4 Cyclic Voltammetry

Slow cyclic voltammetry experiments were performed in the usual way in order to relate processes arising on the discharge half-cycle to those on the recharge half-cycle. The current responses in these experiments were related to the in situ optical absorption experiments as discussed below. All electrochemical and spectro-electrochemical experiments were conducted at 298K.

3. RESULTS AND DISCUSSION

3.1 Spectro-electrochemical Behavior

Because of the extensive XANES and EXAFS data obtained in this work, space allows only some selected complementary results from the UV optical experiments to be shown here. However, a more detailed paper on the latter results will be published elsewhere.

Results of the optical absorption experiments are shown in relation to the cyclic voltammograms for discharge and rech arge of CM MnOz electrodes as these results provide, in part, the basis for discussion of the electrochemical behavior of the CM Mn02 in relation to that of unmodified -y-Mn02 (see below). Figs. l(A) and l(B) show a series of absorption spectra recorded du ring discharge and recharge of a CM Mn02 electrode. The absorption maximum is at 465 nm, characteristic of Mn(III) species [20] produced as a soluble intermediate. The absorbing species increase in concentration initially as the discharge proceeds and are consumed subsequently (by conversion to insoluble Mn(OH)2' deposited on the graphite-powder matrix) as discharge is completed. Closely related behavior is observed on recharge (Fig. 1B) from the Mn(OH)2 state.

It is to be emphasized that here the electrochemically determined discharge/recharge behavior corresponds to ca. 90% of the potential two-electron charge capacity of Mn02, unlike the situation with unmoditied -y-Mn02 where only ca. 30% of the first electron capacity is rechargeable [14-16].

The spectro-electrochemical results show that there is a clear relation between production and consumption of soluble Mn(III) species [21] (probably [20] Mn(OH)6J.) and the electrochemical behavior, e.g. under controlled-potential cycling (cyclic voltammetry experiments). This relation is illustrated in Fig. 2 for the CM Mn02 preparation examined in this work.

315

<0

ci

(A)

Fig. 1 The change of optical absorption spectrum of Mn(IlI) species with percentage utilization of discharge capacity; (A) in the discharge process and (B) in the rech arge process, at a CM MnOiLonza graphite mixture (1:4).

The results lend important and direct support for the involvement of the so-called "heterogeneous mechanism" of Mn02 e1ischarge and rech arge [1,7] in parallel with the "homogeneous mechanism", also e1iscussed in the literature [1-6] where Mn4 + ions in the Mn02

lattice are progressively reduced to Mn3+ and Mn2+ by electron transfer through the semiconducting oxide lattice while 0 2- species are converted to OB" species by a proton injection and hopping process to preserve charge balance.

From an electrochemical point of view, it has previously been supposed [1,4], on the basis of the progressively declining potential observed e1uring first-electron e1ischarge of MnOz (cf. Fig. 3), that the homogeneous e"/H+ solid-state transfer process predominates, and this, in fact, seems to apply to y-Mn02 • With the CM Mn02, on the other hand, it appears that the heterogeneous mechanism already provieles an alternative reduction pathway right from the beginning of discharge anel this suggests that the presence of the Bi dopant (amongst other effects) allows Mn02 to be discharged anel rechargeel to a greater extent via the soluble Mn(lJl) intermediate pathway than through solid-state intermediates.

316

.-I

7 ;----..

N 0 c 5 :2 Ol

'-../ 3 « "" >... +-'

Ul C <1> 0-1

+-' C <1> -3 L L ~

(A)

U - 5 L--...i---L--'----L-'-....I.--,---.l....-..'-----'---'-----'----'--L--->---.I

-0.6-0.5-0.4-0.~0.2-0.1 0.0 0.1 0.2 Potential vs HgjHgO j V

0.4 r----------=:..;--'=--~-----, 1 .1 :2

« ""0.3 c o

+-' Cl.. o 0.2 Ul

..Cl o x 0.1 o :2

.-. / \

/ \. , \

( B )

,

1.0"" 0.9 ..--­

o 0.80 0.70

x 0.6 _____ 0.5 '-.-/ 0.4 :§ 0.34-0.2 ~

u 0.1 c

O. 0 '--'--'---'-~_'__..I.._J...........J.__'____L........::::_-'---.l..........>--l 0 . 0 8 -0.6-0.5-0.4-0.~0.2-0.1 0.0 0.1 0.2

Potential vs HgjHgO j V

Fig. 2 Correlation of anodic and cathodic curves of a cyclic-voltammogram (A) at 0.5 mV S·I

with changes of maximum optical absorption at 465 nm (B) at various discharge and recharge potentials. CM Mn02/Lonza graphite= 1 :4; solid !ine: discharge; dashed line: rech arge.

Fig. 3 shows the discharg~'curve for -y-Mn02 cornpared with that of CM Mn02 at two discharge rates. The dec!ining curves 1,2 for y-Mn02 reflect the predominant role of the homogeneous, "solid-solution" pathway while curves 3 and 4 for the CM Mn02, exhibiting a predominant flat region, correspond to the much enhanced role of the heterogeneous process (2 phases co existent near equilibriurn) associated with the presence of the dopant.

0.2

~ 0.0 :r: "-01

~ -0.2 >

> ~-0.4

-0.6

0.1

2

0.2 0.3 0.4 Copocity/ Ah/ 9 Mn02

C/l0 rote C/5 rote

0.5

317

0.6

Fig. 3 Comparison of constant-current discharge curves for '}'-Mn02 (curves 1 and 2) (I.C. sampie No. 2) with those for CM Mn02 (curves 3 and 4) at discharge rates CIS (dashed lines) and C/lO (solid lines), respectively.

3.2 X-ray Absorption Results

A cursery examination of the XAS data indicated that the ex situ and in situ data were identical at the Bi L3 edge for the undischarged CM material and CM material that had been discharged to the one-electron (formaJly Mn(III» level. This inspection of the ex situ data also revealed that when the material had been discharged to the two-electron level, the Bi was in the form of partially oxidized Bi metaI. Under similar circumstances, the in situ data indicated the presence of unoxidized Bi, Le. the Bi (III) in the original preparation of the CM Mn02 had become reduced to Bi (0). However, the ex situ data at the Mn K-edge indicated that all discharged Mn materials had become reoxidized, presumably by O2 of the air, during sampie preparation. Accordingly, both the in situ and ex situ Bi data for the oxidized, Le. charged, materials are considered reliable. Similarly, the only reliable ex situ Mn data are those for the undischarged materials. Only in situ Bi data are reliable for completely discharged electrodes. In the case of the Mn results, none of the ex situ data are to be considered reliable for electrodes that have undergone any degree of discharge, i.e. reduction, because of the possibility of aerial reoxidation. This indicates the importance of conducting the complementary in situ measurements.

3.2.1 Manganese XANES

Figure 4 shows normalized XANES spectra for manganese oxides in which Mn is in various oxidation states. The materials include MnO, Mnp3' MnJ0 4 and Mn02 ("CMD", a battery grade, chemically prepared '}'-Mn02). There is a progressive shift in edge energy with increase in oxidation state.

318

1.2

z 1.0 o i= 0.. Cl::: 0.8 o (/)

CD <C 0 0 .6 w N :J ~0.4 Cl::: o z 0.2

I I ,

I

\ I \

I

\ ~"" .

:'\ .' ~"'" .: ji '\ ..... : \ . - ....

I.. ' , ....... , ,

I ;; \ , : , .. '.: I / .:

I . : I . :

I " .... I :

I " : I , : I ' :

1/' .: " .:

0.0 ===-__ ..J.----'-_d.o:....=--'-----L..._-'-----L._-'----l._--'-_I..---L----I

-20 -10 o 10 20 30 40 50 ENERGY TO Mn EDGE (+6539 EV)

Fig. 4 Comparison of near-edge Mn XAS for CMD Mn02 with those for Mn30 4, Mn20 3 and MnO.

Figure 5 shows comparisons of ex situ normalized XANES spectra for Mn20 3, two types of CM Mn02 and two types of ')'-Mn02 ("CMD" and "EMD", an electrolytically prepared battery grade material). The shift in edge positions of the CM Mn02 to lower energies may be due to partial replacement of Mn4+ sites by Mn3 + sites upon the introduction of Bi III. An undoped Mn02 sampIe, prepared by a similar precipitation procedure, had a spectrum almost identical to the 14: 1 doped material. The more intense white line of the CM Mn02 suggests a higher symmetry in the MnOö octahedra. This may be due to the absence of corner-shared octahedra in the structure.

z o e:: P-

1.50

1.25

a:: 1.00 o cn m ..;

fj 0.75 N :3 ..; ::0 @5 0.50 z

0.25

0.00

Mn02 XANES AND PREPARATION METHOD

1'. '.

,,: ... '

}{!-'-i",< : J/f

.... J~/ . : 11 . /'

: "

-20 -10 0 10 20 30 ENERGY RELATIVE TO Mn K EDGE (6539 eV)

319

Fig. 5 Comparison of near-edge Mn XAS for Mn20 3( ••• ), CM Mn02 (8.7:1) (- - - -), CM Mn02 (14:1) (-.-._), CMD Mn02 ( ••••• ) and EMD Mn02 (--).

Figures 6 and 7 show in situ Mn XANES spectra for electrodes at various stages of charge and discharge. The data indicate that, in the fully discharged state, the Mn is in the Mn(II) state, as might be expected. The data also suggests that, at intermediate potentials, there is correspondingly a mixture of oxidation states of Mn.

4.0 ,-.....-r-r-..,-,..--.,.-.--,--r-r--r-,--,--,

3.5 -O.5V ----.

(J) +-'

.'5 3.0 1-----

C .,g 2.5 1=---­

____ ... -O.4V(II)

---_-l -O.4V(I)

:ö .... ~2.0 f.----c o ~1 51-----.... . o (J)

~1.0 1=----

0.51=----

o . 0 '--'......J.......L.....l.......<.....J.......L.....J.........--'--'--'--'--.J

-20-10 0 10 20 30 40 50 Energy to Mn edge/( +6539 ev)

-O.35V

-O.3V

-O.2V

Fig. 6 Series of in situ XANES for CM Mn02 at 6 electrode potentials in the discharge direction.

320

4.0

3.5 ~

2 .§ 3.0 +0.1 V >-. O.OV ~

~ 2.5 .D -0.1 V ~

~2.0 c 0

%1.5 -0.15 V(II) 0 ({J -0.15 V(I) ~ 1.0

0.5

o. 0 '---'---'-~-'-~'--'--'--'--'-~C-...c--' -20-10 0 10 20 30 40 50

Energy to Mn edge/( +6539 ev)

Fig. 7 Series of in situ XANES for CM MnOz at 5 potentials in the rech arge process (potential range is different from that in Fig. 7 due to irreversibility between discharge and rech arge (see Fig.2A).

3.2.2 Bismuth XANES

Figure 8 shows the effect of oxidation state on the Bi XANES. There is a regular shift in edge energy in going from Bi to Bi(III) and up to Bi(V). In the case of NaBiü:;, there is a pre-edge shoulder due to transitions into empty 6s states. Figure 9 shows ex situ XANES data for the CM Mn02 electrode materials at various stages of discharge. Reference data for Bi foil and Biz0 3 are shown for comparison and clearly indicate at least partial reduction to Bi metal in the completely discharged electrodes. The overall edge in the oxidized material is, however, somewhat narrower than for Biz0 3• This suggests a higher degree of symmetry around the Bi than that found in Biz0 3 • The Bi in the material is in the Bi(lII) state in the undischarged material and in CM electrodes discharged to the one-electron level.

3.2.3 Bismuth EXAFS

Figure lO(a) shows the raw ex situ Bi EXAFS data obtained for an undischarged electrode at 298K. Almost identical data were obtained at 77K. This was true for both the undischarged material and material that had been discharged to the one-electron level. The lack of variation of the EXAFS with temperature indicates interestingly a high degree of static disorder. However, change of temperature had a major effect on the Bi EXAFS for the case of the completely discharged electrode.

1.25

z 1.00 o ~ c.. Cl:: o gJ 0.75 < Cl W N :J ~ 0.50 Cl:: o Z

0.25

0.00

EFFECT OF OXIDATION STATE ON Bi XANES

(a)

, , , ,

I I

, ,

, I

I

, I

_.;: .... "" ...

-20 -10 0 10 20 30 ENERGY RELATIVE TO Bi L3 EDGE (13419 eV)

321

Fig. 8 Dependence of XAS of Bi on its oxidation state: ---Bi(Q) --Bi20 3 ; ••••• NaBi03.

1.25

z 1.00 o ~ c.. Cl:: o gJ 0.75 < Cl W N :J ~ 0.50 Cl:: o z

0.25

0.00

EFFECT OF DISCHARGE ON Bi XANES

(b) --

-20 -10 0 10 20 30 ENERGY RELATIVE TO Bi L3 EDGE (13419 eV)

Fig. 9 Effect of charge state (during discharge) of the CM MnOz electrode on the Bi XANES: ---fully charged; ..• 1 e/Mn discharge; -. -' - 2e discharge; -Biz0 3 reference; ----Bi(Q).

322

0.03 EXAFS FOR Bi/MnOa• EX SITU 29BK

0.02 (a) 0.01

rn r..- 0.00 < >< ~

-0.01

-0.02

-0.03 0 2.5 5 7.5 10 12.5 15

k (RECIPROCAL ÄNGSTROM)

IN SITU Bi EXAFS ON ELECTRODE. CHARGED AT +0.2 V 0.03

D.02

(b) 0.01

rn r..- 0.00 < >< ~

-0.01

-0.02

-0.03 0 2.5 5 7.5 10 12.5 15

k (RECIPROCAL ANGSTROM)

Fig. 10 Raw Bi EXAFS data; Ca) für ex situ undischarged CM electrüde 298 K; (b) in situ CM electrode recharged tü 0.2 V.

323

Figure 11 shows Fourier transforms of the ex situ Bi data for the undischarged electrodes and electrodes discharged to the one-, and two- electron levels. Changes in the RDF with discharge are c1early observable. A comparison of the RDF of the completely discharged electrode material at 77K with that found for a Bi foil at 77K c1early reveals the presence of Bi metal species. The major differences between the undischarged material and that discharged to the one-electron level are seen at around 3A.

Table 1. Fourier transform parameters for isolating Bi-O contributions in Bi doped MnDz and in doing Fourier transforms on reference material.

Material k" &(k1) &(A)

Ex Situ

Undischarged,298K 3 2.68-12.85 0-2.24

1e discharge, 298K 3 2.63-12.0 0-2.42

Undischarged,77K 3 2.63-12.5 0-2.22

le discharge, 77K 3 2.80-12.2 0-2.22

In Situ

Charged 3 2.63-12.7 0-2.22

Discharged to -0.8 V 3 2.60-15.8

Recharged to 0.2 V 3 2.60-12.2 0-2.22

Reference Materials

PbO (tetragonal red) 3 2.87-17.3 0-2.40

Bi foil (electroplated) 3 2.60-15.9

Figure 1O(b) shows the raw in situ Bi EXAFS data for an electrode recharged to .. 0.2 V. This is in a potential region where the Bi is oxidized but there is negligible oxidation (rech arge) of discharged Mn material. The data are to be regarded as of high quality.

The first check of the Bi EXAFS data was made by comparing the Fourier transforms for the Bi in the undischarged electrodes and in Bi20 3, is shown in Fig. 12. The Bi is c1early not present as Bi20 3 but must be associated in some way with the Mn02 structure.

324

EFFECT OF CHARGE STATE ON Bi RDF 2.0

tü 1.5 (a) Q :::> e-~ Cl < ;::a ::::s Il::

1.0 0 1<. cn z ..: a:: e-

0.5

2 4. 6 RADIAL CO ORDINATE (ANGSTROMS)

8

Fig. 11 Comparison between RDF's for Bi in CM MnOz in three states of charge: --fuJly charged; --- le discharge; •••• 2e discharge . .1.k = 2.6-12.9 k', kJ weighted.

1.5

RDF FOR BisOs AND UNDISCHARGED ELECTRODE

" " , . , ' I I

(b)

0.0 rra~-,-~~L--,-~~:tt:::~a~~:C...d o 2 4 u 6 B

RADIAL CO ORDINATE (ANGSTROMS)

Fig. 12 RDF's for Bi in Biz0 3 and in the undischarged electrode; --- Bi in electrode; --­Biz0 3 • .1.k = 2.6-14 kt, kJ weighted.

Figure 13(a) shows a comparison of the Fourier transform for the Bi EXAFS in undischarged electrodes at 77 K and 298 K. The data plots are almost identical. The Fourier trans form for the in situ data in Fig. 1O(b) is shown in Fig. 13(b). The data in Fig. 13(b) closely resemble the ex situ data obtained for the CM material after discharge to the one­electron level (see Fig. 11). Integration of slow-sweep current-potential profiles (cf. Fig. 2A) indicates that, at this potential, the electrode is charged to the one-electron level.

325

EXAFS Analysis of Bi-O Coordination

Preliminary analysis of the first peak in the Fourier trans form of the Bi EXAFS c1ata for the oxidizecl materials indicated that it corresponded to Bi-O coorclination shells. Considerable c1ifficulty was encountered, however, in generating good Bi-O phase and amplitude data.

2.0

0.5

2.0

rz:I 1.5 Cl :> f-o Z Cl < ::z ::z P:::

1.0 0 r.. rn Z

~ f-o

0.5

ROF FOR UNDISCHARGED Bi/MnOe 298 K AND 77K

" " I

(a)

2 4 ~ 6 RADIAL CO ORDINATE (ANCSTROMS)

ROF FOR IN SITU ELECfRODE RECHARCEO AT 0.2 V

(b)

8

24 0 68 RADIAL COORDINATE (ANCSTROMS)

Fig. 13 Fourier transforms of EXAFS data; (a) ex situ undischarged electrode at 77 K (---) and 298 K (--), (b) in situ electrode recharged to 0.2 V. Fourier transform parameters are given in Table 1.

326

k-SPACE FIT Bi-O DATA. EX SITU 77K 2 rT-r~rT~~-r~~~~~~~~~~~~~

" (a)

-1

2 4 6 8 k (RECIPROCAL ÄNGSTROM)

10

k-SPACE FIT FOR Bi-O DATA. CHARCED AT +0.2 V

(b)

-1

o 2 4 6 8 10 12 k (RECIPROCAL ÄNGSTROM)

Fig. 14 Fits in k-space tü Bi-O clata: (a) ex situ clata für unclischargeclelectrocle at 77K; (b) in situ electrocle rechargecl to +0.2 V. The clata are k3 weightecl anel the fitting parameters are given in Table 2. (--) experimental clata, (---) fittecl.

327

r-SPACE REAL FIT FOR Bi-O. EX SITU 77K

1.5

(a)

0.0 o 24. 6 B

RADIAL CO ORDINATE (ANGSTROMS)

r-SPACE REAL FIT FOR Bi-O. CHARGED AT +0.2 V

1.5

(b)

0.0 o 24 0 6

RADIAL CO ORDINATE (ANGSTROMS) B

Fig. 15 Fits in r-space for Bi-O data; for (a) data in Fig. 14(a), (b) data in Fig. 14 (b). The transforms are k3 weighted; Llk = 3.7-9.5 k l • (-) experimental data; (---) titted.

328

Table 2. Fitting parameters for the two Bi-O shells in Bi doped Mn~ materials

Material Calculated Parameters N R(A) Äa2(A2) ÄEo(eV)

Ex Situ

Undischarged,298K (1) 4.0 2.20 -0.00075 -0.65 (2) 4.0 2.32 -0.0036 18.9

Undischarged,77K (1) 4.0 2.17 -0.001 1.68 (2) 2.0 2.31 -0.004 18.9

le discharge, 298K (1) 4.0 2.16 -0.0002 6.28 (2) 2.0 2.26 -0.007 18.2

le discharge, 77K (1) 4.0 2.19 -0.004 3.16 (2) 2.0 2.30 -0.007 18.2

In Situ

Charged (1) 4.0 2.19 -0.0026 1.68 (2) 2.0 2.31 -0.006 18.9

Recharged, +0.2 V (1) 4.0 2.16 -0.001 2.24 (2) 2.0 2.29 -0.005 18.3

Initial attempts to generate these from the FEFF program of Rehr [27] failed to yield reasonable results. Data for comparative purposes were obtained on BiOCI at 77K. However, it was difficult to separate the Bi-O and Bi-Cl contributions. Finally, it was decided to use red tetragonal PbO as a reference material. Pure red PbO was prepared and data were obtained at 77K. The phase purity was checked by X-ray diffraction.

The Fourier transform parameters used in isolating the Bi-O contributions in Bi-doped Mn02 are given in Table 1. The parameters used in analyzing the PbO reference material are also given there.

All attempts to fit the data to a single-shell fit failed. Also examinations of Bi-O phase­corrected Fourier transforms indicated the presence of more than one shell. Accordingly, a two-shell fit was tried. Excellent fits were then obtained both in k-space and r-space for the two-shell modelling. These results are presented in Figs. 14 and 15. Excellent fits were also obtained for the imaginary part of the Fourier transform. Table 2 lists the fitting parameters for the two-shell fits. The fitting procedures for two-shell fits have been described in detail elsewhere [22]. All fits gave two shells with coordination numbers elose to 4.0 and 2.0. In the final fit, these were fixed at these respective values. When this was done, the fitting parameters were almost identical for kl and k3 weighted fits.

2.0

0.5

RDF FOR Bi FOIL AND Bi IN SITU DISCHARGED TO -0.8 V

" II I I I I

I

24 0 6 RADIAL CO ORDINATE (ANGSTROMS)

B

Fig. 16 Fourier transforms for electrodeposited Bi (--) and Bi in CM MnO" electrode discharged to -0.8 V.

329

Attel11pts were made to fit the second peak in the Fourier transforl11 tor the chargecl material, at about 3A. The backtransformed data indicated that this peak was most likely duc to aBi-Mn interaction, since the amplitude of the JCl weighteel EXAFS peaked sYllll11etrically at k==8k1• ABi-Mn refe,ence was generated using the FEFF program. Reasonably good fits in k-space were obtained but only with N==O.5 amI R==3.53A. The low value of N==0.5 cmIlel signify local coordination clisoreler generateel by the presence of Bi in the Mn oxide structure. However, the clifficulty in generating a reliable Bi-O reference using the FEFF progralll casts cloubts on these results. ihere are irregularities in the core-at0l11 phase shifts far high-Z atoms such as Pb anel Bi; these would have to be checked against the FEFF theory hefore any firmer conclusions can be drawn.

It is interesting that the seconel-peak contribution largely elisappears at the one-electron elischarge level anel when the electrode is rechargecl at 0.2 V. (See Figs. 11 amI 13(b)). Tbis suggests several possibilities: one is that, at this potential, the Bi-O species are only weakly associated with the Mn02 structure. ThllS, there is the possibility that the recharge promotion or "catalytic effect" of the Bi on the initial stages of rech arge might arise from an adsorption effect at the swjaces of MnO" particles.

Figure 16 shows a comparison of the Fourier transfarm of the in situ EXAFS for a deeply clischargecl electrocle and tor an electroplated Bi layer on Grafoil. The Bi is clearly present in the metallic state in the reclucecl material. However, the deet'ease in the amplitucle of the Fourier transforlll strongly suggests that very small particles are involvecl.

3.2.4 In Situ Manganese EXAFS

Although our data analysis of the in süu manganese EXAFS is not, at the moment, complete, the Fourier transforms of the clata are shown for comparison. Figures 17(A) and 17(B) show the progress of the Mn RDF profiles with various stages of clischarge. The data for the undischargecl amI completely clischargecl procluct are similar to those founcl tor the

330

0.20

r.:l 0.15 Q ::> E-o Z C < :::11 :::11 0.10 Il::

~ CI)

z < Il:: E-o

0.05

0.00

0.20

r.:I 0.15 § E-o Z C < :::11

~ 0.10

~ 0.05

0.00

0

o

IN SITU RDF Mn0alBi AND AT -0.25 V,-0.30 AND -0.35 V

(a)

" " 1..' ,: "J " '.1 , , f , '.

,

2 4 0 6 RADIAL CO ORDINATE (ANGSTROMS)

RDF CHARGED Mn0l!' AT -0.4 V (2) AND AT -0.5 V

",i

,. . , I'

:~ , : .,

(h)

24 0 6 RADIAL CO ORDINATE (ANGSTROMS)

B

B

Fig. 17 Fourier transforms of the in situ Mn EXAFS data during discharge; Ca) undischarged electrode (--), after discharge to -0.25 V (----),-0.3 V ( .... ) and -0.35 V (-.-.-~; (b) undischarged electrode (--), first scan at -0.4 V (----), second scan at -0.4 V ( ..... ), after discharge to -0.5 V (-.-.-). ~k=2.7-11.3 kt, k1 weighted.

IN SITU RDFs DURING CHARGE, -0.15 V (I&III), 0.1 V, 0.2 V 0.20

[:l 0.15 ~ e-. 52 t!) -< ::::iI

~ 0.10

~ 0.05

0.00

0.20

[:l 0.15 ~ e-. 52 t!) -< ::::iI

~ 0.10 o ~

~ e-.

0.05

0.00

i

I ,

o

f\

I I

, I

I ,

, , , , , :, , I

!: :' I

: ~~

I I I

(a)

24. 6 RADIAL CO ORDINATE (ANGSTROMS)

B

RDF UNDISCHARGED AND RECHARGED ELECTRODE (0.2 V)

(b)

o 2 4 0 6 RADIAL CO ORDINATE (ANGSTROMS)

B

331

Fig. 18 Fourier transform for in situ Mn EXAFS data during CM Mn02 recharge; (a) first scan at -0.15 V (--), second scan at -0.15 V (----), at 0.1 V ( .... ), at 0.2 V (-.-.-). (b) comparison of elata for undischarged electrode anel electroele recharged to 0.2 V (---), Ak=2.7-11.3 k\ k1 weighted.

332

Ni.O.OH/Ni(OH)2 couple [22], in that the reduced material has all the features of the oxidized material but with peaks at larger R values. This is consistent with a layer-type Mn02 being reduced to a layer-type Mn(OH)2 with extended Mn-O and Mn-Mn bond lengths.

The reduced amplitude of the peaks for the material at the intermediate stage of diseharge suggests a highly disordered material. Figure 19(B) shows a eomparison of the data for an eleetrode in undischarged and reeharged states. The results are identical and thus elegantly eonfirm the reversibility of the system, espeeially in respeet of structure ehanges. Figure 19(A) shows the development of the EXAFS pattern during charge. At -0.15 V, the disappearanee of the discharge product and the formation of the corresponding eharged produet can be cIearly seen. Onee again the data indicate that the intermediate, presumably some Mn(I1I) species, has a much weaker EXAFS pattern than either the charged material or the eompletely discharged material. Elucidation of this will require further data analysis and extension of the experimental work.

4. CONCLUSIONS

a) The uv-visible in situ speetrophotometry gives clear proof of the role of a soluble intermediate (MnIII species) in both the proeesses of discharge and reeharge of CM Mn~.

b) During discharge and recharge the quantities of soluble intermediate generated (and consumed) are substantially greater for the CM Mn02 than for nblank-n, or )'-Mn02• It is eoncIuded from this result that one of the effects of Bi dopant is to promote the heterogeneous over the homogeneous pathway.

c) XANES results give charaeterizations of the states of oxidation of both Mn and Bi in the proeesses of discharge and rech arge of CM Mn02 •

d) The ex situ EXAFS resuIts give speeitle information on the Bi-O, the Bi-Mn and Mn-Mn coordination distances through the evaluated radial strueture functions.

5. ACKNOWLEDGEMENTS

Grateful acknowledgement is made to the Natural Sciences and Engineering Research Council for support of this work on a Strategie Grant. D. Qu acknowledges the award of an Ontario Graduate Seholarship. Both the Ottawa authors are much indebted to the their co-author, J. McBreen, for arranging for the XAS experiments to be carried out at the Brookhaven synchrotron and for his personal involvement in their execution and in discussions on the interpretation of results obtained. We also thank Dr. L. Bai of the laboratory at Ottawa for useful discussions.

333

6. REFERENCES

[1] Kozawa A. and Yeager J.F., (1968), J. Electrochem. Soc., 115, 1003.

[2] Kozawa A. and Powers R.A., (1966) J. Electrochem. Soc., ill870.

[3] Kozawa A. and Powers R.A., (1967), Electrochem. Tech., ~, 535.

[4] Kozawa A. and Powers, R.A., (1968), J. Electrochem. Soc., 115, 122.

[5] Kozawa A. and Kagaku B., (1983), BMRA Symposium, Brussels 38.

[6] Kozawa A. and Yeager J.F., (1965), J. Electrochem. Soc., ill, 959.

[7] Kozawa A. and Powers R.A., (1972), J. Chem. Educ., 49, 587.

[8] Ruetschi P., (1976), J. Electrochem. Soc., 12,495.

[9] McBreen J. in Collins D.H., (1975), (Eds.), Power Sources, Vol. 5, Academic Press, London, paper no. 31, p. 525.

[10] Yao Y.F., (1985), U.S. Patent No. 4,520,005, May 28; see also Dzieciuch M.A., Gupta N. and Wroblowa H.S., (1988), J. Electrochem. Soc., 135, 2415.

[11] Dzieciuch M.A., Gupta N. and Wroblowa H.S., (1988), J. Electrochem. Soc., 135, 2415.

[12] Yao, Y.F., Gupta N. and Wroblowa H.S., (1987), J. Electroanal. Chem., 223, 107.

[13] Wroblowa H.S. and Gupta N. (1987), J. Electroanal. Chem., 238b, 93.

[14] Kordesch K., Gsellmann J., Peri M., Tomantschger K. and Chemelli R., (1981), Electrochimica Acta, 26, 1495.

[15] Kordesch K., (1983), BMRA Symposium, Brussels 81.

[16] Kordesch K., Daniel-Ivad J., Kahraman E., Mussnig R. and Toriser W., (1991), Paper #10052. 26th International Energy Conversion Engineering Conference, Boston, Massachusetts.

[17] Sharpe L.R., Heineman W.R. and Eider R.C., (1990), Chem. Rev., 90, 705.

[18] Eisenberger P. and Kincaid B.M., (1978), Science, 200, 1441.

[19] McBreen J., O'Grady W.E., Pandya K.I., Hoffman R.W. and Sayers D.E., (1987), Langmuir, 1, 428 .

334

[20] MeBreen J., Qu D. and Conway B.E., to be published.

[21] Pandya K.I., O'Grady W.E., Corrigan D.A., MeBreen 1. and Hoffman R.W., (1990), J. Phys. Chern., 94, 21.

[22] Pandya K.I., Hoffman R.W., Me Breen J. and O'Grady W.E., (1990), J. Eleetroehern. Soe., 137,383.

[23] Qu D.Y., Conway B.E., Bai L., Zhou Y.H. and Adams W.A., (1993), 1. Applied Eleetroehern. in press.

[24] Lott K.A.K. and Syrnons M.C.R., (1959),1. Chern. Soc., 829.

[25] Kozawa A., Kalnoki-kis T. and Yeager J.F., (1966),1. Eleetrochern. Soc., lli, 405.

[26] Godart C., Latroehe M., Tretigny C., and Levy-Clement C., (1992), Phys. Stat. Sol. (a) 132, 253.

[27] Rehu 1.1., Albers R.C. and Mastre de Leon 1., Physica B., 158,417 (1989).

EXAFS STUDIES OF FILM COATED ELECTRODES

R.C. EIder, Lee R. Sharpe, David H. Igo, Robert O. Rigney and Villiam R. Heineman

Department of Chemistry University of Cincinnati Cincinnati, OH 45221-0172, USA

ABSTRACT. EXAFS has been used to study structural and redox changes in transition metal complexes as films coating electrodes or contained in films. Ve have used copper complexes of 2,9-dimethylphenanthroline embedded in Nafion films to study the change from four to five coordination which occurs on oxidation of Cu(I) to Cu(II). The increase in coordination number presumably occurs from the addition of a water ligand on rearrangement from tetrahedral to trigonal bipyramidal geometry. Vhen similar experiments are performed with the analogous bathocuproine ligand, with phenyl sulfonate substituents, in a poly(dimethyldiallylammonium) chloride film, the copper can still be oxidized from I to 11. However in this case, there is no increase in the coordination number. Other studies were made on Prussian Blue and Ruthenium Pur pIe films. In these films multiple oxidation states are also present and the EXAFS were measured for each. In the latter case the iron centers are redox active whereas the ruthenium centers are not. EXAFS studies were performed at both the Fe and Ru K edges. These studies were carried out at SSRL, NSLS and CHESS.

1. Introduction

X-ray absorption spectroscopy of electrochemically genera ted species involves the combination of spectroelectrochemical methods developed by Professor Heineman and his students and those of X-ray absorption spectroscopy, as imp1emented by Professor Elderand his students.

Spectroelectrochemistry combines an electrochemical technique with a spectroscopic technique (1). In a typical experiment, electrochemistry is used to control the oxidation state of an electroactive species in solution or on an electrode surface while the system is monitored spectroscopically. Spectroelectrochemical techniques have the potential to provide more information about a particular system than either or both of the component techniques used separately.

Thin-layer spectroelectrochemistry takes advantage of the speed wi th which complete electrolysis can be achieved in a thin-layer of solution or in a thin polymer film on an electrode surface with mass

335

C. A. Melendres andA. Tadjeddine (eds.), Synchrotron Techniques in Inteifacial Electrochemistry 335-348. © 1994 Kluwer Academic Publishers.

336

transfer by diffusion. Formal redox potentials, electrode mechanisms, and spectra of a wide range of redox species (organic, inorganic, and biological) have been determined by thin-layer spectroelectrochemistry (1). Techniques have been developed which combine electrochemistry with spectroscopic measurements in the ultraviolet (2,3), visible (3-5), and infrared (2) regions of the spectrum.

X-ray absorption spectroscopy (XAS) enables the determination of the oxidation state of an absorbing atom in a compound, the type and number of atoms bound to the absorbing atom, and the corresponding bond lengths (6-8). Two regions of the spectrum are of particular interest. The edge region (X-ray absorption near edge structure, or XANES) corresponds to promotion of a core electron to an unoccupied atomic or ~olecular orbital or to the continuum. This region has the potential to provide information about the oxidation state of the absorbing atom and the identity of coordinating atoms. The extended X-ray absorption fine structure (EXAFS) region covers approximately 1000 eV beyond the edge and is characterized by aseries of oscillations in the X-ray absorption coefficient,~. EXAFS contains information about the identity and number of neighboring atoms and their distances from the absorber.

Although EXAFS has thus far seen relatively limited application to the study of electrode processes, the field is growing, since it provides valuable structural information that is otherwise unavailable. We have recently reviewed EXAFS spectroelectrochemistry (9); two other reviews have recently appeared (10,11).

Our early work in this field dealt with: (a) development of thin­layer cells that are specifically adapted to the particular constraints of EXAFS spectroscopy and thus enable rapid and exhaustive generation of specific oxidation states of metal complexes for obtaining EXAFS spectra, (b) evaluation of pertinent experimental parameters, such as accessible concentration ranges and reproducibility, (c) application to coordination compounds and bioinorganic materials in which a media tor­titrant is needed to enhance electrolysis, (d) demonstration of in situ electrochemistry to remedy reduction by X-ray genera ted hydrated electrons, and (e) examination of the effect of oxidation state on bond length and coordination number in selected systems (12-14). Our more recent work, which is described in the next sections, has focused mainly on the development of EXAFS spectroelectrochemistry for the study of: (a) metal ions that are immobilized in thin polymer films on electrode surfaces, (b) electroactive films on electrode surfaces and (c) electronically conducting polymer electrodes.

2. Coordination Compounds in Ionically Conducting Polymer Films

A major objective of our recent work has been to demonstrate that EXAFS spectroelectrochemistry is applicable to the study of metal complexes immobilized in polymer films on electrode surfaces. We have demonstrated this with two ionic polymers: Nafion and poly(dimethyldiallylammonium chloride).

337

2.1. Cu(dmp)2+1 ,+2 in Solution and in Nafion.

Perhaps the most extensively studied polymer for modified electrodes is the water insoluble cation exchange membrane Nafion. The practical importance of Nafion lies partly in its use as a protective coatin~. As an overlying membrane, Nation provides both size exclusion and a charge selective barrier. The size selectivity of the polymer matrix is determined in part by the hydrocarbon chain length. The ability to change the molecular weight of the Nafion repeat unit allows partitioning into the film based on size. The sulfonate group enables the polymer to selectively allow both neutral and posi tively charged materials of the appropriate size to permeate through or be incorporated into the matrix (15). Electrodes coated wi th Nafion can then be used to competitively select and detect charged and neutral electroactive materials. A good example is i ts use as a coating for ultra-microelectrodes for the in vive detection of neurotransmitters (16) .

We have examined [CuI (dmp)2]+ (dmp 2,9-dimethyl-1,10-phenanthroline) incorporated into a Nafion polymer (17). The spectroelectrochemical cell had a gold film electrode vapor deposited on a Mylar sheet. Co-dissolved colloidal graph i te and Nafion in ethanol/ethylacetate was sprayed onto the gold to give a -0.1 mm-thick polymer film. The Nafion-coated electrode was then immersed in electrolyte containing [CuI (dmp)2]+' which partitioned from aqueous solution into the negatively charged Nafion. High quality EXAFS spectra were obtained for [CuI (dmp)2]+ and [cuII (dmp)2]2+ in the Nafion film by measuring the fluorescence signal passing back through the electrode, which was held at either an oxidizing or reducing potential.

We reported tha t the Cu-N bond length decreased from 2. 06A to 2.02A and the coordination number increased from 4 to 5 when the CuII species was generated by application of a sufficiently positive potential to the gold film electrode. This increase in coordination number is attributed to a change of coordination geometry from a tetrahedral (4-coordinate) to a trigonal bipyramidal (5-cooräinate) arrangement. The added ligand is most likely coordinated through oxygen, ei ther from a sulfonate group in Nafion or a water molecule present in the polymer matrix. This experiment clearly showed that EXAFS has sufficient sensitivity for the study of metal ions in thin polymer films on electrode surfaces.

2.2. Cu(bcp-S)2-3 ,-2 in Solution and in p(DMDAAC).

Electrochemical cells in which the traditional supporting electrolyte consists of ionically-conducting, semi-rigid, polymer films for solid­s ta te vol tammetry have been reported (18-20). Such elec trochemical cells enable gas phase detection of electroactive species (21,22). The water soluble polymer, poly(dimethyldiallylammonium chloride) or p(DMDAAC) is an example of this type of polymer. The uses and charateristics of p(DMDAAC) as an ionically conducting polymer film have been investigated extensively in our laboratories (20,22-25).

338

Structural Formula 1 p(DMDAAC)

Pt Aux.

Mylar

o /' Mylar

Figure 1: The constant humidity cell for examination of metal complexes in p(DMDAAC).

In our EXAFS spectroelectrochemistry experiment, p(DMDAAC) not only physically contains the metal complex in the X-ray beam but also acts as the electrolyte during electrochemical conversion from one oxidation state to another (25). The electrochemical cell (Fig. 1) consists of a gold minigrid optically transparent thin-layer working electrode (OTTLE), two platinum foil auxiliary electrodes, and an oxidized silver wire reference electrode. The electrochemical cell components are attached to a Mylar window. A solution of the complex and p(DMDAAC) -is cas t on to the OTTLE and allowed to dry. In order to provide the required ionic conductivity, the completed electrochemical ce 11 is encased in achamber with a saturated aqueous potassium nitrate solution (-2 mL) that maintains a constant relative humidity of 92.5% (26). This cell can then be placed perpendicular to the light path of our diode array spectrometer for uv-visible spectroelectrochemistry or posi tioned 45 0 wi th respect to the X-ray beam inside our Lytle X-ray fluorescence detector for obtaining X-ray absorption spectra.

The cOnstant humidi ty electrochemical cell provides some unique advantages: (a) the experiment is done in the absence of support ing electrolyte and (b) the metal complex is concentrated within a thin polymer film (-0.05 mm). Vith no supporting electrolyte, the behavior of the complex can be analyzed much more simply in the presence of only the polymer film. This also minimizes attenuation of the X-ray be am by electrolyte absorption. Concentration of the complex at the electrode surface increases the amount of material probed by the X-ray beam, thus increasing the signal to noise ratio. Vith this electrochemical cell design we have shown that within the polymer film complete electrolysis of the complex occurs. This was demonstrated via double potential step chronoamperometry in conjunction with visible spectroscopy experiments (25) • In addit ion, we have shown tha t high qua li ty X-ray fluorescence spectra of a copper complex incorporated into a p(DMDAAC) film can be obtained.

339

In these experiments, the electroactive molecule was [CuI(bcp­S)2]-3, where bcp-s = 2,9-dimethyl-4,7-diphenyl-1,10-phenanthroline disulfonic acid disodium salt. This complex possesses a nega ti ve charge, thus allowing it to electrostatically interact with the positive charge of p(DMDAAC). The electrochemistry of the complex was evaluated in the semi-rigid polymer film and was found to exhibi t an increased degree of reversibility as compared to that in solution.

The abili ty to access each oxidation state of the com[!lex allows analysis of the structural characteristics of both the CuI and CuII oxidation states while in p(DMDAAC). The structures were determined from analysis of our EXAFS spectroelectrochemistry experiment (25), and compared to the results for solution. Results of these experiments are presented in Table 1, employing CuI(dmp)2BF4' where dmp = 2,9-dimethyl-1, 10-phenan throline, as a model compound. Since, dmp and bcp-s have identical phenanthroline backbones, [CuI(dmp)2]BF4 should be an ideal model for Cu-N interaction in Cu(bcp-s)2' Vhen comparing results obtained with Cu(bcp-s)2 with those of Cu(dmp)2 solution (17), the Cu-N bond distance remains approximately constant while Cu(dmp)2 showed a decrease of -0.04 X. As with Cu(dmp)2, the number of atoms coordinated to copper increases by about 25% when the CuI complex is oxidized in solution. The results for coordination number have been scaled to a value of four for the CuI complexes and the highly correlated, disorder parameter has been held constant. Thus in solution, Cu(bcp-s)2 seems to convert from a four-coordinate, tetrahedral complex to a five-coordinate, trigonal bipyramid on oxidation. However, in the polymer film, no increase in coordination number is observed on oxidation from CuI to CuII .

Table 1 - Cu(bcp-s)2 in solution and p(DMDAAC) Results

Solution (8.7 mM) Complex in Film BL CN BL CN

CuI 2.05 4.0* 2.06 4.0* CuII 2.04 5.0 2.06 4.0

*The values of the coordination number for both CuI complexes have been scaled to 4.0 for the expected tetrahedral geometry.

Our results for [CuI (bcp-2)2]-3/[CuII (bcp-S)2]-2 in p(DMDAAC) are extremely interesting with respect to the effect of the polymer environment on the structures of the two redox forms of the copper complex: namely, in aqueous solution [CuI (bcP-S)2]-3 exists as a tetrahedral complex which converts, with the addition of a fifth ligand, to a trigonal bipyramid when oxidation to [CuII(bcp-S)2(H20)]-2 occurs; whereas, in the polymerie film, p(DMDAAC), the tetrahedral structure is observed for both oxidation states. The polymer environment apparently prevents the structural change associated with electron transfer that occurs in aqueous environment. Thus, a seemingly innocuous polymer with no coordination sites, whose primary

340

purpose in the electrochemical cell is simply to provide an ionically conducting medium (i.e. supporting electrolyte) to support charge flow in the electrochemical cell, appears to significantly alter the chemistry of a metal complex.

We attribute this structural anomaly to the role that bulky, negatively charged [CuI(bcP-s)21-3/[CuII(bcp-S)21-2 plays in electrostatically cross-linking the positively charged DMDAAC polymer. The effect of electrostatic cross-linking is dramatically observable in a swelling experiment. [CuI (bcp-S)21-3 i5 added to an aqueous solution in which a DMDAAC-coated electrode has been immersed and allowed to swell. Addition of complex to the solution causes the polymer film to shrink by a factor of 10 or more. As the complex diffuses into the film, it ionically interacts with the quaternary ammonium sites, thereby, cross-linking the polymer with an accompanying "squeezing out of water." Apparently, the ionic attraction between positively charged polymer sites and the negatively charged sulfonic acid sites on the ligands is sufficiently strong to hold the complex in a relatively fixed tetrahedral geometry when the copper center is oxidized to CuII . The size of the bcp-s ligands should also affect the ability of the complex to reorganize. In order for the complex to assurne trigonal bipyramidal geometry, the bulky ligands mus t reorient, placing one ni trogen of a bcp-s ligand in an axial coordination si te. For the ligand to move, it must essentially drag along large amounts of polymer as it assumes the new geometry, or substantially increases the separation between anionic and cationic sites.

3. Metal Ions in Electroactive Polymer Films on Conducting Hetals.

Electrodes modified with iron hexacyanometalate films have interesting electrochemical and spectral properties (27). An iron hexacyanometalate film is potentially useful in electrochromic display devices. We have investigated the structures of iron hexacyanoferrate and iron hexacyanoruthenate, termed Prussian Blue and Ruthenium Purpie, respectively, with synchrotron X-ray absorption spectroscopy.

3.1. Prussian Blue

X-ray powder diffraction was first used to examine the structures of Prussian Blue (PB) precipi tates (28). Four structures were reported: Prussian Blue (PB), MFeIII FeII (CN)6; Everitt's Salt (ES), M2Fe U FeII (CN)6; and Berlin Green (BG), Ml/3FeIIIFe~3III, Fe~/.3II(CN)6. PB prepared in the absence of alkali metals, Fe 4 II[FeII (CN)6h, was also reported (28). This last form has tradi tionally been termed "insoluble" PB, while the form including alkali cations has been termed "soluble" PB, although both forms are actually nearly insoluble. X-ray crystallography has also been used to characterize the structure of the "insoluble" crystalline complex (29). Four different iron coordination environments were found. Upon first inspection, the iron coordination structure of PB appears too heterogeneous to prac tically analyze by EXAFS techniques, yet there is a distinct Fe-Fe separation of 5.08 Z. This distance is associated with the Fe-CsN-Fe linkage within the cage

341

structure. The colinearity of the Fe-C!!N-Fe array allows direct observation of the separation of the iron centers within the cage of PB at a distance not normally achievable wi th conventional room temperature EXAFS analysis (30).

An electrodeposi ted film of PB on an electrode and its electro­chromic behavior was first reported in 1978. Energy dispersive X-ray analysis, Mossbauer, infrared, and optical absorption spectroscopic investigations have confirmed that the structure of insoluble PB is similar to that of electrochemically deposited material. Upon reduction alkali cations can be found in the film. Although the structure of the PB form of the film is known, li ttle or no direct information has been obtained concerning the in si tu behavior of the electrochemically deposited maferial and the structure of its oxidized and reduced forms. EXAFS analysis allows relatively direct evaluation of the distances between the iron and other atoms of the cage. Electrochemically genera ted forms of PB include the totally reduced Everitt's Salt (ES) and its partially oxidized state, Berlin Green (BG). The electrochemical properties of the compound in its respective states have been described in great detail (31); a representation of the electrochemical reaction is given below:

{FeIII[Fe~~~,Fei~3(CN)6]1-1/3

BG

{Fe II [Fe(CN)6]1-2

ES

Importantly, the various oxidation states are easily accessed and the products are stable over a long period of time if potential control is maintained.

Our investigations have involved the electrochemical deposition of PB onto a gold film electrode (300 K thick, vapor-deposi ted onto Mylar) incorporated into a thin-layer EXAFS spectroelectrochemical cello described in our recent review (9). The gold film not only provides a suitable electrode material and physical support for PB, but it is transparent in both the visible and X-ray spectral regions. High quality X-ray fluorescence spectra werecollected on each of the three oxidation states. The Fourier transforms of the EXAFS (Fig. 2) indicate Fe-Fe interactions. The peak appears at a shorter distance (::4.5) than the crystal value of 5.08 K due to the EXAFS phase shift, as expected. A marked decrease in amplitude of the Fe-Fe peak at ::4.5 K is seen for ES compared to PB or BG. The decrease of the iron backscattering interaction is attributed to distortions of the cage structure of PB upon i ts reduction to ES. Such distortions probably occur since the cage has to incorporate hydrated cations to counter­balance the excess nega ti ve charge tha t resul ts from reducing the charge on iron. That increased disorder occurs is further supported by analysis of the amplitude of the back-transform function

342

of the radial distribution peak backscattering interaction (Fig. 3). charge of the iron wi thin the film, envelope amplitude decreases markedly.

r 0.96 T

M 0.72 A G 0.48

x10~. 24

3.0 4.2 5.4 R

Figure 2: Psued.o Radial Distributi.on Functi.on from the Fourier transforms .of the BG (solid), PB (d.otted) and ES (dashed) EXAFS.

3.2. Ruthenium PurpIe

associated with the Fe-Fe Upon decreasing the .overall

fr.om BG t.o PB and t.o ES, the

E X A F 5 * K * * 3

, , I

2.52 r

~ dl.{I.,ß 0.84 . ' , , , I I , '''''AJV'~ I

, , I I

-0.84 ' , ,

-2.52 -,

" ,

3.0 7.0 11.0 K

Figure 3: Filtered EXAFS .of the BG (s.olid), PB (d.otted) and ES (dashed) resulting fr.om the Fe-Fe interacti.on.

Tle have als.o investigated a PB structural anal.og, namely Ruthenium Pur pIe (RP), in which the hexacyanoferrate in PB is substituted with hexacyan.oruthenate in RP. Thus the two transiti.on metal sites are distinguishable in RP. M.oreover, electr.odes m.odified with RP maintain electr.ochemical properties similar t.o that .of PB (32);.a representati.on .of the electr.ochemical reacti.on is given bel.ow:

Ruthenium PurpIe, RP Ruthenium Tlhite, RTl

High quality X-ray flu.orescence spectra were .obtained at the ruthenium and ir.on regi.ons für an electr.ode m.odified with a RP film. Electr.ochemically genera ted structures .of the film, the RP form and the t.otally reduced RTl form, were examined by XANES and EXAFS meth.ods. The t.otally .oxidized f.orm was not investigated.

Examinati.on .of the ruthenium XANES regi.on für the RP and RTl forms (Fig. 4), reveals n.o apparent changes. H.owever, examinati.on of the ir.on XANES regi.on (Fig. 5) sh.ows a 4 eV decrease in edge energy between RP and RTl. Thus the ir.on red.ox center is inv.olved in the electr.ochemical reducti.on mechanism, and n.ot the hexacyan.oruthenate species. This has been verified in PB films by M.ossbauer spectr.osc.opy.

N 0.88 0 R M 0.66 A l 0.44

0.22

22.1 22.1 22.2 KEV

Figure 4: Ruthenium XANES for RP (solid) and RW (dotted).

N 0 R M A L

1.60 -

1. 20

0.80

0.40

7.1 7.1 7.2 KEV

Figure 5: Iron XANES for RP (solid) and RW (dotted).

343

Fourier transforms were calculated for the ruthenium EXAFS (Fig. 6) and the iron EXAFS (Fig. 7) for both oxidation states of each of the films to produce pseudo-radial distribution functions (PROFs). Examination of the ruthenium PROF for the RP form indicates Ru-C=N-Fe interactions. That this is a highly ordered, linear array is indicated by the presence of the third peak (Fe backscatterer near 4.5 A) described previously in the PB study. Upon reduction to RW, the order (linearity) is lost as indicated by the disappearance of the Fe peak in the ruthenium PROF for RW. Only small variations are seen in the first and second peaks assigned to the carbon and nitrogen atoms, respectively, in the cyanide linkage. Subsequent analysis has shown the Fe peak to reappear in the ruthenium PROF for RP after cycling the film as many as 50 times between RP and RW forms. Assuming that the

r 7.20 T

M 5.40 A G 3.60

1.80

0.0 3.0 6.0 R

Figure 6: Fourier trans form of Ruthenium EXAFS for RP (solid) and RW (dotted).

r T 1.76

M 1.32 A G 0.88

X10A.44

0.0 3.0 R

6.0

Figure 7: Fourier transform of Iron EXAFS for RP (solid) and RW (dotted) .

344

Ru-GeN-Fe array is quite linear in RP, then the changes noted in the ruthenium PRDF for RW are due most likely to both moving the iron off colinearity with the still linear Ru-GeN and also increasing the Ru-Fe distance disorder . When cycled back to RP, Fe is reposi tioned again into the linear array. This has previously been ascribed in the PB study as due to incorporation of hydra ted cations into the caged structure.

The iron PRDFs for RP and RW (Fig. 7), again indicate the linearity of the Fe-NaG-Ru array by the presence of the third peak (Ru). Interestingly, upon reduction from RP to RW, a drastic decrease in the magnitudes of all three peaks in the iron PRDFs is evident. Again this indicates that the iron has moved out of the linear array. However, that alone would not explain the large decrease seen in the magni tude of the first peak. Two variables can cause such a dramatic effect: changes in the coordination number and/or the Debye-Waller factor (disorder parameter) a. After cycling the film 50 times between the RP and RW forms, the peaks in the iron PRDF for RP returned at the magnitudes formerly seen for the uncycled iron PRDF for RP. This suggests that the coordination number remained constant throughout the reduction/oxidation of the film. This means that actual loss and subsequent return of ligands at the Fe site is unlikely to occur when the film is cycled between the RP and the RW forms. Thus disorder around iron must significantly increase in going to the reduced RW form.

Fe-Ru distances were calculated for both the RP and the RW forms. These gave 5.03 Z for RP and 5.02 Z for RW. Using the Ru EXAFS, we were able to calculate a Ru-Fe distance only for the RP form, 5.02 A. For RW there was no discernable Fe peak to fit. Thus, on average, the Ru-Fe distance appears to remain constant in both the RP and RW forms. However, note that the fit value is only an average of the actual values.

If the cage were to remain intact except for motions of the iron atoms away from the corners of the square array, as shown for a two­dimensional case in figure 8, one should still see the same Ru-Fe average distance. However, the Fe peak in the Ru EXAFS would be weaker due to increased disorder and loss of colinearity in the RW case. For the Fe EXAFS all three peaks would be weaker for the RW case, since the colineari ty of all the second and third iron neighbors would be lost and the disorder would most affect the first shell neighbors.

The rationale for such motions could be the repulsion between the alkali cations and the Fe(II) ions of the RW form. The alkali ions must move into the cage to provide charge neutrality, and also the Fe(II) ions are expected to be more loosely bound to the cyanide anions in the RW form compared to the Fe(III) ions of the RP form. While such an explanation is highly speculative, it does have the virtue of accounting for all of the observations.

345

Figure 8: A) A Two-dimensional segment of the RP structure.

B) A two-dimensional model of the structural changes occur­ring on reduction of iron to give R\.T and the uptake of Na+ ions for charge neutrality.

4. Acknowledgement \.Te thank the Air Force Office of Scientific Research for support provided by grant AFOSR 88-0089.

5. References

1) Heineman, \.T.R.; Hawkridge, F.M.; Blount, H.N., in Electroanalytical Chemistry; (A.J. Bard, ed.) Marcel Dekker, New York, 1984; Vol. 13, pp. 1-113.

2) Heineman, Y.R.; Burnett, J.N.; Murray, R.Y.: Optically Transparent Thin-Iayer Electrodes: Ninhydrin Reduction in an Infrared Cello Anal. Chem. 40, 1974-1978 (1968).

3) Anderson, C.\.T.; Halsall, H.B.; Heineman, \.T.R.: A Small-Volume Thin-Iayer Spectroelectrochemical Cell for the Study of Biological Components. Anal. Biochem. 93, 366-372 (1979).

4) Heineman, \.T.R.; Norris, B.J.; Goelz, J.F.: Measurement of Enzyme EO, by Optically Transparent Thin Layer Electrochemical Cells. Anal. Chem. 47, 79-84 (1975).

5) Murray, R.Y.; Heineman, Y.R.; O'Dom, G.\.T.: Transparent Thin Layer Electrochemical Cell. 1666-1668 (1967).

An Optically Anal. Chem. 39,

346

6) Cramer, S.P.; Hodgson, K.O., X-ray Absorption Spectroscopy: a New Structural Method and .Hs Applications to Bioinorganic Chemistry, in Progress in Inorganic Chemistry, Vol. 25, (S. J. Lippard, ed.), pp. 1-39 (1979).

7) Eisenberger, P.; Kincaid, B.M.: EXAFS: New Horizons in Structure Determinations. Science, 200, 1441-1447 (1978).

8) Teo, B.K.: Chemical Applications of Extended X-ray Absorption Fine Structure (EXAFS) Spectroscopy. Acc. Chem. Res. 13, 412-419 (1980).

9) Sharpe, L.R.; EIder, R.C.; Heineman, W.R.: EXAFS Spectroelectro­chemistry. Chem. Rev. 90 705-722 (1990).

10) James Robinson: X-ray Techniques in Spectroelectrochemistry, Theory and Practice, (R.J. Gayle, ed.) Plenum Press, New York, pp. 9-40 (1988).

11) Dewald, H.D.: Use of EXAFS to Probe Electrode-Solution Interfaces Electroanalysis (N.Y.) 3 145-155 (1991).

12)

13)

Dewald, H.D.; Watkins, J.W.; EIder, R.C.; Heineman, W.R.: Development of Extended X-ray Absorption Fine Structure Spectroelectrochemistry and its Application to Structural Studies of Transition-Metals in Aqueous Solution. Anal. Chem. 58, 2968-2975 (1986).

Smith, D.A.; EIder, R.C.; Heineman, Absorption Fine Structure Thin-Layer Anal. Chem. 57 2361-2365 (1985).

W. R. : Extended X-ray Spectroelectrochemistry.

14) Smith, D.A.; EIder, R.C.; Heineman, W.R.: Direct Determination of Fe-C Bond Lengths in Iron(II) and Iron(III) Solutions Using EXAFS Spectroelectrochemistry. J. Am. Chem. Soc. 106, 3053-3054 (1984).

15) Whiteley, L.D.; Martin, C.R.: Perfluorosulfonate Ionomer Film Coated Electrodes as Electrochemical Sensors: Fundamental Investigations. Anal. Chem. 59, 1746-1751 (1987).

16) Kristensen, E.W.; Kuhr, W.G.; Wightman, Characterization of Perfluorinated Ion Microvoltammetric Elec trodes for in Vi vo Use. 1752-1757 (1987).

R.M.: Temporal Exchange Coated

Anal. Chem. 59,

17) EIder, R.C.; Lunte, C.E.; Rahman, A.F.M.M.; Kirchhoff, J.R.; Dewald, H.D.; Heineman, W.R.: In Situ Observation of Copper Redox in a Polymer Modified Electrode Using EXAFS Spectroelectrochem­istry. Electtoanal. Chem. 240, 361-361 (1988).

347

18) Jernigan, J.C.; Chidsey, C.E.D.; Murray, R.V.: Eleetroehemistry of Polymer Films Not Immersed in Solution: Electron Transfer on an Ion Budget. J. Am. Chem. Soe. 107, 2824-2826 (1985).

19) Oliver, B.N.; Egekeze, J.O.; Murray, R.W.: "Solid-State" Voltammetry of a Protein in a Polymer Solvent. J. Am. Chem. Soe. 110, 2321-2322 (1988).

20) Tieman, R.S.; Igo, D.H.; Heineman, W.R.; Johnson, J.; Seguin, R.: Fabrieation and Charaeterization of a Platinum/Ceramie Eleetroehemieal Sensor. Sensors and Aetuators, !!, 5, 121-127 (1991) .

21) Geng, L.; Reed, A.; Kim, M.H.; Woster, T.T.; Oliver, B.N.; Egekeze, J.; Kennedy, R.T.; Jorgenson, J.Y.; Pareher, J.F.; Murray, R.Y.: Chemieal Phenomena in Solid-State Voltammetry in Polymer Solvents. J. Am. Chem. Soe. 111, 1614-1619 (1989).

22) Tieman, R.S.; Heineman, Y.R.; Johnson, J.; Sequin, R.: Oxygen Sensors Based on the Ionieally Conductive Polymer Poly(Dimethyldiallylammonium Chloride). Sensors and Aetuators, !!, 8, 199-204 (1992).

23) Oe Castro, E.S.; Huber, E.Y.; Villarroel, D.; Galiatsatos, C.; Mark, J.E.; Murray, P.T.; Heineman, Y.R.: Electrodes with Polymer Network Films Formed by y-Irradiation Cross-Linking. Anal. Chem. 59, 134-139 (1984).

24) Huber, E.Y.; Heineman, Y.R.: Role of Monomer in y-Irradiated Oimethyldiallylammonium Chloride Modified Eleetrodes. Anal. Chem. 60, 2467-2472 (1988).

25) Igo, D.H.; EIder, R.C.; Heineman, Y.R.: Solid-State EXAFS Speetroelectrochemistry: The Effects of Supporting Electrolyte on the Structure of Cu(bcp-s)2. J. Electroanal. Chem., 314, 45-57 (1991) •

26) Lange's Handbook of Chemistry (Dean, J. A., ed.) (1979) 12, Chapter 10, 84.

27) Ellis, 0.; Eckhoff, M.; Neff, V.D.: Eleetroehromism in the Mixed­Valence Hexaeyanides. 1. Voltammetrie and Speetral Studies of the Oxidation and Reduction of Thin Films of Prussian Blue. J. Phys. Chem. 85, 1225-1231 (1981).

28) Keggan, J.F.; Miles, F .. O.: Struetures and Formulae of the Prussian Blues and Related Compounds. Nature (London) 137, 577-578, (1936).

29) Buser, H.J.; Sehwarzenbach, D.; Petter, Y.; Ludi, A.: -The Crystal Str.ucture of Prussian Blue: Fe4[Fe(CN)613 ·XH20. Inorg. Chem. 16, 2704-2710 (1977).

348

30) Teo, B.-K.: Novel Method for Angle Determination by EXAFS via a New Multiple-Scattering Formalism. J. Am. Chem. Soc. 103, 3990-4001 (1981).

31) Lundgren, C.A.j Murray, R.~.: Observations on the Composition of Prussian Blue Films and Their Electrochemistry. Inorg Chem. 27, 933-939 (1988).

32) Itaya, K.j Ataka, T.j Toshima, S.: Electrochemical Preparation of a Prussian Blue Analogue: Iron-Ruthenium Cyanide. J. Am. Chem. Soc. 104, 3751-3752.

ELECTRODE-ELECTROLYTE INTERFACES INVESTIGATED WITH X­RAY STANDING WAVES: CU(111)/Pb,TI

J. ZEGENHAGENb), G. MATERLIKa), J.P. DIRKsa), M. SCHMÄHa) a) Hamburger Synchrotronstrahlungslabor HASYLAB am Deutschen Elektronen-Synchrotron DESY Notkestr. 85 D-22603 Hamburg Germany

b) Max-Planck-Institut für Festkörperforschung Heisenbergstr. 1 D-70569 Stuttgart Germany

ABSTRACT. We report on the results of the in-situ structural characterization of underpotential deposited Tl and Pb on Cu(ll1) electrodes with x-ray standing waves. These measurements show that submonolayer amounts of Tl and Pb are adsorbed at a distance normal to the Cu(111) surface, which is consistent with a threefold coordinated adsorption site for both, Tl and Pb. On oxidized Cu(111) surfaces, the adsorbate relaxes inward by ab out 0.3 A in both cases. This can be under­stood by oxygen incorporation into the Cu(ll1) surface for which a model is proposed. Oxidation and reduction of the Cu electrodes can be controlled by the electrode potential and were monitored by cyclic voltammetry. The x-ray standing wave technique is explained briefly in particular with respect to its application towards the characterization of electrode-electrolyte interfaces.

1. Introduction

The x-my standing wave (XSW) technique couples x-my interference, typically occurring during dynamical diffraction, and spectroscopic tools for chemical analysis such as x-ray photoelectron spectroscopy (XPS), Auger electron spectroscopy (AES), and energy dis­persive x-ray fiuorescence spectroscopy (EDX). The basic idea is to excite atoms with a spatially modulated interference field which can be scanned continuously across the atomic sites. By recording the direct inelastically sca~tered electrons (XPS) or radiative (EDX) or non-radiative deexitation channels (AES) as a function of the controlled shift of the nodal and anti-nodal planes of the interference pattern, chemical species can be identified qual­itatively and quantitatively, and their spatial site can be determined. This is illustrated for the case of adsorbed atoms in fig. 1 for two different positions of the anti-nodal planes of the wavefield. In one case (left) the adsorbate atoms Al and in the other case (right) the substrate bulk atoms and adsorbate atoms A2 are excited and emit element specific fiuorescence photons.

As early as 1956, Knowles combined neutron interference and inelastic scattering using

349

C. A. Melendres andA. Tadjeddine (eds.), Synchrotron Techniques in Interfacial Electrochemistry 349-369. © 1994 Kluwer Academic Publishers.

350

Figure 1: Scheme of the x-ray standing wave technique. Two different kinds (element Al and A2) of adsorbate atoms are located on a cubic substrate (element B). On the left side of the figure, the maxima of the wavefield intensity- coincide with the location (center) of Al from whieh thus a maxi~um of fluorescence is observed. The positions of A2 and B coincide with the intensity minima of the wavefield and they are not photoexcited. On the right hand side of the figure, the situation is reversed and a maximum of fluorescence intensity is observed from A2 and Band no fluorescence originates from atoms Al.

the wavefield of Bragg reflected neutrons to excite a nuclear reaction.1 In 1964, Batterman2

obtained evidence for t.he formation of an x-ray standing wave within a Ge crystal. He monitored the Ge-K fluorescence as a function of reflection angle within the range of strong Bragg reflection. Since the location ofthe Ge atoms within the crystal are, of course, known, he could determine on the movement of the wavefield. From the fluorescence response he deduced that the wavefield coincides with the Miller planes of the crystal at the low angle side of the range of total reflection whereas at the high angle side the antinodes are located on these planes. How to locate impurity atoms within a crystallattice by monitoring their fluoresence response as a function of reflection angle was demonstrated by Batterman in 19693 and more convincingly in 1974 by Golovchenko et al.4

The first adsorbate was studied with XSW in 1980 by Cowan et al.5 : Br adsorbed on Si(111) from a methanol/Br solution. It was not only the first studied adsorbate system but the measurement was also performed in-situ, Le. on the immersed crystal, under presence of the liquid layer.

In 1984, the first successful application of synchrotron radiation (SR) to XSW studies by Materlik and Zegenhagen6 was reported. Exploiting the properties of SR opened the way to fullydevelop the rieh potential of the XSW method. When compared to x-ray anodes, the main advantages of SR in this field are (i) a tremendous gain in brightness, Le. a high intensity emitted into a narrow angle which is almost comparable to the width of single crystal reflection curves, (ii) a continuous spectrum which offers a free choice of excitation energy, and (iii) linearly polarized radiation. All three properties mentioned

351

above had been instrumental to the success of the present study. Because of high source intensity, measuring times were short enough to prevent electrode poisoning. Furthermore, the strong collimation allows a variety of crystal combinations to be used and the diffraction plane spacing of the monochromator does not have to match the diffraction plane spacing of the sampie crystal (for a discussion see Zegenhagen et al.7 ). Here we used a Ge(220), Si(220) monochromator and Cu(l11) sampie crystals. The importance of (ii) and (iii) is stressed in sections 2.3 and 3.

In 1985 the first step towards the structural characterization of the solid-electrolyte interface with XSW was carried out by Materlik et al.. 8 This study of underpotentially deposited Tl on Cu(ll1) was important in two respects: On the one hand, it was the first reported measurement with XSW using a metal crystal (and metal crystals are of course the standard electrodes in electrochemistry). On the other hand, Tl was deposited from an electrolyte under potentiostatic control with additional characterization of the interface by cyclic voltammetry (CV). The XSW study itself, however, was performed ex-situ, on the emersed electrodes. The first XSW in-situ study on the same system was reported in 1987.9

Compared to the number of analytical tools available for structural characterization of surfaces under ultra-high-vacuum (UHV) conditions, the number of methods applicable to the solid electrolyte interface is small due to the presence ofthe liquid. Practically all particle probes are excluded. Because of the high-penetration power, x-ray methods thus play an important role. We deern it informative to illustrate the features of the XSW technique by comparison with other x-ray techniques like surface x-ray diffraction (SXD) and (surface) extended x-ray adsorption fine structure «S)EXAFS) which are treated in some applications in this proceedings as weIl.

Unique to the XSW technique is that independent structural information can be obtained simultaneously for a large number of elements during one measurement. In contrast to other methods, in particular to SXD, an adsorbate can be very dilute ($ 0.01 ML) and very small areas of the surface can be examined. Unlike SEXAFS measurement, which provides information ab out distances of neighbors of a particular atomic species, we obtain the adsorbate registry with respect to the crystal lattice (or the distance to the surface) of the electrode with XSW. The electrolyte layer on the electrode can be relatively thick in contrast to the requirements for SXD which requires a thin layer because of the grazing incidence. Last but not least, also unique to the XSW technique, we can probe the atomic distribution profile in front of an electrode on length scales ranging from 0.01 A to about 1000A. A good demonstration for probing large length scales was the investigation of the diffuse double layer in front of an electrode by Bedzyk et al.lO

The present paper has two intimately connected goals. Firstly, we want to report results of investigations of the adsorption of Pb and Tl on Cu(ll1) surfaces from an electrolyte. We describe the outcome of cyclic voltammetry but most importantly XSW measurements performed on the working electrodes in-situ, under potentiostatic control. However, sec­ondly, without going into too much detail, we want to give the reader an impression of the features, experimental requirements, and the potential of the XSW technique for interfacial electrochemistry.

In the following we begin with an outline of the principles of the XSW technique (sec. 2). This includes a mathematical description of the interference field, its formation and properties. Furthermore, we present a brief treatment elucidating the quality of structural information obtained with XSW and some peculiarities arising from XSW measurements in

352

Figure 2: Schematic representation of the formation of an x-ray interference field by Bragg refiection from a single crystal Cu(111) electrode under the presence of an electrolyte (see text for further details).

case a liquid layer is covering the surface. Seetion 3 is dedicated to experimental procedures, describing briefly the basic XSW set-up including the thin-film cell and the set-up for cyclic voltammetry. In sec. 4 we report experiments of the underpotential deposition (UPD) of Tl and Pb on CU(111). These are results obtained with CV (sec. 4.1) and XSW (sec. 4.2) . They are discussed in sec. 5. We summarize and conclude in sec. 6.

The experimental work in the case of Cu(111)/Tlll and Cu(111)/Pb12 which we are reporting about here were part of Hamburg University Diploma theses in physics conducted at the Hamburg Synchrotron Radiation Laboratory HASYLAB, Germany.

2. X-ray Standing Wave Technique

2.1 X-RAY STANDING WAVES

A planar x-ray interference field (standing wave) can be generated via the superposition of two coherent, plane waves t:~ and EH, with

t:~.H = EO.HeXp27ri (vt - /CO.Hf) , (1)

traveling in different directions as indicated in fig. 2. We consider them to be linearly polarized and their E-1ield vectors to be colinear. Their propagation vectors with I/Col = I/CHI = )-1 are related by

(2)

For the complex amplitudes (we omit the vector symbols for the E-fields in the following) we write

EH = VREoexp(iv), (3)

353

where v'ii relates the amplitudes and v defines their phase relationship. With

(4)

we obtain for the normalized intensity iw of the whole wavefield

i{t = U* /IEoI2 = 1 + R + 2v'iicos (v - 211'Hr). (5)

The intensity is spatially modulated since H . r is a dimensionless real number. In the direction parallel (normal) to H the modulation is maximal (zero), Le. nodes and antinodes of the wavefield are located on planes. The spacing d. of the planes of equal intensity is determined by H, i.e.

d. = H-1, (6)

with H = IHI. Whether the nodal or antinodal planes are located at a particular position r is determined by the phase v. From fig. 2 we can deduce

sin0 = (~) /ko = >./(2d.) (7)

with ko = Ikol. Thus the smallest possible spacing of the wavefield d.,min depends on >., Le.

d.,min = >'/2 for 0 = 900 , (8)

whereas the largest spacing of the wavefield d.,mo.x approaches infinity (for fixed >'), Le.

d.,mo.x -+ 00 for 0 -+ o. (9)

We can produce the x-ray interference field or standing wave via diffraction. Commonly used is Bragg reflection from single crystals and in this case we can employ quite a variety of H-values, Le. diffraction vectors where typically H-l ;S 5 A. The Cu(l11) reflecivity R, phase v and wavefield intensity for several positions with respect to the Cu lattice and a (111) reflection, Le. H = (111), are shown in fig. 3.

In case of Bragg diffraction from artificial multilayers,13,14 all H -vectors (diffraction vec­tors) are colinear and d. :5 dL where dL is the layer spacing which is typicallyaround 30 A. For reflection from mirror surfaces,15 the H-vectors are oriented normal to the surface and d. :2: >'/(20c) where 0 c is the critical angle for total reflection. For example if 0 c = 10 and >. = 1 A, d. :2: 50 A.

2.2 STRUCTURAL ANALYSIS WITH X-RAY STANDING WAVES

In the dipole approximation (DA) for the photoabsorption process, the radial extent of the initial electronic state (the final state of the electron is usually a continuum state) is neglected. In case of confined core levels (K,L) and hard x-rays (>' '" 1 A) the DA describes the absorption process almost exactly. This means that the strength of absorption and thus also the intensity of the photoelectron emission or subsequent decay processes such as x-ray fluorescence or Auger electron emission is directly proportional to the wavefield intensity at the center TA of the excited atom.

354

(cl <111> ct>~11 Tt

W <101> cn 05 « >- T1(111)8-::I: I-a.. 00

0 d111=2.09 Ä 05

>- 3 1-0(111)--I- 00

Ci.) 2 Z w

r I-Z ~ n 0 0

0 50 ANGLE e-eB (~rad)

Figure 3: a) Calculated refiectivity R and phase v for (111) refiection Ul = (111) from Cu and E-y = 15.3 keV. b) Wavefield intensity iw for certain positions 4>~11 (see sec. 2.2) which is the position along the (111) direction normalized by the wavefield spacing d. = dll1 = 2.09 A. c) Cross-sectional view of the Cu fcc lattice.

The only important scattering signal for the present study is x-ray fiuorescence since electrons cannot penetrate the liquid layer. The normalized fiuorescence yield YJ! from one atom at position TA within the range of the interference field is thus

YJ! = 1 + R + 2VRcos (v - 211"HTA) , (10)

Le. the same expression as eq. (5). The superscript indicates that the yield is characteristic for the interference field generated by H which is a diffraction vector in case of Bragg refiection from a single crystal.

In a typical XSW experiment, the number NA of atoms studied is larger than 1013 cm-2 ,

Le. > 1010. To take this into account, we write the fiuorescence yield as

(11)

with NA

P(r) = NÄ1 L 6(r- Ti), (12) ;=1

where Ti denotes the position ofthe i-th atom. Erquation (11) can be parametrized as

YJ! = 1 + R + 2VRftI cos (v - 211"4>~) . (13)

355

Because of thermal vibrations, the Delta-functions in eq. (12) become Gaussian functions (in the harmonic approximation). It is straightforward to show that f[l and </>!! represent amplitude and phase of the .ii-component of the Fourier decomposition of P(T) .. 16 A function given by eq. (13) is fit ted to the fiuorescence data with the two fitting parameters f[l and </>!! plus a third fitting parameter IFT for overall normalization. IFT is proportional to the adsorbate coverage (see sec. 2.3). The coherent position </>!! can be converted to a distance value zH via

(14)

where dH is the spacing of the wavefield planes or diffraction planes for the H -vectors employed. In the following, dH = dU1 = 2.09Ä for Cu.

It is important to keep in mind that the XSW technique is not a diffraction technique. However, the analogy between Fourier analysis and XSW analysis was already pointed out by Hertel et al. It suggests that we can compare the information gained by XSW with the information gained by e.g. (surface) x-ray diffraction. This comparison will be carried out briefiy in the following.

We can write the x-ray scattering amplitude, which is proportional to the structure factor in case of a periodic object, for a distribution of nT different elements as

nT

A(.ii) = L AI'(H), (15) 1'=1

We express AI'(H) as NI'

AI'(H) = L fl'(H) exp -27riHf, (16) i=1

where p denotes a particular element (NI' atoms) with the form factor fl'(H). The complex scattering amplitude AI'(H) of the p-th sublattice can also be written as

(17)

The coherent position </>!!'I' for a particular element p which we determine with XSW can be expressed as

(18)

The coherent fraction f!!# for the p-th element is propertional to the magnitude of AI'(H), i.e.,

H -1 -1 - -fe,1' = NI' fl' (H)IAI'(H)I· (19)

Thus, the more individual elements contribute to an adsorbate surface structure the higher is the possible gain in information by XSW compared e.g. with diffraction exper­iments. In case of single crystals, H corresponds to strong substrate refiections and thus to H-values which are usually omitted in case of surface x-ray diffraction. For a detailed description of the XSW analysis see ZegenhagenP

2.3 XSW ANALYSIS FOR A SOLID-ELECTROLYTE INTERFACE

There are two aspects which have to be taken into account if the adsorbate under study is

356

not at a solid-vacuum but at a solid-liquid interface. (i) If atoms of the same kind which is adsorbed at the interface are present in the liquid phase, these will be excited by the penetrating x-ray beam and contribute to the fiuorescence yield. (ii) If the sheet of liquid is scattering strongly, this scattered radiation could excite the atoms under study. The experimental conditions should be chosen that (i) and (ii) can be neglected. If not, the interpretation of I! and 4// in terms of adsorbate structure could be wrong. Or even worse, the fiuorescence yield could not be expressed by eq. (13) at all. We first illustrate this for the case of (i).

In case we study underpotential deposition of ions/atoms A from an electrolyte in-situ, a certain number of atoms N AS will be part of the interface whether they are specifically adsorbed or are part of the diffuse double layer. However, a certain density PA(r') of ions/atoms will remain in the bulk part of solution. In the following we will assume PA( r') = const. The situation of an interference field created at a solid-liquid interface is depicted in fig. 2. We denote the fiuoresence intensity from atoms excited by the incoming beam as In, the fiuoresence excited by the (coherent) interference field as IFC and the fiuoresence excited by the refiected, exiting beam as IFE. Thus, the total fiuoresence intensity is

(20)

With the beam crossection Q the number of atoms per cm in solution is

(21)

With the linear coeflicient of absorption of the electrolyte J' and the liquid-Iayer thickness d we can express IFI as

(22)

for an electrolyte of sizeable J'd. If J'd -+ 0,

(23)

where I is a proportionality constant taking into account absorption cross-section, fiuores­cence prob ability etc. For the scattering from the wavefield we obtain

(24)

where i!t is given by eq. (5). We introduce the expressions

(25)

or (26)

for J'd -+ 0 and a = exp (-J'd(sin 0)-1) . (27)

Thus, we obtain IFE = IRaNs. (28)

357

The fiuorescence yield can then be expressed as

H Ns+ NAS NAS r;:; H YF = IFT/ (fNT) = 1 + Ra NT + a NT 2v Rfc cos<p. (29)

In eq. (29) we used (30)

and <p = v - 21r4>~. (31)

To understand eq. (29) better, we discuss two limiting cases. For a ~ 1 (pd ~ 0) we obtain

YJ! = 1 + R + ~:2VRf[l cos<p. (32)

While the true coherent fraction from the ions/atoms at the interface is f!f, the experi­mentally determined fraction is reduced by the factor NAS/NT which is the smaller, the higher the concentration of ions/atoms in the bulk of the solution. The other limiting case is a strongly absorbing electrolyte. We take as an example a = ! and Ns ~ NAS. Thus we obtain

2 1 r;:; YJ! = 1 + "3R + "32v RftI cos<p. (33)

In this case not only the measured coherent fraction is reduced, but the whole functional form of the fiuorescence yield as given by eq. (13) is changed due to the prefactor of the refiectivity R. In order not to further increase the length of the present discussion we neglected in the above mathematical description the absorption of the outgoing fiuorescence in the liquid which enhances the effect expressed in eq. (33).

To avoid complications for XSW in-situ studies it is advisable to keep the layer of elec­trolyte thin enough that it is only weakly absorbing and furthermore to keep the ion/atom concentration of the species A of interest in the bulk part of the solution low such that eq. (13) or at least eq. (32) still holds.

Let us now turn to the second point (ii). In fig. 4 part of a photon spectrum from a Cu surface covered with electrolyte layers of different thicknesses recorded with a Si(Li) solid state detector is shown. The TDS and in particular the Compton peak becomes very prominent with increasing thickness of the electrolyte layer. If the energy of this scattered radiation ia above the absorption edge of the ions/atoms of interest, e.g. Pb in the case shown, it can contribute to the excitation of the Pb fiuorescence and distort the functional form of the fiuorescence yield. There are two ways of rendering this problem negligible: firstly, tune the excitation energy directly above the absorption edge of the atom of interest which, of course, requires SR. In this case, Compton scattering is too low in energy to excite fiuorescence. Secondly, chose the liquid layer reasonably thin in order to minimize the Compton scattering.

3. Experimental

The experimental set-up for XSW measurements is shown in fig. 5a with the in-situ electro­chemical cell in fig. 5b. It was discussed in detail by Zegenhagen et al.7 Sychrotron radiation

358

Cf) I­Z :::l o c..>

1

Cu-Ka (a)

A-K~ Comptonline

" /OS

Cu-Ka (b)

Cu-Kp comPton~TOS 1\

comPtonline-"~ (c) TOS

Cu-Ka

Compton-

ShOUldej Cu-Kp \ A Pb-La1

8 10 12

ENERGY (keV)

Figure 4: Sections of spectra recorded from 30 Cu(111) electrode with 30 Si(Li) solid state detector. Excitation energy E-r = 13.1 keV, 0B = 13.10 • The Cu-K fluorescence was suppressed by 30 0.1 mm thick Al absorber in front of the Si(Li) detector. 30) no electrolyte in front ofthe surface, b) 50 J-Lm thick electrolyte, c) 175 J-Lm thick electrolyte with Pb ions in solution.

from 30 bending magnet of the DORIS storage-ring at DESY was used_ The experiments were carried out at the ROEMO station of the Hamburg Synchrotron Radiation Labora­tory HASYLAB. The monochromator selected energies were optimized for the adsorbates (Pb and Tl) under study, Le. right above the Pb (Tl) L3 absorption edge. It generated 30

'pseudo' plane wave, linearly polarized in the reflection plane. A Si(Li) solid state detector monitored the fluorescence yield and 30 Na.! (Tl) detector recorded the reflectivity from the Cu substrate crystal. Compton scattering in the spectra ofthe Si(Li) detector was reduced additionally by making proper use of the horizontal polarization of the SR. The Si(Li) was located sideways in which direction the Compton and elastic scattering, exhibiting dipole­like character, is minimal. The x-ray excitation energies for the Cu(111 )/Tl (E-r = 12.8 keV) and Cu(111)/Pb (E-r = 13.1 keV) were chosen just above the Tl (EL3 = 12.66 keV) and Pb (EL3 = 13.04 keV) L3 absorption edges such that Compton scattered radiation could not excite Tl or Pb fluorescence radiation. A number of design consideration were met for the in-situ cell to make the XSW studies feasible.

SR « ~J) ""

'7 Me

(a)

(b)

Im I surtoce I!\ nun ,....-~--

I

s C,,-ELECTROOE

M lorwindow

r?~~~'"'"1C;~ Crystat

to~ , to~ft'l'l'nce t>it>ctrode

Z ter eteclrodE' -~lectro!yte

In

.... ,,-,-/Older forCu --.... - Electrical contoct

359

Figure 5: Experimental arrangement. (a) Schematic representation of the XSW set-up at the ROEMO station at HASYLAB.6 The double·crystal monochromator (MC) with a Ge(220) first crystal and an asymmetrically cut Si(220) as a second crystal is located 35 m from the source point and generates a plane wave Hke beam ('pseudo' plane wave) of energy E'Y' It is confined by slit Sand reflected by the sampie crystal, 2 m apart from the monochromator. NaI and Si(Li) detector monitor the Bragg reflected and scattered photons. (b) The in-situ cell. The electrolyte is confined to the face of the Cu(lll) electrode by teflon lips, almost touching the crystal and a 4 /Lm thin mylar window_ The thickness of the electrolyte layer was adjustable by the action of a peristaltic pump or by supplying slight press ure via the valve at the bottom of the cello

360

a) The sheet of electrolyte above the crystal surface was thin (~ 50 /Lm) and its thickness could be varied (cf. fig. 5b).

b) It was possible to reduce the ion/atom concentration in the bulk solution to an ac­ceptably low value.

c) Only the polished18 (111) face of the cylidrical, Czochralski grown19 Cu crystals were exposed to the electrolyte to allow a straightforward interpretation of the voltammo­grams.

d) Cu easily oxidized in aqueous solution in the presence of dissolved oxygen. 8,9 To prevent this, the oxygen content in the electrolyte was controllable and could be reduced to a negligible level.

e) The geometrical construction of the cell allowed movement of the Si(Li) solid state detector as close as possible to the Cu surface in order to accept photons from the surface in a large solid angle.

A cell design which meets an these requirements is shown in fig. 5b. To reduce the amount of dissolved oxygen (d), the whole cell including electrolyte supply bottles, reference and counter electrodes was placed in a box filled with an appropriate gas (mostly N2).

The scan of the (Bragg-) reflectivity profile of the sampie crystal for an XSW mea­surement can be performed by scanning the sampie in angle 0 or the monochromator in energy,20 for a discussion see e.g. Zegenhagen et al .. 1 In the present case it was crucial to keep the sampie at rest and do it the latter way.

Cyclic voltammetric measurements were performed with a three-electrode system under the control of a standard solid-state potentiostat.

4. Results

4.1 CYCLIC VOLTAMMETRY

4.1.1. Tl on Ou(111). For the deposition of Tl on Cu(111) a 0.5 M Na2S04 supporting electrolyte was used. The chemical grade was suprapure and triply distilled, respectively. Before attempting deposition of Tl, the native oxygen layer of the Cu( 111) electrode had to be removed. This was usually performed by cycling the electrode in carefully deoxygenated (by bubbling purified nitrogen) electrolyte solution. The oxygen removal could be traced by the development of the voltammogram with time as it is shown in fig. 6. After anodic and cathodic charge were balanced, indicative of a clean Cu(l11) surface, deoxygenated ThS04 (puratronic) was added. It should be noted already here that oxygen removal from the Cu-surface could not be achieved when ThS04 was immediately added to the solution since adsorbed Tl obviously is able to inhibit Cu-oxide reduction. For deposition of Tl, the Tl2S04 concentration chosen was 0.5 X 10-3 M. For an in-situ study and electrolyte films above the surface of ~50 /Lm this is a much too high concentration. Thus, after Tl depostion at a particular potential and prior to an XSW measurement, the electrolyte in front of the electrode was exchanged with pure supporting electrolyte while preserving potentiostatic control. Due to this 'rinse' the bulk stripping peak was significantly decreased while the underpotential, monolayer desorption peak remained unchanged as it is shown in fig. 7. Analysis of the voltammograros yielded -0.54 V and -0.46 V for the Tl UPD adsorption

-« 2; 10 ~ 0 Z -10 w a: a: :::> ü -50

0.5 M Na2S04 50 mVs-1

---------.. .... _ .... "":":"".7:-:7".~-: ...................... . C:-----

-1.0 -0.8 -0.6 -0.4

POTENTIAL vs SCE (V)

361

Figure 6: Voltammograms demonstrating the process of cleaning the eu(111) electrodes from oxic1e layers in the in-situ cell by cycling in pure, deoxygenated supporting elf'drolyt.e.

-« ::::l. -~ Z w a: a: :::> ü

50

10 0

-10

\/

STRIPPING BEFORE_fI RINSE / \

r' I/I I AFTER I \/RINSE / /\

I \ I \

;/ '" i"'-_ .... -:-:"': ................................... .

-1.0 -0.8 -0.6 -0.4

POTENTIAL vs SCE (V)

Figure 7: Voltammograms recorded in the in-situ cello Dashed curve with 0.5 mM Tl2S04 in solution. Solid curve after exchanging the electrolyte in the in-situ cell with clean supporting electrolyte sub se quent to bulk deposition while maintaining potential control.

362

and desorption peaks, respectively if the electrolyte was carefully deoxygenated. With in­creasing oxygen concentration in solution the adsorption peak at -0.54 V developed an increasing shoulder at -0.59 V. Similarly, the UPD desorption peak developed a shoulder at around -0.70 V to -0.77 V. The initial peaks, indicative for a oxide free surface, never vanished completely indicating that Tl adsorption occurs on both oxidized and oxide free Cu surface areas.

4.1.2. Pb on Cu (111). For the UPD of Pb on Cu(H1) we chose Pb(CI04 h and a 0.5 M NaCI04 supporting electrolyte. In this environment the reduction of the oxidized Cu(111) surface turned out to be difficult. The reduction process could strongly be enhanced by decreasing the pR value of the solution via addition of RCI04 • This is demonstrated in fig. 8. With decreasing pR, the peak marking the reduction of the Cu surface moves to

-~~ t--c pH::::4.5 (j -<{ • ;;:=z~~ :::l.

_~~I:-L pH:52.5 - -~ =-z t pH=1.8 (c) w -----

-10 a: t a:

~ 0.5 M NaCI04 Ü -50 + HCI04

-1.0 -0.8 -0.6 -0.4 -0.2 0 POTENTIAL vs SeE (V)

Figure 8: Oxidation and reduction (arrow) of Cu(111) as a function of hydrogen ion con­centration in solution. The onset of hydrogen development is also indicated.

an increasingly positive potential. Just as in case of Cu(H1)jTI, the oxide reduction was inhibited if a relatively high concentration of the metal ions (Pb) were present in the solu­tion. Different from the case Cu(111)jTI, continuous cycling of the Cu(lH) surface in the Pb met al-ion containing, deoxygenated solution lead to the reoxidation of the previously reduced surface. This is shown in fig. 9. While the Pb stripping peak for the mostly oxide free Cu(Hl) surface appears at -0.18 V, a shoulder at -0.25 V indicative for Pb adsorp­tion on oxidized Cu(H1) shows up with extended cycling time. A typical voltammogram, obtained prior to XSW measurements, is shown in fig. 10.

363

40

- 0 « :::l. -.... -40 Z w

-0.25 ....... 0.5 M NaCI04 ce (c) ce 40 5 x 10-5 M Pb (C104)2 ::J C,) 1 n HC104

0 pH=2.1

-40

-0.6 -0.4 -0.2 -0.6 -0.4 -0.2

POTENTIAL vs seE (V)

Figure 9: Reoxidation of Cu(l11) as a function of cycling time. (a) The cleaned electrode shows a Pb stripping peak at -0.18 eV. With time a shoulder at -0.25 V characteristic for an oxidized (111) surface, becomes the dominant peak.

4.2 XSW RESULTS

4.2.1. Tl on Cu (111). The results of two XSW measurements are shown in fig. 11. The measurements were performed with the potential held at -1.00 V and -0.7 V vs SCE for curve 1 and 2, respectively. Hut it is more important that in case of curve 1, the electrolyte had been deoxygenated by N2 bubbling as carefully as possible whereasin case of curve 2 oxygen was still in solution. For a total of 12 XSW measurements carried out for electrode potentialsbetween -0.35 V and -1.00 V vs SCE no dependency of 4P on the potential could be found. However, different 4P values were observed depending on the oxygen concentration in solution. As limitingvalues, 4P = 2.64A and 4P = 2.28 A could be distinguished for low and high oxygen concentration in solution, respectively. The maximum observed coherent fractions were [111 = 0.92 ± 0.15 and fJl1 = 0.35 ± 0.07 in the former and in the latter case, respectively. Tl coverages never exceeded 0.4 ML. Surprisingly, almost 1/4 ML Tl was found to be adsorbed on the surface at -0.40 V which is 140 mV positive with respect to the UPD adsorption peak (-0.54 V) and 60 mV positive with respect to the UPD desorption peak ( -0.46 V), Le. a potential were no charge transfer can be observed in the voltammograms. The measured position was the same as the one observed at lower potential values. Changing the potential to -0.35 V eventually led to the expected result that no Tl could be observed on the Cu(l11) surface. Generally, the Tl coverage determined from the fluorescence intensity1 were larger than the values determined from the charge transfer (Tl stripping peak) in the voltammograms.

364

-~ z w a:

40

o

a: -40 ::> Ü

-80

0.5 M NaCI04 5 x 10-5 M Pb (CI04)2 pH=1.8

-0.8 -0.6 -0.4 -0.2

POTENTIAL vs SeE (V)

Figure 10: Typical voltammogram obtained prior to XSW measurement for Pb on Cu(lll).

4.2.2. Pb on Cu(111). A total of 16 XSW measurements were performed for electrode potentials ranging from -0.31 V to -0.70 V where the UPD adsorption and desorption peak was observed at -0.33 V and -0.24 V, respectively (compare fig. 10). The results of two XSW measurements are shown in fig. 12. Curve 2 was recorded for an electrode potential of -0.33 V yielding z~t1 = 2.28 A ± 0.05 A and curve 3 for -0.58 V yielding z~t1 = 2.59 A ± 0.04 1. This dependency of Pb position on the electrode potential was characteristic for all XSW measurements. For those performed elose to the UPD adsorption peak at around -0.33 V (A in fig. 10) an average position of (zW) = (2.32 ± 0.07) A = Z1

was observed. For measurements performed at an electrode potential < -0.58 V (B in fig. 10), an average position of (4\1) = (2.54 ± 0.03) A = Z2 was obtained. In the intermediate potential range, the observed position varied between these extreme cases. The coherent fractions determined were smaller than 0.6. The coverage as determined from the Pb fiuorescence, which ineludes Pb stillieft in the solution (see sec. 2.3), ranged from 0.3 ML to 1.2 ML whereas the coverage determined from the charge transfer from the voltammogram never exceeded 0.5 ML. As in the case Cu(lll)/TI the coverages calculated from the Pb fiuorescence were always larger than the values deduced from the charge transfer (Pb stripping peak) in the voltammograms.

365

> z111 =2.23 A f~11=0.19 I- 2 ..... -r.*-.~ . CI) ..J........ '-. Z . . ..... W T·.i-f I-

@ Z 0 Cl

f~ll =0.35 W 3 z111 =2.64 A N ....J <C 2 ~ a: (111) REFL. 0

"x"x--x--x- -x--x. Z "X -x-,x_.

0 .- ... - "x X--x_.

0 50 100

ANGLE I S-Ss (~rad)

Figure H: Two results of XSW measurements. E-y = 12.8 keV at ab out 30 mA DORIS e--beam current, 0.4 ML Tl. About 10 mm2 ofthe Cu surface were exposed to the x-ray beam. Shown are Cu(H1) reflectivity and two Tl La fluorescence yield curves obtained with no (1) and some oxygen (2) in solution.

5. DiscussioD

Two different positions normal to the Cu(lH) surface could be identified both for Tl on Cu(H1) and Pb on Cu(H1). In case of Tl, the observed positions could be associated with Tl adsorption on an oxidized Cu(H1) surface (zi.P = 2.28 A) and mostly oxygen free surface «(z}F) = 2.58 ± 0.05 A). We discussed this already in an earlier publication.9 The same kind of strikingly different Tl positions were also found in an ex-situ study if the Tl adsorbate was similarly prepared.7,8 The Tl position determined for the (mostly) oxygen free surface can be understood by Tl adsorbing in a threefold (zlll) coordinated geometry. Assuming an unrelaxed Cu(H1) surface and bondlengths determined by the Cu and Tl covalent radii which are 1.28 A and 1.73 A, respectively, ZJll = 2.62 A for the threefold site, z~ll = 2.72 A for the twofold site, whereas a onefold, atop site would lead to zlll = 3.01 A.

Assuming again bondlengths determined by covalent radii (TPb = 1.74 A) we can calculate for Pb on eu(H1) zlll = 3.02 A, z~l1 = 2.74 A, and zlll = 2.63 A. Thus, the position measured for deposition elose to the Nernst potential (;S -0.58 V), z~},l = 2.54 ± 0.03 A, is explainable with Pb in a threefold adsorption site, possibly with a small contraction in bondlength. The other limiting position observed for potentials at the UPD potential (~ -0.33 V) is too small to be explained by the above calculated values. It can, however,

366

>­~ Cf) z w ~ z Cl W N ...J « ~ a: o z

3

2

o~----~----~----~----~ o 30

ANGLE 60 90 120

, a - as (~rad)

Figure 12: Two results of XSW measurements. E-y = 13.1 keV at about 50 mA stored DORIS e--beam current. Shown are Cu(111) refiection curve (1) and two Pb La fiuores­cence yield curves obtained for two different electrode potentials. For furt her details see text.

be understood by assuming Pb adsorption on an oxidized Cu(I11) surface. Obviously, oxidation of the Cu(111) surface can happen with a NaC104/Pb(C104h electrolyte even for a well deoxygenated solution. The peak labeled 3 in fig. 10 is associated with reduction of the oxide layer. The onset of the reduction (2 in fig. 10) occurs at '" -0.55 V. Thus, at potentials< -0.55 V, large parts of the surface may be oxygen free whereas there seems to be no way of preventing oxidation at potentials> -0.55 V. This is in contrast to the observation for the system NaS04/TIS04. A model for adsorption of Pb and Tl on an oxidized Cu(I11) surface is shown in fig. 13. Because the incorporated oxygen (0) pushes the Cu surface atoms outward, the metal atoms can relax downward further, leading to the observed small zll1-values.

In the case of Cu(I11)/Pb, the fact that the x-ray fiuorescence analysis always revealed higher coverages than is calculated from the charge transfer in the voltammograms could be explained by a certain amount of Pb in solution. However, the same observation was made in the case of Cu(111 )/TI where the Tl in solution was rinsed away via exchange with pure supporting electrolyte. Thus, these findings maybe related to the fact, that Tl was found to be specifically adsorbed at a potential (-0.40 eV) where no charge transfer was visible in the voltammograms.

367

6. Summaryand Conclusion

These experimental studies c1early demonstrate the potential of x-ray standing wave mea­surements using synchrotron radiation for in-situ structural studies of electrochemical in­terfaces. For the case of Tl and Pb adsorption from different electrolytes, in-situ position determination was carried out for the adsorbate coordinate normal to the (111) Cu surface. The influence of oxygen for this highly reactive system was c1early revealed a.s wen as the influence of the H+ concentration in solution, Le. the pH-value, on the oxidation which obviously also strongly influences the surface structure. To determine three coordinates of adsorption sites unambigously, measurements with diffraction planes oblique to the surface are required. However, the present measurements already exc1ude onefold adsorption sites of Pb and Tl and favor a threefold coordination. To avoid the influence of oxygen, stud­ies on Au electrodes are under way.21 The average measuring time for one XSW scan of about 45 minutes for the present study can be reduced by using the high brightness of SR from wigglers or espeeially undulators. Thus, systematical studies of potential dependen­eies of the adsorbate structure will be possible, small spots from less perfect crystals may be used and because of the high angular collimation, many ii vectors, Le. strong substrate reflections, may be employed without the necessity of changing monochromator crystals.

(a) /-, I TI \ \ J

(b)

t<111>

._.jAd -C--1 Ll R .d111

@<111>

Figure 13: Model for the adsorption geometry of Tl (the same holds for Pb) on oxidized Cu(l11) in (a) side view and (b) top view.

368

References

1. Knowles, J.W. (1956) 'Anomalous absorption of slow neutrons and x-rays in nearly perfeet single crystals', Acta Cryst. 9, 61 - 69.

2. Batterman, B.W. (1964) 'Effect of dynamical diffraction in x-ray fluorescence scatter­ing', Phys. Rev. 133, A759 - A764.

3. Batterman, B.W. (1969) 'Detection of foreign atom sites by their x-ray fluorescence scattering', Phys. Rev. Lett. 22, 703 - 705.

4. Golovchenko, J.A., Batterman, B.W., and Brown W.L. (1974) 'Observation of inter­nal x-ray wave fields during Bragg diffraction with an application to impurity lattice location', Phys. Rev. BIO, 4239 - 4243.

5. Cowan, P.L., Golovchenko, J.A., and Robbins, M.F. (1980) 'X-ray standing waves at crystal surfaces', Phys. Rev. Lett. 44, 1680 - 1683.

6. Materlik, G. and Zegenhagen J. (1984) 'X-ray standing wave analysis with synchrotron radiation applied for surface and bulk systems', Phys. Lett. 104A, 47 -50.

7. Zegenhagen, J., Materlik, G., and Uelhoff, W. (1990) 'X-ray standing wave analysis of highly perfeet Cu crystals and electrodeposited submonolayers of Cd and Tl on Cu surfaces', J. of X-ray Science and Technology 2, 214 - 239.

8. Materlik, G., Zegenhagen, J., and Uelhoff, W. (1985) 'X-ray standing-wave fluorescence analysis of electrodeposited Tl on clean and oxygen-reconstructed Cu( 111)', Phys. Rev. B 32,5502 - 5505.

9. Materlik, G., Schmäh, M., Zegenhagen, J., and Uelhoff, W. (1987) 'Structure deter­mination of adsorbates on single crystal electrodes with x-ray standing waves', Ber. Bunsenges. Phys. Chem. 91, 292 - 296.

10. Bedzyk, M.J., Bommarito, G.M., Caffrey, M., and Penner, T.L. (1990) 'Diffuse-double layer at a membrane-aqueous interface measured with x-ray standing waves', Science 248,52 - 56.

11. Schmäh, M. (1985) 'In situ Strukturuntersuchungen von elektrochemisch abgeschi~­denen Thallium-Adsorbaten auf einer Kupfer-(111)-Oberfläche mit Röntgeninterferenz­feldern, DESY, Internal Report HASYLAB 85-11, unpublished.

12. Dirks, J.-P. (1988), 'In-situ Strukturuntersuchung von elektrochemisch abgeschie­denen Blei-Adsorbaten auf einer Kupfer-(111 )-Oberfläche mit stehenden Röntgen­wellenfeldern', DESY, Internal Report HASYLAB 88-9, unpublished.

13. Barbee, T.W., jr. and Warburton, W.K. (1984) 'X-ray evanescent- and standing-wave fluorescence studies using a layered synthetic microstructure', Materials Letters 3, 17 - 23.

14. Bedzyk, M.J., Bilderback, D., White, J., Abruiia H.D., and Bommarito, G.M. (1986) 'Probing electrochemical interfaces with x-ray standing waves', J. of Phys. Chem. 90, 4926 - 4928.

369

15. Bedzyk, M.J., Bommarito, G.M., and Schildkraut, J.S. (1989) 'X-ray standing waves at a reflecting mirror surface', Phys. Rev. Lett. 62, 1376 - 1379.

16. Hertel, N., Materlik, G., and Zegenhagen, J. (1985) 'X-ray standing wave analysis of bismuth implanted in Si(llO)', Z. Phys. B - Condensed Matter 58, 199 - 204.

17. Zegenhagen, J. (1993) 'Surface structure determination with x-ray standing waves', Surf. Sei. Reports, in press.

18. Fehmer H. and Uelhoff W. (1969) 'A machine for simultaneous electrolytical polishing and flattening with rotating cathode', Journal of Seientific Instruments 2, 767 - 770; Fehmer H. and UelhoffW. (1969) 'On the strainfree preparation of copper single crystal sampies with low dislocation densities', Journal of Seientific Instruments 2, 771- 775.

19. Fehmer H. and Uelhoff W. (1972) 'Die Züchtung versetzungsfreier Kupfereinkristalle' , Journal of Crystal Growth 13/14, 257 - 261.

20. Funke, P. and Materlik, G. (1985) 'X-ray standing wave fluorescence measurements in ultra-high vacuum: adsorption of Br on Si(111)-(lxl)', Solid State Commun. 54, 921 - 923.

21. Bommarito, G.M., Acevedo, D., Rodriguez, J.F., Abruiia, H.D., Gog, T., Materlik, G. (1993) 'X-ray wave studies of underpotentially deposited metal monolayers', these proceedings.

X-ray Standing Wave Studies of Underpotentially Deposited Metal Monolayers

G. M. Bommarito, D. Acevedo, J. F. Rodriguez and H. D. Abrufia* Department of Chemistry Baker Laboratory, Comell University Ithaca, New York 14853-1301

and T. Gog and G. Materlik HASYLAB at DESY Notkestraße 85 2000 Hamburg 52 Germany

ABS1RACf The x-ray standing wave technique has been employed in the study of the structure of underpotentially deposited (UPD) copper on an iodine covered platinum surface and of copper on a Au(lOO) single crystal e1ectrode. For Cu UPD on Pt, surface coverage isotherms derived from both electrochemical and x-ray measurements are compared. The growth mode of the copper ad-layer appears to be strongly influenced by the electrode's surface morphology. For Cu UPD on Au(lOO) the coherence of the adlayer is strongly dependent on the mode of deposition.

1. Introduction The process of underpotential deposition (UPD) of metals has been extensively

studied during the past two decades due to its theoretical and practical importance.[1] In this process, submonolayer to monolayer amounts of a metal can be electrodeposited on a foreign metal substrate in a quantifiable and reproducible fashion prior to bulk deposition. Numerous eleetroehemical and speetroscopie techniques have been utilized to probe the mechanism(s) of formation, and the struetural properties of UPD layers. Conventional eleetrochemical methods have been used to obtain thermodynamie and kinetic information about the UPD process [1-3]. Although electrochemical methods are invaluable in controlling and measuring thermodynamic parameters, structural inferenees are always indireet and often model dependent.

Surfaee sensitive ultra high vacuum techniques have been employed in the study of such systems and much information has been obtained from them [4]. However, the fact that these studies are inherently ex-situ raises some fundamental questions as to their applieability.

In recent years, the use of atornie resolution mieroseopie teehniques has provided the means to obtain in-situ direct atomic structural information from UPD systems. Scanning tunneling and atomic force microscopy have been recently employed

371

C. A. Melendres andA. Tadjeddine (eds.), Synchrotron Techniques in Inteifacial Electrochemistry 371-385. © 1994 Kluwer Academic Publishers.

372

in the study of UPD processes [5]. These studies have shown that, in general, the UPD process occurs in a well-defined manner and that the structures observed from these experiments are similar to those observed in vacuum. As was the case in the ex-situ experiments, these techniques provide information only for the deposited layer.

Recently, in situ x-ray spectroscopic and diffraction techniques have provided unique atomic resolution structural information about UPD systems [6]. Extended x-ray absorption fine structure (EXAFS) and x-ray absorption near edge structure (XANES) have been widely used to study various UPD systems [7], providing information about the local structure atomic environment and the oxidation state of the adsorbed species. Furthermore, surface x-ray scattering measurements have been used to study the in­plane structure ofvarious UPD systems [8]. X-ray standing waves (XSW) [9-12] have also been utilized to probe the structure of UPD layers and were the first x-ray experiments to demonstrate the applicability of these techniques in-situ and ex-situ [9]. In addition, this technique allows one to obtain information pertaining to the distribution of species, including the diffuse layer, in a direction normal to the substrate's surface.

In this paper, we present the results of aseries of x-ray standing wave experiments aimed at probing the potential dependent structural nature of the underpotential deposition of copper on an iodine covered platinum surface and of copper underpotentially deposited on a Au(lOO) single crystal electrode.

2. Theoretical Background X-ray standing waves are generated when coherently related incident and

reflected plane waves interfere.[13] The standing wave electric field intensity is given by:

(1)

where eo an d ~ are the incident and reflected plane waves if their respective

wavevectors ko and kR lie in the x-z plane with the z axis normal to the substrate's surface. Q = ko - kR is the momentum transfer with a magnitude given by:

IQ I = Q = 2 sin 80" = 1/D, (2)

-" At the Bragg angle (8B), the scattering vector is a reciprocal lattice vector H with

IHI=l/dH where dH is the substrate's characteristic d-spacing ofreflection. The angular

dependence of equation (1) is contained within the variables R(8) anct-v(8) which correspond, respectively, to the intensity and the phase of the reflected wave relative to the incident one. During specular reflection (total extern al reflection) [14] and Bragg diffraction, a strong and well-defined standing wave field is generated. In addition, as the angle of incidence 8 is scanned across these reflection regimes, there is a change of

1t in the relative phase v(8) , causing the nodal and antinodal planes of the standing wave field to move inward in a direction normal to the substrate's surface (we confine our discussion here to the case where the diffraction planes of the substrate are parallel to the surface). Since the photoelectric effect for core electrons is directly proportional, in the dipole approximation, to the electric field intensity at the center of an atom, the emission yield (i.e. the fluorescence yield) from the atoms in an overlayer or in a distribution of species above the substrate's surface will be uniquely modulated as a

373

function of S. To ca1culate this yield, the standing wave electric field intensity I(z,S) must be integrated over the entire distribution N(z):

Y(z,S) = jI(Z,S) f(z) dz (3)

Conventionally, XSWs are generated by dynamical Bragg diffraction from perfect single crystals [13,15]. In this work, we are interested in studying structural changes not only for an atomic overlayer but also for extended distributions of species (on the order of tens to hundreds of A). Thus, depending on the length scale of interest, we have employed two different substrates. In the study of Cu UPD on an iodine covered Pt surface we employed PtfC layered synthetic microstructures (LSM) with characteristic d-spacings of the order of 40A as both the electrode and the diffracting structure. LSMs are artificial, depth-periodic structures [16], prepared by depositing altemating layers of high and low electron density elements, thus creating a superlattice structure with diffraction planes centered in the high electron density layers. The XSW technique using LSMs has been applied in several studies [10-12], and we refer the reader to these references for further details.

In the ca se of copper UPD on gold, we employed a single crystal Au(100) grown from the melt and prepared so as to have a very low density of dislocations. [17].

3. Experimental Experiments of Cu UPD on an iodine treated Pt surface were carried out at the

Comell High Energy Synchrotron Source (CHESS) using the B2 beam line employing a double-crystal Si(111) monochromator.

The electrochemical cell, housed inside an aluminum holder, consisted of a cylindrical Teflon body with feedthroughs for electrolytes and electrode connections. The cell was thoroughly cleaned prior to use. The filling and rinsing of the cell with electrolyte was accomplished with pressurized glass vessels through the fluid feedthroughs. A thin layer of solution (approx. 1-3 mm thick) was trapped between the electrode, and a 6.35 mm thick polypropylene film which was held in place by a Teflon ring. All the electrochernical measurements were conducted with the polypropylene film distended by the addition of excess bulk electrolyte into the cello The thin layer was then restored by removing excess electrolyte. Potential control of the electrode was retained through filling and rinsing stages. All applied potentials are reported with respect to a AgfAgCl reference electrode.

Platinumlcarbon LSMs of dimensions 15 mm by 20 mm were obtained from Ovonic Synthetic Materials Co. (Troy, MI). The LSMs used had d-spacings of 39.7 A or 41.4 A, and consisted of 200 layer pairs of platinum and carbon with platinum as the outermost layer, deposited on a 0.015 in. thick Si(lll) substrate.

Solutions were prepared with ultrapure reagents (Aldrich, Baker, Alfa) and pyrolytically distilled water (PDW). Prior to use, solutions were degassed for over 30 min. with high purity nitrogen which was passed through hydrocarbon and oxygen traps. The electrolyte was 0.10 M sulfuric acid (Baker Ultrex) containing lx10-4 M copper sulfate (Aldrich Gold Label) and was prepared using pyrolytically distilled water.

The Pt/C LSM was cleaned by aseries of oxidation-reduction cycles (at 20mV/sec) in pure supporting electrolyte (O.IM sulfuric acid) followed by formation of the iodine ad-Iayer which was formed by contacting the electrode with a ImM solution of NaI in O.IM sulfuric acid for 15 min. Afterwards, the electrode was rinsed with supporting electrolyte. Prior to copper deposition, electrolyte solution was added to the cell so that the polypropylene film distented somewhat, thus allowing the UPD layer to

374

be deposited from bulk: electrolyte. The monolayer was deposited from bulk: electrolyte because of the low copper concentration. Deposition was carried out at constant potential for 15 min. after which the current had decayed to background levels. Deposition potentials of +0.45, +0.25, +0.20, +0.15 and +0.10 vs Ag/AgCI were employed and these corresponded to copper coverages of 0, 1/4, 1/2, 3/4 and a fuH monolayer, respectively. After deposition, part of the electrolyte solution was withdrawn, leaving only a thin layer of electrolyte, whose thickness we estimate from reflectivity measurements to be of the order of 5 microns, between the electrode and the polypropylene film. The amount of copper ions contained within the thin layer represents about 2-5% of the amount electrodeposited on the surface. As a result, no interference from copper in solution was anticipated.

For each XSW scan, an energy-dispersed fluorescence spectrum at a given angular position was recorded into 256 channels of a LeCroy histogramming memory module. A typical sean eonsisted of 64 points over angular ranges of 10 mrad and 3.75 mrad for the specular reflection and Bragg diffraction regions respeetively, and took approximately 20 min. to complete. Approximately 2 min/point of data were collected for each potential studied.

In the study of Cu UPD on Au(l00), the electrode was plaeed in a thoroughly degassed solution ofO.1M H2S04 and the potential was seanned at 20 mV/sec until the characteristie voltammetrie profile was obtained[18]. The electrode was then plaeed in a thoroughly degassed solution of O.lM H2S04 containing copper at either ImM or 50~M eoneentration and the potential was scanned over the UPD region [19]. The potential was held at the desired value for a preseribed amount of time until the current had deeayed to background levels. The electrode was removed from the solution under potential control and rinsed thoroughly with water. It was subsequently mounted on a Huber Euler eradle where the XSW experiments were earried out.

The XSW studies of Cu UPD on Au(lOO) were carried out at the Hamburger Synchrotronstrahlungslabor (HASYLAB) on Beam line Römo 1. A Ge(220) and an asymmetrically cut (17°) Si(220) double crystal monochromator were used to seleet an ineident energy of 1O.54keV. This value was sufficient to excite CuKa fluoreseence, but below that of all the gold L edges, thus minimizing the background signal. Data aequisition was done with the pro gram SPECTRA in eonjunetion with LeCroy histogramming memory modules.

It should be emphasized that while the studies of Cu UPD on an iodine treated platinum surface were carried out in-situ, the work on Cu UPD on Au(lOO) was ex-situ.

4. Results 4.1 X-Ray Standing Wave Study of Cu UPD on an Iodine Treated Pt Surfaee: 4.1.1 Surface coverage Isotherms:

In these experiments, a platinum/earbon LSM is used as the diffraeting substrate and working electrode. In sulfuric acid media, the voltammetry due to the platinum surfaee of the LSM exhibited only one pronouneed (the so-ealled weakly bound) hydrogen adsorption peak. Such behavior has been previously shown to be charaeteristic of a clean well-ordered Pt(lll) eleetrode that has been eyc1ed into the oxide region a few times to yield a Pt(lll) surface with nearlyrandomly distributed monatomic steps [20]. In fact, the voltammetry for copper UPD on the iodine treated LSM (Figure 1) was virtually identieal to that of an iodine coated Pt(lll) electrode that was treated as mentioned above. As mentioned previously, x-ray measurements were carried out at applied potentials of +0.45, +0.25, +0.20, +0.15 and +0.10 eorresponding to copper surfaee coverages of approximately 0, 1/4, 1/2, 3/4 and 1 monolayer, respeetively.

0.0 +0.20 +0.40 E vs AgjAgCI

Figure 1

+0.60

Cydic voltammogram at 20m V /s for the UPD of copper on an iodine covered platinum surface of a PtfC LSM in contact with a O.lM H2S04 solution

containing copper at a concentration of O.lmM.

375

Copper surface coverages were determined from both electrochernical and x-ray fluorescence measurements. Electrochemically, the coverage was determined from integration of the area under the voltammetrie wave. Surface eoverage isotherms from two different sets of experiments were obtained. In the first, the deposited copper was stripped in the presence of bulk copper whereas in the second case the electrode was rinsed three times with supporting electrolyte containing no copper. Comparing these isotherms, we observe a loss of deposited copper, after rinsing, that is coverage dependent. At full monolayer coverage the loss was only 16% whereas at sub­monolayer coverages of 3/4, 1/2 and 1/4 the losses were 47, 55 and 62%, respectively.

In the x-ray measurements, the coverage isotherms were determined from the off-Bragg fluorescenee yield data of the XSW measurements. (Note: Such off-Bragg, i.e. away from the Bragg angle, yield experiments essentially measure all the copper species contained within the thin layer of liquid trapped between the e1ectrode and a polypropylene film which serves as a window.) We carried out rinsing and no rinsing experiments equivalent to the ones described above. Again, we observe a drastic loss of surface coverage after rinsing the electrode. However, the fractional losses are considerably larger than those measured electrochemically. In an attempt to compare these experiments, the x-ray derived isotherms have been plotted on an absolute sca1e versusthe electrochernical ones normalizing the two data sets at only one point: +O.lOV, for the rinsing experiments (Figure 2). We note that the coverage isotherms for the rinsing case for both the x-ray and electrochemical experiments are in excellent agreement, but when we compare the results of experiments where the elecrrode had not been rinsed, the x-ray measurements indicate the presence of a considerable amount of electrochemieally inactive copper, above and beyond the bulk copper present in solution. In addition, XSW measurements corresponding to this coverage place this excess copper at the solid/solution interface. Furthermore, even at applied potentials of +0.45 V, where no electrodeposition has yet occurred, we observe an amount of copper equivalent to approximately 20% of a monolayer. Finally, we note that the iodine

376

coverage, as determined from x-ray fluorescence, is constant throughout the experiment (bottom panel Figure 2). This last observation is consistent with previous UHV and electrochemical studies by Hubbard and co-workers.[21]

::1 6 (t) tJI) cd H (t)

> 0

U (t) u ~ H ;::l

C/)

;::l U

1.5

X-Ray Data: • Rinsing \1 No Rinsing

Electrochem. Data

1.0 T Rinsing o No Rinsing

Residual Data: • (\1-0)

0.5

O.ol-.....L---L---L-----'--=:z...----j \---I-----l-----+--:;:::=:==::::=!:==::-l1.50 ~

@ co

1--'V=~:::::=~~====::::~==::;II.OO 2 §

L-~' =-_---".:.,,' ,.,,--_--;:-,,-;;-_----,-:;;-:;-;~--:-;;~0.50 '-' +0.10 +0.20 +0.30 +0.40 +0.50

Potential (V vs Ag/AgCI)

Figure 2 X-ray and electrochemical derived isotherms plotted on an absolute coverage

scale after the two data sets were normalized at one point, +0.1 V after rinsing. Bottom panel: Normalized iodine fluorescence as a function of applied potential.

XSW experiments were carried out at the same potentials as before and under conditions of Bragg diffraction and specular reflection. This allows for adetermination of the distribution of interfacial species on two different length sc ales. The results of these measurements are consistent with having a deposited layer of copper Qn the platinum surface and, in the case of the no-rinsing experiments, an additional amount of copper is present in a region proximal to the electrode surface. These results are fully consistent with those derived from the previously mentioned isotherms.

4.1.2 Reflectivity measurements Reflectivity measurements were carried out to characterize important structural

features of the substrate. Fram a reflectivity measurement one can determine the thickness ofthe thin solution layer trapped between the LSM and the polyprapylene film encapsulating the electrochemical cell, and the LSM's interfacial and surface roughness.

377

Figure 3 shows the angular dependence of the measured absolute Bragg reflectivity for a platinum/carbon LSM under a solution layer 0.98 11m thick. From fits to the Bragg reflectivity measurements we determined the interfacial roughness to be 6.8±0.SA.. Assuming this roughness value for the surface, produced a good fit to the specular reflectivity data as weIl. We can compare this reflectivity determined surface roughness to that expected for a surface with a random distribution of monoatomic steps. In this case we assurne a Gaussian distribution whose half-width, G, is representative of the rms surface roughness in atomic units for this model surface. If we take this value and multiply it by the c10sest packing distance for platin um (2.26 A), in order to place the probability function p(z) on an ängstrom sc ale, we obtain a value of 6.28 A, which is in

excellent agreement with the value of 6.8±O.sA found by fitting the specular reflection profile. This correlation indicates that a randomly monoatomic stepped surface is a reasonable model for the surface of the PtfC LSM. This result is also consistent with the voltanm1etric results previously mentioned.

> -' () ()

0.4 t I [

ü.2 L

......... _.+.\ ..... .

17 18

e (mrad)

Figure 3

o E~p. Refl. - O'b6.8A

t(sot:)=9 80 OA 0''=0.0.'1. t(~ol. )=9800A O''=O.OA t(sol.)=52245A

19 20

Angular dependence of the measured absolute Bragg reflectivity (filled circles) for a PtfC LSM under a solution layer 9800A thick, encapsulated by a 6.35mm

polypropylene film. The dashed line represents the theoretical prediction for the reflectivity when the interfacial roughness of the substrate is neglected. The dotted line represents a calculation attempting to fit the experimental data by increasing the

solution layer thickness, whi1e continuing to neglect the interfacial roughness. The solid line represents the best fit to the data when inc1uding interfacial roughness.

378

4.1.3 XSW measurements: We now turn to an analysis of the standing wave fluorescence data

corresponding to the rinsing experiments discussed above. Specifically, we measured the standing wave profiles for both specular reflection and Bragg diffraction after deposition at potentials of +0.25, +0.20, +0.15 and +0.10 V and rinsing the electrode with clean supporting electrolyte (no copper present in solution) while maintaining potential control over the system at all times. The background subtracted Cu Ku XSW fluorescence yield was extracted from each fluorescence spectrum (in energy dispersed form) by fitting to a Gaussian on a quadratic background. These data were then '1.,2 fitted to the theoretical yields. The free parameters in these fits were: the distribution's peak position with respect to the substrate's surface, the distribution's FWHM, and a normalizing constant directly proportional to the distribution's area. In addition, XSW data from the specular and Bragg reflection regions were fitted simultaneously. The best theoretical fits are shown as solid lines on Figure 4 for applied potentials of +0.10 and +0.20 V.

G (mrad)

Figure 4 The XSW fluorescence profiles for both the specular reflectivity and Bragg

diffraction regimes for +O.lOV and +0.20V after rinsing the electrode surface with clean electrolyte. A magnified view of the Bragg data is shown in the in sets. Also

shown at the bottom, is the compIete reflectivity profile. Fits of the data were performed over the entire angular range simultaneously, and are plotted as solid lines.

379

In specular reflection XSW, the first antinode reaches the surface at the critical angle. At the LSM's critical angle the standing wave period Dc is about 100 A. Keeping the above discussion in mind, we note that in all cases studied, the Cu fluorescence yie1d peaks at the critical angle of the LSM. This means that a narrow distribution of copper exists at the LSM surface for an of the potentials investigated. However, the period of the standing wave in this regime is large and limits the resolution to which we can determine the distribution's position and width.

To improve the resolution we can make use of the XSW measurements in the Bragg regime, where the periodicity of the standing wave is essentially equivalent to the LSM's d-spacing. Referring to the insets of Figure 4, we observe rather different XSW profiles as a function of applied potential. The expected yield for a random distribution is proportional to (1 +Reflectivity) but in an cases studied, the fluorescence peak amplitude to background ratio is well beyond this random limit, indicating that the copper distribution is fairly narrow on the length scale of the standing wave period which in this ca se is about 40A. The changes in the shape of each standing wave profile are representative of changes in the position of this overlayer with potential. Fitting XSW data generated in the specular reflection and the Bragg diffraction regimes simultaneously, allows us to probe the same distribution of species on two rather different length scales and two different z-scale origins, leading to an unambiguous result.

Figure 5 summarizes the standing wave results in terms of the distribution profiles at each potential studied. In the main panel an distribution profiles are normalized to the same peak intensity, while in the inset each distribution is plotted in terms of its relative area. Also shown is the surface density profile of the LSM on a normalized scale as determined from reflectivity data and from which we determined a surface roughness of 6.8A. Note that the origin of the z scale is defined to be where bulk platinum begins.

z

I

1.0 ~ !

?insing Experiments

10

Solution

20 30

z (..\.r..g.)

Figure 5 Copper concentration profiles (after rinsing) vs. distance z normal to the LSM's surface, derived from the analysis of the standing wave data. In the main panel all distributions are normalized so that the peak concentration is one. The inset

shows the same concentration profiles in terms of their relative areas.

380

An especially revealing way of presenting the data is to plot the center of mass (i.e. the z-position where we reach 50% of the total amount of copper) of each distribution as a function of surface coverage. This is so because the center of mass is dependent on both the peak position and the FWHM in a given distribution. In order to explain the changes we observe in this parameter we need to consider the surface morphology in terms of the Gaussian model we have chosen to fit the reflectivity data. In this model we consider the fractional concentration of surface sites as a function of position along the z-scale (the Gaussian's area is normalized to one). In addition, we sectioned this concentration profile into bins with a width approximately equal to the closest packing distance for platinum (2.26 Ä) in order to introduce a finite size effecL If open surface sites were occupied in a random mode, one would expect a homogeneous copper distribution whose center of mass was always at the same z position, namely the center of the Gaussian representing the surface sites concentration profile (i.e. the position with the largest density of open sites). At the opposite extreme, we have a model in which open surface sites are occupied sequentially with the deepest (dosest to z=O Ä) ones first. In this case, the center of mass position would vary with the copper surface coverage. Both of these models are plotted, along with the experimental data in Figure 6 as a function of surface coverage. It is immediately clear that the observed results are in excellent agreement with the model that involves sequential filling of available surface sites with the deepest ones being occupied first. This finding implies that the more favorable surface sites for deposition are the ones closest to the platinum bulk lattice, either because the substrate-deposit interactions are maximized at these sites, or because the interaction with the electric fields present at the interface is greatest at these locations. In addition, deposited copper atoms either diffuse to these positions or the deposition process itself is "catalyzed" by these particular sites.

A - - -- ------- --

8 ~ ----V4

~

Z B 6

4

2 0.00 0.25 0.50

Coverage (ML)

Figure 6

diI: y

0.75

Variation of the center of mass in the copper ad-Iayer as a function of surface coverage. Curve A represents the expected variation in the center of mass for a model in which filling of the surface sites is random, whereas B represents the

variation expected for a model where the surface sites are filled sequentially starting with the deepest ones (dosest to z=O Ä) first. Points are experimental data.

381

Furthermore, one needs to consider whether the nature of the deposition process is coverage dependent, since lateral interactions among deposited atoms might become more important as the coverage is increased, and what structural role iodine might play. It is also unclear what structural effect rinsing the electrode surface with pure electrolyte has on the UPD layer. It is likely that some structural rearrangement will be triggered by this rinsing proeedure.

4.2 Cu UPD on Au(lOO): As mentioned earlier, these studies were carried out ex-situ so that the eleetrode

was removed from the eleetrolytic solution during the XSW study. Prior to any eleetroehemieal studies, a reflectivity profile around the Au(200)

refleetion was reeorded at an ineident energy of 1O.54KeV resulting in a weIl defined reflectivity eurve [22] (Figure 7 A) and aseertaining the nearly perfeet quality of the gold eleetrode used in this study. In addition, the width of the refleetion curve compared very weIl with the ealeulated value.

>-. .t:: 1.0 > ...... ....... t) ~

cg 0.5 P:::

F = 0.64±D.06 P = 0.89±D.02

-100 0 100 200 300

ce - eB) !1-rad

Figure 7

2.0 §J 1.5 ~

lZl ()

1.0 ~ (1)

~ ....... ~ p...

XSW measurement of copper eleetrodeposited on a Au(lOO) eleetrode from a O.1M H2S04 solution containing 50llM eopper sulfate. Shown are (A) the

reflectivity (0) and (B) the Cu Ku fluoreseenee (.) along with fitted theoretieallines.

The electrochemical response of the Au(100) e1ectrode was then obtained in O.1M H2S04 until the characteristic voltammetrie response for a clean and weIl ordered surface was obtained [18]. Afterwards, the electrode was transferred (with a protective drop of electrolyte) to a copper solution (lmM or 501lM) in O.1M H2S 04 where electrodeposition was carried out. In the first case, deposition was carried out from the 1mM solution of copper and a full monolayer was deposited by holding the potential at +O.lOV for 3 min. The eleetrochemical response for monolayer deposition and stripping was weIl behaved (Figure 8A). The eleetrode was removed from solution (under potential control) and rinsed with water. It was then transferred to a special holder and mounted on an Eulerian eradle where XSW experiments were carried out at

382

A

E vs i\g/i\gCI

~ 0.30 +0.10

I I

I I I I I I I I I + 1.00 +0.50 0.0

E vs Ag/AgCl

Figure 8 A. Cyclic voltammogram at 20 m V/s for a Au(lOO) electrode in contact

with a O.IM H2S04 solution containing copper at a concentration of ImM. E. Cyclic voltammogram at 20 mV/s for a Au(lOO) electrode in contact

with a O.IM H2S04 solution containing copper at a concentration of 50~M after holding the potential at +0.1 OV for 3 minutes.

an incident energy of about 10.54KeV where copper fluorescence could be excited while avoiding any of the L edges of gold. The fluorescence intensity was monitored as function of the angle of incidence around the Au(200) reflection. The fluorescence yield obtained followed closely the reflectivity profile (1 + Reflectivity), indicative of an incoherent (random) distribution of the electrodeposited copper adlayer.

Subsequently, the experiment was repeated after electrodeposition of a sub­monolayer (ca. OAOML) amount of copper from the ImM copper solution. Contrary to the previous case, a weH defined modulation in the fluorescence yield was obtained and from a fit of the measured fluorescence to a theoretical yield function, a coherent position of 0.9 was determined, together with a coherent fraction of abaut 50%.

Finally, deposition of a copper monolayer was performed from a dilute (50~M) solution of cop per. In this case, no discernible deposition peak was observed (as anticipated) whereas the stripping response was extremely sharp (Figure 8B) indicative of a very wen defined structure. XSW measurements on this samp1e again exhibited a very weIl defined modulation in the x-ray fluorescence intensity. Again, from a fit of the measured fluorescence to a theoretical yield function, a coherent position of 0.89±.02 was determined with a coherent fraction of 0.64±0.06 (Figure 7B).

Taking into consideration the radii of gold and copper atoms and the measured coherent position, the copper ad-atoms would appear to be located at four-fold hollow sites on the gold Iattice with a Cu-Au distance of 2.73±O.03Ä (Figure 9). Models where the copper ad-atorns occupied either a-top or bridge sites (Figure 9) were

Z Bridge Site

Figure 9 Pictorial representation of the structure of electrodeposited copper on a A u(1 00)

surface comparing the copper ad-atoms occupying either bridge or four-fold hollow sites.

383

inconsistent with experimental findings . Although additional measurements will be necessary to unambiguously make this assignmem, our results are fully consistent with such an adlayer structure. It should also be mentioned that our findings are at odds with the EXAFS data of Tadjeddine and Tourillon who propose that, at monolayer coverage, the copper atoms occupy a-top positions [23].

It is clear that the structure of the copper ad-Iayer is very strongly dependent on the mode of deposition and this may provide a means of controlling interfacial structure. It will be of great interest to determine if the deposition of bulk amounts of copper under similar conditions gives rise to a coherent deposit and if so, to determine the extent of such coherence. Such studies will be the subject of future experiments.

5. Conclusions We have been able to study in situ, the underpotential deposition of copper on an

iodine covered platinum/carbon layered synthetic microstructure, using XSWs generated by specular (total external) reflection and Bragg diffraction. The equilibrium structure of the UPD layer after rinsing of the electrode surface with pure electrolyte is one where the deposited copper density is highest for those surface sites closest to the bulk platin um lattice. In addition, we were able to follow potential dependent changes in the copper surface coverage as determined by independent electrochemical and x-ray measurements. There is excellent agreement between x-ray and electrochemical data for the case of rinsing of the e1ectrode. However, x-ray derived isotherms, in the case of no rinsing reveal the presence of a large excess of electrochemically inactive cop per at the solid/solution interface, when compared to the corresponding electrochemically derived isotherms.

In the case of Cu UPD on Au (100) the structure and the coherence of the ad­layer are very strongly dependent on the deposition conditions with a much more ordered and coherent deposit being obtained under slow deposition (from dilute solution) conditions.

384

6. Acknowledgment. This work was supported by the U.S. Anny Research Office, the Office of Naval

Research and the GerrnanFederal Ministry for Science and Technology. D.A. acknowledges a MARC NlH fellowship. J.F.R. acknowledges support by the Ford Foundation. H.D.A. acknowledges support by the J. S. Guggenheirn Foundation during a visit to HASYLAB. The authors gratefully acknowledge Dr. W. Uelhoff for kindly providing the Au(lOO) single crystal and A. Fattah for assistance in preparation of the crystal. Contributions rnade by Donna Taylor and Howell Yee are gratefully acknowledged.

7. References 1. a. Kolb, D.M., in H. Gerisher and C. Tobias, eds., Advances in Electrochernistry

and Electrochernical Enginneering, Vol. 11, J. Wiley and Sons, New York, 1978.

b. Adzic, R.; Isr. J. Chern. 1979, lß., 166. c. Adzic, R., in H. Gerisher and C. Tobias, eds., Advances in Electrochernistry

and Electrochernical Enginneering, Vol. 13, J. Wiley and Sons, New York, 1985.

d. Juttner, K.; Lorenz, W.J.; Zeit. Physik. Chernie, 1980, 122, 163. e. Lorenz, W.J.; Hermann, H.D.; Wuthrich, N.; Hilbert, F.; J. Electrochern.

Soc. 1974, 121, 1167. f. Szabo, S.; Int. Tev. Phys. Chern. 1991, 10,207.

2. a. Schultze, J.W.; Dickertrnann, D.; Syrnp. Faraday Soc. 1977, 12,36. b. Salvarezza, R.C.; Vasquez Moll, D.V.; Giordano, M.C.; Arvia, A.J.; J.

Electroanal. Chern. 1986, 213, 301. c. Parajon Costa, B.; Canullo, J; Vasquez Moll, D.V.; Salvarezza, R.C.;

Giordano, M.C.; Arvia, A.J.; J. Electroanal. Chern. 1988, 244, 261. 3. a. Schultze, J.W.; Dickertrnann, D.; Surf. Sci. 1976,.5..4., 489.

b. Bewick, A.; Thornas, B.J.; Electroanal. Chern. 1976, 70, 239. 4. a. Hubbard, A.T.; Accts. Chern. Res., 1980, 13,987.

b. Yeager, E.B.; J. Electroanal. Chern., 1981, m, 1600. c. Ross, P.N.; Surf. Sci., 1981, 102,463. d. Kolb, D.M.; Zeit. Physik. Chemie N.F., 1987, 154, 179. e. Hubbard, A.T.; Chern. Rev., 1988, 88, 633. f. Beckrnann, H.O.; Gerisher, H.; Kolb, D.M.; Lehnpfuhl, G.; Syrnp. Faraday

Soc. 1977, 12,51. 5. a. Magnussen,O.M.; Hotlos, J.; Nichols, R.J.; Kolb, D.M.; Behrn, R.J.; Phys.

Rev. Lett. 1990, 64, 2929. b. Manne, S.; Hansrna, P.K.; Massie, J.; Elings, V.B.; Gewirth, A.A.; Science,

1991, 251, 183. c. Hachiya, T.; Honbo, H.; Itaya, K.; J. Electroanal. Chern. 1991, ill, 275. d. Magnussen, O.M.; Hotlos, J.; Beitel, G.; Kolb, D.M; Behrn, R.J.; J. Vac.

Sci. Tech. B, 1991,2.,969. e. Chen, C-H.; Vesecky, S. M.; Gewirth, A. A.; J. Arn. Chern. Soc. 1992, 114,

451. 6. Abrufia, H. D. ed. Electrochernical Interfaces: Modem Techniques for In-Situ

Interface Characterization, VCH, New York, N.Y. 1991. 7. a. Abrufia, H.D.; White, J.H.; Albarelli, M.J.; Bommarito, G.M.; Bedzyk, M.J.;

McMillan, M.J.; J. Phys. Chern. 1988,2,2, 7045. b. Tourillon, G.; Guay, D.; Tadjeddine, A.; J. Electroanal. Chern. 1990, 289,

263.

385

c. Tadjeddine, A. J.; Guay, D.; Ladouceur, M.; Tourillon, G.; Phys. Rev. Lett. 1991, 66, 2235.

d. Sarnant, M. G.; Borges, G. L.; Gordon, J. G.; Melroy, O. R.; Blurn, L.; J. Arner. Chern. Soc. 1987, 109,5970.

8. a. Sarnant, M.G.; Toney, M.F.; Borges, G.L.; Blurn, L.; Melroy, O.R.; J. Phys. Chern. 1988, 92, 220.

b. Toney, M.F.; Gordon, J.G.; Sarnant, M.G.; Borges, G.L.; Wiesler, D.G.; Yee, D.; Sorensen, L.B.; Langrnuir, 1991,1,796.

9. a. Materlik, G.; Zegenhagen, J.; Uelhoff, W.; Phys. Rev. B, 1985, 32, 5502. b. Materlik, G.; Schrnah, M.; Zegenhagen, J.; Uelhoff, W.; Ber. Bunsenges.

Phys. Chern., 1987, 91, 292. c. Zegenhagen, J.; Materlik, G.; Uelhoff, W.; X-Ray Sei. Teeh. 1990,2:,214.

10. Bedzyk, M.J.; Bilderback, D.; White, J.H.; Abruiia, H.D.; Bornrnarito, G.M.; J. Phys. Chern., 1986, 90, 4926.

11. a. Abruiia, H. D.; Bornmarito, G. M.; Aeevedo, D.; Scienee, 1990, 250, 69. b. Bornrnarito, G.M.; White, J.H.; Abruiia, H.D.; J. Phys. Chern. 1990, 94,

8280. 12. Bedzyk, M.J.; Bornrnarito, G.M.; Caffrey, M.; Penner, T.; Scienee, 1990, 52,

248. 13. a. Batterman, B.W.; Cole, H.; Rev. Mod. Phys., 1964, 36,681.

b. Batterman, B.W.; Phys. Rev., 1964, 133, A759. 14. Bedzyk, M.J.; Bilderbach, D.H.; Bommarito, G.M.; Caffrey, M.; Schildkraut,

J.J.; Scienee, 1988, 241, 1788. 15. a. P.L. Cowan; J.A. Golovchenko; M.F. Robbins; Phys. Rev. Leu. 1980, 44,

1680. b. J.A. Golovehenko; J.R. Patel; D.R. Kaplan; P.L. Cowan; M.J. Bedzyk; Phys.

Rev. Lett. 1982, 49, 560. c. Materlik, G.; Zegenhagen, J.; Phys. Lett, 1984, 104A, 47.

16. Underwood, J.H.; Barbee, T.W.; in AlP Conf. Proc., 75, 170, D.T. Atwood, B.L. Henke, eds., AlP, New York, 1981.

17. a. Fehrner, H.; Uelhoff, W.; J. Sei. Instrurn. 1969,2:,767. b. Fehrner, H.; Uelhoff, W.; J. Sei. Instrurn. 1969,2:, 771.

18. Angerstein-Kozlowska, Conway, B. E.; Hamelin, A.; Stoicoviciu, L.; Eleetroehirn. Acta 1986, 31, 1051.

19. Kolb, D. M.; Al Jaaf-Golze, K.; Zei, M. S.; DECHEMA-Monographien Bd. 102 VCH Winheim, 1986.

20. Aberdarn, D.; Durand, R.; Faure, R.; EI-Ornar, F.; Surf. Sei. 1986, 171,303. 21. a. Stiekney, J.L.; Rosasco, S.D.; Song, D.; Soriaga, M.P.; Hubbard, A.T.; Surf.

Sei. 1983, 130, 326. b. Hubbard, A.T.; Stickney, J.L.; Rosasco, S.D.; Soriaga, M.P.; Song, D.; J.

Eleetroanal. Chern. 1983, 150, 165. e. Stickney, J.L.; Rosasco, S.D.; Hubbard, A.T.; J. Electroehern. Soc. 1984,

131,260. 22. Jarnes, R.W.; The Optieal Principles of the Diffraction of X-rays, Oxbow Press,

Woodbridge, Connecticut, 1982. 23. Tourillon, G.; Guay, D.; Tadjeddine, A.; J. Electroanal. Chern. 1990, 289, 263.

TUE APPLICATION OF INFRARED SYNCHROTRON RADIATION TO THE STUDY OF INTERFACIAL VIBRATIONAL MODES

CAROL J. HIRSCHMUGL AND GWYN P. WILLIAMS National Synchrotron Light SOUTce. Brookhaven National Laboratory. Upton. New York. 11973. USA

ABSTRACf. Synchrotron radiation provides an extremely bright broad-band source in the infrared which is ideally suited to the study of surface and interface vibrational modes in the range 50-3000 ern-I. Thus it covers the important range of molecule-substrate interactions, as weH as overlapping with the more easily accessible near-ir region where molecular internal modes are found. Compared to standard broadband infrared sources such as globars, not only is it 1000 times brighter, but its emittance matches the phase-space of the electrochemical cell leading to fuH utilization of this brightness advantage. In addition, the source is more stable than water-cooled globars in vacuum for both short-term and long-term fiuctuations. Thus one can worlc at high resolution and use isotopic shifts to identify and study very weak modes.

We will summarize the properties of synchrotron radiation in the infrared, in particular point­ing out the distinct differences between this and the x-ray region. We will use experimental data in discussing important issues of signal/IlOise ratios and will address the unique problems and advantages of the synchrotron source. We will emphasize the important considerations necessary for developing new facilities. This analysis will then lead to a discussion of phase­space matching to electrochemical ceHs, and to other surfaces in vacuum.

Finally we will show several examples of the application of infrared synchrotron radiation to surface vibrational spectroscopy. The examples will all be for metal crystal surfaces in ultra­high vacuum and will include CO/Cu(100) and (111) and CO/KICU(100). The experiments will show how the stability of the synchrotron source allows subtle changes in the background to be observed in addition to the discrete vibrational modes. These changes are due to elec­tronic states induced by the adsorbate. In some cases we have seen interferences between these and the discrete vibrational modes, leading to a breakdown of the dipole selection rules, and the observation of additional modes. These important experiments serve as sensitivity limit indicators and thus as a guide to future applications in the field of electrochemistry.

1. Introduction

We have used infrared synchrotron radiation to study vibrational modes of adsorbates on metal surfaces in ultrahigh vacuum. The experiments can also be extended to interfaces in electro­chemical ceHs as reported elsewhere in this conference[l]. Signal to noise is critical in the experiments that we report here, since some of the vibrational modes give a refiectivity change of less than 0.1 %.

387

C. A. Melendres andA. Tadjeddine (eds.). Synchrotron Techniques in lnteifacial Electrochemistry 387-399. © 1994 Kluwer Academic Publishers.

388

In this paper we report in some detail the critical components of the synchrotron radiation extraction and delivery system, pointing out differences between a beamline of this type and a more conventional x-ray one. We discuss in detail how instabilities in the electron beam position in the ring translate into noise in the data, and we show how these have been minim­ized at the NSLS facility.

Finally we present data for CO/Cu(lll), and for O/K/Cu(lOO) as examples of the capa­bility of the system as a probe of interfacial vibrational modes. Such studies yield data not only about the energetics of such modes, but about surface species and bonding geometries as weIl as the dynamics.

2. Infrared Synchrotron Radiation

2.1 BRIGHTNESS AND FLUX

Calculations of infrared synchrotron radiation for adipoie source can be relatively easily car­ried out on a PC using the algorithms of Kostroun[2] applied to the equations given in Hulbert and Weber[3]. Tbe long wavelength approximations for the emitted flux presented in Dunean and Williams[4] are also useful as a guide, and the treatment of the source term is valuable for brightness calculations. One finds three important facts from such a calculation. 1. Synchro­tron radiation is several orders of magnitude brighter than a conventional globar source. 2. For wavelengths longer than 100 microns, (frequencies less than 100 em-I ), synchrotron radiation also rves more photons per second, even into its limited angle-area product (emittance) of -10- mm2 steradians than a 10 mm2 2000K globar does into its -100 mm2 steradian emit­tance. 3. Synchrotron radiation has very large natural opening angles in the infrared, which are at least 100 times larger than those found for x-rays from the same storage rings.

Tbe situation is summarized in Fig. I, in which we have also included the free electron lasers CLIO at LURE, Orsay, France and Santa Barbara, USA for comparison. In Fig. 1 we have deliberately restricted the 2000K, 10 mm2 globar to emission angles of O.IXO.l radians. This was chosen to closely match the acceptance of the systems that we will be studying.

As mentioned in the introduction, however, it is signal to noise which is more critical than flux or brightness. Due to fundamental detector noise limits the S/N ratio is improved if the delivered flux can be increased without introducing additional noise of its OWll. It is this last issue that we now concentrate on.

2.2 SYNCHROlRON RADIA nON "NOISE".

Source fluctuations can have several origins, and for athermal source the most notable is from temperature fiuctuations. For a synchrotron source, the output is strict1y proportional to the stored beam current which decays in a slow exponential fashion, and cannot fluctuate up and down. Tbere are fluctuations, however, in signal detected from synchrotron radiation experi­ments. These arise from motions of the electron beam in the storage ring and consequent

389

changes in the collection efficiency. These beam motions are responsible for rnuch of the noise observed in synchrotron radiation experiments of all kinds.

1020 CUO 1 0.3% lIlI J r-----,-----=i . . Enon Laser L ____ _

Globar

2000K 10mm2 .1 by .1 rads.

0.01% BW

108 L--L--'--'-...L.J..Ju..LL_'---'---'--''-'-'-'-'.L-----'--'---'....ULLLLI

101 102 103

Wavenumbers [ern-I]

Fig. 1. Average tlux as a function of wavelength for various infrared sources including a con­ventional globar, the National Synchrotron Light Source, and various IR free electron lasers at University of Califomia at Santa Barbara, Los Alarnos, Exxon Research, Annandale, NJ, USA and at LURE, Orsay, France (CLIO).

In the case of the NSLS, these rnovernents of the bearn tumed out to be critical to our experiments, since we were trying to rneasure retlectivity changes of <0.1 % in sorne cases. We also found that they varied with the time of day, being less pronounced at night and at weekends thus giving some confidence that they rnight be able to be dealt with. If signals frorn pick up electrodes (pue's) around the ring are recorded and frequency analyzed, then it is found that most of the bearn motions are at frequencies below 100Hz, with a few distinct higher frequency ones, such as at 360 Hz. The abundance of low frequency peaks isa conse­quence of the fact that the alurninum vacuum charnber screens out and reduces the effects of unintentionally applied rnagnetic fields of higher frequeneies. Nevertheless some higher fre­queney peaks, whieh were easily seen in our early data taken with the rapid-sean Miehelson interferometer (see later), were found to be coming from dipole and quadrupole power sup­plies, and these were replaced.

390

When the pue's are recorded every 20 seconds or so, we find that the beam movements are typicaIly of the order of ±5 microns in the vertical and ±25 microns in the horizontal. In addition to this there is a horizontal drift of up to 0.5mm as the beam decays from 1000 mA to 200 mA. The beam size as seen by the infrared experiment is typically 1mm vertically x3mm horizontally. These source motions give a reproducibility in the data or "noise" of aroWld 1% as determined from the difference between 2 successive spectta, each averaged for 270 seconds. Due to other drifts, averaging for longer periods of time does not improve the situa­tion.

To take care of the 10wer frequency ßuctuations, a feedback system was built and installed on the ring[5]. ACUlally two independent systems take care of the vertical and hor­izontal. The effects of these feedback systems are shown in Fig. 2 in which we have plotted the beam position as a function of time in both the vertical and horizontal over an 8 hour period. Several times during this period the feedback systems were tumed on and off to demonstrate what would happen to the beam in the absence of any control. With the respec­tive feedback on, the beam position is held close to position "zero". The horizontal and verti­cal systems are completely independent of each other and this can be seen in Fig. 2 when the horizontal feedback was U1med off at -11:40, and there was no effect on the vertical position. 1be vertical feedback system confines the vertical beam motion to ±3 microns over several hours, while the horizontal motion is limited to ±1O microns, also over several hours. All of the horizontal drift with beam current is removed, as are any sudden jumps in vertical position. For example the vertical jump between 11:30 anc;t 12:00 was due to a 1° temperature change in the ring cooling water. Repeatibility is also good from fill to fill.

l00~------~--------~~-----,

so -".-

-500

-750 L... _____ -'-_~ ___ _'_ __ ~~__'

12 15

Time [Hours]

Fig. 2. Effects of feedback on electron beam position in the NSLS 750 MeV electron storage ring. The feedback system was cycled on and off throughout the day. Tbe beam position is held near zero with feedback on, but drifts considerably without feedback control.

391

Our surface science experiment is more sensitive 10 the vertical motions due 10 our par­ticular experimental geometry. With the vertical feedback on. we reliablyacbieve a repeatabil­ity of 0.1 %. With both feedback systems operating, we acbieve a repeatability of much less than 0.1 % even down 10 100 cm -1 and in 270 seconds of seanning. However notice that the feedback systems play more of a role in controlling beam drift and jumps than ImS beam movement. Thus they ensure that 110 sudden changes in signal occur, and they optimize the utilization of the beam.

2.3 nIE MICHELSON RAPID-SCAN INTERFEROMETER

In order 10 fully understand how we have been able 10 acbieve the signal 10 nQise values wbich allow us a confidence in our spectra of <0.1 % (in spectra taken over a 3 minute sean period), we now deseribe the spectrometer in more detail. 'These issues have also been described in more detail by Hirschmugl[6].

Expt. + Oetector v=100Hz

Fixed Mirror

1*--+---0 + U2 -+----1\:

Ä. = 100 Ilm (say) Source

Moving Mirror 5 mmlsec

Fig. 3. Schematic of a Michelson Interferometer with a rapidly seanning milTOr. The effect is a modulation of the incoming beam at a frequency proportional to wavelength.

392

Dur beam1ine uses an interferometer, a schematic of which is shown in Fig. 3. It is of the conventional amplitude dividing Michelson type - in our case a modified commercial Nicolet model 20F. This instrument is of the rapid-scan type[71. in which the moving mirror moves at a speed of typically 5 mm/sec. Thus the path difference is changed by 1 cm/sec and the signals are modulated at the detector in the audio frequency range without the need for a chopper which would halve the signal available at the detector. Wavelengths of 100 microns are modulated at 100 Hz, of 10 microns at 1000 Hz and so on. Fourier analysis of the inter­ferogram generated while the mirror scans the required distance can then be understood as a transform from the time domain 10 the frequency domain. In practice we co-add up to 1000 such interferograms.

Thus it is easy 10 see that at this scanning velocity, any beam motion occurring at fre­quencies less than 100 Hz will appear in the spectrum at wavelengths less than 100 microns. The data is collected on a time scale short compared with the beam noise and most of the latter is thus automatically eliminated.

2.4 nIE BEAM IDcrRAcnON SYSTEM

Fig. 4. Schematic of the infrared beam extraction system at the NSLS. This design allows light to be collected into an angle of almost lOOxIOO milliradians.

393

In order to deal with the very large opening angles for infrared synchrotron radiation discussed earlier, we modified the NSLS VUV storage ring as shown in Fig. 4. The extraction mirror, in this case SiC mounted in a water-cooled copper block, is placed in a special chamber immediately downstream of adipoie, and thus close to the electron beam itself. Special radio-frequency damping resistors beneath the mirror "spoil" the resonance of the cavity, so as to minimize the effect on the beam.

The arrangement shown allows 90 mradians x 90 mradians to be collected from the source. Aseries of mirrors [8] delivers this to the interferometer as a 4 cm dia. collimated beam.

3. Surface Vibrational Spectroscopy

3.1 GENERAL

Energy absorbed

r Adsorbate /

Y Substrate

! Energy

Transfer

Fig. 5. Geometry of the surface vibrational spectroscopy experiment for metal surfaces in vacuum. Electric fields parallel to the surface undergo a phase change, giving a net field of zero immediately above the surface.

The general principle of the surface vibrational spectroscopy experiment is shown in Fig. 5. Light refiected at grazing angles from metal surfaces excites vibrational modes perpendicular to the surface - the so-called surface dipole selection rule [9, 10]. If we look at a specific sys­tem, namely CO adsorbed in atop positions on low index Cu surfaces, then Fig. 6 shows that there are 4 possible vibrational modes. Since only the carbon-metal and the internal C-O stretch modes are dipole allowed, then in a retlectivity experiment such as that shown sehern at­ically in Fig. 7, we would expect a transfer of energy from the incoming beam to these vibra­tional modes with a corresponding loss at certain frequencies in the retlected beam. In fact we will see that for this system there is a breakdown of the dipole selection rule at low frequen­cies and the spectrum obtained yields richer information.

394

t 0 0

A1 V1 A1 V2

C C

~ ~ @ @ @ @ @ @

0 ~ O~

V3 E V4

E ~ C C ~

@ @ @ @ @ @

Fig. 6. The four possible vibrational modes for a CO molecule adsorbed in an atop position on Cu. vI and V2 are the only dipole active modes.

~ ·00 1-,,---., c: Q)

C a:

Frequency (cm-1)

Fig. 7. Schematic of the refiectivity experiment, in which the discrete vibrational modes remove intensity from the refiected beam as ShOWll.

395

Tbe adsorbate-substrate modes are of particular interest as they can in principle yield information not only about the adsorbing species but also about the adsorption site symmetry, and more importantly about energy transfer mechanisms or dynamics. Tbese adsorbate­substrate modes are all at low energies, or long wavelengths in the far infrared beyond 10 microns, and particularly benefit from the use of synchrotron radiation as can be seen from Fig. 1. Tbe throughput of the sampie matches the synchrotron emittance weIl. For electro­chemical systems whose electrolytes are highly absorbing the synchrotron offers another advantage. In this case the additional light available from the synchrotron source boosts the detected signal levels weil above detec10r noise thresholds. We should also point out that unfortunately there are many water vapor lines in this spectral region, necessitating the use of vacuum for the whole beamline and adding to the complexity of studying electrochemical sys­tems since the windows have 10 provide an interface to vacuum.

One can in general determine 4 things from peaks observed in a surface vibrational spec­troscopy experiment. From the frequencies one can learn about dipole-dipole interactions, self image shifts, chemical shifts, identification (a1omic, molecular intermediate etc.) and molecular symmetry. Tbe peak widths contain information about the dynamics, phonon or electron-hole coupling, vibrational phase relaxation and inhomogeneous broadening. The shapes yield infor­mation about possible adiabatic and other energy coupling mechanisms and the intensity gives the dynamic dipole moment. In many cases it is of interest to investigate the behavior of peaks as conditions change at the surface. Isotopic substitutions are helpful in identifying some of the modes and interactions.

3.2 RESULTS FOR CO, ü+K ON Cu.

In Fig. 8 we show actual data for CO/CU(111) in which the differences from the schematic of Fig. 7 are immediately apparent. We have plotted the ratio of the reflectivity of a CO covered surface to a clean surface for 2 coverages. If the adsorbate had no effect, then we would have a straight line at 100%. Two things are immediately clear - there is a broadband background change and a large peak at around 2100 em-1• Note that this spectrum was measured with the same Si beamsplitter in the Nicolet 20F. Two detectors were used - a Cu doped Ge detector above and a B doped Si bolometer below 600 em -1. Tbe optical arrangement is actually optimized for the region below 500 em -1 and in fact one can see that the noise in this region is considerably lower than elsewhere. In fact the variations in the background in the region from 500-1000 em-1 are probably due 10 noise. On the figure we point out the usuallimit of "conventional" surface vibrational spectroscopy at around 800 em -1. Tbe absolute limit of non-synchrotron spectroscopy is shown at around 350 em-I , the lowest frequency for excita­tion of carriers in pho1o-conductive detectors of higher sensitivity but not necessarily better signal 10 noise.

If we zoom in on the region below 400 em-I we see the spectrum shown in Fig. 9. Here we see two features on the sloping background. Tbe features, however, are extremely weak compared with the >5% of the 2100 em -1 feature of the carbon-oxygen mode and are only visible due 10 the extremely low noise level of 0.01 %. In Fig. 9 each spectrum was aver­aged for - 10 minutes for the background and the same time for the CO covered sampie. We identify the peaks as the carbon-metal stretch at 346 em -1 and the hindered (dipole-forbidden) rotation at 284 em-1• The latter appears as more of a peak than a dip. We attribute its

396

appearance to the background change and to an interference with electrons excited in the bulk[ 11]. The assignments are based on the isotopic shifts shown in the additional 2 spectra. Notice how the frequency of the hindered rotation mode is independent of oxygen mass - as expected for a bending mode if the molecule is adsorbed linearly atop with the carbon adjacent to the metal.

100 ~-- - -~ ---- - ---- ~-- ---- ---------.----

I ~: .~~-

95 t-0: "-0: <l

90 -

o

o

i 1 " . ' r.-~ ... 0

... ! C

•• Cu

,

Data -- Drude Theary

(Perssan)

" .

'<-Usuol limit of surfac~ , vibrationol sp~ctroscopy

, , I+- Limi t cf conventianal I "non-synchrotron" sRectroscopy.

500

so

• 0 t

• C i • Cu 2000

I ! 250 300

-

-

Fig. 8. Infrared reflectivity spectra from a Cu(l11) surface for 2 coverages of CO. The discrete vibrational modes are seen superimposed on a background which changes with cover­age.

We make one remarlc based on the above and on other work that we and others have car­ried out on surface vibrationaI spectroscopy. Most of the internal molecular modes for molecules lie weil above 1000 em -1, while most of the metal-molecule modes He below 400 em-I . Thus there are relatively few modes in the range 400-1000 em- I . Thus studies which extended conventional spectroscopy to lower frequencies were frequently abandoned with no new peaks seen. With the synchrotron, however, we see a whole new region full of adsorbate-substrate vibrations.

The spectra in Fig. 9 show the present limits of the technique in tenns of signal to noise, and the ability to observe weak transitions. Not an of the vibrational modes are this weak, and in Fig. 10 (left) we show data laken as a function of time during the exposure to oxygen of a potassium multilayer on Cu(100)[12]. The potassium oxide vibration at 350 em-I has a mag­nitude of 9%. Annealing of this potassium superoxide layer yields the spectrum shown on the right in Fig. 10 in which we see the fonnation of cuprous oxide, CU20.

397

Isotope Shirt for e(2x2) COjCu(lOO)

225 250 275 300 325 350 375 400 Wavenumbers (ern-I]

Fig. 9. Far infrared refiection spectra for 3 isotopes of CO/Cu(111). The peak at the left is the hindered rotational mode, while the dip is the carbon metal stretch mode. This assignment is based on the observed shifts as discussed in the text.

340

200 150 300 450 600

Wavenumbers [em -lJ

347 0/":/CU(100)

350

350 500 l"avenumbe:3 {em -lJ

621

650

Fig. 10. Far infrared refiection spectra from a Cu(100) surface covered with a K multilayer. On the left is aseries of spectra taken as a function of exposure to oxygen. On the right is the effect of annealing 10 various temperatures and the formation of CU20.

398

4. Conclusions.

We have attempted to show the capabilities of the new technique of low frequency surface vibrational spectroscopy using synchrotron radiation. In doing so we have diseussed in detail the measures that are essential to the success of the experiments. To summarize, these are:

• 1. A careful calculation of the source emittance and the provision of large enough extrac­tion angles.

• 2. Careful attention to the stability of the electron beam in the storage ring. • 3. Rapid sean interferometry with beam splitters covering as wide a range as possible.

• 4. A vacuum light path for the complete optical system . • 5. Resolution of typically 1 cm -1.

The U4IR beamline at the NSLS has demonstrated the capability of observing 10% of a monolayer of CO/CU via the carbon-metal mode. We are undertaking an extensive program to study the low frequency modes of CO on Cu and Ni on the low index faces. These experi­ments can be compared with cluster[13] and surface scattering theories[14] whose goal is to model the dynamics of these systems.

ACKNOWLEDGEMENTS. We are extremely grateful to Yves Chabal, Fritz Hoffmann, Paul Dumas, Tammy Bush, Wendy Walter and Dennis Carlson, for discussions and assistance in the conduct of the experiments. We thank Steve Kramer, Om Singh, Li-Hua Yu, Sam Krin­sky, Herb Langenbach and Yong-Nian Tang for discussions regardiilg the storage ring and the feedback and data acquisition systems. Research at the Brookhaven National Laboratory was perfonned under the auspices of the U.S. Department of Energy Contract No. DE-AC02-76CHOOO16.

References

1. A. Russell et al. these proceedings. 2. V.O. Kostroun, Nucl. Instr. & Meth. 172 371 (1980). 3. S.L. Hulbert and J.M. Weber, Nucl. Instr. & Meth. A319 25 (1992). 4. W.D. Duncan and G.P. Williams, Applied Optics 22 2914 (1983). 5. L.H. Yu, E. Bozoki, J. Galayda, S. Krinsky and G. Vignola, Nucl. Instr. & Meth. A284 268 (1989). 6. C.J. Hirschmugl, Nucl. Inst. & Meth. A319 245 (1992). 7. R. Madden and J. Strong, J. Opt. Soc. Am. 44 352 (1954). 8. G.P. Williams, P.Z. Takacs, R.W. Klaftky and M. Shleifer, Nucl. Instr. & Methods A246 165 (1986). 9. F.M. Hoffmann, Surface Science Reports 3 107 (1983). 10. Y.J. Chabal, Surface Science Reports 8 211 (1988). 11. C.J. Hirschmugl, G.P. Williams, F.M. Hoffinann and Y.J. Chabal, Physical Review Letters

399

6S 408 (1990). 12. F.M. Hoffmann, K. Lin, R. Tobin, C.J. Hirschmugl, G.P. Williamsand P. Dumas, Surface Science Letters 275 L675 (1992). 13. M. Head-Gordon and J. Tully, Physical Review B46 1853 (1992). 14. RN.J. Persson, Chemical Physics Letters 197 7 (1992).

FOURIER TRANSFORM INFRARED COMBINED WITH SYNCHROTRON RADIATION FOR PROBING THE ELECTROCHEMICALINTERFACE

Y-L. MATHIS, K. MURAKOSHI*, A. TADJEDDINE* AND P. ROY Laboratoire pour l'Utilisation de Rayonnement Electromagnetique, bat. 209 D Universite Paris-Sud, 91405 Orsay Cedex, France *also at Laboratoire d'Electrochimie Interfaciale du CNRS, 1 place Aristide Briand,92195 Meudon Cedex

ABSTRACT. A complete spectroscopic system combining the use of Synchrotron Radiation as a broad band source with a Fourier Transform interferometer and an in-situ electrochemical ceH is described here. The extraction and optical system as weH as the general principle of Fourier Transform Infrared Spectroscopy (FTIR) are presented. Preliminary measurements on the adsorption of CN- on platinum are discussed on the basis of classical electrochemical voltammograms as wen as mid infrared and far infrared absorption spectroscopy.

Introduction

Progress in fuHy understanding the electronic transfer at the electrochemical interface depends to a large extent on the development of new spectroscopic methods. Ideally such methods should allow the determination of the interface structure and of the nature of interaction among its various components which in turn depend on the nature of the electrode itself, its structure, the nature of the solvent, the ions presents in the solution and the potential applied on the electrode. While recent progress in X-ray absorption techniques have allowed the determination of the crystallographic structure of the layers adsorbed at the electrode (see A. Tadjeddine communication presented in the same conference), the understanding of the nature of the interaction governing the different constituents involved remains largely unsolved. The intermolecular couplings among them are medium to weak interactions and therefore may couple to the far infrared region of the spectrum (see figure 1). Moreover the vibrational, rotational and translational modes of the constituents may be shifted at the interface as a result of different intermolecular interactions.

Here we discuss the recent progress in the use of FTIR combined with far infrared synchrotron radiation sources, to study the structure and dynamics of the electrochemical interface.

1. Synchrotron radiation as an infrared source

Radiation emerging from ben ding magnets in synchrotron and storage rings is increasingly utilised as a far-infrared source. InfraRed Synchrotron Radiation (IRSR)

401

C. A. Melendres andA. Tadjeddine (eds.), Synchrotron Techniques in Interfacial Electrochemistry 401-420. © 1994 Kluwer Academic Publishers.

402

Wavelength ( J1m )

r r I 10 100 1000

EJeclronic tran ition in Solid SemiconduClOI'S Gap , Supraconductors Gap

lntennolecular Vibration

IntramolecuJar Vibration

I 1000 1000 100 10

Energy (on- J )

Figure 1. The main interactions and phenomena causing absorption in the low energy part 0/ the spectrum.

facilities are operating at the NSLS VUV in Brookhavenl (U.S.A.) and on the UVSOR2 and SPring-8 facilities3 (Japan). Further beamlines are under construction or testing in Daresbury (UK)4 at the MAXI ring of Lund5 (Sweden), and again at Brookhaven. Indeed, such sources are by two or three orders of magnitude brighter than a black body emitter at 2000 K, at all infrared wavelengths. On the other hand, the total photon flux f (number of photonlsecond/unit bandwidth) is considerably greater than that emitted by a black body for wavelengths A. > 100 mm onlyl,2. Therefore, use of IRSR from bending magnets has been proven 10 be advantageous for measurements on small sampies and/or in the 10000-100 ~m (1-100 ern-I) range. The gain in brightness in this spectral region is of interest in a number of areas, as for example studies of excitation spectra in crystals under pressure, interface properties of small size singlecrystal ,determinations of the superconducting gap in thin single crystals, etc.

403

1.1. SYNCHROlRON RADIATION EMITIED FROM WIGGLERS

An alternative source of IRSR is a wiggler, which may present advantages in terms of flexibility, photon flux, and extraction optics. Such a choice has been done at the Super­ACO ring of Orsay, where the infrared beamline Spectroscopie dans l'InfraRouge Lointain (SIRLOIN) is now completed.

In a wiggler or an undulator, relativistic eIectrons (or positrons) travel through a periodic magnetic structure (usually varying sinusoidally and in the vertical direction). All available undulators and wigglers (including the SU3 used for SIRLOIN) have been developed for the production of short wavelength UV or X-Rays. In these ranges the understanding of such devices is not complete but has been greatly enhanced by the general availability of software such as Shadow7. In the following section we briefly list the general properties of wiggler emission before presenting the system developed for extracting them and using them for Fourier Transform infrared spectroscopy. Wiggler radiation properties can be summarised as follows :

1.1.1. Undulator description

The relativistic charged particles are characterised by the y parameter, which is the ratio of their total energy E to their rest energy : y = EI mc2. The electron motion in the undulator shows a maximum angular deflection 0 proportional to the transverse magnetic field and to the length of one period of the undulator and inversely proportional to the energy of these relativistic charged particles.

The deflection parameter K links these two quantities : 0 = K I y, K = 0.934 Äu(cm) Bo(T), where Äu is the magnetic period in cm and Bo is the maximum magnetic field in tesla. The calculation consists in performing a correct convolution between the line frequency spectrum at an angle e caused by the periodicity of the undulator, COk: = k co 1(1 - ß cos e), the spectral distribution of the energy radiated by a particle into a solid angle dn and a frequency interval deo

1.1.2. Spectral range of the radiation

Alferov and co-workers8 have shown that the deflection parameter K is characteristic of the spectral range emitted by a given undulator or wiggler. In the case K>I, interference effeets are less important and the photons amplitudes add incoherently. The speetral distribution is given by 2 times the number of periods multiplied by the appropriate formulas for equivalent dipole magnets.

1.1.3. Angular distribution ofthe radiation

As in the case of the dipole radiation, the emission is coneentrated in a narrow angular range approximately given by 1 I y for very short wavelengths. In the infrared the emission cone is mueh larger and for wavelength of 100 microns the eone approaches 100 mrad for most synchrotron emission.

1.1.4. Degree of polarisation

For symmetrie undulators, the particle radiation is linearly polarised in the plane of the charged partieles. At large colleetion angles (100 mrad) it becomes only partially polarised. Moreover the extraction optics tend to reduce greatly the degree of polarisation.

404

As is evident from this list, the infra.red part of the synchrotron radiation spectrum is quite different from the emission in the classical spectral range (UV and X-Ray). In the following section we briefly present the SU3 wiggler characteristics as weIl as a calculation of the emitted intensity.

1.2. THE SU3 WIGGLER OF SUPER-ACO

The intensity emitted in the infra.red by undulators has been evaluated for the insertion device SU3 of Super-ACO. The Super-ACO ring is schematically represented in figure 2. Positioned on one of its straight section is the SU3 wiggler. This insertion device is 3096 mm long, has 24 periods (N = 24) each measuring 12.9 cm (Au = 0.129 m) and its central position is located at 3.3 m from the extraction mirror. Its working K value is 6 and the positrons circulating in the ring have an energy of 0.8 Ge V consequently the 'Y parameter is 1565. In the course of the beamline development we have designed a computer program allowing the evaluation of the intensity as weIl as the spatial distribution of the infrared emitted by the SU3 wiggler. This program also provides an evaluation ofthe ratio ofthe infrared actually collected by the extraction system (described in the next seetion). This calculation takes into account the fact that the field in the undulator varies as a sine function and that the source is actually a 3 m long stick and that the photons are not collected at the infinite. It also assumes that coherence effects are negligible due either to the very !arge divergence and to the length of the source.

Figure 2. Top view 01 the Super-ACO ring in Orsay. The SIRLOIN beamline is located inside the ring.

405

The total photon flux emitted by the SU3 undulator is expectedto be greater than for IR radiation extracted from a ben ding magnet of the same storage ring (50 mrad). The geometrical distribution of the radiation intensity, effectively available at the extraction mirrors of SIRLOIN after taking into account the geometrical constraint of the undulator, is pictorially represented for a wavelength of 100 microns in figure 3. Therein, each darker tone of grey corresponds to an intensity loss of 12.5 % compared to its maximum value (white). The dotted lines represent the plane mirrors with aseparation of 16 mm (the standard separation is 10 mm). It can be noticed that the far infrared intensity distribution presents a natural minimum at the centre of the ring, which makes the geometry employed here particularly suitable for the extraction of IRSR. This intensity distribution calculated for all the infrared range have allowed the evaluation of the fraction of photon flux which is intercepted and transmitted to the beamline. This fraction turns out to be about 80 % for 100 microns and is reduced to about 30 % for a wavelength of 1 micron. In figure 4, this photon flux is plotted as a function of the radiation wavelength for different values of the gap between the two mirrors as described in the next section. For comparison, the flux emitted by a black body at 2000 K is also shown. Details of this calculation is available in reference 6.

·60 · 40 · 20 o 20 40 60

Figure 3. The calculated spatial distribution 0/100 j.lm photons at the extraction mirrors. The dotted points represents the contour 0/ the mirrors.

406

o ........ (fJ

........ (fJ

.s 1011 o

.s::. a..

1

16

·24 >':. /' /

!l = 30 mm

L

1---

-=.-=.--=.-::::.---::::--.... - .-......;:: -- .-,,;;::::::,.... '""'Si!! .-

.... black body

10 100 1000 Wavelength (11m)

Figure 4. Infrared photon flux calculated for a black body at 2000 K (dotted Une) and for IRSR produced by the SV3 wiggler with extraction mirrors as described in the following seetion.

1.3. THE SIRLOIN BEAMLINE

SIRLOIN is a beamline exploiting the infrared photons emitted by the SU3 undulator of Super-ACO. The extraction of the infrared photons is schematically represented in figure 5.

In order to collect the infrared photons we exploit the fact that the long wavelength are emitted in a very large solid angle in contrast to the higher energy photons emitted in a more collimated beam. By letting the small angles ('I' = ± 5 mrad) pass freely and by collecting only the larger angles using aplane mirror, the line can be used in parallel with the SU3 beamline positioned downstream.

The infrared beamline is composed of three elements : the undulator (described previously), the optics system (presented in the following paragraph) and the spectroscopic assembly whose principle and description are summarised in the next section.

Undulator

407

Diamond window

141==f=r~~==~~~,-.- I.R. to experiment

UV and XR to experiment

Figure 5. Schematic showing principle for extraction of infrared photons for SIRLOIN, the IRSR Une on the Super-ACO storage ring of LURE. Wide-angle, low energy radiation produced by positrons in the undulator U is reflected vertically by plane mirrors , while the positrons and the hard part of the spectrum passes in between.

The SIRLOIN optics is schematically shown in figure 6. Its extracting mirror is assembled directly on one of the eight straight sections of the Super-ACO storage ring, which contains the variable-gap undulator SU3 (U). The extracting optics is actuaHy composed of two gold-coated, copper plane mirrors (MI. MI') placed at 45° with respect to the ring plane. The mirrors are water-cooled and are kept at a certain separation Ll in order to let positrons, X-rays, and UV radiation pass through. In such a way the IR beamline does not interfere with the existing ones, and in standard conditions the beam lifetime is not appreciably affected.

In order to coHect a large fraction of the emitted radiation, the gap Ll has to be as small as possible without affecting the normal working conditions of the ring. Moreover, it is desirable to set the mirror separation Ll at its maximum value during the injection of positrons into the ring, when the positron beam is not weH coHimated. The mirror displacements are performed by a remote-controHed device. It is designed to move either mirror horizontally while keeping the two parallel within a few milliradians. Tests have shown that the gap can be as smaH as 20 mm without affecting the normal working conditions of the ring. Infrared and visible radiation coHected is then focused onto the type 2A natural diamond window by cylindrical mirrors M2 and M3 which have their axes in the horizontal and vertical planes, respectively. The window is 200 11m thick, has a usable diameter of 10 mm and is wedged. The window separates the ultrahigh vacuum zone from the remainder of the line, which is equipped with plane (MS, M6) and toroidal (M4, M7) mirrors. This section is kept at "" 10-6 torr in order to strongly reduce the infrared absorption due to water and C02. The line ends in correspondence with the emission port of a rapid scanning Bomem DA8 3X interferometer. The centre of the undulator (conside­red as the source point) is focused on the beamsplitter while the MI mirrors (considered as coHimators) are focused onto a 4 mm aperture. From there on, synchrotron radiation follows exactly the same optical path as the interna! sources of the interferometer.

408

Figure 6. Schematic view 0/ SIRLOIN, the IRSR line on the Super-ACO storage ring 0/ LURE. The beam is /ocused by two cylindrical mirrors on the diamond window It is brought to and matched to the entrance optics 0/ the interferometer by aseries 0/ Jour mirrors The window has an aperture 0/ 1 cm, is wedged and separates the ultrahigh­vacuum section 0/ the line from the lower vacuum 0/ the interferometer.

2. FTIR spectroscopy and the experimental set-up

Vibrational spectroscopy is a weH established technique for studying the structure of moleeules as weH as determining the composition of an unknown material. The two principal types of vibrational spectroscopy Raman and infrared measure molecular vibration al frequencies and differ mostly in their sensitivity to different types of vibrations. While infrared responds to any transition modifying its electric dipole moment, Raman is sensitive to transitions involving a change in electronic polarisability.

The flrst element of an IR spectrometer is a broadband source which is commonly a quartz lamp (visible and near infrared), glowbar (in the mid and far infrared), mercury lamp (in the far infrared) or synchrotron radiation (covering the total spectral range). The second element is the optics assembly used for coHecting infrared light and directing it

409

through a sampie. In conventional dispersive spectrometers this function is accomplished by spreading the light spatially using a dispersive grating and then selecting a narrow range of energy with a slit. Collection of a spectrum in a broad energy range will require several measurements of the absorption of this quasi-monochromatic light. Tbe alternative approach employed in the infrared is to pass the broadband infrared beam through an interferometer before sending it through the sampie to a high sensitivity detector.

2.1. THE BASIC PRINCIPLES OF FTIR

All interferometers exploit the ability of a photon beam separated into different path to interfere constructively or destructively depending on their wavelength. This interference pattern contains information on the spectra1 distribution. Tbe most common interferometer in the infrared uses the Michelson principle. It basically consists in dividing the amplitude of the incoming radiation in two equivalent beams using a beamsplitter. Bach beam trajectory comprises one reflection on the beamsplitter and one passage through it as weH as one ret1ection on a "perfect" mirror. Tbe trajectory in each "arm" are therefore exactly equivalent apart from the fact that one of the mirror is mobile while the other one is fixed. Tbe recombination of the two beams occurs at the detector. When the path length traverse by light in the two arms are eqUal both beams combine constructively independently of the wavelength. This position of the moving mirror is referred to as the zero path difference or ZPD. Moreover, as a function of this position, the interference of a given wavelength will be successively constructive and destructive, with the frequency of the modulation depending on the wavelength. Tbe frequency of the (cosine) function of the intensity versus the optiCal pathlength difference is the reciprocal of the associated wavelength (i.e. the wavenumber : V). Tberefore the function describing the total intensity versus the optical path difference (interferogram) is a superposition of cosine functions and the spectrum itself (intensity vs. wavelength) can be obtained by a mathematical operation: the Fourier transform of the interferogram. Tbe result of this operation is a function which is referred to as the single beam spectrum. For low absorbance sampies such as electrochemical interfaces it is not directly interpretable. Instead one usually presents the corrected spectrum calculated from dividing the single beam spectrum by a reference spectrum.

Interferometry has a number of advantages over dispersive spectroscopy, the main one. being the multiplex advantage i.e. the simultaneous detection of the complete spectral distribution. Another important advantage is the result of the large solid angle that the absence of entrance and exit slits allows. Moreover, based on the simultaneous use of a wavelength standard (usually an He-Ne laser) the position of the moving mirror is known with aprecision far more accurate than what is obtained with a dispersive instrument. Finally the only signal that will contribute to the spectrum is the modulated signal which me ans that any background signal will not contribute to the spectrum after the Fourier transformation. This property is especially important in the far infrared where the 300 K ambient intensity contributes strongly to the total intensity. All these advantages as weIl as the theory of Fourier Transform spectroscopy are described in more detail by Griffiths and co-workers9,lO.

2.2. THE SIRLOIN SPECTROSCOPIC SYSTEM

As mentioned earlier the SIRLOIN beamline ends at the emission port of a Bomem DAS 3X interferometer. At present the projects under development in the far infrared region of the spectrum are listed figure 7.

410

10000 1000

10

RESEARCH THENES

Wavelengch (JUTI) 100

100

~MIf1Y (=-1)

10

1000 10000

Figure 7. The six main scientijic axis developed on the far infrared beamline SIRLOIN.

Clearly, the spectroscopic system needs to be very versatile in order to accommodate many different experimental set-up including electrochemistry cello The modified Bomem DA8 interferometer has the following optical features :

1. The sampie compartment accommodates different optical setups adapted for the various experiments. Two plane mirrors with fine alignment are used for displacing the focusing point from the centre of the sampie compartment to the electrochemical interface.

2. The selection of the source (synchrotron radiation, glowbar, quartz, mercury lamp), the iris aperture (from 0.5 to 10 mm), the filter, the output be am port (five possible positions) and the vacuum operation are computer controlled.

3. Various detectors encompass the wide range of conditions. The best detector yielding highest signal to noise ratio is the liquid helium-cooled bolometer for the region 80 -700 cm- l and pumped liquid helium-cooled bolometer for the lower energy range. The DTGS detector used in the mid and far infrared is much easier to use but is less sensitive by orders of magnitude.

4. Different beamsplitters are necessary to cover the complete infrared energy range: KBr in the mid infrared region and different thicknesses of mylar pellicules in the far infrared.

5. The entire interferometer including the beam switching compartment, the sampie compartment and the detector housing can be evacuated to a vacuum better than 10-3 torr by turbomolecular pumping and cold trapping. This vacuum allows a direct opening of the interferometer on the beamline pumped at 10-7 without Use of windows. This differential pumping reduces interference from very strong water vapour lines as purging with dry nitrogen is much less effective in the far infrared region.

411

6. The spectrometer system can achieve a resolving power of a 1 in 1()6. Analysis of electrochemistry sampies near ambient temperatures with band structure narrower than 2 cm-l is not expected.

3. The electrochemical assembly

3.1. IN SITU ELECTROCHEMIS1RY EXPERIMENTAL REQUIREMENTS

The main goal of infrared absorption study is to see how the structure of the species is affected by a change ofpotentialll,12. These changes include reconstruction ofthe surface itself, adsorption of ions, solvent molecules or solvent multimers, etc... All these absorbates are present with densities typical of a surface and therefore the associated signal is weak compared with other absorption levels. Superimposed on this weak signal are strong sources of absorption. The first is the solvent absorption of the radiation, the second is the absorption by the window of the cell, the third is the absorption due to the residual gas and the fourth is the absorption due to the bulk of the electrode. Other important causes of lowering the signal-to-noise ratio are the direct reflection on the surface of the cell window as well as the multiple reflection caused by its two parallel faces. Although these absorption will always be present there are partial solutions to reduce them to a working level in many practical cases.

The passage of the infrared beam through a 10 ~ layer of water causes absorption varying from 50% (around 100 cm-l ) to 99% (around 300 cm-l). Considering that a typical cell will have a distance of about 50 ~ between the window and the electrode, passage of the IR beam at 45° imply that these cells will not work in the regions where water presents strong resonances and therefore the use of a non aqueous solvent will sometimes be necessary. Other solutions are ex-situ measurements which is not the subject of this study and the use of a layer of solvent as thin as a few microns possible with precision machining. Such thin layer of water may reduce the absorption to a working level, höwever as for ex-situ measurements one drawback is that in such situation the control of the electrode potential becomes very approximate. The two passes of the infrared beam through the window is a very strong constraint and the choice of window material, its thickness and the reflection geometry is critical. Indeed the material has to be transparent in the region of interest, has to support the vacuum of the sampIe compartment on one side and the pressure of the electrolyte solution on the other side without leaking. Its size and its refractive index should allow grazing (or pseudo grazing) incidence and its shape and orientation should prevent the reflection and multiple reflections to reach the detector. Finally most electrochemists need to keep it at a reasonable price. To our knowledge there are no perfect solutions. Minimizing the signal due to direct reflection on the surface of the window can be achieved by setting its surface non parallel to the surface of the electrode; however this will also substantially increase the layer of solvent and therefore is only possible in regions where solvent is minimal. Finally a window with a wedge will prevent multiple reflections but it is only possible in regions where the window material does not absorb substantially.

A voiding absorption by the residual gas is mandatory to achieve signal to noise ratlos in the 103 range (.ARIR=lQ-4). The absorption in a nitrogen purged sampIe compartment also prevents signal to noise larger than 1 Q3. A careful insulation of the electrochemical cell is possible and this, combined with strong pumping of the sampIe compartment of the

412

interferometer (vacuum better than 10-3 torr) will prevent any absorption due to 1eaks coming from the cell itself or due to the residual gas. Finally the absorption caused by the bulk: of the electrode itself can be reduced by using grazing incidence and polarised light. The com bination of all these sources of undesired absorption on the top of the weak: signal due to the electrolyte interface imposes a very stringent requirement in order to achieve a high signal to noise ratio. The procedure to achieve stability which would allow measurements will be described in a following section after abrief description of the electrochemical cell used in conjunction with the interferometer.

3.2 ELECrROCHEMICAL CELL AND ADAPTING OPTICS

The various constraints described in the previous seetion have been considered in designing a cell that allows measurements of the optical reflectivity of the electrochemical interface at various potentials. The main requirement is the possibility of achieving simultaneous measurements of the infrared absorption with classical voltammetry without any need for removing the electrode from the working position.

pu~

... c: ., E 1:: ..

atmosphere

~ --------------~~~1 o <> ~

~

,_ solution input I chame! tor t I>-S;"~ reference ~ iS' i ~ ~. ~

worl<iI1g .. ectrOlle

Cl. Q

~~~------------~~~ Q

Figure 8. Vertical sectional view 0/ the inside 0/ the sam pie compartment containing the optical set and the hermetically sealed celL

413

The electrochemical cell is made of Kel-F and equipped with a 400 JlIn thick Silicon window (figure 8). This thin window has good transmittance throughout the whole infrared range and is available in !arge sizes. Its main drawback is that it is not transparent in the visible range and moreover, it tends to deform preventing a uniformly thick solvent layer. In the worse cases, this deformation may cause leaking. To prevent this, the window has to be held firmly by a rigid frame. The gap between the optical window and the working electrode surface can be adjusted from about 5 mm to contact condition. The electrolyte solution can be transferred to the thin layer compartment of the cell from a supply flask through a Teflon tube by positive Ar pressure. The potential of the worlcing electrode is measured relative to a reference electrode. Bright ring platinum wire used as counter electrode was placed behind the thin layer compartment. In the experiment presented in the next section, a smooth well-defmed platinum polycrystalline plate was used as working electrode. Its area is approximately 1 cm2. A AglO.OlM AgN03 reference electrode was used, and all potentials reported here are referred to this reference. Platinum plate was used as counter electrode. The electrolyte solution was prepared from reagent grade chemieals and Millipore-Q purified water under pure Ar atmosphere. The solution was deoxygenated with Ar prior to use. The hermetically sealed electrochemical cell is precisely positioned in the stainless steel sampie eompartment of the interferometer with a piereed aluminium plate equipped with a Viton seal. The modulated beam from the interferometer is focused in the sampie compartment with f/4 optics. The use of a flat mirror (adjustable in three directions and two translations) moves the focusing point onto the electrode surface with an angle of incidence of approximately 60°. A second flat adjustable mirror brings baek the beam in the horizontal direetion. The optical path is eompleted with the focusing optics of the detector.

3.3. DATA ACQUISmON AND ANALYSIS

Fourier Transform InfraRed speetroseopy ean be a powerful tool for studying the orientation and dynamics of speeies at the electroehemieal interface provided that it has suffieient sensitivity to alterations at the surfaee. To aehieve the highest sensitivity the difference speetroseopy in which the sampie absorption spectrum is normalised by a referenee speetrum is required. It is also mandatory to use the most appropriate beamsplitter, detector and ftlter. For the data presented in the next section, 12 JlIn and 25 JlIn mylar films as weIl as a KBr beamsplitter were used for aecessing frequencies ranging from 50 to 220 ern-I, 30 to 130 cm-l and 1900 to 2300 ern-I, respeetively. The resolution was 4 ern-I. For reversible systems (not described here), a new reference was taken after eaeh speetrum. This proeedure allows better signal to noise ratios as it compensates for variation in the intensity of the source. However this method is not adapted for non reversible system sueh as the measurements deseribed here. In this ease each spectrum eolleeted at a given potential of the working eleetrode was obtained by averaging 400 interferometer scans. For each of them a referenee speetrum was also measured. The differenee speetrum was obtained by dividing the reflectance speetra eolleeted at these two different potentials (eq. 1) .

.clR. _ RsampIe- Rreference

R - Rreference (1)

Following this procedure we obtained noise to signal ratios smaller than 0.005 % at 60 ern-I. This noise level is eomparable with that of ordinary IRRAS (InfraRed Absorption Refleetion Speetroseopy) system in UHV and makes surface analysis possible.

414

4. CN- adsorbed on Pt polycrystalline surface

As explained previously the detection of vibrations in the far-infrared region is difficult as it requires large signal to noise ratios. Therefore choosing strongly adsorbed species with relatively large effective charges is required as a test system to be investigated. In this seetion we discuss CN- adsorption on a platinum electrode. The basic electrochemical behaviour of the system was first investigated by cyclic voltammetry and voltammograms were taken continuously during the experiment to verify the cell conditions. This procedure is described in section 4.1, the infrared measurements and analysis in section 4.2 and those of the far infrared in section 4.3, respectively.

(a) f

(b)

-1.0 -0.5 o 0.5 1.0,

POTENTIAL I V vs. Ag/Ag+

Figure 9. Cyclic voltammograms 0/ platinum electrode in 0.3 M NaCl04 and 20 mM KCN. Scan rate is 50 mV/sec, a) vertical scale 1 j.JA/division, b) vertical scale 2 j.JA/division.

4.1. ELECTROCHEMICAL CLASSICAL MEASUREMENTS

The degassed electrolyte solution of 0.3 M NaCI04 and 20 mM KCN was transferred to the cell equipped with a polycrystalline Pt electrode. The voltarnmogram of the Pt electrode is shown in figure 9. In this experiment the negative potential limit was set in

415

such a way that a further decrease would have produced hydrogen evolution. Positive potential limit was varied. Only double layer charging current is observed as the positive potential limit is below 0 V. It indicates that hydrogen and oxygen adsorption is prevented due to the CN- layer on the surface. Positive potential excursion above 0 V results in an increase of oxidation current and also of reduction current peaked around -0.5 V in cathodic scanning. This current reflects oxygen adsorption-desorption behaviour on the Pt surface after remove of the adsorbed CN- ion. These results suggest that the Pt surface is covered with CN- as far as the electrode potential is kept between -0.9 V and 0 V. After the scanning or holding the potential more positive than 0 V, CN- should be oxidized and the amount of oxygen adsorption should increase. It is expected that such a behaviour of the adsorbed species on the surface will be manifested in the vibrational features of the infrared spectrum.

4.2. MID INFRARED ABSORPTION MEASUREMENTS

For this set of measurements each spectrum in the mid infrared region was collected by averaging 400 interferometer seans with aresolution of 4 cm-l while the potential of working electrode was kept constant. The reference absorption spectrum was measured at -0.9 V. All absorption spectra were obtained after bringing the potential from -0.9 V to the potential of the measurements. This procedure corresponds to the voltammograms presented figure 9. The measurements have been performed at potentials varying from 0 V to 1.0 V by steps of 0.1 V. These cycles illustrate the non reversibility of the system. The difference spectrum was then obtained by ratioing the reflectance at the two potentials as in equation (1).

Measurements of reference spectra on the platinum surface in 0.3 M NaCI04 and 20 mM KCN were carried out in the wavenumber region extending between 1900 cm-l and 2300 cm-l , and showed a pronounced dip in the region between 2140 cm-l and 2200 cm1 (see figure 10). Clearly, as the electrode potential increases from 0 V to 0.4 V, the band at 2170 cm-1 grows. On the other hand, increasing the potential from 0.4 V to 1.0 V causes the structure to disappear. Moreover, a more pronounced structure at the same wavenumber is present on the spectrum at -0.9 V, measured after keeping the electrode at 1.0 V. The peak position does not depend on the potential significantly. The frequency of the peak is assigned to CN- vibration and as it is potential dependant it corresponds to CN­adsorbed on platinum surface. The integrated absorption intensity in the potential around 0.4 V and at -0.9 V reflects the number of adsorbed CN- ions on the surface. The integrated intensity as a function of electrode potential is shown in figure 11. It illustrates quantitatively that the number of adsorbed CN- ions on surface increases from 0 V up to 0.4 V, and then decreases and disappears around 1.0 V, assuming that transition moment of CN- adsorbed does not change with potential.

Figure 11 also shows the potential dependence of the total charge to reduce oxygen adsorbed on platinum surface, which was calculated from the integration of reduction peaks in the voltammogram presented in figure 9. This curve illustrates that the quantity of charge starts to increase at approximately 0.3 V and reaches 223 JlC cm-2 at 1.0 V. Note that the charge necessary to reduce monolayer adsorbed oxygen on platinum is 220 mC cm-2. The correspondence between the position of the maximum coverage of adsorbed CN- and the threshold for the increase of charge to reduce adsorbed oxygen confirrns that CN- is removed by the adsorption of oxygen. Moreover, the potential dependence of the charge reflects the total number of adsorbed CN- on the surface. The

416

amount of CN- begins to decrease at 0.4 V and is at half intensity around 0.6 V. At 1.0 V the adsorbed CN- seems to be almost completely removed.

10.02 % Pt I--C---N IRel. = -0.9 vi

~ -J, 1.0 V

0.9 V

0.8 V

0.7 V

0.6 V

0.5 V

0.4 V

CI: 0.3 V -CI:

<I

0.2 V

0.1 V

0.0 V

-_ ... __ .-.... - ...... _-~"

.... -0.9 V ".

(after 1 cycle) \ \

•....

2140 2150 2160 2170

Wavenumber I cm"

Figure 10. Absorption-rejlection spectrum in the mid infrared region.

At positive potential, surface oxidation also proceeds. The adsorbed oxygen should displace adsorbed CN- from the surface. The decrease in the number of adsorbed CN- is indicated by the decrease of the integrated intensity of the spectrum. At 1.0 V, CN- ions

417

are totally removed. When the electrode potential is brought back to -0.9 Vagain, surface oxygen desorbes and CN- ions are readsorbed. The number of CN- ions bonding under this condition is larger than that at 0.4 V. This result suggests that after CN- sites have been formed, their readsorption is greatly facilitated and full coverage of CN- occurs at 10wer potential.

40xl0· 3 ..-----.-----r-----r-----,----/-.... -;[1 ..

Ia-f I ptl

30

i. /'

20 j/

!{.

// 10

.//

l· 0.0 0.2 0.4 0.6 0.8 1.0

Potential I V vs. Ag/Ag +

200 ~

" :T Dl ca ..

Figure 11. Potential dependence 0/ integrated intensity 0/ CN- absorption band in the mid infrared (-... ) and reduction cu"ent to adsorbed oxygen (-Li-).

4.3. FAR INFRARED ABSORPTION MEASUREMENTS

Measurements of reference spectra in the far infrared region were carried out by using the same system. Figure 12 shows typical spectra measured at three potentials. Absorption structure peaks positioned around 270 cm-1 is observed for potentials between 0 V and 0.6 V. Absorption intensity is maximum at 0.4 V. The coincidence of the potential dependence of the reference spectrum in mid and far infrared region supports the hypothesis that the adsorption structure observed at 270 cm-1 in the present system should be due to vibration between platinum and C of the CN- ion. The vibration between CN­and a silver surface was observed by SERS at 220 cm-1 approximately13. The difference in the band position in this system should reflect characteristics of bonding between platinum meta! and CN- in the electrolyte solution.

418

The doublet structure ("peak and dip") observed between 240 cm-1 and 320 cm-1 may be the signature of the two different adsorption states : N bound and C bound. It seems that as potential is decreased the structure moves from predominant C adsorption to a situation where both C- end and N- end adsorption coexist. This is supported by Sum Frequency Generation studies14 of adsorbed CN- on Pt in the mid IR.

er er <I

la'f / ptl IAel. = -0.9 vi

Pt L._C- N

r,t 0.9 V

-0.2 V

180 200 220 240 260 280 300 320 340

Wavenumber I cm- 1

Figure 12. Absorption-rejlection spectra in the far infrared region.

419

Conclusion and future prospects

A novel synchrotron infrared spectroelectrochemica1 system developed for the study of adsorption at the electrochemical interface is now completed at LURE. This facility allows simultaneous electrochemical measurements as well as infrared reflectivity of the electrode interface. It has provided new information on the interaction between surface and adsorbed moleeules or ions and complements information obtained by elassical electrochemical measurements.

Future experimental improvements include arevision of the photon extraction optics and mirror fme adjustments as well as a feedback system on the position of the positron beam aimed at improving the signal to noise ratio. The spectral system available now in the mid to far infrared range will be extended up to visible wavelengths. Once the technique has been applied to some model systems such as the CN--on-platinum described here, we plan to extend it to the study of more complex reactions, e.g., in catalysis where it could improve the understanding of the role of adsorption of specific species on the inhibition or acceleration of electrochemical reactions.

Acknowledgements

The development of the infrared beamline was only possible in elose collaboration with the technical staff of the LURE facility ; the authors are sincerely grateful to them. We also want to express our gratitude to the Director of LURE who always showed great support and encouragement Mr. Flammang (L.E.I. Meudon) and Joel Jaffre (CPMA Orsay) are acknowledged for their contrlbution to the construction of the electrochemical cell and for the development of the optical reflectivity setup. We also thank P. Calvani, S. Lupi, B. Tremblay, M. Costa and A. Gerschel for stimulating discussions and assistance in data acquisition. This work was supported in part by the Chemistry Department of CNRS, the Ministry of Research and Technology and the Institut de Physique MoIeculaire at Universite d'Orsay. One of us (KM.) wishes to thank the CNRS for a senior visiting fellowship at LURE during the period 1992-1993.

References

1. G.P. Williams, Nuc!. Instr. and Meth. A291, 8 (1990). 2. T. Nanba. J. Yarwood, T. Shuttleworth, and J.E. Hasted, Int. J.lnfrared &

MM Waves 7, 759 (1986). 3. M. Hara, S. H. Be, I. Takeshita and T. Nanba, Rev. Sei. Instrum. 63 (1), 1543

(1992). 4. D. A. Slater, P. Hollins, M.A. chesters, J. Pritchard, D.H. Martin, M. Surman,

D.A. Shaw, I. Munro, Rev. Sei. Instrum. 63 (1), 1547 (1992). 5. Max Lab activity report 1991, National Laboratory Lund, Sweden. 6. P. Roy, Y.L. Mathis, A. Gerschel, J.P. Marx, J. Michaut, B. Lagarde and

P. Calvani, Nuc!. Instr. and Meth. A325, 568 (1993). 7. B. Lai and F. Cerrina. Nuc!. Instr. and Meth., A 246 (1986) 337 8. M. Barthes, C. Bazin, M.E. Couprie, A. Dael, C. Evesque and C. Humbert,

Int. Rep. Super-ACO/87-63,1987.

420

9. P.R. Griffiths, "Chernical Infrared Fourier Transform Spectroscopy", Chern. Anal. Ser. Monogr. Anal. Chern. Appl.~, (1975)

10. P.R. Griffiths, J.A. de Haseth "Fourier Transform Infrared Spectrornetry", Chern. Anal.Sec. Monogr. Anal. Chern. Appl..8.3., (1986)

11. K. Ashley, Spectroscopy 5., (1990) 22 12. K. Ashley, S. Paris, Chern. Rev . .8..8., (1988), 673 13. P. Koetz and E. Yeager, 1. Electroanal. Chern. m (1981),335 14. A. Tadjeddine and P. Guyot-Sionnest, Electrochirnica Acta ~ (11-12), (1991),

1849

Far-Infrared Synchrotron Radiation and the Electrochemical Interface

Andrea E. Russen

Chemistry Department, University of Liverpool, P.O. Box 147, Liverpool L69 3BX, U.K.

William O'Grady

Code 6170, Naval Research Laboratory, Washington, D.C. 20375-5000, U.S.A.

Abstract: The advantagesand application of synchrotron radiation in the far-infrared for the in situ study of the electrode/electrolyte interface are described. Results are presented for studies of Ag(lll)/aqueous electrolyte systems. In 0.1 mol dm-3 HCI04, vibrational bands were observed which may be attributed to the Ag-Cl stretching mode of low coverage adsorbed chloride, a product of perchlorate reduction. In 0.1 mol dm-3 NaF, bands attributed to Ag-O vibrations of bound water below the pzc are observed. The results are discussed in relation to the development of a picture of the effects of adsorbed anions on the interfacial structure of water.

1.0 Introduction

In the last two decades developments in in situ spectroscopic techniques have contributed significantly to improvements in the understanding of the microscopic structure of the electrode/electrolyte solution interface. The union of such in situ techniques and synchrotron radiation has created yet more opportunity for advancement, as documented by this workshop. Of the spectral regions accessible with synchrotron radiation, the far-infrared is possibly the least wen utilized. However, in situ far-infrared studies provide the ability to directly probe the molecular characteristics of the interface and the structure of interfacial water.

The infrared region may be conveniently divided in to three sections which each probe different aspects of the interfacial structure; far- «75Ocm-I ), mid- (75O-4OOOcm- I ), and near-infrared (>4OOOcm -1). Of the infrared regions, the mid-infrared has proved the most fruitful. Using external reflectance techniques, mid-infrared studies have provided information regarding the identity and orientation of adsorbed molecules, and evidence of the strength of the electric field at the electrochemical interface, the electrochemical Stark effect (1). The vibrations ofthe metal-adsorbate bond, however, are oflow frequency, in the far-infrared « 750 ern-I). Thus, to obtain direct information on the metal-adsorbate bond and to probe the electric field strength within the first 1-2 A of the metal surface in situ far-infrared studies are required.

421

C. A. Melendres and A. Tadjeddine (eds.), Synchrotron Techniques in Interfacial Electrochemistry 421-431. © 1994 Kluwer Academic Publishers.

422

Probably the most weH studied adsorbate system is carbon monoxide adsorbed on a platinum electrode surface. In the mid-infrared, the v(CsO) stretching vibration is probed. The frequency of this vibration is sensitive to the local environment surrounding the CO molecule; adsorption site, solvent, and applied electrode potential. Thus, CO has been used to probe the environment of the electrochemical interface at a distance of ~2A from the meta! surface. In the far-infrared, the v(Pt-C) stretching vibration would be studied .. This will allow the strength of the electric field and other aspects of the chemical environment within the first few angstroms of the metal surface to be probed. Such information has not been available from other techniques.

1.1 Synchrotron Sources

Reflectance measurements, such as those used in many in situ electrochemical studies, are brightness limited. Brightness is defined by the etendue, the conserved product of the image or source size and beam diver~ence. Conventional thermal far­infrared sources have an etendue on the order of 1 cm sr. This may be compared to the throughput limiting element of the in situ infrared reflectance experiment, the reflection off the electrode, which has an etendue on the order of 10-4 cm2 sr. The mismatch between these two values means that the efficiency of the experiment is greatly reduced and, thus, the acquisition time to obtain a spectrum of an adequate signal to noise ratio becomes on the order of an hour. In the far-infrared region synchrotron radiation is between 100 and 1000 times brighter than standard black body sources (2,3). The match between the light in the far-infrared from a synchrotron source such as the far-infrared line at the National Synchrotron Light Source (NSLS) and in situ reflectance measurements results in a coupling efficiency of 1. This may be compared to a coupling efficiency of 8xlQ- 4 obtained for a 2000 °K black body source. Thus, with far-infrared radiation from asynchrotron source in situ reflectance studies of the electrode/electrolyte interface become possible.

2.0 Applications of Far-Infrared Synchrotron Radiation

In the following sections two examples of the use of far-infrared synchrotron radiation as a source for the study of the Ag( 111) electrode/aqueous electrolyte interface are presented.

2.1 IR Reflectance Experiments

The in situ spectroelectrochemical measurements described below were conducted using synchrotron radiation in the spectral range from 25 to 800 cm- I on the infrared beam line, U4-IR, at the National Synchrotron Light Source. The vacuum of the vacuum­ultraviolet storage ring was separated from that of the interferometer, a Nicolet 20F vacuum bench, by a diamond window. High density polyethylene windows mounted on the sampie chamber, electrochemical cell, and the window of the liquid helium-cooled bolometer, limited the measurements to below 370 cm-I .

As in mid-infrared SNIFTIRS (subtractively normalized interfacial Fourier transform infrared spectroscopy) or PDIRS (potential difference infrared spectroscopy) (4) experiments, a thin layer cell was used to reduce the amount of infrared radiation loss due to solvent absorption. The cell design closely followed that reported by Roe et al. (5, 6) with the replacement of the calcium fluoride window by a flat window made from 0.03 inch thick high density polyethylene. The palladium/hydrogen reference electrode,

423

chosen to minimize chloride contamination, and platinum wire counter electrodes were brought in through ports in the cell body. (Potentials reported have been converted to the SCE scale). The silver single crystal working electrode was oriented by Laue x-ray back diffraction, cut and then polished with diamond suspensions and alumina down to 0.3 ttm to obtain a mirror finish. Prior to each experiment the crystal was chemically polished using a chromium trioxidelhydrochloric acid etch procedure (7) followed by washing in concentrated sulfuric acid and thoroughly rinsing with triply distilled water (8). The crystal was then mounted in a Kel-F holder (9) and placed into the filled cello The electrolyte had been de-aerated for 15 minutes with argon in the cell and argon flow through the cell was sustained during introduction of the working electrode to the cello

Individual spectra were collected by poising the electrode at the potential of interest and collecting 1024 interferometer scans with a mirror velocity of 1 cm sec- l and a resolution of 6 cm-1. Each measurement took approximately 5 minutes. The SNIFTIRS difference spectrum was then obtained by ratioing the reflectance at two potentials as folIows:

AR R= [1]

Because of this sign convention, bands pointing down are due to species which are present in a greater concentration at the sampie potential, R2, while those pointing up are due to species at the base potential, R 1. The reflectances used in the calculation above were obtained by dividing the raw reflectance spectra collected at each potential by the ring current of the VUV storage ring. This procedure serves to normalize the data for the decay in source intensity with time.

2.2 Perchlorate Reduction on Ag( 111)

The interactions of ions with the electrode surface play an important role in determining the structure of the electrochemical double layer. In situ far-infrared studies provide the opportunity to probe the metal-specific adsorbed ion vibrations. These vibrational modes have previously been investigated using surface enhanced Raman scattering (SERS). However, to observe the SERS effect the electrode surface must be roughened, often in a chloride electrolyte solution. Using synchrotron radiation and the SNIFTIRS technique, the adsorption of ions on the electrode surface may be studied on flat single crystal electrodes.

Dur first studies of ion adsorption were of perchlorate and chloride ions on Ag( 111) (10). The spectra shown in figures 1 to 3 were obtained for a Ag(l11) single crystal electrode in 0.1 mol dm-3 HCI04. The reference spectrum, at -O.3V vs. SCE, was collected at the beginning of the potential sequence. Sampie spectra were then collected during several potential excursions from -0.3 to -0.65 and back to -0.2 V vs. SCE. The resulting difference spectra were then generated. In the figures the spectra have been staggered to allow the differences between them to be observed. The regular, sharp; narrow features in the spectra may be attributed to fluctuations in the water vapour in the path of the IR radiation.

The band centered at 190 cm-1 in figures 1 through 3 has been attributed to the formation of adsorbed chloride from the slow reduction of perchlorate in the hydrogen

424

evolution region by an ec mecbanism. Cyclic voltammograms obtained for a silver working electrode in 0.1 mol dm-3 HCI04 at various sweep rates are sbown in figure 4. Tbe voltammogram obtained at 50 mV sec-1 is in good agreement witb those in tbe literature (11), while at tbe slower sweep rates tbe current crosses over that ofthe forward sweep (towards negative potentials) during tbe reverse sweep (towards positive potentials). Similar behavior has been reported for slow sweep rates in percbloric acid at Ru (12), Ir (13), Pt (14, 15), and Rh (13, 16) electrodes and is attributed to the slow reduction of perchlorate ions in the hydrogen evolution region generating chloride at the electrode surface by an ec mecbanism. Tbe SNIFfIRS in situ reflectance method used in this study is rougbly equivalent to a 0.2 mV sec-1 sweep rate (0.1 V steps every 8 minutes).

I2XIq-J

75 125 175 225 275 325

wavenumbers (em·I )

Figure 1. Difference spectra fOT Ag(111) in 0.1 mol dm-3 HCl040btained during the first cathodic excursion. Reference potential is -O.3V and sampie potentials from top to bottom are -0.4, '(}.5, -0.6, and -0.65V vs. SCE.

I 2"Iq-J

75 125 175 225 275 325

wavenumbers (em·I )

Figure 2. As in figure 1. Obtained during the reverse excursion. Sampie potentials from top to bottom are -0.65, -0.6, -0.4, -0.2V VS. SCE.

I2x1t;3

75 125 175 225 275 325 wavenumbers (cm·1)

Figure 3. As in figure 1. Obtained after second cathodic excursion. Sampie potential is -0.65 V vs. SCE.

sOmY/s.e

2(}".Y/s.e

lOmY/s.e

-0.4 -0.3 -0.2 -0.1 0.0 E(J?Jls.SCE

Figure 4. Cyc1ic voltammograms for Ag in 0.1 mol dm-3 HCI04

425

The vibrational band is only observed to increase in intensity. as the potential approaches the hydrogen evolution region, fig. 1, and remains constant in intensity if the potential is retumed to more positive values, fig. 2. On a second excursion towards negative potentials, the band again increases in intensity, fig. 3. The sampie potential could not be made more negative to determine whether the adsorbed chloride desorbs below the pzc of Ag(1l1), -O.695V vs. SCE (17), because of the possibility of the generation of bubbles in the thin layer.

The frequency of the v(Ag-Cl) stretching vibration observed in this experiment, 190 ern-I, may then be compared to that obtained by various SERS studies (18-21). In the SERS experiments with the v(Ag-Cl) band ranges between 215 and 240 ern-I. The discrepancy between the SERS results and those obtained in the SNIFfIRS experiment may be attributed to the difference in coverage of adsorbed chloride in the two

426

experiments. The frequencies of the vibrations of an adsorbed moleeule are dependent upon the local environment surrounding the moleeule. Dipolar interactions between adjacent adsorbates can result in a shift in the vibrational frequencies of the adsorbed molecules. The dipole term arises from the dipole formed by the adsorbed ion and the corresponding image charge within the metal.

Numerous theoretical calculations of the chloride adsorbed on silver system have been conducted. These calculations have shown that the dipolar coupling between adjacent adsorbed chloride ions influences the vibrational frequency of the v(Ag-CI) mode. Illas et al. (22) calculated the vibrational frequency for the motion of one chloride ion perpendicular to a AgS cluster, a low coverage limit, by SCF ab initio methods. The adsorbed chloride ion was found to retain acharge of 0.75 e-. A vibrational frequency of 143 ern-I was obtained, but was not able to account for lateral dipolar interactions between neighboring adsorbed chloride ions. Nichols and Hexter (23) calculated v(Ag­Cl) for chloride adsorbed on Ag(I00) using a model which includes the dipolar coupling terms. At the limit of zero coverage they obtained a frequency of 159 ern-I. As the coverage of adsorbed chloride was increased the calculated vibrational frequency increased to a value of 241 cm- l at full coverage. In comparing their SCF ab initio calculation to the results of Nichols and Hexter's molecular dipole moment model, Illas et al. concluded that the effect of lateral interactions is responsible for the vibrational frequency dependence on the coverage. In a SEXAFS (surface extended x-ray absorption fine-structure) study, Lamble et al. (24) determined that the lateral Cl-CI nearest neighbor interactions for chlorine adsorbed on Ag(11l) do not become appreciable until chlorine coverages of 213 of a monolayer are reached.

The 190 cm- l band reported for adsorbed chloride species observed from the reduction of perchlorate in this in situ far-infrared study is, therefore, attributed to coverages below 2/3 of a monolayer. The -230 cm- l band observed in SERS experiments is most likely due to adsorbed chloride at coverages greater than 2/3 of a monolayer, where lateral interactions playa significant role. Isotopic substitution experiments and experiments with mixed electrolytes, «0.1 - x)mol dm-3 HCI04 + (x)mol dm-3 HCI) are in progress to confirm the conclusions ofthis preliminary study.

2.3 Water adsorption on Ag(III): The Structure ofthe Double Layer

In addition to providing the ability to probe the vibrations of the metal-adsorbate bond, in situ far-infrared measurements are able to probe the vibrational modes of water. In the far-infrared region, the vibrations of water are very sensitive to the phase, liquid or the various solid phases. The hindered translation, Vr, shifts from 193 ern-I to 229 ern-I, and the librational mode, VL, from 300-900 cm- l to 200 ern-I, on going from liquid water to ice (25). The metal-water modes of adsorbed water may also be probed.

The electrochemical double layer is thought to be made up of an oriented layer of solvent/water moleeules and specifically and non-specifically adsorbed ions. In the classical model of the interface, the orientation of adsorbed water moleeules is thought to change with applied potential; dipoles oriented towards the metal surface at potentials above the pzc (oxygen end down) and away from the metal at potentials below the pzc (hydrogen end down). More recently, molecular dynamics simulations have suggested that the water molecules in the first bilayer adopt an ordered structure which is similar to ice-I (26). This perturbation of the liquid water structure decays over a distance of -lnm, so that the next layer of water molecules displays a much more liquid like structure. In

427

situ infrared experiments, by directly probing the vibrations of adsorbed and unbound water in the double layer, should be able to determine the extent to which these theories are correct.

An experiment which probed the potential dependence of the structure of water at the electrode surface was the study of Ag(lll) in 0.1 mol dm-3 NaF with and without 0.001 mol dm-3 NaCl (27). At potentials above the pzc chloride ions are known to specifically adsorb on silver, while fluoride ions are only slightly specifically adsorbed (17). When an ion is specifically adsorbed it displaces water molecules from the metal surface. By comparing spectra obtained on either side of the pzc, the bonding of water to the metal electrode surface may then be probed.

The set of spectra shown in figures 5 and 6 illustrate the effect of making the potential more negative on the adsorrtion of water on Ag(lll) in 0.1 mol dm-3 NaF. One upward pointing band at 322 cm- is observed. The direction of the band indicates that the species responsible is present in a greater concentration/coverage at the reference potential, -1.0 V vs. SCE. The pzc of Ag( 111) is -0.695 V as stated above. Therefore, the band is unlikely to be associated with an anionic adsorbate. The band is assigned to the v(Ag-O) stretching vibration of adsorbedlbound water based on comparison to UHV EELS (electron energy loss spectroscopy) measurements of water adsorption on Ag(110) (27,28). In the EELS studies the v(Ag-O) stretching vibration of adsorbed H20 and OH were reported at 320 and 280 cm-1, respectively. As in the case of chloride adsorption described above, isotopic substitution experiments are in progress to confirm this band assignment. We are also extending the spectral region accessible in this experiment beyond 375cm-1 to minimize the influence of the decreased signal levels above 3OOcm-1

associated with the detector and window cut-off.

I I x10·Z

25 75 125 175 225 275 325

wavenumbers (em·l )

Figure 5. Difference spectra far Ag(lll) in 0.1 mol dm-3 NaF. Reference potential is -1.0 V vs. SeE. Sampie potentials as in figure.

As seen in figure 5, the intensity of the upward pointing band decreases as the sampie potential is made more negative until the sampie potential is -0.7 V vs. SCE and continuing down to -1.1 V. Below -0.7 V, the sampie and reference spectra are enough alike that the band at 322 cm-1 is barely visible. Upon retuming to more positive potentials, as in figure 6, the band at 322 cm-1 reappears at sampie potentials as negative

428

as -i.OV vs. seE. The intensity of the band then continues to increase as the sampIe potential is made increasingly positive. These results indicate that the coverage of water on the silver surface is greater at the reference potential, -1.0 V than at the sampIe potentials more positive than -0.7 V, which is approximately the pzc. The re-appearance of the band at sampIe potentials as negative as -1.0 V on the reverse sweep then indicates that the coverage of adsorbed water on the surface increases again after going to -1.1 V. This increased coverage may be attributed to the removal of residual more strongly bound adsorbed fluoride ions or other anionic impurities.

I lxIO·]

25 75 125 175 225 275 325

wavenumbers (cm·l )

Figure 6. As in figure S. Spectra collected in the order -0.5 V, -1.0 V (reference), -1.1 V, -1.0 V, -0.9 V vs. SeE.

We were unable to determine the identity of the species which displaces the adsorbed water moleeules at potentials positive of the pzc. If fluoride ions were specifically adsorbed at potentials above the pzc a downward pointing band for the v(Ag-F) stretching mode would be expected at 325 ern-I. The frequencies of the metai-haiide stretching mode for chloride, bromide, and iodide adsorbed on silver electrodes have been reported using SERS (21). If this series of vibrational frequencies is extrapolated to fluoride, by assuming that the force constant remains constant and using the mass of fluoride for the reduced mass, the predicted v(Ag-F) stretching frequency for adsorbed fluoride is 325 ern-I. This frequency is nearly coincident with the 322 cm-I band attributed to v(Ag-O) above. The resolution of the U4-IR beam line is only 6 cm-I at present, due to the configuration of the diamond window, and the width of the bands observed for vibrations of adsorbates in solution is considerably wider than the 3 cm- I difference. Therefore, we would not be able to resolve the v(Ag-O) and v(Ag-F) modes for adsorbed water and fluoride, respectively, if they were both present. However, these two vibrations are expected to show opposite trends and signs as the potential is made more negative. The data is limited to an upper limit of 350 ern-I. Thus, no definitive statement may be made regarding the presence or absence of a band which may be attributed to adsorbed fluoride.

The lower coverage of adsorbed water above the pzc may be a result of i) the displacement of water from the electrode surface by the adsorption of fluoride ions, or ii) by the breaking of the silver-water bond to allow solvation of fluoride ions brought nearer

429

to the surface, but not forming a bond (specifically adsorbed) with the silver surface. Additional spectral resolution will be required to distinguish between the two possibilities.

The observation of the v(Ag-O) stretching vibration of adsorbedlbound water in NaF solution indicates that the additional water molecules are bound to the electrode surface via the oxygen atom in this system, even at potentials negative of the pzc. This oxygen atom down-orientation at potentials negative of the pzc is opposite that predicted by electrostatics and intuition. The orientation of the O-H bonds of the adsorbed water molecules can not be determined from the observation of the v(Ag-O) mode alone. The surface selection rule for IR reflectance measurements at metal surfaces states, that a component of the dipole moment change associated with the vibrational mode must be perpendicular to the electrode surface for the mode to be observed. However, the v(Ag-0) stretching vibration of oxygen end down, adsorbed water will always have a component of the dipole moment change perpendicular to the metal surface, regardless of the orientation O-H bonds of the water molecule.

-0.8

25 75 125 175 225 275 325

wavenumbers (cm- l )

Figure 7. Difference spectm for Ag(III) in 0.1 mol dm-3 NaF plus 0.001 mol dm-3 NaCI.. Reference potential is -1.4 V vs_ SCE. Sampie potentials as in figure.

The effect of adding lxlO-3 mol dm-3 chloride ions to the sodium fluoride electrolyte solution is shown in figure 7. For a potential excursion from -0.5 V to -1.4 V vs. SCE no absorption bands are observed in the difference spectra. The ripples in the spectra may be attributed to the vibrations of water vapor in the spectrometer sampie chamber. The lack of any bands in the spectra indicates that no change occurs over the potential region investigated. It was expected that any adsorbed chloride ions would desorb below the pzc, and by using -1.4 V as the reference potential, a downward pointing band for the v(Ag-CI) stretching vibration of adsorbed chloride is expected at 214--24Ocm-1 (18--21). A conclusion which may be drawn from the spectra in figure 7 is that in a solution containing 10-3 mol dm-3 chloride, the chloride remains adsorbed on the electrode surface at potentials as far negative as -1.4 V. Perhaps the electrode surface has been converted to silver chloride as opposed to specifically adsorbed chloride ions in the thin layer cell used for these experiments. At greater concentrations of chloride ions the electrode

430

surface becomes dull and discolored. Similar spectra are obtained in chloride contaminated 0.1 mol dm -3 NaCI04 and HCI04 electrolyte solutions.

3.0 General Conclusions and Future Developments

The results presented in this paper represent the first applications of synchrotron radiation in the far-infrared to the in situ study of the electrode/electrolyte interface. The preliminary experiments described have provided new information regarding the adsorption of anions and the interactions of water with the electrode surface.

In the experiments conducted using Ag(111) electrodes so far, no absorption bands have been observed which may be attributed to the frustrated translation of water in either a liquid or ice-like state. Because the SNIFTIRS spectra are difference spectra, the lack of absorption bands only indicates that no change occurred. If there were a potential dependence to the formation of an ice-like structure at the metal-solution interface, an absorption band would be expected at ~250 ern-I, as has been observed in UHV studies (30). Silver does not interact with water as strongly as platinum, the metal commonly used in the molecular dynamics simulations which have provided the indication that water may form an ice-like layer at the metal surface, and, therefore, may not induce as much order in the adjacent water layers. Experiments on Pt( 111) may indeed prove to be much different.

Access to the far-infrared region of the spectrum provided by the use of synchrotron sources should continue to provide much exciting new information regarding the structure of the innermost regions of the electrochemical interface.

4.0 Acknowledgements

The work discussed was supported by the Office of Naval Research, Washington. Work at the U4-IR beam line was supported by the U.S. Department of Energy under contract DE-AC02-76CHOOOI6. Experiments were conducted with the assistance of Andrew S. Lin, Gwyn P. Williams, and Carol Hirschmugl. Figures 1 through 4 are reprinted with the kind permission of the Royal Society of Chemistry, and figures 5 through 7 with perrnission of Elsevier Sequoia, S.A.

5.0 References

1. See for example the recent review: K. Ashley, Spectroscopy, 5 (1990) 22. 2. G.P. Williams, Nucl. Instr. Meth. Phys. Res., A291 (1990) 8. 3. G.P. Williams, P.Z. Tackas, R.W. Klaffy, M. Shleifer, Nuc. Inst. and Meth. A, 246

(1986) 165. 4. For an overview of in situ infrared techniques see: S.M. Stole, D.D. Popenoe,

M.D. Porter, in "Electroehemical Interfaces", H.D. Abruna, Ed., VCH Publishers, Ine., N.Y., 1991, pages. 341-412.

5. D.K. Roe, J.K. Sass, D.S. Bethune, A.C. Luntz, J. Electroanal. Chem., 216 (1987) 293.

6. D.S. Bethune, A.C. Luntz, J.K. Sass, D.K. Roe, Surf. Sci., 197 (1988) 44. 7. T. Kurasawa, Patent Japan, 35-5619,23 May 1960. 8. A. Hamelin, L. Stoieoviciu, L. Doubova, S. Trasatti, Surf. Sci., 201 (1988) L498. 9. B.D. Cahan, H.M. Villullas, J. Electroanal. Chem., 307 (1991) 263.

10. A.E. Russell, G.P. Williarns, A.S. Lin, W.E. O'Grady, J. Eleetroanal. Chern., in press.

431

11. A. Harnelin, S. Morin, J. Rieher, J. Lipkowski, J. Eleetroanal. Chern., 285 (1990) 249.

12. MaJ. Gonzalez Tejara, F. Colorn Polo, An. Quirn, 80 (1984) 219. 13. M. Sanchez Cruz, Ma.J. Gonzalez Tejara, Ma.C. Villarnanan, Electrochirnica Acta,

30 (1985) 1563. 14. Sa.Ya. Vasina, O.A. Petrii, Electrokhirniya, 6 (1970) 242. 15. G. Horanyi, G. Vertes, J. Electroanal. Chern., 64 (1975) 252. 16. C.K. Rhee, M. Wasberg. G. Horanyi, A. Wieckowski, J. Electroanal. Chern., 291

(1990) 28l. 17. G. Valette, J. E1eetroanal. Chern., 269 (1989) 191. 18. M. Fleisehrnann, J. Robinson, R Waser, J. EleetroanaI. Chern., 117 (1981) 257. 19. B. Pettinger, M.R. Philpott, J.G. Gordon 11, J. Chern. Phys., 74 (1981) 934. 20. B. Pettinger, M.R. Philpott, J.G. Gordon 11, J. Phys. Chern., 85 (1981) 2746. 21. P. Gao, MJ. Weaver, J. Phys. Chern., 90 (1986) 4057. 22. F. IIIas, J. Rubio, J.M. Rieart, J.A. Garrido, J. E1eetroanaI. Chern., 200 (1986) 47. 23. H. Niehols, R.M. Hexter, J. Chern. Phys., 74 (1981) 2059. 24. G.M. Larnble, R.S. Brooks, S. FeITer, D.A. King, Phys. Rev.B., 34 (1986) 2975. 25. D. Eisenberg, W. Kauzrnan, "The Structure and Properties ofWater", Oxford Univ.

Press, 1969. 26. K. Raghavan, K. Foster, K. Motakabbir, M. Berkowitz, J. Chern. Phys., 94 (1991)

2110. 27. A.E. Russell, A.S. Lin, and W.E. O'Grady, J. Chern .. Soc., Faraday Trans., 89

(1993) 195. 28. E.M. Stuve, RJ. Madix, Surf. Sei., 111 (1981) 11. 29. K. Bange, T.E. Madey, J.K. Sass, E.M. Stuve, Surf. Sei., 183 (1987) 334. 30~ N. Kizhakevariam, E.M. Stuve, R. Dohl, Oelze, J. Chern. Phys., 94 (1991) 670.

TUE ADSORPTION OF CO AND U20 ON POLYCRYSTALLINE GOLD, AS STUDIED BYSYNCHROTRONINFRAREDSPECTROSCOPY

B. BEDEN(+), C.A. MELENDRES(*), G.A. BOWMAKER(X), C. LIU(*), and V.A. MARONI(*). (*) Materials Science Division, Argonne National Laboratory, Argonne, IL 60439 (USA). r+) Laboratoire de Chimie 1, URA CNRS D0350, Universite de Poitiers, 86022 Poitiers (France). (X) Department o/Chemistry, University 0/ Auckland, Auckland (New Zealand).

ABSTRACT. Synchrotron infrared spectra of CO and H20 adsorbed on a polycrystalline (pe) gold surface have been obtained for the tirst time. Assignments were confirmed through isotopic shifts measurements with 13CO and D20. Surprisingly, CO was found to adsorb on (pe) gold over a wide range of temperatures weIl above the limit of a few K generally accepted in the literature. CO adsorption occurred on mostly atop sites characterized by a strong C-O stretching mode at 2111 ern-I. A maximum absorption intensity of 3 % was measured at 132 K and a surface coverage of .5 to 20 L. But despite repeated efIorts, no reprodUClble Au-C stretching vibrations were detected.

Water adsorbed molecularly and associatively on (pe) gold at low temperatures, i.e., 100 to 140 K. For example, at 117 K and a coverage ofO.3 L, IR bands were observed at ca. 3380 and ca. 1650 ern-I, corresponding to the OH stretching and H-O-H bending modes, respectively. Bands attributed to frustrated translations and frustrated rotations were also observed at 230-290 cm-1, and 810-900 ern-I, respectively. The spectra are suggestive of the formation of (HOH)n clusters at 10w coverages, but at higher coverages, multilayer formation with a water structure similar to ice is evident.

Coadsorption experiments were also carried out. Tiny traces of water were found to irreversibly displace adsorbed CO from the (pe) gold surfaces.

1. Introduction.

Carbon monoxide is an interesting test molecule that is currently used to probe the reproducibility of catalytic (or electrocatalytic) surfaces in terms of adsorption sites [1]. The literature is extremely large on the subject as far as platinum is concemed, but there are only a few papers dealing with the adsorption of CO on gold, simply due to the fact that there is a prevailing notion that "CO practically does not adsorb on gold".

ActuaIly, in contrast with platinum, the behaviour of CO seems to be different if one compares its adsorption at the solid-gas interphase with that at the electrode-solution interface. It is now weil established that dissolved CO adsorbs quite weil on gold electrodes, either in acid medium [1 ][2][3][4][5] or in neutral and aIkaline media [2][6], conversely to the solid gas interface where weakly adsorbed species have been observed on various gold substrates [7][8][9][10], but only under restricted experimental conditions, among which temperature seems to play the most important roie.

Such a difference of behaviour between the two interfaces may suggest a possible strong influence of water coadsorbate and therefore, the first aim of the present work was to investigate carefully the influence of traces of water on the adsorption of CO at the solid-gas interface. But, to be complete, the adsorption of CO aIone on gold was first investigated, followed by that of water on gold, and finaIly the coadsorption of the two species. The second aim of this work was to explore the use of synchrotron infrared spectroscopy, especiaIly in the far infrared, where this new

433 C. A. Melendres andA. Tadjeddine (eds.), Synchrotron Techniques in Inteifacial Electrochemistry 433-449. © 1994 Kluwer Academic Publishers.

434

technique is reported to be significantly more sensitive because of the high energy of the infrared synchrotron source relative to commercially available laboratory sources [12][13].

2. Experimental

Infrared measurements were made using the reflection mode (lRRAS) on a polycrystalline gold substrate maintained in an ultra high vacuum chamber (UHV). All experiments were carried out at the National Synchrotron Light Source (Brookhaven National Laboratory, Upton, New York), using the U4IR infrared beamline developed by G.P. Williams et al. [12][13]. The sampie was mounted on a special cryogenic holder, so that all manipulations and experiments (thermal treatments, Ne+ sputtering, Auger and mass spectroscopies, LEED, IRRAS, ... ) could be done without removing it from the chamber.

The U4IR synchrotron beam provides an extremely bright broad band source, which can be used not only in the far infrared (180 - 600 ern-I, with a bolometer as a detector), but also in the mid-infrared (320 - 4000 ern-I, using a Cu-Ge detector). A Nicolet 20F vacuum interferometer mounted at theend ofthe infrared beamline was used for spectral acquisitions.

The preparation ofthe gold sampie has already been described elsewhere [14]. Special care was taken to avoid contamination of the gold surface by carbon and oxygen, and the water level ofthe UHV chamber was kept as Iowas possible (i.e., below 10-11 torr).

3. Results

3.1. ADSORPTION OF CO ON GOLD SUBSTRATES

3.1.1. Literature Survey. As stated in the introduction, CO is believed to adsorb weakly on gold. On rough surfaces such as silica-supported gold, only one CO stretching band, near 2115 ern-I, was observed by Yates [8]. It is interesting to note that this band shifted to 2125 ern-I, ifthe gold was not fully reduced and to 2145 ern-I in the presence of coadsorbed oxygen. With evaporated gold films, the situation was different. It has been shown by Dumas et al. [9] that gold films deposited at temperatures below 290 K allow the adsorption of CO wben the cold metal was put into contact with CO gas at very low temperatures (i.e., a few K). They observed two infrared bands, at ca. 2125 and 2143 ern-I, whicb were assigned to chemisorption and physisorption, respectively. The 2125 ern-I band was shown to shift to 2110 ern-I when the CO exposure was increased from 0.01 to 0.6 L. The physisorbed CO band appeared only at CO exposures greater than 0.4 L. Conversely, gold films deposited at temperatures higher than 290 K did not exhibit any band due to chemisorbed CO under all temperature conditions.

CO adsorption was also studied on weIl defined Au(llO) single crystal surfaces. Outka and Madix [10] concluded that the desorption activation energy ofCO was less than 8.0 kcal mol­l, which suggests a rather weak interaction. Similarly, no reaction with oxygen adatorns was observed at temperatures down to 125 K.

Experiments have also been carried out at electrochemical interfaces usuig potential modulated infrared spectroscopies EMIRS, PMFTIRS, SNIFTIRS, as weil as SERS (for a review of these techniques, their definition and their application to the electrode-solution interface, see for instance reference [15]). CO is known to adsorb linearlyon atop sites when (pe) gold electrodes are immersed in acid solutions. This leads to a weak band at around 2105 ern-I, when the applied potential is elose to the pzc; both linear and bridge configurations (the latter at around 1950 ern-I) are found when the solution is alkaline [2][3][4][5]. The lower frequency band increases in intensity with pR and becomes dominant in strongly alkaline solutions. As pointed out by Weaver et al. [4], the fact that the frequencies due to the C-O stretching modes of CO adsorbed on gold are

435

aImost the same at the solid-gas and at the electrode-solution interphases indicates that the nature of the bonding is essentially the same, at least for the linear configuration (i.e., with the carbon bonded to a single gold atom). This is not surprising since the surface potential due to water molecules is minimized at the pzc, which makes the two interfaces more comparable.

3.2.2. COads on (pe) Gold: Results ofthe Present Study. As the pretreatment ofthe gold sampie was expected to be very critical, great care was taken to ensure reproducible conditions. In particular, aseries of experiments was done on surfaces cleaned with different conditions of annealing, sputtering, and flashing temperatures. In each case, the CO doses ranged from 0.3 L to 20 L and experiments were carried out either at increasing dosing rates or at a fixed dosing rate. Similarly, the temperature at which CO was introduced to the UHV chamber varied from 112 K to 300 K.

Using the equipment described in the experimental section, we obtained spectra of CO adsorbed on a clean polycrystalline gold sampie (previously sputtered with Ne+) under conditions for which it has never been observed before.

w u z a: f-f-

I: U1 Z a: 0: t- ~ 0

rn

r--.------r-----,-----~------r-----~----_r----_.------r__;

a) --b)

200 600

BolDmeter ----I i

1000 1400 1800

CuGe

2111 cm-1

leo exp.=5LI

1150 K I

2200 2600 3000 3400 j.JAVENUMBEA / cm-1

Fig.l. Synchrotron infrared spectra of CO adsorbed on a polycrysta1line gold substrate (see text for pretreatments), at 150 K. Two detectors were used to cover the whole range of wavenumbers, as indiC3ted : (a) background; (b) in presence of 5L CO.

Figure 1 illustrates the results obtained under optimized conditions. The experiment was carried out at 150 K, with a 5 L fixed dose of CO. A strong single narrow band was observed at ca. 2111 ern-I. Under these conditions, the absorption intensity reaches nearly 3 %, which is not very fur from that of 3.5 % observed for the adsorption of CO on atop sites of platinum [16]. For comparison, the background (i.e., the spectrum before introducing CO) is given in Fig.l curve a. It allows a test ofthe noise level and ofthe contamination ofthe surface by adsorbates. As expected

436

from the energy curve of the source, the noise level increases with frequency but remains satisfactory in the mid-infrared range. In particular, no water is detected, since any traces ofwater would give rise to a strongabsorption band in thc O-H stretching region, near 3400 cm- l (see scction 3.2).

aJ m

aJ

lD Wm u z ([ I-~

1-. _Ln :;:Gl Ul Z

iE,,: 1- cn ~m

o o 2-:10 cm-1

15 L

20 L

m ~1~9-5-0----2'6-o-0----2CT1S-0----2'i-o-o---~-,ri5~O----2'2-0-0---2-2r/5~L--l WAVENUMBER /cm-1

Fig.2. Influence ofthe CO exposure on the CO peak intensity at 150 K and 5L to 20L CO exposures.

Increasing the CO exposure from 5L to 20 L does not significantly change the shape ofthe peak [Fig.2]. There is a slight asymmetry at the foot of the peak which exhibits a weak shoulder near 2080 em-1 as weIl as an oceasional small positive contribution on the opposite side of the shoulder. Under our experimental eonditions, the linewidth at half-height can be estimated to be slightly less than 13 ern-I, i.e., signifieantly less than that ofthe band observed by Dumas et al. [9] on gold deposits, whieh was about 17 ern-I. This rather narrow peak is probably an indication that CO adsorbs on smooth (pe) gold only on one type of site, if one ignores the very small contribution noticed at the lower energy side ofthe main peak.

In agreement with Dumas et a1. [9], the CO stretching band shifts towards lower frequency with exposurc, especially at low exposures [Fig.3]. The intensity of the shift also depends on the temperature, as seen in Fig.4, where a clear maximum is found near 130-135 K.

band eantre lern-1

2118

13 180 K ~------~Br-------~9~-------D

2114

2110~ I 130 K

160K

118 K 2106L-------~------~--------~-------L------~

o 5 10 15

exposure /L 20 25

Fig.3. Frequency shift of the stretching band due to adsorbed CO on (pe) gold, vs. CO exposure.

100 120 140 160 180 200

Temperature IK

Fig.4. temperature dependence of the intensity of the stretching band of CO adsorbed on (pe) gold.

437

An experiment was also carried out using 13eo instead of l2eo. The exact downshift of frequency for the isotopic exchange was observed.

The lower frequency region was carefully checked using a bolometer detector, whose noise level is very low. Features were seen near 390 - 400 cm-l in some measurements, but they were

438

not always reproducible. Experiments with 13C180 did not allow us to draw any conclusions. It could be that the features are due to instabilities and not to any contribution from the Au-C stretch. If one looks at the literature, weak bands due to M-C stretches have been characterized with other metals and othcr tcchniques like Raman spectroscopy. For example, Cu-CO has been measured at 355 cm-1 [17], Ni-CO at 475 cm-1 [18][19] and Pt-CO at 460-466 ern-I [19][20]. A good discussion ofwhy a band such as the M-CO stretch may or may not be detected, can be found in Ref. [21]. For instance, the possibility that the M-C bond lies more or less parallel to the surface (thus indetectable according to the so-called surface selection rule), cannot be excluded at least at low coverages. It may also be that the intensity of the band is too low to be extracted from the noise [16].

This problem has recently been reviewed by Schwerdtfeger and Bowrnaker [23], who also discussed several models for adsorbed CO on gold, including a bent configuration. Whatever the explanation, the fact that the C=O stretch appears extremely intense, while the studied Au-C seems to be undetectable under our present experimental conditions is in fairly good agreement with SERS observations at the electrochemical interface of CO adsorbed on rough gold electrodes in perchloric acid medium by Weaver et aI. [5].

3.2. ADSORPTIONOFWATERONC,oLD.

w u z ([ r r :l::: (f")

Z ([ 0: r

~ o

6)

~.~W"'v-~~ W"-.,r.~ .5U+OH) ~ \817 1650

l' Frustra~ed 230 Rotation

Frustrated

~ ,I rn Translation

200 600

Hp exp.=OJL

T = 117 K t + V( H)

1000 1400 1800 2200 2600 3000 3400 3800 hlAVENUMBEA /cm-1

Fig.5. Synchrotron infrared spectra of H20 adsorbed. at a polycrystalline gold substrate (see text for pretreatments) : at 117 K. Two detectors were used to cover the whole range of wavenumbers, as indicated in Fig. I ; (a) background; (b) in presence ofO.3 L H20.

3.2.1. Literature Survey. Water is known to adsorb on metals [23]. The main vibrational bands are identified as the O-H stretching modes near 3300 - 3750 cm-1 and an H-OH scissoring mode near 1620-1660 cm- I . At lower frequencies, frustrated rotations (with some rotational contribution) are

439

observed in the 250 - 900 cm- l range and frustration translations (with some translation contribution) in the 50 -250 cm-l range.

The following questions, which are still under debate, are relevant whenever the adsorption of water on metals is considered :

LJJ U Z ([ r­r-

Tr.

i) Does H20 adsorb molecularly or dissociatively? ii) Does H20ads form ice, or ice-like structures ? iii) What is the structure ofthe successive layers ofwater? iv) Is the structure temperature dependent ?

812 Rot.

al

t 1480

o H [,( '0/ )

'-----y------'

frustr. modes 2481

v (0-0)

v (O-H) 3352

200 600 1000 1400 1800 2200 2600 3000 3400 3800 WAVENUMBER / cm-1

Fig.6. Synchrotron infrared spectra of D20 adsorbed at a polycrystalline gold substrate, at 116 K. Two detectors were used to cover the whole range of wavenumbers, as indicated. a) background; b) low D20 exposures; c) higher D20 exposure. The O-H stretch is due to remaining traces of water in the UlN chamber, which was impossible to remove at the time scale ofthe experiment (this would probably need a few days). This explains also the DOH bending mode at 1480 cm-1 (formed by a fast exchange process).

Although the subject is of great interest for electrochemists, the huge absorption of the infrared beam by bulk water has certainly limited the studies of water adsorption at the electrode -aqueous solution interface. However, despite the strong attenuation of the beam, the H20 adsorption on noble metal electrodes (Pt, Rh) can be successfully investigated in the mid-infrared, as shown for the first time by Bewick and Russell [24], then by Kunimatsu [25] for the adsorption of water on polycrystalline gold electrodes. Of course, due to the high energy of the beam, there is no doubt that synchrotron infrared radiation offers new promises to extend the study down to a few tens ofwavenumbers. Recent attempts have been made in that direction [26][27J,

440

3.2.2. Water Advorption on (pe) Gold: Results ofthe Present Study. Experiments were carried out at increasing exposures with dosages ranging from 0.1 L to 20 L of H20. Temperature was generally fixed in the range 11 0 to 130 K. Comparative experiments were also done with D20 under the same experimental conditions. Preliminary results have alrearly been presented [14].

0.5 L

1.0 L

w

~ 1.5 L ([ f-­f--~

~ 2.0 L z ([ er: f--

t 3325 ,3395

~----~----+-~

2500 3000 3500 4000 WAVENUMBER /cm-1

Fig.7. Dependence of the OH stretching modes of H20 adsorbed on (pe) gold at 120 K, on water exposure.

Figure 5 gives a reconstructed spectrum (obtained using the two detectors), that covers the whole spectral region from 190 to 4000 ern-I. At 117 K and 0.3 Lexposure ofH20, all the modes are detected. The strongest bands, slightly below 3400 ern-I are due to OH stretching modes. Conversely, the intensity ofthe H-OH bending mode at ca. 1650 cm-1 is certainly weaker than expected. medium intensity frustrated rotations absorb near 817 ern-I, while frustrated

441

translations, although weak, are clearly detected at 230 ern-I, thanks to the high SIN ratio of the bolometer.

All assignments were eheeked using isotopie exchange with D20. An example is given in Fig.6 for the whole spectrum. All bands shift towards lower wavenumbers, but to a different extent, as expected. More details are given in the following paragraphs and figures.

Figure 7 shows the dependence ofthe OH stretching mode on water exposure level. OftOO two OH modes at 3325 and 3395 ern-I, the latter beeomes dominant at high exposures. It is interesting to note that there is a shoulder at around 3200 ern-I, more clearly visible at I L H20 exposure, whieh stays practieally eonstant at all exposures. As discussed previously [14], such low frequency OH modes are encountered in ice-like structures. Thus the contribution from ordered layers ofwater on the gold surface is a likely origin ofthis band.

Fig.8. Synchrotron infrared spectra of H20 adsorbed on a polycrystalline gold substrate, at 118 K, showing the regions of the scissoring mode and of the frustrated rotation modes, as a function of water exposure.

442

Band eentre lem-1 1050~--------------------------·--------------~

1000 +++ (shoulder)

950

H20 900

850

--'---(Shoulder) ~ 800 ~ 020 750

700L---------~--------~-----------L---------~ o 2

exposure IL 3 4

Fig.9. Frustrated rotation bands observed with H20 and D20 and their dependence on water exposure.

w u z ([ I­I-

L ')) Z ([

ce I-

Hfl Dose

r\~~~~0.1L

0.3 L

0.6 L

to L

200 300 400 500 600 ~JAVENUMBER I cm-1

Fig.IO. Synchrotron infrared spectra of H20 adsorbed at a polycrystalline gold substrate, at 118 K, showing the regions for the frustrated translation modes, as a function of water exposure.

443

Figure 8 gives spectra in the mid-infrared region, from 600 ern-I to 2000 ern-i. Tbe band at around 1650 ern-I, due tothe seissoring mode ofthe moleeule, is a proofofthe non dissoeiative adsorption of water on gold. Tbe reason why it is weak at iow exposures is not cJear yet, but might indicate that the orientation ehanges with exposure. For instance the plane containing the water molecule eould be tilted at low exposures and be perpendicular to the reflection plane at high exposures.

Tbe intensity of the band assigned to frustrated rotations is mueh stronger. InitiaIly observed at 817 em-I (Fig.5), it shifts to 870 ern-I at I L H20 exposure, then rcaches 892 em- l at 4 L. Tbe shape ofthe band aIways shows some asymmetry, especiaIlyon lowenergy side.

With D20, there is a downshift ofthe main band. From Fig.9, it is estimated to be about 60 to 80 ern-I, i.e., roughly the magnitude observed by EELS measurements during the coadsorption ofH20 and D20 on Pt at 100 K [22b]. Tbe region of frustrated translation modes is detailed in Fig.1O. Investigations were carried out using the bolometer detector over the range of 170 to 650 ern-I, and two bands were c1early detected, at 230 and 290 ern-I. Tbe positions of these bands move only slightly with inereasing exposures. However, their relative intensities do ehange with exposure, so that the 290 em-I band becomes a shoulder of the main band at I L H20. The fact that the 290 ern-I band remains approximately constant suggests that it corresponds to a monolayer of water having a partieular structure.

w u z a: l-r--;:: IH201 ({)

z a: er: I-

~ IH20 + DtJI T = 111 K

A-0--2~3~O- '--2~9-0--3~5-0---'t~i""" 0-WAVENUMBER

/(rn-1

Fig.ll. (a) Synchrotron infrared spectra of H20 adsorbed on a polycrystalline gold substrate; (b) Synchrotron infrared spectra of CO + H20 coadsorption on the same polycrysta1line gold substrate at 150 K.

444

Working with a mixture H20 + D20 results in an additional band at 212 cm-l (Fig.ll). In that case, the downshift due to the isotopic exchange can be estimated to be about 15 to 20 cm­I, (Fig.l2), i.e., again in agreement with literature [23b].

At this stage of the work, it can be said that our experiments demonstrate that water adsorbs non-dissociatively on gold. Even at low exposures the spectra exhibit features that correspond to structured water, most probably with contributions from ice-like layers. In all cases, high exposures lead to multilayered water formation. Moreover, vibrational spectra for small (H20)n water clusters have been calculated by Knochenmuss and Leutuyler [27] in the self­consistent field approximation. According to the authors, the OH stretch should decrease approximately from 3600 cm-l (for water dimers) to 3200 ern-I for larger clusters (n = 8), whereas the rotational band, initially at ca. 750 cm-I with n = 2, should move positively with increasing cluster size up to n = 5 (reaching about 975 ern-I), then should decrease again down to ca. 840 cm-1 at n = 8. From our experiments and the frequencies ofthe bands we have observed, it seems reasonable to suppose that water clusters having 4 or 5 units are the ones most likely to be present on the gold surface under our conditions.

Band centra Icm-1 350~------------------------------------------.

300 ..

250 H20 + ;j<: :I<

~---*' + + + +

ra/ 9 8 -EI

200 D20

150 1

0 2 ::5 4

exposure IL

Fig.12. Frustrated translation bands observed with H20 and D20 and their dependence on exposure.

3.3. COADSORPTION CO + H20

Carbon monoxide and water coadsorption experiments are interesting to consider. They provide information on the nature of the interaction between the two types of adsorbates. For instance, Wagner et al. [28] showed by TDS and HREELS that the CO-H20 interactions are repulsive on rhodium, whereas they are attractive on platinum. From IRRAS measurements, Tomquist et al. [29] concluded that, on platinum, the type of interaction depends mostlyon the CO coverage. Thus, at high CO coverages water adsorbs as clusters on top of a saturated CO adlayer without perturbations, while at low CO coverages the H20 adsorption results in the disappearance

445

of the linear CO band. A similar disappearance of the CO band was also observed for the coadsorption ofCO + H20 on Al substrates [2I].

Due to the diffieulty of totally evaeuating the water from the UHV ehamber in a short time, the coadsorption experiment had to be performed as follows : after ehecking the cleanliness of the chamber and of the gold surface, CO was introduced first under the conditions which allow a maximum coverage, i.e., leading to a stretching COads band intensity ofnearly 3 %. A reference spectrum was then collected. Then, a dose of water was introduced, without removing CO from the chamber. A new spectrum was taken and subtracted from the reference.

Figure 13b shows the difference spectrum resulting from the above experiment. The CO peak points upwards because of the subtraction technique. Its high intensity indieates that CO has been totally removed by water. The OH bands points downwards, with a maximum at 3266 ern-I, and a shoulder at higher frequeneies near 3420 ern-I. Thus, compared to the spectrum for the adsorption of pure H20 (Fig.13a), the OH bands are clearly downshifted when the adsorption occurs on a surface previously covered by CO. Conversely, the CO band is only slightly shifted towards higher wavenumbers (byabout 2 cm-I).

a)

.~~{It~Jlf'

2113

pure HZO

co + H20 coadsorption

3266

llOO <: 160 ;~620 3080 35\,· C 4 JOD

~/cm-1

Fig.13. a) pure H20 adsorption; b) CO + H20 coadsorption experiment. see text for the sign ofbands.

446

It can be tentatively suggested that the removal of an ordered layer of CO (for instance a 2x2 structure on the (Ill) facets) leaves the surface with a layer of structured ice-like water. In other words, the adsorption of water displaces CO but keeps the fingerprint of the previous CO adlayer. Of course, another explanation is that in the presence of H20 the CO species become so tilted on the surface that they are no longer detectable by infrared spectroscopy. Such a situation would need a strong attractive interaction between CO and H20 species. However, the upwardgoing CO band, which indicates the removal of CO species, occurs at 2113 cm- I , i.e., essentially unchanged compared to the frequency observed with CO alone.

4. Conclusions.

4.1. ONTIIE TECHNIQUE.

Using a synchrotron generated beam for infrared spectroscopy is definitely an exciting idea, even through it is far from a routine technique. The present results substantiate the feasibility of synchrotron infrared spectroscopy, not only in the far infrared., but also in the mid-infrared., where the source has still enough energy to allow the recording of spectra with high signal to noise.

As expected the energy is very high in the far infrared. However, at present on U4IR, some limitations arise from beam instability. These problems would have to be solved in order to reach the 10-3 absorbance sca1e which is necessary for the detection of weak signals due to adsorbates at the submonolayer sca1e. In particular, this limit is extremely critical for further work at the electrode-solution interface.

4.2. ON TIIE ADSORPTION OF CO AND H20 ON GOLD.

Very surprisingly, CO was found to adsorb on smooth polycrystalline gold over a wide range of temperatures where it has not been detected before. A maximum intensity of nearly 3% was obtained., i.e., comparable to that of the adsorption of a monolayer of CO on platinum. The single C=O stretching mode, at 211I cm-I, shows that CO adsorbs on atop sites with no contribution from higher energy sites. But, despite efforts, no reliable signals corresponding to Au­C stretches were detected. Either they are too weak, or they are in a configuration which make them undetectable because of the surface selection rule.

As expected., water was found to adsorb on polycrystalline gold, exhibiting features which correspond to structured water. Frustrated rotations as weIl as frustrated translations were clearly observed (the latter modes because of the high energy of the synchrotron source in the far infrared). The features associated with the frustrated translation modes suggest that at least two different layers of water are formed.

Coadsorption experiments of CO and water have led us to conelude that a monolayer, or nearly so, of COads can be easily removed by water (i.e., the strong C=O stretching mode disappears without any shift). Simultaneously, the coadsorbed water layer exhibits an ice-like structure.

Further experiments should now be extended down to the far infrared; 10 cm-1 can be reached with a polyethylene window. However, the problem is not there. Tbe greatest difficulty will come from the inability to check at the same time the mid-infrared region (it is cut off by polyethylene). In the present work we found it very convenient to cover all the infrared range up to 4000 ern-I . It allowed us in particular to check the absence of adsorbed water on the sampie, just by looking for the very sensitive OH stretching bands at ca. 3400 ern-I. Similarly, a quick look to the C=O band at ca. 2111 ern-I was obviously the best test for the clean1iness of the sampie and of the reproducibility of the CO adsorption experiment.

447

The last point of comment is that it is still not clearly understood why CO is so easily rernoved by water at the solid-gas interface, whereas it can be detected at a gold electrode surface immersed in an aqueous solution, i.e., in an environment of excess water molecules. The answer might rely on the influence of the potential applied to the gold electrode. There is no doubt that a positive potential favours the adsorption of CO molecules. Once the compact CO monolayer is formed, it probably prevents the water molecules from reaching the surface, which is thus dehydrated. The further layers of water that adsorb on top of the CO layer, probably do not interact strongly with it.

Acknowledgments.

Financial support for this research was provided mainly by the Division of Materials Sciences, Office of Basic Energy Sciences, US Department of Energy, under contract W -31-109-Eng.38. The work was carried out at the National Synchrotron Light Source at Brookhaven National Laboratory. Other source of supports were the NATO Collaborative Research Grants program (grant n0920512), the "Ministere des Affaires Etrangeres" (France), and the Research Grants Committee, University of Auckland (New Zea1and).

The authors would like to thank G.P. Williams and CJ. Hirschmugl (BNSLS lab) for their assistance in the operation of the U4IR beamline and stimulating discussions. We thank also S. Johnson and M. Pankach, of Argonne, for their help in designing and testing the UlN sampie holder.

References

1. See for instance: a) Hoffinann, F.M., (1983) "Infrared reflection-absorption spectroscopy of adsorbed molecules", Surr. Sci Rept. 3, 107-192. b) Sheppard, N. and Nguyen, T.T., (1969) "Chap. 2 : Vibrational spectra of carbon monoxide chemisorbed on the surfaces of metal catalysts. Suggested scheme of interpretation", in "Advances in Infrared and Raman Spectroscopy", Vol. 5., Clarke, R.J. and Hester, R.E. (eds.), Heyden, London, pp.67-148. c) Beden, B., Lamy, C., de Tacconi, N.R. and Arvia, A.J. (1990) "The electrooxidation of CO : a test reaction in electrocatalysis", Electrochim. Acta, 35, 691-704.

2. Beden, B., Bewick, A., Kunimatsu, K. and Lamy, C., (1982) "Infrared study of adsorbed species on electrodes: adsorption of carbon monoxide on Pt, Rh and Au", J. Electroanal. Chem., 142, 345-356.

3. Nakajima, H., Kita, H., Kunimatsu, K. and Aramata, A., (1986) " Infrared spectra of carbon monoxide adsorbed on a smooth gold electrode. Part I : EMIRS spectra in acid and alkaline solutions", J. Electroanal. Chern., 201,175-186.

4. Tadayyoni, M. and Weaver, MJ., (1986) "Adsorption and electrooxidation of carbon monoxide at the gold-aqueous interface as studied by surface-enhanced Raman spectroscopy", Langmuir, 2, 179-183.

448

5. Leung, L.W.H. and Weaver, M.1., (1987) "Extending surface-enhanced Raman spectroscopy to transition-meta! surfaces : carhon monoxide adsorption and electrooxidation on platinum -and palladium-coatedgold electrodes", J. Am. Chern. Soc., 109, 113-5119.

6. Cbang, S.C., Hamelin, A. and Weaver, M.1., (1990) "Reactive and inhibiting adsorbates for the catalytie electrooxidation of carhon monoxide on gold(21O) as eharacterized by surface infrared spectroscopy", Surf. Sei., 239, L543-L547.

7. Ikezawa, Y., Saito, H., Matsubayashi, H. and Toda, G., (1988) "Comparative study of CO adsorbed on Pt, Pd, Au and Ag electrodes in neutral solutions by IR reflectance absorption spectroseopy", J. Electroanal. Chem., 252, 395-402.

8. Yates, D.1.C., (1969) "Spectroscopic investigations ofgold surfaces", J. CoUoid Interface Sei., 29, 194-204.

9. Dumas, P., Tobin, R.G. and Richards, P.L., (1986) "Study of adsorption states and interactions of CO on evaporated noble meta! surfaces by infrared absorption spectroscopy. II) gold and copper", Surf. Sei., 171, 579-599.

10. Outka, D.A. and Madix, R.1., (1987) "The oxidation of carbon monoxide on the Au(llO) surface", Surf. Sei., 179,351-360.

11. Schmitz, P.1., Kang, H.C., Leung, w.Y. and Thiel, P.A., (1991) "Growth mode and CO adsorption properties of Au film$ on Pd(llO)", SUff Sei., 248, 287-294.

12. Williams, G.P., Hirschmugl, C.1., Kneedler, E.M., Sullivan, E.A., Siddons, D.P., Chabal., Y.1., Hoffinan, F. and Moeller, K.D., (1989) "Infrared Synchrotron Radiation Measurements at Brookhaven", Rev. Sci. Instr. 60, 2176-2178.

13. Hirschmugl, C.1., Williams, G.P., Hoffinann, F.M. and Chabal, Y.1., (1990) "Adsorbate­Substrate Resonant Interactions observed for CO on Cu( 1 00) and (111) in the far IR using synchrotron radiation". J. Electron. Spectrosc. Relat. Phenom., 54/55, 109-114.

14. Melendres, C.A., Beden. B., Bowmaker, G.A., Liu, C. and Maroni, V.A., (1993) "Synchrotron infrared spectroscopy of H20 adsorbed on polycrystalline gold", LangnlUir (in press).

15. Beden, B., Leger, J.M. and Lamy, c., (1992) "Electrocata1ytic oxidation of oxygenated aliphatic organie compounds at noble metal electrodes" in "Modem Aspects of Electrochemistry", Bockris, J.O'M., Conway, B.E., and White, R.E. (eds.), Vo122, chapter 2, pp. 97-264.

16. Tüshaus, M., Schweizer, E., HoUins, P. and Bradshaw, A.M., (1987) "Yet another vibrational study ofthe adsorption system Pt(I11)-CO", J. Electron. Relat. Phenom., 44, 305-316.

17. Akemann, W. and Otto, A., (1991) "Vibrational modes ofCO adsorbed on disordered copper films", J. Raman Spec., 22, 797-803.

449

18. Ryberg, R., (1989) "Infrared spectroscopy of molecules adsorbed on metal surfaces" in "Advances in Chemical Physics", Prigogine, I. and Stuart, A, (eds.), Vol. LXXVI, 1. Wiley Intersci., pp 1-44.

19. Hoge, D., Tüshaus, M., Schweizer, E. and Bradshaw, AM., (1988) "The metal carbon stretch in the vibrational spectrum of CO adsorbed on Pt(lll)", Chern. Phys. Letters, 151, 230-235.

20. Malik, LI. and Trenary, M., (1989) "Infrared reflection-absorption study of the adsorbate­substrate stretch ofCO on Pt(1I l)", Surf. Sei., 214, L237-L245.

21. Jacobi, K., Bertolo, M., Geng, P., Hansen, W., Schreiner, 1. and Astaldi, C., (1991) ''H20 and CO + H20 co-adsorbate on the Al(lll) surface at low ternperature", Surf Sci., 245,72-84.

22. See for instance: a) Thiel, PA and Madey, T.E., (1987) "The interaction ofwater with solid surfaces : Fundamental aspects", Surf. Sei. Rep., 8, 211-385. b) Sexton, BA, (1981) "Identification of adsorbed speeies at metal surfaces by energy loss spectroscopy (EELS)", Appl. Phys., A26, 1-18. c) Stuve, E.M., Madix, RJ. and Sexton, BA, (1981) "The adsorption and reaction ofwater on clean and oxygen covered Ag(llO)", Surf. Sei., 111, 11-25.

23. Schwerdtfeger, P. and Bowmaker, G.A, "The importance ofrelativistie and correlation effects in group II dipole polarizabilities and weak bonding in monocarbonyl compounds" , J. Chern. Phys.,(in press).

24. Bewick, A and Russell, 1.W., (1982) "Structural investigation by infrared spectroscopy of adsorbed hydrogen on platinum", 1. Electroanal. Chern., 132, 329-344.

25. Kunimatsu, K. and Bewick, A, (1986) "Electrochemica1ly Modulated Infrared Spectroscopy of adsorbed water in the inner part of the double layer. Part I : O-H stretching spectra ofwater on gold in I M HCI04", Indian J. tech., 24, 407-412.

26. Russell, AE., Willianis G.P., Lin, AS. and O'Grady, W.E., (1993) "In situ Far-infrared evidence for a potential dependence of silver-water interactions", J. Electroanal. Chern., 356, 309-315.

27. Melendres, CA, Beden, B., Liu, C., Bowmaker, GA and Maroni, VA, (work in progress).

28. Knochenmuss, R. and Leutwyler, S., (1992) "Structures and vibrational spectra of water clusters in the self-consistent field approximation", 1. Chem. Phys., 96, 5233-5244.

29. Wagner, F.T., Moylan, T.E. and Schmieg, SJ., (1988) "Hydrophilie versus hydrophobie coadsorption : carbon monoxide and water on Rh(I11) versus Pt(lll)", Surf. Sci., 195, 403-428.

30. Tornquist, WJ. and Griffin, G.L., (1986) "Infrared reflectance absorbance spectroscopy of coadsorbed CO and H20 on Pt(lll) surfaees", 1. Vac. Sci. Technol., A4, 1437-1441.

LA YERED SEMICONDUCTOR/ELECTROL YTE MODEL INTERFACES INVESTIGATED IN UHV BY SYNCHROTRON INDUCED PHOTOELECTRON SPECTROSCOPY

T. MAYER and W. JAEGERMANN Hahn-Meitner-Institut .. Abteilung Solare Energetik, Glienicker Str. 100, 14109 Berlin, Germany

ABSTRACf. The use of synchrotron induced photoelectron spectroscopy is presented for the investigation of semiconductor/electrolyte model interfaces. The model interfaces are prepared by adsorbing electrolyte components e.g. H20 and Br2/Na onto layered semiconductor van der Waals surfaces. The electric potential distributions measured for the model interfaces are rclated to semiconductor/ electrolyte junctions.

1. Introduction

The solid/c1ectrolyte interface is at prescnt intensively investigated with a variety of interface sensitive spcctroseopic techniques in order to complement standard electrochemical techniques. The aim cf a11 these investigations is to elucidate the geometric and electronic structure of the interface on a molecular level and to reconsider the accepted electrochemical models of electric potential distributions and related charge transfer processes. For a complete understanding of interfacial electrochemistry the chemical composition and structure of the solid electrode, the distribution of ions, molecules, and solvent moleeules in the different double layers of the electrolyte, as weH as the nature of specific surface interactions must be known in molecular detail.

The use of synchrotron radiation offers new possibilities for the in-situ investigation of the electrochemical interface as is discussed in other papers in this volume. However, ex-situ ultra-high-vacuum (UHV) investigations mayaiso contribute to a better understanding of interfacial electrochemistry. Various UHV surface science techniques and especially photoelectron spectroscopy have been applied for the ex-situ analysis of electrochemical interfaces. The related problem of emersion and transfer into UHV has been discussed with respect to the possibilities and !imitations of this approach [1,2]. Synchrotron radiation as a photoemission source, however, has not been used for ex-situ experiments. Evidently the more severe restrietions of obtaining uncontaminated surfaces, because of the increased surface sensitivity, have not been solved satisfactorily up 10 now.

451

C. A. Melendres and A. Tadjeddine (eds.), Synchrotron Techniques in Inteifacial Electrochemistry 451-468. © 1994 Kluwer Academic Publishers.

452

Here we report on an alternative approach to investigating the properties of solid/electrolyte interfaces. It is based on the pioneering work of Sass and coworkers who started 10 model the electrochemical interface in UHV by adsorbing electrolyte components onto defined metal surfaces [3,4]. With this approach all modem techniques of surface science can be applied in combination to investigate nearly all aspects of surface interactions regarding the chemical, structural, and electronic properties. It was shown that a elose correspondence between the work function change of such model experiments and the change in the PZC within the electrolyte could be obtained [4]. Evidently, these model experiments give significant results in relation to interfacial electrochemistry. Because of the success of this approach more peoplc started to investigate the interaction of model electrolytes on inert metal substrates [5-7]. However, comparable studies of simulating semiconductor/electrolyte interfaces were not performed apart from our early experimental investigation on the interaction of halogens and H2Ü, respectively, on layered chalcogenides [9-11], FeS2 [12], and CuInSe2 [13]. In this paper we will first discuss the background of semiconductor/electrolyte model experiments and the possible information which may be obtained by photoelectron spectroscopy using synchrotron radiation (section 2). Then we will report on a systematic investigation of H2Ü adsorption on p- and n-doped layered chalcogenides (section 3) and the Br2 and Na/Br2 coadsorption on n-WSe2 (section 4). Finally, our approach of simulating semiconductor/electrolyte interfaces will be summarized in section 5.

2.1 THE SEMICONDUCTOR/ELECTROLYTE INTERFACE

The formation of semiconductor/elcctrolyte junctions according to accepted theories [14-17] will only be briefly summarizcd. The main emphasis is to relate the concepts of electrochcmistry to the concepts of solid state physics and surface science and vice versa. The contact is formed by achieving clectronic equilibrium between the electrochemical potential TJeE1 for elcctrons in the electrolyte solution and Ep or TJese of the electrons in the semiconductor (Fig. 1). TJe corresponds to the binding energy of an electron which is transferred from astate at rest in vacuum E:ac, the zero of electron energy, into the corresponding phase (semiconductor or electrolyte). lle is equivalent to the energetic position of the Fermi energy Ep within the solid:

(1)

where J.le is the chemical potential of electrons within the phase, X is the surface dipole contribution passing the vacuum-solid interface and ,'I' is the electrostatic potential just outside the phase ( .. 10-4 cm) [18,19] Couter or Volta potential), which approaches 0 for vanishing excess charge cr on the phase. The sum X + 'I' = <I> is the inner or Galvani potential. The so called "real potential" <Xe (of the electron within the phase) can be measured experimentally (section 2.2) as it equals -<1>, the work function of the solid. For thermodynamic equilibrium between the phases, the difference in electrochemical potentials equals the contact potential difference (Galvani potential difference) 1J~ - Tl!C is compensated by an electrostatic potential EI,1.SC'I' (Volta potential difference) due to

453

exchanged excess charge on the semiconductor ase and the electrolyte 0EI and by changes of the dipole potentials Elf,SCX at the phase boundary:

ase + OE! = 0 (2)

(3)

The Volta potential difference can be measured by using an appropriate reference electrode. The Galvani potential of the different phases cannot be measured as the dipole contribution X to the vacuum is not known. The problem in eq (3) is the different reference levels for EpSC and of the electrolyte redox potential E(red/ox). [20-22]. The electric potentials for solid state contacts as weil as for semiconductor/adsorbate interfaces are referred to the vacuum level (absolute scale E~AJ. The electrochemical potentials are usually referred to the redox potential of an ideally unpolarizable reference electrode, for example the saturated calomel electrode (SeE) or the normal hydrogen electrode (electrochemical scale). The two sc ales are related to each other by E~AC = - E(NHE) - 4.5. E(NHE) refers to the redox potential measured versus the normal hydrogen electrode (NHE) as reference. The scaling value of 4.5 is only an approximate number that is often used. Recent calculations give values of 4.44 [23] but experimental values of 4.7 or 4.8 have been proposed in the literature [24-26]. Based on this scaling it is possible to relate energy values obtained from measurements in clectrolytes to measurements in UHV. The different values in the scaling of the absolute value of the electrochemical potentials result from the unknown value of X and different estimations of the changes f,X due to junction formation.

Semiconductor Electrolyte

sc

eB t------'-~~ "'ox

-"'red ~~~~+l

Pigure 1. Schcmatic formation of Schottky barricr-like contacts at semiconductor/electrolyte interfaces. The outcr boundaries refcr to the vacuum interface. At the junction, the different potential contributions are schematically shown. In addition, the Gaussian distribution of occupied and unoccupied states of a redox couple is dcpictcd.

454

In the Schottky limit of semiconductor/electrolyte junctions (in analogy to semiconductor/metal interfaces [27]) the dipole potential change EI.1SCx is neglected and the potential drop in the space charge layer of the semiconductor (band bending eVb) is equal to the work function difference .1<1> given by:

e SC.1E1\jI = <1>SC - <1>EI = .1<1> = eVb (4)

However, in reality the electrochemical double layers formed at the semiconductor/electrolyte interface change the electron affinities of each phase by e .1X and thus the relative position of the energy levels. As a consequence the contact potential drop SC.1E1 \jI inside the semiconductor space charge layer (band bending Vb) is changed as shown in Fig. 1:

(5)

Tbe changes in surface dipole contributions .1X result from different dipole layers: 1.) the change of the surface potential of the semiconductor from the vacuum interface to a contact phase .1XSC. 2.) the contributions of the inner Helmholtz (IHP) e .1<PIHP, outer Helmholtz (OHP) e .1<POHP and Gouy-Chapman (GC) e .1<po-c double layers from adsorbed or electrostatically bound electrolyte components carrying ionic charge or dipoles. These double layers induce an additional potential drop of unknown quantity across the electrochemical double layer (Fig. 1). Tbe contact potential difference of the semiconductor to the redox potential in solution is thus composed of several potential terms:

Tl~ - Tl~c = .1<1>= eSC .1E1\jI + e EI.1SCX = eVb + e.1xSC + e .1<PIHP + e .1<POHP + e .1<po-c (6)

Only the first two components mayaiso be found in solid state semiconductor/metal Schottky barriers [ 27].

For many semiconductor/electrolyte interfaces the observed experimental behaviour deviates even more from the picture discussed above as surfacelinterface states are not considered up to now. The deviation is related to the electronic influence of extrinsic surface states. Tbis influence is only expected in that energy range where EF passes the energetic position of the surface states Ess. In this case the charge neutrality condition for electrochemical contacts is modified taking into account the charge stored in the surface states:

OSC + oss + OE! = O. (7)

When the available surface state density exceeds typical charge carrier numbers of 1011 -13cm-2, which are exchanged for the formation of space charge layers,the band bending is fixed (the semiconductor Fermi level is pinned to the energy level of the surface states) when electronic equilibrium is established (Bardeen limit) [27). The contact potential drops complctely across the electrostatic doublc layer between the surface and the electrolyte

455

(=,6,cp) by changing the occupancy of surface states. In most real cases the situation is more complex as different types and lower concentrations of surface states may be involved and the contact potential difference is distributed inside the semiconductor and across the double layer.

In Fig. 1 also the density of states distributions D(E) of the redox active electron states within the electrolyte are given according to the usually found representations [14-17]. They are based on a Maxwell-Boltzmann distribution of thermally populated vibrational modes of the coordination sphere around the redox active components, which result in a Gaussian distribution of electronic energies:

(8)

_ (E - EOox)2 Dox(E) = Cox . Wox(E) = Cox . (4 1t kT A.ox) 1/2 exp - 4 kT Aox (9)

where C is the concentration of the redox-active species, EO is the most probable energy position, and A. is the reorganisation energy.

2.2 PHOTOELECTRON SPECTROSCOPY

The density of states distribution and electric potential distribution at interfaces can be directly measured by photoelectron spectroscopy [28-32]. In photoelectron spectroscopy (UPS, (S)XPS) high energy photons are used to emit electrons from the valence band states or core levels of the sampie. Usually a division is made between XPS (excitation by X-rays, most often used sources: Mg Ka. : 1253.6 eV and Al Ka. : 1486.6 eV) probing inner orbitals and the valence band region and UPS (excitation by UV-light, most often used sources: He I : 21.22 eV and He II : 40.82 eV) probing mostly the valence band region. Such a division is outdatcd by the development of continous light sources from synchrotrons which provide photons in the energy range of 10-10000 eV. The emitted photoelectrons are analyzed with respect to their current density in dependcnce of their kinetic energy Ekin and emission direction I(Ekin,e,cp), (electron energy distribution curve, EDC). For solids Ekin(i) is given by the equation:

(10)

Ip(i) denotes the ionization potential and EB(i) the binding energy of the electronic state (i) referred to EF and <I> is the work function .

For semiconductor sampIes EF is situated in the forbidden energy gap and therefore is not defined in the spectrum via the onset of photoemission as for metallic sampIes. The first expected contribution to the emitted photoelectrons usually originates from the valence band maximum EYB. The position of EF can be defined via the Fermi edge of metallic

456

sampIes, if electricaI contact of the semiconductor to the reference is established. In this way the energetic distance of the valence band maximum to Ep can be determined, which for flatband conditions aIlows one to determine the doping of the semiconductor. The work function of the semiconductor <llsc is defined by the width of the EDC (L\E = Eso - EYB):

(11)

where the term (EYB- EF) gives the distance of Ep from the edge of the vaIence band EYB. As the position of EF at the semiconductor surface is dependent on the energetic condition at the interface, changcs in <llsc and EB F are observed in the photoelectron spectrum (see Fig.2) according to eq. 12:

(12)

eL\x describes changes due to surface dipoles aItering the electron affinity and thus only the secondary electron onset ESO ( Fig. 2b).

a) E.(VB)

b) :§" "E

" e !!!.

f .!!!

c) .E~~~~ ____ ~-=~==~~ __ ~~~~

bin ding energy (arb. unils) ""=Ea= OeV

Fig. 2. Schematic changes of energetic conditions at semiconductor interfaces and its consequence for the valence band spectra. The changes in surface dipole only shift the electron affinity by L\X and therefore <llSC. Band bending eVb shifts both EBF and <llsc.

457

The term e Vb considers band bending which shifts the whole EDC and thus changes the binding energy EBF and the secondary electron onset Eso (corresponding to the work function <1» by the same amount (Fig. 18c). These values can be measured due to the high surface sensitivity of the photoemission techniques (IOA vs typically 100 A - 10000A for the width of the space charge layers). The different effects usually do not occur separately but adsorbates may change the several surface potentials simultaneously. Whereas it is not possible to separate them by Kelvin probe experiments (measuring only <1», in the photoemission spectra .M> is determined by LlEso and ~BP is only determined by eVb. For n-type semiconductors an upward bending of the energy bands corresponds to a displacement of the EDC towards Ep decreasing EB p as shown in Fig.2. For p-type semiconductors band bending leads to increased values of EBP.

Photoemission measurements on semiconductor/adsorbate interfaces may answer many questions conceming the distribution of electronic states and the potential distribution across the interface; one would also like to know these for electrochemical interfaces. In which way the investigation of model electrolytes in UHV may contribute to this problem will be presented in the next chapters.

3. H20 Adsorption on layered semiconductors

The He I spcctra of n and p type WSe2 (0001) in the course of H20 adsorption at 100 K (low temperature, LT) are shown in Fig. 3 [33]. Due to adsorption of H20 extra emission features appear at around 8, 11, and 14 eV which are assigned to the 1b1, 3al' and 1b2 orbitals of molecularly adsorbed H20 [34,35]. The energy differences of the orbitals as measured by SXPS indicate that HzO is adsorbed in ice-clusters [35]. Also the intensity increase of the HzO emission lines and the attenuation of the substrate emission suggest adsorption of H20 clusters, which is often found for molecular adsorption on mther weakly interacting substrates [14-17]. In XPS or (S)XPS the substrate emission peaks do not show any chemical shifLs due to surface reaction. When the WSez sampies are annealed at mom temperature (RT) the molecularly adsorbcd HzO is completely desorbed restoring the spectra obtained for the clean surface.

The spectral series on p-WSe2 (Fig. 3) shows a strong shift of the valence band maximum to larger binding energies with increasing HzO coverage which is due to band bending by formation of a depletion layer (saturation value 0.8 eV). In addition the work function as determined from the secondary electron onset is drastically decreased (1.45 eV). The binding energy and work function shifts 6<1> are summarized in Fig. 4a. The difference between the curves is due to the change of electron affinity (0.65 eV) induced by dipole changes 6X at the semiconductor/adsorbate interface. The changes are established very slowly which is related to the strong clustering on the van der Waals plane [17]. For n-WSe2 only a slight increase of the measured binding energies is observed (0.2 eV) probably due to the formation of an accumulation layer (compare Fig 4b). In addition we observe a considerable decrease of elcctron affinity as for the p-type sampie (Fig. 4b).

458

10.5L

7.5L

6.5L

6L

-20 ·15 -10 -5

binding energy I eV

Fig. 3. VaJcnce band spectra of the WSe2/H20 interface for increasing amounts of the adsorbate. The change of band bending eVb and work function cI> is indicated.

[eV) 0.0 0.

-0.2

-04

-0.6

-0.

·1.0

·1.2

·1.4 ~ .. • "'<I>

·1.6

4 6

exposure I L

p-WSe,/H,O (WSB)

10

[eV)r-------------, n-WSez /H20

0.0 • (WS5)

-0.2

-0.4

• "'<I> -0.6 I • ev~1 -0.' ""'r--,--,------r---,-,---,J

3 exposure/L

Fig. 4. Change of interface potentials at the WSe2fH20 interface, a) p-WSe2, b) n-WSe2.

The SXP-spectra of the n-InSe (0001) surface in the course of H20 adsorption at LT also leads to extra emission features of molecularly adsorbed H20. Again a clustering of the adsorbed H20 is evident, and areaction of the adsorbate with the substrate can be excluded based on SXPS of the core levels. At RT H20 is completely desorbed restoring the original spectra. As for WSe2 the work function is considerably decreased by 0.9 eV due to adsorption. A small increase of binding energy (0.2 eV) indicates the formation of a weak accumulation layer (Fig. 5a). Unfonunately the available p-type InSe sampies show a strong shift in the position of the Fermi-level to n-type conductivity after cooling the sampies to LT. For this reason we studied the adsorption of H20 on p-GaSe which is isoelectronic to InSe. Again H20 forms molecularly adsorbed clusters on the (0001) plane without reacting

459

with the substrate. The changes of binding energy and work function are shown in Fig. Sb. Similar to the previous case we observe a strong decrease in work function (0.9 eV) but in contrast to WSe2 the binding energy shift due to band bendi\lg is considerably reduced (0.3 eV).

[eV]

0.0 _ ... InSe/H20 [eV] • 0.0 .... • • p-GaSe/H2O

·0.2 , • • • -0.2 ~ • • • (GaSe3) • • • •

-0.4 • •• ,_ ev~, ·0.4 • • -0.6 • M' ·0.6 ..

~ ·0.8 -0.8 • • • - eVbb • • -1.0 ·1.0

4 6 '0 3 4 exposure/L exposure/L

Fig. 5. Change of interface potentials at the n-InSe/H20 (a) and p-GaSe/H20 (b) interfaces.

On WSe2 adsorbed H20 clearly behaves as electron donor in our adsorption experiments. A strong downward bending of energy bands is only observed with p-type sam pies whereas on n-type sampies mostly the electron affinity due to the H20 induced dipole is changed. These results are in good comparison to results for n- and p-type MoS2 [37], which also show band bending only with p-type sampies. A schematic model for the formation of the donating extrinsic surface state is shown in Fig. 6. For the observed band bending on p-type layered substrates electrons have to be donated from occupied electron states above the semiconductor Fermi level into the extended semiconductor space charge region and to occupy the acceptor levels. As the occupied orbitals of adsorbed H20 molecules are weH beyond EFSC (the positions are given by the valence band spectra) a direct electron transfer is not possible. Since the dissociation of H20 can be excluded we suggest the formation of an extrinsic surface state by the electronic interaction of the occupied lbl and 3al H20 MO's (02p based) [38] with the valence band states of the semiconductor substrate (W5dz2 based) as shown in Fig. 6. Such interaction of occupied states does not lead to binding interaction in molecules but may result in weak. bonding when electrons can be transferred to the solid from the antibonding states [39]. Interestingly we observe a much larger effect (0.7 eV) on p-WSe2 which agrees with the published value for p-MoS2 of 0.5 eV [2,10] compared to the considerably reduced value of 0.3 eV for p­GaSe. The observed difference in the amount of band bending seems to be determined by electronic factors as the (0001) van der Waals plane is structurally similar for all investigated layered materials. However, the electronic structure of the compounds is considerably different. The group VI B chalcogenides are d-band semiconductors with the valence and conduction band edges, which are mainly involved in charge transfer interaction with adsorbates, derived from metal d-states. As a consequence we expect a stronger electronic coupling of the H20 molecular orbitals to the semiconductor states as shown in Fig. 6. The valence band edge of GaSe are derived from the bonding a (Ga4pz)

460

molecular orbital of the Ga pairs and the Se4p levels [38]. The electronic coupling of these valence band electron states to the H20 orbitals is probably much weaker. As a consequence we expect a less pronounced charge transfer from adsorbed H20 to the semiconductor.

----=Er-------"E=t----=E+--.-Evac

3a1 --3a1

vac 1b2 Et,=17.5 eV

DOS (adsorbate) (surface molecule)

<1>=5.2 eV

W 5dxz.yz/Se 4p

W5dN,xy ___ EF

W5dl

Se 4p W 6s/5d

/-"-~_ Se 4s

DOS (semiconductor)

Fig. 6. LeAO diagram for the formation of an extrinsic donor surface state at WSe2IH20 interfaces.

In relation to semiconductor/electrolyte interfaces it is interesting to note that strong band bending is observed far H20 adsorption on the (0001) plane of the p-type MX2 semiconductors. Evidently the van der Waals plane cannot be considered to be completely inert with respect to electronic interactions with the solvent molecules. Our results seem to indicate that there is at least a weak electronic overlap between the adsorbate and the semiconductor conduction band states. As these are derived from the metal d-orbitals we take this as evidence for non-negligible metal electron density outside the chalcogenide plane.

Finally, the dipole contribution ö,X at the adsorbate interface will be considered. If the photoemission spectra of the H20/P-WSe2 interface in Fig. 3 are plotted with reference to the vacuum level (considering the changed work function 4», the ionisation potential of the adsorbate levels remains constant. We have observed a similar behaviour for all H20/layered chalcogenide interfaces studied. The H20 emission lines are always the same when referred to the vacuum level and are only shifted to the free molecule ionization potential of H20 vapor by a typical relaxation energy of 1.2 eV. The resulting interface potential distribution of the p-WSe2lH20 interface is schematically shown in Fig. 7 for the two reference levels. In Fig. 7a (reference level EF as measured in the photoelectron spectra) the band bending eVb due to charge transfer from the H20 induced surface state leads to

461

increased binding energies with increased coverage e. The increase in EBF of the H20 levels is even larger. If the energy levels are referred to the position of the valence band maximum (Fig. 7b) and thus to Evac• the dipole change .1X at the interface becomes evident. As the H20 levels remain constant vs. Evac• we conelude that .1X is changed at the interface between the SC and the adsorbate and may result mostly from a changed surface dipole of the substrate .1XSC due to the adsorbed ice-Iayer. Most of the H20 molecules are outside this double layer.

a) Reference level: EF

CB--~==-levbb(e)

EF_ VB

- --,,: b) Reference level: valence band maximum

Fig. 7. Schematic representation of interface potentials at thc WSe2fH20 interface referred 10 EF (a) and referred 10 EVB (b).

4. Br2 and Br2/Na adsorption on n-WSe2

The valence band photoelectron spectra (synchrotron induced. hv=2l eV) of the WSe:z!Br2, Na coadsorbate interface at different stages of the experiment are shown in Fig. 8 [40]. Br2 is molecularly adsorbed on the van der Waals (0001) plane. The typical valence band features of the Br2 emission lines are marked and assigned according to literature values and gas phase photoelectron spectra (EB F values: 1tg = 3.0 eV; 1tu = 5.2 eV; Og = 6.4 eV) [41,42]. In good agreement with previous experiments on halogen adsorption on layered semiconductors [9.10] we observe for different n-doping density a shift of the Fermi level always elose to the valence band edge (inversion) as a result of band bending eVb induced by the adsorbed electron acceptor Br2. EBF increases up to 0.9 eV for n=1017 cm-3 [10], whereas in the experiment presented (Fig. 8) eVb is only 0.5 eV due to the low doping of the semieonductor (n=1015 cm-3) . The work function <11 is increased by 1.7 eV and 2.0 eV, respectively. The substrate does not react with the Br2 overlayer, as is evident from the core level emission lines (e. g. W 4f bands, Fig. 9). We only observe a decrease of EBF due to band bending. The Br 3d emission lines are shown in Fig. 9. For the molecularly adsorbed

462

Br2 we observe the expected 3d 5/2-312 doublet (EBF 3dS/2 around 69 eV). The actually measured ESF values change with substrate band bending. For some

.~ " ~ i!:' .~

.~

·15 ·10 -5 binding energy / eV

Fig. 8. Valence banG spectra (hv=21 eV) ofthe WSc}/Br'}/Na interface.

Br 3d core level

WSe2 / Sr2 J Na WS17

hv"" 100eV

Na

Smin

3mi"

WS·2 /Br2 /Na hv::: 80 eV

WS17

Na/min

24 annealed

300K

16

12

9

7

5 3

Br2

1.5L

O.75L

cle8ved

Na/min 2'

annealed

16

12 7 5 3 1

1.5LBr2 1min

Sr2

1.5L

O.75L

F:=:~:::::=;==~~"i'== ..... - ...... --l cleaved -36 -35 -34 -33 -32 -31 -30 -29

binding energy

cleaved

-71 -70 -69 -68 -67 binding energy / eV

Fig. 9. Core level spectra of thc WSe2iSr'}/Na interface.

463

adsorption experiments we also observe a weak shoulder at 67.4 eV [40] which we attribute

to adsorbed B{ [43]. It is formed by electron transfer from the n-type semiconductor to the adsorbed elcctron acccptor Br2 to establish thermodynamic (electronic) equilibrium.

Tbe surface concentration of the reduced species (Bn was increased by coadsorbed Na. The valence band region (Fig. 8) shows a gradual transformation of the molecularly adsorbed Br2 to an adsorbed Br--species for increasing doses of Na. Tbe EBF values of NaBr are markcd and the assignment is given according to literature [41,42] (Pxy = 4.5 eV, pz = 5.9 eV). As the substrate core levels still exclude any chemical re action the Br formation is assigned to areaction of BQ with the coadsorbed Na. The Na 2p core level emission (EBF= 31.1 eV) is typical for ionic Na+ (compare [44]). For the final step of Na adsorption the molccularly adsorbed Br2 is completely transformed to NaBr and at this point all emission lines shift back to the original EB F values before the adsorption experiment was started (Fig. 8-10). These shifts indicate that the electron transfer from the n-type semiconductor to the adsorbcd elcctron acceptor Br2 is reversed and the band bending is reduced again. After annealing the sampie at room temperature for 12 h we obtain the original flatband conditions with NaBr staying adsorbed on the surface. If the adsorption experiment is pcrformed only with Br2 we observe a complete desorption of the physisorbed Br2 and of the adsorbed B{ and also in this case the original flatband condition is obtained again [10].

[eV] r--------__ :---=--~------_, 1.2

1.0

0.8

0.6

0.4

0.2

0.0 -w;~---_;_------,;-------------' I I

0.75L 1.5L Br2

I I I I I I Na! min 1 5 9 12 16 20

Fig. 10. Change of surface potentials of thc WSe2lBr2fNa interface.

Tbe above results give evidence that adsorption of Br2 as weH as the coadsorption Br2/Na onto n-WSe2 (0001) surfaces lead to energetic conditions which may be qualitatively related to the semiconductor/clectrolyte interface containing a Br:zjBr- redox couple. As discussed in section 2.1, thermodynamic equilibrium between the semiconductor and the adsorbate (electrolyte) is attained when the Fermi level within the semiconductor E~c == T\!C coincides with the electrochemical potentialT\~ of the adsorbate (electrolyte). For the adsorbate T\~ is determined by the energetic position, distribution, and occupation of the highest occupied molecular orbital of the reduced species (HOMO) and lowest unoccupied MO (LUMO) of the oxidized species (Fig. 1). For an electron transfer from the n-type semiconductor to the

464

adsorbate unoccupied eleclron states should be available below E~c. The LUMO of Br2 as electron accepting electron state is the 2au antibonding level. Its energy can be estimated to be 3.0 eV above the HOMO (ltg) based on recent electron energy loss data of molecular Br2 adsorbed on Fe [45]. Thus we conclude, in contrast to the H20 adsorption experiment, that '1 direct charge transfer from the n-type semiconduetor (with a low TJe or <1» will be possible to the adsorbcd Brz (with a high electron affinity of the LUMO in the adsorbate stage) as shown in Fig. 11. The energetic differeI}ce HOMO (ltg) to LUMO (au) of about 3 eV corresponds reasonably to the measured Es value of the ltg level of Br2 of 3.0 eV which implies that E~c is on the same energetic position as the 2au level of Br2' Therefore it may be concluded that electronic equilibrium of the WSe2l'Br2 adsorption interface is determined by the energetic position of this level (Fig. 12).

Fig. 11. Schematic representation of the charge transfer proccss to adsorbcd Br2 leading to Br- formation.

Tbe overall shift of the work function <I> of the semiconductor which is the sum of band bending (0.9 eV for high doped and 0.4 eV for low doped WSe2) and the increased electron affinity (1.1 eV for high doped and 0.8 eV for low doped WSe2) amounts to an overall change .MI> of 2.0 and 1.7 eV, respectivcly, and to a final <I> for the adsorbate interface of around 6.0 eV (4.0 + 2.0 eV for high dopcd, 4.5 + 1.7 eV for low doped WSe2)' We may now compare this value to the redox potential of a Br2/Br- redox couple referred to the vacuum scale. For electrolytes the standard redox potential EO and the

concentration (activity a) of reduced (Br- ) and oxidized (Br2) components are important for the definition of the electrochemical potential according to the Nernst equation:

(13)

The value of Eu for liquid Br2 is 1.1 V vs. NHE [45) and E6(~JBd is approximately 5.5 to 5.8 eV which is in good agreement with the above determined value for E~c in equilibrium with adsorbed Br2 (Fig. 12). We take this as evidcnce that aqsorbed Br2 on WSe2 (0001) can be approximatcd as redox couple (Fig. 15). 11~ '" 11 {Br;/Br-J is evidently determined by the energetic position of 2au. In the next step we would like to estimate the concentration dependence of E~c of the adsorbate interface with changed Br2/Br- concentrations. The existence of oxidized (Br2) and reduced (Br") species on the surface is formally equivalent

465

to a surface redox couple. In our experiments the surface concentration of Br- is increased by co adsorption of Na. When the low temperature (100 K) of the given experiment is considered one would expect about 10 mV change of equilibrium potential for every decade change of activity (concentration) ratio based on the Nemst equation. We can estimate the number of adsorbate species which may be involved in charge transfer. Forming a depletion layer of 1 eV only involves about 1013 cm-2 which is about 1 % of a monolayer. The additional potential drop due to the formation of a strong inversion (degeneration) layer involves 1014 cm-2 (about 10% of a monolayer) due to the sm aller distance of this double layer. The coadsoption of Na and reaction with the adsorbed Br2 involves about one monolayer (1015 cm-2). Comparing these concentration differences it is elear that changes of energetic conditions (e. g. band bending) can hardly be measured in such type of model electrolytes formed by co adsorption as they are below 0.1 eV. Only for the extreme case when nearlJ; a1l of the adsorbed Br2 is reduced by the coadsorbed Na will a back shift of EBr:z/Br- and EF C be observed experimentally. The constant level observed for EFSC for a wide Na codeposition range is taken as additional evidence that the surface co adsorption system behaves qualitativcly as a redox couple according to the Nerst eq. 13 (constant eVb in Fig. 10).

a) belore adsorption b) after adsorption c) electrolyte

Evac = 0

n - WSe2 WSe2' Br2 Br2 ' Br-

2 <I>cleaved <I>adsorbed 4.5eV

3 Mdo L\1tx_ .

4 > .l!!

E~ed/Ox ~ 5 MJ' 1:/ ~ 6

"t2ZZl

0> -Dau-c '6 7 .S: .0

EF 8 b 1tg L... _____

Fig. 12. Schematic representation of energctic interface conditions at the WSe2fBr2fBr- interface; a) befoTe the ad~orption experiment, b) after elcctronic equilibrum (see text for details).

In contrast to H20 adsorption the shifts of adsorbate E~ and surface levels of the semiconductor substrate are nearly parallcl (compare Fig. 8,9) as shown in Fig. 13a. We conelude from this that the contact formed between the semiconductor and the adsorbate phase (surface redox couple) behaves elose to the Schottky approximation (..:\XSC = 0) [27]. The changes in work function <I> due to dipole changes ..:\X do not occur in major parts at the interface betwecn the semiconductor and the adsorbate but at the adsorbate/vacuum interface (Fig. 13b). This result implies that the adsorbate Br2 is within the dipole induced by itself as also shown in Fig. 13b. A significant dipole drop between the semiconductor and the adsorbate phase can be excluded since the eVb and EBF shifts of the adsorbate levels are nearly parallcl.

466

a) Reference level: E F

CB __ ~"""'==-_ Ef

VB--",,~====--

b)Aeference level: valence band maximum E,,,,, 8>0

__ rr,

--rr,

Fig. 13. Schemalic repreSenlation of interface potentials at the WSe2fBQ interface referred to EF (a) and referrcd to EYB (b).

5. Summary and ConcIusions

Our resulls show that adsorption experiments to model semiconductor/electrolyte interfaces may provide much information on semiconductor/adsorbate interactions which are important in electrochemistry. Surface reactions, the related formation of surface states, and their consequences on interface energetics and especially band bending can easily be idcntified [38]. Non-reactivc interfaces charge transfer processes which induce band bending can also be idcntified and related to electrochemical interfaces as demonstrated for H20 and Br2/Na adsorption. The comparison of the Br2/Na co adsorption to Br2/Br electrolytes suggests a qualitative correspondence of the model c1ectrolytes to electrolyte solutions. We feel that a more quantitative comparison may be possible for more advanced adsorbate phases which approach the electrolytc composition (solvent addition) and state variables (e.g. room tempcrature). The invcstigations of semiconductor/model electrolyte interfaces offcr a specific advantage compared 10 metal electrolyte interfaces. The adsorbate induced changes due 10 charge transfer (Volta potential difference) will change the semiconductor space charge layer; the relatcd band bending eVb can be measured. The additional dipole changes I'1X can thus be dcduced from the measured value of 1'1tt>. At meta! surfaces this differentiation is not possible as the contributions of I'1X and 1'1'1' to 1'1tt> cannot be separatcd.

One may expcct to dClermine the dcnsity of states distribution (DOS) of the electrolyte directly by using photoelcctron spectroscopy. The DOS of the adsorbate electron states are directly given by the additional photoemission intensity in the valence band region (neglecting final state effects) (Fig. 1). In our experiments we did not get any dear evidence for a high concenlralion of occupied eleclron states induced by the adsorbate around the Fermi level. Due to the high photoemission intensity of the substrate in the EBF range of interest we were not able LO obtain reasonable difference spectra up tilI now. In addition, we still do not know how the low sampie temperature and the clustering of our adsorbates lead

467

to deviations in relation to elcctrolytcs. For a more realistic model of the electrolyte the solvent is also a crucial ingredient. Finally and probably most important, the halogens are no easy "reversible" rcdox couples (two equivalent redox steps involved) which may be described by the reorganisationaI broadening of only one electron state (2ou) due to its interaction with iLs surrounding [41 j. 1t is not possible to decide on these aspects of electrochemical theory based on the simulation experiments presented here. But as the surface spcctroscopies and espccially photoelectron spectroscopy may provide very interesting taols for a microscopic evaluation of interfaces we feel it would be worthwhile to investigate these qucstions in more detail. The ncxt steps evidently should involve the coadsorption of a solvent and probably thc investigation of "easier" (one equivalent) electrolyte/elcctrode models.

Acknowledgements

This work was supported by a grant of the BMFT. We would like to thank C. Pettenkofer and J. Lehmann from the HMI and the BESSY staff for their valuable assistence in the use of synchrotron radiation.

Referenees

1. D. M. Kolb, Z. Phys. Chern. 154, 179 (1987) 2. R. Kötz, in C. Gutierrez and C. Melcndres (eds), Spectroscopic and Diffraction

Teehniques in Interfaeial Eleetroehemistry, NATO ASI Scries C320, Kluwer, Dordrecht (1990)

3. E. M. Stuve, K. Bange, and J. K. Sass, in A. F. Silva (ed) , Trends in Interfacial Eleetroehemistry, D. Reidcl, Dordreeht (1986)

4. K. Bange, B. Straehler, J. K. Sass, R. Parsons, J. Electroanal Chem. 229. 87 (1987) 5. F. T. Wagner and T. E. Moylan, Surf. Sei. 206, 187 (1988) 6. T. Solomun, K. Cristmann, and H. Baumgärtel, J. Phys. Chem. 93. 7199 (1989) 7. T. Solomun, K. Cristrnann, A Neumann, and H. Baumgärtel, J. Electroanal Chem

1Q2., 95 (1991) 9. W. Jaegermann, Chern. Phys. Leu. 126,301 (1986); Ber. Bunsenges. Phys. Chem.

92, 537 (1988) 10. T. Mayer, C. Pettenkofer, and W. Jaegermann, 1. Phys. : Condens. Matter 3, Sl61

(1991 ) 11. W. Jaegermann and C. Pettenkofer, Ber. Bunsenges. Phys. Chem. 92. 1354 (1988) 12. C. Pettenkofer, W. Jaegermann, and M. Bronold, Ber. Bunsenges. Phys. Chem. 92,

1354(1991) 13. M. Sander, W. Jaegermann, and HJ. Lewerenz, J. Phys. Chem. 96.782 (1992) 14. H. Gerischer, in Physical Chernistry, An Advanced Treatise Vo. 9A, H. Eyring, D.

Henderson, W. Jost (eds.) Acadernic Press, New York (1970) 15. S.R. Morrison, Electrochernistry at Serniconductar and Oxidized Metal Electrodes,

Plenum Press, New York (1980) 16. V.A. Myamlin and Y.V. Plcskov, Electrochemistry of Semiconductors, Plenum Press,

New York (1967) 17. R. Mcmming, in Eleclroanal. Chem. Vol. 11, AJ. Bard (cd.) M. Dekker, New York

(1979) 18. S. Trasatti, in J.O'M Bockris, B. E. Conway, E. Yeager (eds) Comprehensive Treatise

of Electrochemistry, Plenum, New York (1980)

468

19. R. Parsons, in J.O'M Bockris, B. Conway, (eds) Modern Aspects ofEleetrochemistry, Butterwoth, London (1954)

20. S. U. M. Khan and J.O'M Bockris, J. Phys, Chem. 87, 2599 (1983) 21. H. Geriseher and W. Ekardt, Appl. Phys. Leu. 43, 393 (1983) 22. H. Reiss, J. Eleetroehem Soe. 135, 247C (1988); J. Phys. Chern. 89. 3783 (1985) 23. S. Trasatti, J. Eleetroanal. Chem. 209,417 (1986) 24. R. Gomer and G. Tryson, J. Chem Phys. 66,4413 (1977) 25. W. N. Hansen and D. M. Kolb, J. Electroanal. Chem. 100, 493 (1979) 26. E. R. Kötz, H. Neff, and K. Müller, J. Eleetroanal Chem. 215,331 (1986) 27. E. H. Rhoderiek and R. H. Williams, Metal-Semiconduetor Contacts, 2nd ed.,

Oxford Scienee Publ., Oxford (1988) 28. G. Ertl and J. Küppers, Low Energy Electrons and Surfaee Chernistry, Verlag

Chemie, Weinheim (1985). 29. H Cardona and L Ley, (eds.), Photoemission in Solids (Vol 1,2), Springer Verlag,

Berlin (1978). 30. B. Feuerbach, B.FiLton, and R. F. Willis (eds.) Photoemission and Tbe Electronic

Properties of Surfaees,Wiley, New York (1978). 31. E. E. Koch, Handbook of Synehroton Radiation,Yol la, Ib, North Holland,

Amsterdam, (1983). 32. D. Briggs (ed.), Handbook of X-ray and UV Photoeleetron Speetroseopy, Heyden,

London (1977). 33. T. Mayer, A. Klein, O. Lang, C. Pettenkofer, and W. Jaegermann, Surf Sei 269/270,

909 (1992) 34. P. A. Thiel and T. E. Madey, Surf. Sci. Rep 7 , 211 (1987) 35. D. Schmeißer, F. J. Himpscl, G. Hollinger, B. Reihl, and K. Jacobi, Phys Rey. B 27

3279 (1983) 36 K. Bange, D. Röhl, D. E. Goder, and J. K. Sass, Vacuum 33 (1983) 757 37. W. Jaegermann and D. Schmeißer, J. Vac. Sci. Teehnol..A5 , 627 (1987) 38. W. Jaegermann, in A. Aruchamy, Photoelectrochemistry and Photovoltaics of

Layered Semieonduetors, Kluwer, Dordreeht, 195 (1992) 39. R. Hoffman, Rev. Mod. Physics 60, 601 (1988) 40. T. Mayer, J. Lehmann, C. Pettenkofer, W. Jaegermann, ehern. Phys. Leu. 198,621

(1992) 41. M. Grunze and PA Dowben, Appl. Surf. Sei. 10,209 (1982) 42. PA Dowben, CRC Crit. Rev. Solid Stare Mater. Sei. 13, 191 (1987) 43. S.R. Morrison in c.G. Seott, C. Reed (eds.), Surfaee Physies of Phosphors and

Semiconductors, Academic Press, New York (1975) 44. A. Schellenberger, R. Schlaf, C. Pettenkofer, and W. Jaegermann, Phys. Rev, B 45,

3538 (1992) 45. PA Dowben, M. Grunze, and S. Varma, Solid State Comm. 57,631 (1986)

FUTURE PROSPECTS FOR TUE APPLICATION OF SYNCHROTRON TECHNIQUES TO INTERFACIAL ELECTROCHEMISTRY

C. A Melendres Argonne National Laboratory Materials Science and Chemical Technology Divisions Argonne, IL 60439, U.S.A

ABSTRACT: Results of this NATO Advanced Research Workshop (ARW) and future prospects for applying synchrotron-based techniques to problems in electrochemical science and technology are briefly summarized. There is general consensus in the workshop that there is great promise for the continued use of present techniques and the development of new ones. Examples of questions that may fmd answers, as weIl as potential research areas, are cited.

The present workshop was intended not only to review the present uses of synchrotron radiation in electrochemical science and technology, but also to look into the future of the field. The author writes with trepidation on the latter subject as he knows fully weIl that most predictions often turn out to be wrong! This may, in fact, be the reason why most authors in this volume, though requested to write a paragraph or two on the future of their technique, did not dare do so. The burden therefore has fallen on the author' s shoulders to try his best here to address the issue at hand.

A summary of the Workshop deliberations will be made briefly. As had been alluded to before in the Preface to this volume, the Workshop was very successful in accomplishing the objectives that the Organizing Committee had set out. The meeting was truly a multidisciplinary one with physicists, chemists, and materials scientists communicating with one another intimately in very pleasant surroundings. The participants represented not only NATO countries, but came as faras Japan and Israel; there were 52 participants, weIl over the average attendance in NATO ARWs. This author personally believes that this kind of meeting is extremely useful not only in promoting elose interaction between scientists, but also in advancing a particular scientific field; hence, it should be encouraged and more workshops should be organized. NATO investment in the conduct of scientific meetings and workshops without doubt go a long way towards peace and understanding among the participants and their home countries.

Most of the lectures presented in the Workshops are collected in this proceedings volume. Thanks to the diligence and cooperation of the authors. The 26 papers provide an excellent overview of the application of synchrotron techniques to problems in electrochemical science and technology. The techniques

469

C. A. Melendres and A. Tadjeddine (eds.), Synchrotron Techniques in Interfacial Electrochemistry 46~74. © 1994 Kluwer Academic Publishers.

470

roughly divide into 4 categories: (1) x-ray scattering and diffraction experiments carried out mainly by physicists in collaboration with electrochemists, (2) x-ray absorption spectroscopy done principally by electrochemists and materials scientists, (3) Infrared spectroscopy with a synchrotron source practiced by both physicists and chemists, and (4) UV photoelectron spectroscopy. The excellent papers by J. Robinson on synchrotron instrumentation and those of R. Parsons (The ElectrodelSolution Interphase) and J. Kruger (Nature of Surface Films) provide very good introduction to the various techniques and applications. The papers illustrate the power of synchrotron methods and delineate classic problems in electrochemical science.

The future outlook in applying synchrotron-based spectroscopic and scattering techniques to the solution of problems in electrochemical science and technology was examined in two round-table discussions of this NATO ARW. The first session was co-chaired by H. D. Abruna (Comell) and J. Robinson (Warwick), while the second was chaired by W. E. O'Grady (NRL). The ideas presented in these discussions are summarized here. It is unfortunate that one cannot capture the atmosphere and sometimes wild discussions that often ensue from such a gathering of top-notch scientists (especially between physicists, chemists, and materials scientists); so, regrettably, this artic1e will have to settle for the drab technical aspects. In addition, the author has taken some liberty in expressing his own personal views on areas that he thinks look promising (from bis own myopic viewpoint), but wbich have not been discussed extensively in the Workshop. The application of synchrotron radiation is continually and rapidly developing so that this attempt to fill-in-the gaps here is by no means complete.

A number of problems that would be amenable to study using synchrotron radiation have already been pointed out by R. Parsonsi) in his chapter in this proceedings volume. There is a general agreement among the participants that a complete description of the structure and properties of the electrochemical interface is needed. Among the questions to ponder are: What atomic or molecular species are present in the interface region? How are these species arranged? How are the valence electrons arranged? What are their atomic and molecular motions? What is the electron distribution in the metal? Of course, one is always interested in understanding the effect of applied electrical potential on the structural parameters as well as their relation to the rates and mechanisms of Faradaic processes occurring at the electrode.

Another major focus is the study of the structure of surface ftlms, i.e., surface phases and oxides that are formed by anodic oxidation and corrosion reactions. Structure and composition of the ftlms as weH as their transport properties and the role of defects continue to be of great interest to the scientific community. The dynamics of electrocrystallization, nucleation, and growth of surface films need better understanding.

This being a workshop dealing mainly with techniques, the participants spent considerable time on discussions regarding the appropriate tools needed to elucidate the various problems of interfacial structure that have been put forward.

471

The general opmlOn is that solvent structure can be fruitfully studied by synchrotron-based techniques such as far infrared spectroscopy (IR), x-ray absorption spectroscopy (EXAFS and XANES), as well as by x-ray diffraction (XRD). Solute concentration and distribution in the double layer region may be inferred from the results of x-ray standing waves (XSW), x-ray reflectivity (XRR), scattering and diffraction measurements. The interactions between the electrode and species adsorbed from the solution could be elucidated by XSW and IR techniques. The role of theory in data interpretation and in modeling structures is also emphasized. The need to compare and confirm surface structural information obtained by synchrotron techniques with the results of Scanning Tunneling Microscope (S1M) measurements and studies using non-linear optical techniques has been pointed out and is worth emphasizing. Information on the dynamics of atomic and molecular motions is of interest and may be gained from Infrared spectroscopy and S1M measurements. The Debye-Waller factor in EXAFS also yields useful dynamical information. Time resolved IR and pump­probe laser experiments coupled with a synchrotron source are currently being pursued and should yield very interesting and valuable information on the dynamics of molecular motions and reactions.

The structure and composition of surface films will continue to be an active area of investigation because of their technological importance. The nature of the surface species and the presence of defects are of prime interest and may be studied using x-ray and neutron reflectivity. The technique of reflection EXAFS or REFLEXAFS developed by Cortes2), et al., is particularly promising because of the ability to measure the film thickness simultaneously and the possibility of profiling the composition of the film as a function of depth. X-ray diffraction, IR, and standing wave techniques are additional versatile techniques for getting compositional information. Structural and morphological information can be obtained by x-ray microscopy and diffraction.

There are a number of other synchrotron based techniques that this author believes will fmd application to problems in electrochemical science and technology. They are in the process of continuing development and should be discovered by electrochemists and engineers in due time. Among these are imaging techniques i.e., x-ray microscopy,3) tomography,3) and, topography4)

Scanning transmission x-ray microscopy (STXM), with aresolution of about 45 nm, has been applied to image the calcium distribution in human cartilage5) and could probably be used also to image manganese sulfide and other inclusions in corroding steel sampIes "in-situ." Similarly, the spatial distribution ofmetal atoms in a supported catalyst system can be determined by x-ray tomography. For example, D' Amico et. al. studied the distribution of platinum catalyst in a zeolite support.6) The technique involved taking radiographs of a material at x-ray energies above and below the platinum Labsorption edge. Projection radiograph maps of the spatial distribution of In (IJI)' are subtracted to yield the metal distribution on the support The technique could be nicely used to study the distribution of supported platinum electrocatalysts for use in fuel cell systems.

472

The technique of topographic x-ray imaging in grazing Bragg-Laue geometry has recently been developed4) and used for the analysis of defects as a function of depth in a 200 nm thick ffixGa1_X layer. The depth resolution of better than 10 nm achieved for a strained layer is remarkable. The technique should find application in the study of defects in electrochemically formed surface ftlms.

Oxidation state mapping using microxanes has been demonstrated recently7) and should find interesting applications in the study of battery electrodes, for example. In this technique, 2-dimensional x-ray fluorescence maps are produced at 2 different monochromatic beam energies chosen to preferentialiy excite different oxidation states of the fluorescing element. The ratio of the images produces an oxidation state image. The mapping of Mn+2 and Mn+4 distribution around a fungus-infected wheat root has been reported to illustrate the technique?)

A very promising and versatile technique that is Just being developed is that of Diffraction Anomalous Fine Structure (DAFS).8),) This technique is something of a hybrid between x-ray absorption spectroscopy and x-ray diffraction. In DAFS, the intensity modulation in a diffraction peak are measured as the incident energy is varied through an absorption edge of one of the elements contributing to the structure factor. The experimental set-up is similar to a conventional diffraction experiment but with an energy-scanned incident beam. The reflection from a crystal plane contains oscillations similar to EXAFS in the energy region after the edge. The power of DAFS comes from the fact that it generalizes EXAFS from a energy scanning spectroscopy to one that uses photon momentum and energy at the same time. This double tunability adds important new capabilities to the chemical selectivity afforded by EXAFS. For example, it is possible to use different diffraction peaks to separately study the different phases of a multiphase material. Moreover, using DAFS spectroscopy with different photon momentum transfers, it is possible to extend the e1emental selectivity of EXAFS to inelude site selectivity. One can envision that the DAFS technique will fmd many applications in the study of battery electrodes "in-situ" during charging and discharging cyeles.

The use of synchrotron-based x-ray scattering techniques have already been amply demonstrated by several papers in this volume. The tunability of synchrotron radiation considerably extends the power of scattering techniques via the phenomena of anomalous scattering Le., the change of the atomic scattering factors with x-ray energy. The absolute position of x-ray absorption edges is sensitive to the chemical state of the absorbing species. Therefore, very elose to the edge, atoms of the same element but with different valences will exhibit different anomalous scattering factors. In principle, one should be able to distinguish the scattering within a molecule by atoms of the same element but different valences. Warner, et. al.,10) have demonstrated that scattering factor differences of from one to two e1ectrons between Fe+2 and Fe+3 in a Fe20 3 can be observed.

We are also using the technique of anomalous x-ray scattering in

473

specular reflectivity studies of copper-on-silicon ftlm electrodes "in-situ" in aqueous solutions as a function of potentialy),12) It is weil known that the extraction of meaningful physical parameters from the theoretical analysis of reflectivity data becomes very cumbersome as the number of layers in a multilayer fIlm increases. Using anomalous x-ray scattering near the copper edge, we are hoping to obtain a better defInition of the interfaces involved and to facilitate the obtaining of unique theoretical fIts to the experimental data.

Small angle x-ray scattering (SAXS) with synchrotron radiation is a technique that has not found much application in electrochemistry so far but which this author believes has tremendous potential. The technique is fairly well developed and is routinely applied to problems in condensed matter physics. 13-16) SAXS is useful for characterizing aggregates of small particles such as carbons, colloids, and alloys. The x-ray scattering pattern allows one to deduce and follow morphological changes, i.e., the size, shape, volume and mass of particles that compose an electrode. The analogue of SAXS in 2 dimension is diffuse x-ray scattering (DXS) from surfaces and interfaces. We are currently applying this technique to the study of localized (pitting) corrosion of metal surfaces "in-situ" under electrochemical potential control. We have recently demonstrated the feasibility of the technique and hope to gain a better understanding of the development of surface and interface roughnesses in electrochemical systems.17)

Moesbauer spectroscopy is another technique that will profit greatly from the use of synchrotron radiation. The application of conventional Moesbauer spectroscopy to electrochemistry is well documented in the literature. The use of synchrotron radiation as a source has the advantage that with its polarization and high intensity the study of small sampies should be possible. It also perrnits the study of nuclei which do not have a radioactive parent The work of Hastings, Siddons, et. al.18) constitutes ademonstration of the technique.

There are many other synchrotron based techniques which use soft x­rays and uv radiation, e.g., photoemission. Without doubt, they are extremely useful in "ex-situ" work under vacuum, like other conventional surface analytical techniques. We have chosen to emphasize here techniques that are capable of "in­situ" use in the electrochemical environment. The papers collected in this volume show some of the applications and techniques. We encourage those who have not used synchrotron techniques to look closely into the use of these very versatile techniques. We wish people success in using them. Have fun!

474

REFERENCES

1. R. Parsons, This volume, Chapter 2, p. 21. 2. R. Cortes, M. Froment, A Hugot Le-Goff, and S. Joiret, Corr. Sei.,.ll. 121

(1990). 3. G.P. Williams, in National Svnchrotron Light Source Annual Report for

1991. S.L. Hulbert and N.M. Lazarz, (Eds.), BNL 52317, Vol. I, April 1992, p.20.

4. M. Dudley, ibid, p. 43. 5. C.J. Buckley, G.F. Foster, R.E.Y.S. All, C. Scotchford, and J. King, in

National Synchrotron Light Source Annual Report for 1991, S.L. Holbert and N.M. Lazarz (Eds.), BNL 52317, Vol. H, April 1992, p. 104.

6. K.L. D' Amico, J.H. Dunsmuir, S.R. Ferguson, and G.B. McVicker, ibid, p. 110.

7. G.P. Williams, in National Synchrotron Light Source Annual Report for 1992. S.L. Holbert and N.M. Lazarz, Eds., BNL 52371, April 1993, p.21.

8. C.E. Bouldin, J. Woicik, H. Stragier, D. Yee, et. al., NSLS AnnuaI Report for 1991. BNL 52317, Vol. H, April 1992, p. 262.

9. J.O. Cross, H. Stragier, L.B. Sorensen, E.C. Bouldin and J. Woicik, ibid. p.263.

10. J.K. Warner, AK. Cheetham, and D.E. Cox, ibid. p. 134 11. C.A Melendres, H. You, V.A Maroni, Z. Nagy and W. Yun, J. of

Electroanal. Chem. 297. 549 (1991). 12. Y. Feng, D. Lee, M. Pankuch, C.A Melendres, and S.K. Sinha, NSLS

Annual Report for 1993. in press. 13. A Guinier and G. Foumet, Small Angle Scattering ofX-rays. J. Wiley, NY

(1955). 14. O. Glatter and O. Kratky, (Eds.), Small Angle X-Ray Scattering. Academic

Press, NY (1982). 15. R. Perret and W. Roland, J. Appl. Cryst.!. 308 (1968). 16. C.H. Tyson and J.R. Marjoran, J. Appl. Cryst.~, 488 (1971). 17. C.A Me1endres, Y. Feng, D. Lee, and S.K. Sinha, paper to be presented at

the March 1994 Meeting of the American Physical Society, Pittsburg, PA 18. J.B. Hastings, D.P. Siddons, U. van Burck, U. Bergmann, and R. Hollatz,

in NSLS Annual Report for 1990, BNL 52272, p.353.

INDEX

absorption coefficient, x-ray adsorption of CO adsorption of hydrogen adsorption of metal atoms adsorption, specific alloying elements in film anodic oxidation anomalous x-ray scattering

backscattering amplitude beam divergence beam extraction system beam lifetime beamline optics beampipe beam position bending magnet bismuth bolometer detector Born approximation Bragg reflection breakdown of passive film bromine buckets bunch size

carbon monoxide cation vacancles charge distribution of ions charge transfer composition of film conformal roughness coordination shells copper complexes copper oxides corrosion crystallinity of films crystallographlc analysis crystal truncation rod

475

182, 183 434 25 26 23 42 119 473

185,251 8 7 7 17 5 7 5,11 320 435 86, 140 70, 75 35,43 461 7 7

388,433 51 28 226 37,38,43 93 187,251 335 295 199 59 174 158

476

cyanide cyclic voltammogram of Au(100)

Debye - Waller factor defects density of states detectors differential capacity diffraction pattern Diffraction Anomalous Fine Structure (DAFS) diffuse layer diffuse scattering dipole moment double layer structure

electroactive polymer films electrochemical cell, transmission geometry electrodeposition electrode - solution interphase electron distribution electronic properties of films electroreduction electrostriction emittance EXAFS

Far Infrared absorption film - coated electrodes fluorescence focussing magnet Fourier filtering Fourier transform Fresnel transmission factor FTIR spectroscopy

Helmholtz plane hydrogen in film

infrared reflectivity infrared spectroscopy with synchrotron radiation inner layer insertion devices instrumentation for synchrotron radiation interface potentials interface roughness interfaces

414 160

187 50 466 18 21 110 472 22 85,89 23 426

340 159 133, 157 21 28 52 281 117 8 41,181,201,215,229,247, 263,265,281,288,295

417,421 335 199,215 6 187 190,192,252 88 408

21 40,60

396,422 387,401 21 2, 12 1 466 87 67

477

interfacial width 99 interferometer 391 interphase 21

kinematical approximation 67,75,110

laser, free electron 17 lattice 6 layered semiconductors 457 layered synthetic microstructures 373 light sources 9 localized corrosion 199,211 low energy electron diffraction 263,267 (LEED)

magnetic properties 40 manganese 329 manganese oxides 311 mechanical properties of films 56 metal oxidation 27 mirrors 18 Moesbauer spectroscopy 40,473 monochromators 18 monolayer 34

nickel hydroxides 248,253 nickel oxides 247

off - specular scattering geometry 100, 104 optical absorption spectrum 315 optical properties of films 52,57 ordering of adsorbed ions 28 orientation of moleeules 28 overlayer 72

pair correlation function 174 parasitic mode 2 particle acceleration 4 passive film 200,205 perchlorate reduction 423 photoelectron 183 photoelectron spectroscopy 451,455 phthalocyanine 281,282 place exchange 27 polarization 403 polymer film 93,336 proton injection 315

478

Prussian Blue

radial structure function radiation spectrum reflectivity REFLEXAFS repassivation ring current rocking cu rves Ruthenium Purple

scattering geometry scattering intensity scattering vector self-affine fractal surface semiconductor-electrolyte Interface Small Angle X-ray ScatterIng (SAXS) solvation of adsorbed ions spectral brightness spectral distribution spectroelectrochemistry specular reflectivity storage ring structure and breakdown structure of film super1attice superstructure surface films surface morphology surface reconstruction surface roughening surface states of semiconductors surface vibrational spectroscopy synchrotron infrared spectroscopy synchrotron radiation synchrotron radiation noise

thickness of films truncation rod

underpotential deposition undulators

valence and bonding state voltammogram

water adsorption

340

251 11 74,85,376 46,295 36 7 145 342

69 71 110 99 451 473 28 10 9 335 100,103,372 2,3 48 44 174 174 33,58 97,98 131 77 28 393 433 1,2,401 388

34 85

113,215,238,263,267,349,371 2,13,15,16,403

56 25

438,440,457

water layers waverector transfer wigglers

X-ray Absorption Near Edge Structure (XANES) X-ray Absorption Spectroscopy (XAS) x-ray diffraction x-ray fluorescence x-ray microscopy x-ray reflectivity x-ray scattering x-ray standing waves x-ray tomography x-ray topography

Yoneda wings

479

116,433 68 2,13,15,403

52,53,181,201,222,260,286,313 181,199,247,249,263,311 157,171 201 471 97, 100, 140 67, 97, 109, 129 349,352,371 471 471

88