Electron and Positron Spectroscopies in Materials Science and Engineering: Materials Science and Technology

10598.

Preface

The present volume, an outgrowth of the 1977 AIME Fall Meeting in Chicago, is concerned with recent advances and present-day limitations of established as well as emerging instrumentation for surface and interface probing useful to metallurgical applications. Most of the chapters of this volume were adapted from papers given by specialists in the various research areas who presented their views on the state of the art, sensitivity, and limitations of their particular techniques.

Three chapters of this volume deal with the more familiar subjects—Auger electron spectroscopy (AES) and electron spectroscopy for chemical analysis (ESCA)—both of which serve to determine the chemistry of the surface. Auger electron spectroscopy has served metallurgists and materials scientists for several years now. Current research in the field emphasizes increasing the precision of AES as a quantitative analysis tool. Although not yet as widely used as AES, ESCA has already shown its potential in many areas of chemistry. Instrumentation for both AES and ESCA is commercially available from a number of manufacturers.

The situation is quite different for two other techniques dealt with in this volume: exo-electrons (more specifically: photostimulated exo-electron emission, or PEEE) and positron annihilation (PA). Application of these techniques to metallurgical problems is very limited at the present time, but their future seems to be very promising.

Finally, three chapters deal with new developments in high-resolution electron microscopy (HREM), analytical electron microscopy (AEM), and extended x-ray absorption fine structure (EXAFS). Although these techniques are not surface tools in the true sense, they have shown great potential in studies of the morphology, chemistry, crystallography, and nature of interfaces, thus providing information on grain or phase boundaries in addition to that obtained by AES and ESCA.

We realize fully that we have not covered all aspects of the field. There are a number of techniques and tools available now or emerging that have not been discussed: secondary ion mass spectroscopy (SIMS), ion scattering spectroscopy (ISS), ion micro mass analysis (IMMA), ultraviolet photoelectron spectroscopy (UPS), low energy electron diffraction (LEED), to name but a few of the important ones. Their omission from this volume is purely a result of space constraints and does not in any way reflect on their potential future applicability. In fact, we believe that no single tool or set of tools will give us a complete answer to all possible surface problems. Thus future volumes will be needed with contributions of the present type discussing the tools neglected here.

To stay somewhat faithful to the title of this volume, we have chosen to arrange the subjects, treated in the present chapters, with respect to the energy regime in which the individual detection methods yield the information. This is schematically shown in Table I, which may serve as a guide. As editors, we certainly hope that these contributions, for which we would like to thank the respective authors, serve as a basis for further discussions and research. As may be seen in the following, some of the topics are quite controversial. All areas are still evolving, and in most cases we have in the truest sense of the word just touched the surface. However, to give some unified overview of the present contributions, we have tried to summarize the highlights of each area briefly in the following paragraphs.

TABLE I

THE VARIOUS TECHNIQUES DISCUSSED IN THE CHAPTERS OF THIS VOLUME. THE ENERGY REGIMES AT WHICH THE PHYSICAL PROCESSES TAKE PLACE

Exo-Electron Emission

The chapters by Buck and Pardee and Baxter are mainly concerned with photostimulated exo-electron emission (PEEE). As far as we can see now, two distinctly different mechanisms (for oxides transparent to the photons) are dominant: (a) an increased photoyield caused by oxide breakup (the oxide acts as an attenuator for the electrons) and (b) an increased photoyield caused by a roughness-enhanced coupling of photons into surface plasmon oscillations. For an improved theoretical understanding of PEEE, one needs more knowledge about the optical constants of the oxide and the metal substrate, as well as work function measurements with and without the oxide. Furthermore, Auger data on the type of oxide and its thickness on the metal are clearly required. Lack of such data leads to speculations that are mostly unfounded.

Positron Annihilation

As pointed out by Tao and Tien and Tien et al., measurements would be suitable for detecting any thermally or mechanically induced changes in the state of the material that lead to changes in these defect contents. However, an a priori knowledge of the nature and details of the alterations in defect contents resulting from these changes would be necessary to fully unravel the mean lifetime data. The current problems associated with the technique are the long times involved in the lifetime measurements, the requirement of stability, and the high cost. Some pointers toward overcoming these limitations are discussed by Tao and Tien.

Auger Electron Spectroscopy

Because of its extensive use, one should view Auger electron spectroscopy (AES) as the standard method for surface chemical characterization. The chapters by Finello and Marcus and Marchut and McMahon show that the sensitivity and spatial resolution of AES are excellent; current research attempts to make it a genuine quantitative analysis tool. Before such a task can be completed, certain questions on the interactions of electrons with the matrix have to be settled. Further complications arise from the ion beam–matrix interaction as one sputters the surface to obtain depth profiles.

Electron Spectroscopy for Chemical Analysis

Electron spectroscopy for chemical analysis (ESCA), or x-ray photon spectroscopy (XPS), has a distinct advantage over AES in that it allows one to determine the valence state of the analyzed species, as discussed by Grant and Housley. The technique is also more applicable to the chemical analysis of polymeric materials than is AES.

Extended X-Ray Absorption Fine Structure

Extended x-ray absorption fine structure (EXAFS) is only beginning to be applied as a tool to determine structural properties in materials science. As indicated by Knapp and Fradin, the usefulness of this technique is its unique capability to yield information on identities, numbers, and radial distances of near-neighbors around the x-ray absorbing atom. The authors discuss various types of sources available to study EXAFS as well as possible applications such as the study of local distortions around defects and thus the atomic structure of ordered and disordered materials.

High Resolution Electron Microscopy

High resolution electron microscopy (HREM) has a great potential for allowing the direct observation of the location of diffracting species, as is described by Gronsky et al. It is a complementary technique to the aforementioned surface techniques in that it provides a real space metallurgical analysis down to the atomic level. Specifically, crystallographic, structural, and chemical information can be obtained from the same region. Therefore, it is particularly useful in following solid state transformations and in analyzing the structure of transformation interfaces and grain boundaries.

Analytical Electron Microscopy

As discussed by Clarke, the analytical electron microscope (AEM) combines the features of the transmission electron microscope with those of the scanning electron microscope and the electron probe microanalyzer. Collecting the characteristics of the various types of emitted radiations (which, in some cases, have been discussed in the previous chapters) allows a complete analysis, both chemical and crystallographic, of selected regions of the material with a spatial resolution of a few tens of nanometers. Minimum detectable mass and minimum detectable mass fraction set limitations to the techniques, which are presently being studied theoretically in more detail.

The editors gratefully acknowledge the efforts of Mrs. Dagmar Buck for preparing the index, Ms. Linda Roberts for reference assistance, and the staff of the Publications Department of the Science Center, Rockwell International, for final typing of the manuscripts.

Mechanistic Basis for Exo-Electron Spectroscopy

OTTO BUCK and WILLIAM J. PARDEE,     SCIENCE CENTER, ROCKWELL INTERNATIONAL, THOUSAND OAKS, CALIFORNIA

Publisher Summary

This chapter reviews the status of intrinsic and extrinsic emissions from metals, which have undergone mechanical treatment in one form or another. It discusses tribostimulated, chemically stimulated, and thermally stimulated emissions. The chapter discusses the interpretation of photostimulated electron emission (PEEE) experiments from the theoretical point of view based on calculations and a series of experimental results. PEEE is a part of electron spectroscopy. The fatigue-enhanced PEEE of oxidized aluminum is enhanced by the absorption of the incident photon by the surface plasmons. Removing the oxide layer further increases the exo-electron emission since the attenuation effect of the oxide layers is absent.

I Introduction

II Review of Mechanically Stimulated Exo-Electron Emission

A Tribostimulated Emission

B Chemically Stimulated Emission

C Thermally Stimulated Emission

D Photostimulated Emission

III Roughness and Oxide Effects on Photoyield of Aluminum

A Theory of Photon Absorption by Surface Roughness on an Oxide Covered Metal

B Correlation of Fatigue and Photoyield

C Effects of Energy and Angle of Incidence of the Photons on PEEE

IV Discussion of Fatigue-Enhanced Photoemission

V Summary and Conclusions

References

I INTRODUCTION

For some time it has been known that mechanical deformation causes an electron emission, defined by Kramer (1950) as exo-electron emission. A general classification of the various types of emissions from metals was given by Ramsey (1974) who suggested the use of the terms intrinsic and extrinsic, depending on whether the observed emissions originate even with a clean metal surface or require an oxide or some other adsorbed layer on top of the metal surface. Himmel and Kelly (1976) proposed to classify the emission phenomenon according to the means by which the emission has been stimulated (chemically, optically, thermally, or mechanically) in which case Ramsey’s definition is still applicable. The present authors prefer this descriptive terminology, although any one single term may not precisely define the type of emission: Combinations of above terms (such as optically and mechanically, as in the case of fatigue-enhanced photo emission, for example) may be more appropriate, as will become evident in the following.

The present chapter reviews the status of intrinsic and extrinsic emissions from metals, which have undergone mechanical treatment in one form or another. Tribostimulated, chemically stimulated, and thermally stimulated emissions will be discussed in Sections II,A–C. For more details on these subjects, the interested reader is also referred to the paper by Himmel and Kelly (1976). In Section II,D, particular emphasis will be placed on the subject of deformation and fatigue-enhanced, photostimulated exo-electron emission—an area which, from the experimental point of view, has progressed remarkably over the last five years. This latter topic actually forms the bulk of the present and the following chapter by Baxter. The reason for the rapid development of this field is quite obvious: As is discussed by Baxter (this volume), it is possible to use photostimulated exo-electron emission (PEEE) as a method to monitor the state of deformation and fatigue of metallic structures. It also appears possible to monitor the size of the plastic zone of a fatigue crack, so that the information obtained by PEEE seems to be useful to advance our present knowledge about fracture mechanics, in which the plastic zone, and its size, plays an important role.

The interpretation of PEEE experiments is still controversial. Section III of the present chapter is an attempt to clarify this complicated subject from the theoretical point of view, based upon calculations (Pardee and Buck, 1976) and a series of experimental results as obtained by the authors recently (Buck et al., 1977b; Pardee and Buck, 1977a). The interpretation given in Section IV attempts to reconcile three important factors: the oxide layer on the surface of the metal, the work function of the material (with and without oxide layer) and the effects of the energy, and (to a limited extent) the angle of incidence of the stimulating photons.

The present authors suggest that, based upon this review, PEEE is a part of electron spectroscopy. At least in this case, the excitation of the emission is triggered by photons (similar to ESCA studies) whose (energy dependent) interactions with the material cause the ejection of the electrons (the well-known photoeffect), which is modified by the presence of surface roughness as shown by Endriz and Spicer (1971a,b) or by oxide breakup as discussed in the following chapter by Baxter. However, in contrast to the photon energy used in ESCA studies, which is of the order of keV, the energies for PEEE are in the range of a few eV only. This energy is just large enough to interact with electrons in the conduction band.

II REVIEW OF MECHANICALLY STIMULATED EXO-ELECTRON EMISSION

A Tribostimulated Emission

Spontaneous exo-electron emission from oxide covered metals undergoing plastic deformation or exposed to surface abrasion (wear) has been observed without any obvious chemical, optical, or thermal stimulation by Gieroszynski and Sujak (1965), Sujak and Gieroszynski (1970), Arnott and Ramsey (1971), Linke (1970), Linke and Wollbrandt (1974), and Rosenblum (1976). This tribostimulated emission appears to be associated mainly with cracking or rupture of a surface oxide film. Deformation studies in high vacuum showed that the emission occurs in bursts only while the material is deformed and at a rate which is proportional to the oxide thickness. The emission is weak with a rate of only about 10²–10³ particles/sec cm². Rosenblum (1976) showed that, in addition to electrons, bursts of positive and negative ions as well as photons can be emitted. He also showed that the maximum energy of the electrons emitted is roughly 6 eV for both NiO and Al2O3 layers. These observations seem to be consistent with a hypothesis, suggested by Arnott and Ramsey (1971), that the emission is caused by an elastic strain energy release during crack propagation in the oxide. If the energy is concentrated within a very localized volume, the lattice temperature could be raised sufficiently to create a microplasma in the oxide layer, the decay of which ejects the electrons. In an older model by Gieroszynski and Sujak (1965), Sujak and Gieroszynski (1970), and recently advocated again by Gieroszynski (1976), it was proposed that a charge separation and thus strong electric fields across a crack in the oxide may account for electron emission. Although Arnott and Ramsey (1971) presented good arguments against this charge separation model, the present authors regard the issue, at least in part, to be unresolved. In a recent review on triboluminescence (defined as the light emitted if certain organic and inorganic materials are stressed to the point of fracture), Walton (1977) discusses both the charge separation and the plasma models: The former is necessary to reconcile the observation of a dielectric breakdown of the surrounding gas and the latter to explain the presence of metal atoms in the gaseous phase (temperatures of close to 10,000°K seem to be possible). Thus it may be seen that both mechanisms are necessary to account for all triboluminescence observations. The similarities to tribostimulated electron emission are quite striking, and thus we have no particular reason to favor one model over the other to explain tribostimulation.

B Chemically Stimulated Emission

When the clean metal surface is exposed to chemically reactive gases, a time dependent electron current may be emitted. Such clean surfaces can be produced by evaporation of metals in a high vacuum and also by abrasion (or wear) of originally oxide covered metals, as has been pointed out by Gesell et al. (1970). Geiger (1955), McCarroll (1969), Gesell et al. (1970), Brus and Comas (1971), Kasemo (1974), and Kasemo and Wallden (1974) noted also that, in addition to electrons, ions and photons or both may be emitted, similarly to the emission by tribostimulation, discussed in Section II,A. Gesell and Arakawa (1972) found good correlations between the electron emission current and the change of the work function during chemisorption, suggesting an emission of electrons by a transfer to a conduction band electron of the energy released during adsorption, as suggested by Steinbruchei and Schmidt (1974).

Recently, Linke and Grabo (1976) found that both the intensity and kinetics of the emission depend mainly on the water vapor pressure present in the vacuum chamber, thus classifying water vapor as a chemically reactive gas. These authors postulated that the energy released by the chemisorption process may be fed into surface charge oscillations of the aluminum, the decay of which ejects the electrons, as suggested by Endriz and Spicer (1971a,b). One may view this model as a refinement of the previously mentioned conduction band model by Steinbruchei and Schmidt (1974). Such a polariton coupling is somewhat different from that in photostimulated exo-electron emission. The photostimulated process, as will be discussed in Section II,D, is a coupling between the radiation field and the conduction electrons and requires roughness to conserve momentum. The chemically stimulated emission coupling, in contrast, is coulombic and does not require roughness.

The effects of oxygen and water vapor on the work function of aluminum deserve some discussion. Linke and Born (1974) and Born and Linke (1976) determined the change of the work function before and during abrasion in high vacuum (10−10 to 10−11 Torr), using the Fowler method. They noted that the work function increased from about 3.2 eV for the oxidized surface to about 4.2 eV for the abraded surface (oxide removed by mechanical means). This means that the oxidized surface has a lower work function than the clean surface if the oxide was formed in the presence of water vapor. This is easily understood as a consequence of the dipole moment of the water molecule built into the oxide layer. It is thought that the first molecular layer on the aluminum surface is, at least in part, made up of oriented water molecules with the dipole tail (the oxygen end) on the metal surface. This acts as a parallel plate capacitor which can do work on an electron, given by the product npe, where n is the number of adsorbed dipoles per unit area, p their dipole moment, and e the electron charge. Thus the electrochemical potential (Fermi energy) of the electrons on the outer surface of the dipole layer is increased by V = np, which is roughly 2 V. As extensively discussed by Tajima (1970), e.g., it is known that an oxide formed in the presence of water vapor will include water molecules, and it is reasonable to expect that they will be, on the average, oriented with the dipole away from the surface. It was also shown by Linke and Born (1974) that exposure of the oxide-free surface to dry oxygen up to partial pressures of 10−2 Torr did not affect the work function, whereas exposure to water vapor (with and without oxygen present) lowered the work function, again to about 3.2 eV. Linke and Born (1974) concluded from these experiments that no oxide is formed without water vapor present. We suggest that there is no particular reason for a dry oxide to change the work function, and the differences, observed by Linke and Born (1974), may be due to just the presence or absence of dipoles in the oxide layer. It seems very desirable to repeat the experiments by Linke and Born (1974) with the surface chemistry monitored by in situ auger electron spectroscopy. The present authors conclude from these results that a lowering of the work function definitely enhances exo-electron emission for certain chemically stimulated processes. It is clear, however, that the energy necessary to eject the electrons will have to be supplied by processes such as described earlier in this section. We would also like to point out that very detailed studies by Grepstad et al. (1976) on the anisotropy of the work function of aluminum have been made recently. These measurements indicate that the work function of clean surfaces is in the range of 4.3 ± 0.1 eV, verifying the value observed by Linke and Born (1974). A lowering of the work function by about 0.2 eV after ion sputtering has also been observed, probably due to a decrease of the order in the surface structure.

C Thermally Stimulated Emission

Oxide covered metals show a thermally stimulated exo-electron current after being mechanically degraded. A typical experiment involves measuring a glow curve of the electron emission current as the temperature is increased. In contrast to photostimulation, repeating the temperature cycle will not produce further emission. Indications are now, as observed by Krylova (1971, 1973, 1974) and Kaambre (1974) and as discussed by Scharmann (1974) and Himmel and Kelly (1976), that even in this case the source of emission can be related to the presence of adsorbed gases. This is not surprising since many chemical species may not have found their lowest state of energy once adsorbed to a surface. Exo-electron emission, as discussed in Section II,B, may thus be retarded until higher temperatures are available. The subject definitely deserves more attention than it has received in the past.

D Photostimulated Emission

Mechanically enhanced, photostimulated electron emission (PEEE), the major area of concern in the present paper, with and without oxide coverage of the surface, has been the subject of numerous experimental and theoretical studies, particularly because the effect shows promise as a nondestructive testing tool. The authors would like to point particularly to the pioneering works regarding PEEE of Endriz and Spicer (1971a,b), Ritchie and Wilems (1969), and Maradudin and Mills (1975), to give some examples. For oxide-free aluminum it was shown that a roughness of the surface enhances the photoeffect (Einstein obtained the Nobel prize in 1921 for the explanation of the nonenhanced photoeffect) due to a coupling of the incident photons into surface plasmon oscillations. In parallel to these developments, Veerman (1969), Baxter (1974, 1975a,b; 1977; this volume), and Baxter and Rouze (1973, 1975, 1976) demonstrated that deformation processes in a metal could be imaged using exo-electron emission. These authors demonstrated that emission occurs preferentially in heavily deformed surface areas. They also observed that uniaxial tensile deformation of oxide covered aluminum enhances the photoyield appreciably whereas compression gives rise to a much smaller enhancement and therefore concluded that the freshly exposed clean surface (due to oxide breakup) is the cause of an exo-electron emission. Even after exposure to air, however, Baxter (1974) noted a residual PEEE which was slightly larger than the one prior to deformation. Baxter (1975b) also pointed out that PEEE after abrasion experiments is strongly time dependent, in contrast to emission after uniaxial deformation where only small time effects occur. He suggested that this difference could be explained in terms of a difference in the sticking coefficients of adsorbed gases to the various crystallographic planes exposed by mechanical abrasion over that on (111) planes, preferentially exposed by uniaxial deformation.

On the other hand, Thompson et al. (1976) noted that in vacuum the increase of the fatigue-enhanced PEEE current of basically oxide-free aluminum was similar to that of aluminum covered with a 6-nm oxide layer. The oxide was removed by ion sputtering which is thought to yield a clean surface and may thus be considered to be a special form of abrasion. These experiments seem to indicate that fatigue-enhanced PEEE cannot be solely understood on the basis of exposure of new, clean surfaces, although it is important simply because the attenuation of photoelectrons in the oxide is diminished.

Interesting results on PEEE from abraded surfaces exposed to low pressure oxygen and water vapor, respectively, have been presented by Linke and Born (1974). As has been mentioned before, the work function was found to stay at about 4.2 eV after abrasion, if the specimens were exposed to oxygen only. After exposure to water vapor, the work function decreased to roughly 3.2 eV. This resulted in photoemission from red light (about 1.8 eV) which, in a two-photon process, may supply about 3.6 eV. Exposure of abraded aluminum to oxygen, on the other hand, had basically no effect on photoyield (indicating very little oxide growth).

Perhaps it will be helpful to discuss briefly the two kinds of effects a dielectric layer can have on photoemission. Scattering processes in the oxide layer will attenuate an electron current as exp(– z’/L(E)), where the attenuation length L depends upon the electron energy E rather strongly (z’ is a coordinate normal to the oxide surface). L decreases with energy over the range considered in the present paper and has a value of about 2 nm at 5 eV, as determined by Smith (1975). Thus a decrease of the oxide thickness from d to (d – Δ) increases the photoyield by a factor exp(Δ/L(E)) which can be very large (typically a factor of 50 for a 10-nm-thick Al2O3 layer). On the other hand, the electron loss as a function of energy in a bulk oxide is proportional to Im(1/ɛ(ω)), where ɛ(ω) is the complex dielectric function of the oxide. With the Fermi energy at the middle of the Al2O3 bandgap, this loss function is very small for electron energies less than about 3.5 eV but will become very large and is continuing to increase by the time the electron energy reaches 4.5 eV. This means that L at 4 eV is much smaller than 2.8 nm so that the photocurrent will be little attenuated for reasonable ranges in oxide thickness. Conversely, changes in the work function affect the photoyield strongly only when the electron energy is not much above the work function (about 0.5–1 eV). Since the work function for clean aluminum is 4.2 eV and that for oxidized aluminum apparently equal or less than that (depending on the presence of water vapor), one can make a qualitative separation of the oxide effects into a work function dominated regime below about 4.2 eV and an attenuation dominated regime above 5 eV, with both effects being about equally significant in the intermediate regime. As will be discussed later, there is an additional effect of the oxide layer upon the energy of the surface plasmon, which is important above 5