AD-A246 329OFFICE OF NAVAL RESEARCH

Contract N00014-91-J-1896

R & T Code 413a006

Technical Report No. 57

Auger Line Shape as a Probe ofElectronic Structure in Covalent Systems

By

Dr. David E. Ramaker

Prepared for Publicatio DTIQ 't'V ELECT!E

in FEB 25199ZlI

Physica) Scripta SYGeorge Washington University

Department of ChemistryWashington, D.C.

January, 1992

Reproduction in whole or in part is permitted forany purpose of the United States Government

* This document has been approved for public releaseand sale; its distribution is unlimited.

92-04211

SsCUIr tLASSIPCATION Or HS.PO

REPORT'DOCUMENTATION PIAGE

23. SECURT CI5IIAIN AUThORJTY 3. TIISTRIUnI O/AVAILABILITY OF REPORTI

2b. DECLASSIFICATION1001NGRAOMS3 ScEDL Approved for public~ release;4. ERFRMIG OGANZATON EPOT NMBE(S)distributibn unlimited

4. PRFOMIN ORANIATIN RPOR NUBER S.MONITORING ORGANIZATION REPORT NUMBER(S)

Technical;IReport # 57

Ea. NAME OF PERFORMING ORGANIZATION 16tL OFFICE SYMBOL 7a. NAME OF MOINITORING £2RGANIZATIONDept. of Chemistry oft appkable)George Washington Univ. Office of Naval Research (Code 418)

Ec. ADDRESS (0y, State. aid ZI Cod) M, ADDRESS(ft, Stat. And ZIP CoCe)

Washington, DC 20052Chmsr Poga800 N. Quincy Street

_________________________________ VA 2221784. NAME OF PUNDING/SPONSOIRING Sb FIESMOL 9. PROCUREMENT INSTRUMENT IDENTIFICATION NUMBER

ORGANIZATION Oft applicawoe

-Office of Naval Rgsaarch7 Contract N00014- 91-J-1.896

8r. ADDRESS(Gfy, Sacvand ZIPCod.) 10. SOURCE OF FUNDING NUMBERS

PROGRAM IPROJECT ITASK V4 ORK UIChemistry Program ELEMENT No. NO. NO. R & IACCESSICNI800 Nth, Quincy, Arlington, VA 22217 1 4l3g006__

11. TITLE (Includo Stairnty CaM6cauo)

Auger Line Shapes as a Probe of Electronic Structure in Covalent Systems

12. PERSONAL AUTHOR(S)Dr. David E. Rasaker

13a. TYPE OF REPORT 13b. TIME COVERED T _ 114. DATE OF REPORT (Yea,.Matu,Day) tS. PAGE COUNTInterim Technical IPROM January' 1992

16. SUPPLEMENTARY NOTATION

Prepared for publication in Physical Scripts

I COSATI CDES -1IL SUBJECT TERMS (Comnuan o ,eir it necwnay and idenitity by block nume'JFIELD GROUP SUB-GROUP

19. ABSTRACT (Co, on revers if nhCawty MWd ideniy by block numaff)our goal has been to develop a generally applicable, semi-empirical approach for quantita-

tively interpreting Auger line shapes for covalently bonded systems. In this regard we exami

four main topics. First, because the Cini-Sawatsky theory was originally derived for initial

filled, single bar~ds in metals, we present our grounds for applying it to covalent systems.

Second, we exam-ne the problem of unfilled bands, and third, emphasize the importance of

satellites. The role of satellites are examined with regard to benizene, graphite, and the ne

carbon mater-ial, C 6, often referred to as "buckyball". Finally we present results of an

application to "car -dic" carbon on Ni, where fundamental new information was obtained concer

inq the nucleation of graphite at higher temperatures. In our discussions, results from the

interpretation of the CXVV Auger line shapes of five different gas phase hydrocarbons (methan

ethane, cyclohexane. benzene, and ethylene), five different solids (polyethylene, diamond,

graphite, "buckyball", and nickel carbides), and a molecularly chesisorbed system (ethylene/N

are included.

,0 DISTRIBUTIONAVAILABITY OF ABSTRACT Zl. ABSTRACT SECURITY CLASSIFICATION

MUN.ASSIIEDJUNUMITED M SAME AS RPT. 03 OneC USERS Unclassified223 NAME OF RESPONSIBLE INDIVIDUAL E22b. TELEPHONE (Indude Area Code) I E~c. OFFICE SYMBOL

Dr. Mark Ross (22 696-4R 4409 1

00 FORM 1473. 8aMAR 83 APR eattiea may tCe use until exhlausted. SECURITY CLASSWFCATION 0- -;41S %,GEAll other emnsans are ezloiete. T F , I-

Physica Scripta. Vol. T36. .0000 1992.

Auger Line Shapes as a Probe of Electronic Structure inCovalent SystemsDavid E. Ramaker

Deparinent of Chenstsry, George Washington University, Washington. DC 20052. USA

Received September 4,1991: accepted September 6,1991

Abstract 1. A "balanced" theoretical approach

Our goal has been to develop a generally applicable, semienprcal A theoretical prescription for generating the principal kvvapproach for quantitatively interpreting Auger hne shapes for covalentlybonded systems. In this regard we exmine four main topics. First, becamuse component of the Auger line shape can generally bethe Cim.Sawatzk" theory was onginally derived for initially filled, sngie expressed by the equation (63,bands in metaa&. we present our grounds for applying it to covalent I BE)=BE,[Put, R(P RA(E+61A, AUAIP 01 (1)system. Second. we examine the problem of unfilled bands. and third. ,, Icmphaue the unportanca of satellites. The role of satellites are examined In eqn. (I) the function A is the Cini-Sawatzky function, (7,with reprd to benstte. graphite. and the new carbon material C60, often 8]referred to as "buckyball'. Finally we present results of an application tocarbidie" carbon on Nt. where fundamental new informint;on was obtained p*p'(E)

concerning the nudetion of rtaphite at higher temperatures. In our dwi A(E, AU, p, p') -T (2)rassions, results from the interpretation of the C KVV Auger line shapes of (I - AUI(E)]I + AUirpp(Ej] (five different gas phase hydrocarbons (methane. ethane, cycloheane, which introduces hole-hole correlation effects, and distortsbenseen, and ethylene), five different solids (polyethylene, ditmond. grtph,ite. "buekybai. and nickel carbides), and a molecularly ehemuorbed the DOS self-fold. AUis the effective hole-ole correlatonsystem (ethylene/,10 are incuded, parameters and I(E) is the Hilbert transform,

Our goal has been to build a generally applicalble, semi. I(E) = J .p* )/(E - e) d. (3)empirical model for quantitatively interpreting Auger line

shapes for covalently bonded systems. This model must The Cini function, which distorts the DOS self.fold forhave a balanced validity and be relatively convenient to use, treatment of Auger line shapes in solids will be discussedso that it can be utilized for a wide variety of materials of more fully below (Section 2). In eq. (1) we (6, 9-11] havetechnological interest. Our objective is to enhance the suit- included additional arguments in A to make explicit theability of Auger spectral line shape analysis as a source of point that the total theoretical kvv line shape is a sum ofelectronic structure information, components with each I' component (e.g. the ss, sp and pp

We know that a complete solution of Schrgdinger's equa- components or more appropriately each multiplet 2s+ L)

tion for a two-hole final state in a molecule or cluster gives having an energy shillt, b , and a hole-hole correlationan excellent interpretation of the Auger line shape. But this parameter, AUu, (these two parameters are defined moreapproach is inconvenient, and can only be performed on the fully in Section 2), and with each component derived from asimplest of molecules. If the interpretation of Auger line fold of the Pt and Pr DOS (e.g. s or p). B is a normalizationshapes is ever going to provide a convenient and useful constant and the R,'s and Pur'S are core hole screeningmeans of extracting electronic structure information on a factors and atomic Auger matrix elements, respectively, towide range of materials, a theory must be provided which is be defined below.widely applicable, convenient to use, and balanced with The factors R, included in eq. (1) are to make the theoryregard to its validity and fundamental justification, Con- consistent with the final state rule for Auger line shapessiderable progress has been made, but much remain; to be [12]. The final state rule indicates that (1) the shape of thedone. individual Irs contributions should reflect the DOS in the

After initially presenting our "balanced" theoretical final state, and (2) the intenstv of each Irs cntributtonapproach. I will examine four topics. More extensive reviews should reflect the electron configuration of the initial state.of itr work have been published elsewhere (I-S. In For the ,evv line shape, the final state has no core hole. InSection 2, I will present our intuitive grounds for applying general it is assumed that the DOS in the final state andthe Cini-Sawatzky theory to covalent systems. This is ground state are similar, so that the spectral shape of p,important. because the theory was originally derived for in- should reflect the ground DOS. However, the imtial state intially filled, single bands in metals. In Section 3, I examine the kvv process has a core hole, therefore the integrated p,the problem of unfilled bands. and in Section 4 the impor- should reflect the electron configuration of the initial coretance of satellites. We examine the role of satellites with hole (CHS) state. The R, factors are defined,regard to benzene, graplite, and me new material, C,,often referred to as 'but.kyball". In Section 5, 1 present Rl= pcAs. dc p(s) de. (4)results of an application to "carbitdic" carbon on Ni, where

fundamental information was obtained from interpretation In most systems, the R, factors are similar so that they areo the line nape. Section 6 contains a brief summary and generally ignored. Effectively this ignores the "static" effects

..; -- -t I - .- -i.r,-,n . f core hole screenine. The R, factors are imoortant oniN

D. .Ramaker

when one orbital momentum component dominates the been shown previously (23] that final states involving thesescreeinig, This apparently occurs in many metals such as in different MO's have different: hole-hole repulsions. There-

Li and Be, and in the L23 VV line shapes for Na. Mg, Al fore. the p DOS must be separated in the Pcc and Pc or p,and Si. where the core-hole screening is dominated by the and p, components. We have given prescriptions for doingvalence s electrons as opposed to the p electrons. However, this previously (24,6].considerable disagreement exists as to the dominant screen- Several argumenits can be given for utilizing semi-ing charge in some zystems. 7or example, Jennison et at. empirically derived DOS, even for simple molecules (6].(13] on the basis of a semi-emptrcal theory argue that the First, most one-electron theoretical calculations do notcore-hole screening involves mainly the s electrons in Be, include electron correlation effects and therefore do not givewhile recent calculations by Almbladh and Morales (14) sufficiently accurate binding energies. Second, the semi-argue that the p electrons dominate. Recent results (14) also empirical DOS include approximatie widths for each orbitalindicate that the total Auger rates calculated from wave feature. Assuming the XES and XPS spectra utilized tofunctions perturbed by a static core hole are a factor of 2-4 obtain the DOS were measured at sufficiently highlarger than those calculated from ground state orbitals. resolution, these widths primarily reflect broadening due toThus static core hole screening is important for all Auger the vibrational state manifold of the final state which projectprocesses, but it is known to significantly alter the Auger onto the core initial state in XES, or the ground state inprofile just for certain systems. An analysis of the profile can PES. The DOS self-fold p*p(E) then has twice the vibra.shed some light on the nature of the core hole screening tional broadening consistent in first approximation with theprocess. Auger two-hole final state.

Consistent with our goal to develop a balanced, consis- Matthew has determined that if the core lifetime is muchtent. and convenient model, we have utilized empirically less than a vibrational period, and if the electron-phonondetermined atomic Auger matrix elements, Pb8, and usually coupling and linear dielectric response are linear, then thean empirical density of states (DOS), p. as well. The latter widths for photoemission and Auger processes are compar-also contains inherent widths, so these are also obtained able for the case of localized holes in ionic crystals [25). Onempirically, the other hand. for rare gases physisorbed on metal surfaces,

The atomic Auger matrix elements have been calculated the Auger width (FWHM) will be about 3 times greaterfor much of the periodic table within a ore-electron than the PES width. Other accurate calculations for the 02Hartree-Fock-Slater approximation by McGu.;e [15] and molecule indicate that the Auger width is about a factor of 2Walters and Bhalla (16], or in a Dirac-.:artrec-Slater greater than the PES width (i.e. that a PES self-fold is aapproximation by Chen and Crasemann C!" 18). A com- reasonable approximation to the Auger widths) (26). Thusplete review of calculated Auger transition probabilities has the self.fold of PES or XES data may not accurately repro-been published [19]. We have shown that the ratios duce the Auger widths, but it appears to be a good firstP,,.P,,, and P,,,/P,, for a filled s or p shell (i.e. s' or ps) approxmiation to them, at least for molecules when vibra-are remarkably constant with Z, the atomic number. tional broadening dominates.However, comparison with experimental ratios reveal that It is important to include the vibrational widths of theelectron correlation effects dramatically alter these ratios (by various orbital states for a quantitative interpretation offactors of two or more) so that the one-electron results are experimertal spectra. at least for molecules. Often sophisti-inadequate for use in line shape analyses. Therefore, etither cated techniques are utilized to obtain the one-electronelectron correlation must be included. such as by per- DOS, without considering the vibrational widths. After cal-forming a configuration interaction (CI calcuiation) or the culating the energies and intensities by theory, a barone-electron results must be scaled (5, 20). For all carbon diagram (i.e. zero widths) is then compared with the expert.compounds we used Pw : Pk,p: Pi,, = 0.8: 0.5: 1 (6]. mental data. Sometimes a constant broadening of all lines is

A basic conceot in AES concerns the nature of the density utilized. Using these procedures, the presence of satellitesof states IDOS) -4flected in the line shape. First, it is well. sometimes has not even been recognized. particularly whenknown that AES samples a site specific DOS, i.e. the DOS the normal kvv line shape accounts for all the major fea-specific to that atomic with the initial core hole. A self- Id tures.of the appropriate one-electron density of states (DOS) on The above gives a summary of what is meant by a "bal-this site. p(E), anced" approach. To obtain an accurat theoretical many-

body DOS. but then to ignore the v~.ths, or to use thep-p(E) = p(E- &)p(e) de. (5) one-electron atomic matrix elements, makes the effort in

obtaining the accurate DOS fruitless. To extract meaningfulthen represents a first approximation to the line shape (21]. information from the Auger line shape, reasonable DOS.Hence. almost all line shape analyses start with a determi. widths, and atomic mairix elements must be utilized. Wenation of the one-electron DOS. have most often obtained these empirically along with

We have often used art empirical procedure to obtain the strong theoretical support.DOS. Photoelectron spectroscopy (x-ray or ultraviolet, XPSor UPS) is often used independently or a combination ofx-ray emission (XES) and XPS. along with theoretical calcu- 2. The Cini-Sawatzky model for covalent systems.ations Furthermore. many of the molecular orbitals (MO's) The Cini-Sawatzky model has been the basis for under-

in t • aikanes have primarily either carbon-carbon (C-Cl standing correlation effects in Auger line shapes. In elemen-r on-hvdrogen iC-H) bonding character (22) or q or tal solids, two parameters determine the degree of

naracer !or the aikenes. In the Auger sDectrum. it has ocanzation of the CVV 1core-valerice-valencet two-nole

. Auger Line Shapes as a "be of Electronic Structure in Covalent Systems 3'

final state. If the effective Coulomb repulsion U_ is largecompared to the band width (U,_ > I"), the line shape will Ugg ' N' GObe atomic-like if r < U_ the line shape will be band-hke. Insystems where U_. 7 r. both atomic- and band-like cotrib- Og .-utons art; evident in the line shape (i e. correlation effects are 0important [7, 8]. N Ugg-Ugg, U e

The results of Cmi and Sawatzky were obtained from uti- g 'lizing the Anderson and Hubbard many-body models. The

Cini expression was derived assuming an initially filledsingle band so that it is not rigorously valid for partiallyfilled bands or degenerate bands (7]. Nevertheless, the Cmi- 0Sawatzky expression, in the absence of a better alternative, BOhas been used for metals, alloys, and insulators, and even Ub ' Nmolecules (i.e. systems for which the bands are not filled andwhich have degenerate bands), with apparently satisfactory Ubb'results. 00 UU.Ub e lThe Cmi-Sawatzky results can be simply understood by Ub-Ub b

considering a cluster LCAO-MO-CI technique wheremolecular orbitals (MO's) are constructed and mixed in a 0 0configuration interaction (CI) (27). For the moment con- AOsider a simple two orbital system which has two holes UN O ,present resulting from the Auger process in an initially filledstate (28]. The holes can be described by one-electron UNN Imolecular orbitals 0i=N(fj +f2) and 0,=N(fi-fA) 0 UNN m.UO = UeX(bonding and antibonding molecular orbitals constructedfrom ± linear combinations of atomic orbitals,f on atom I Fig.. Iluaraton of the W.ecive hole-iole repulions U,,. U,, and U:,or 2) with binding energy +V ande.-_V(e,(fiHf) (U'- AU in eq. (2)] and the corresponding interaction parameter. r. -. oranr 2 H) w hthbindingn er e V ialed the V (cvalnt InItr-V for the NO; amnL GO, BO. and AO refer to group (or eusterl, bondand V = !2(f, I H I f2> where V is called the covalent inter. and atomic oWital rmpmwvlyaction and H is the the proper Hamiltonian operator). Thisgives the Hamiltonian matrix as follows: Here V is the covalent interaction between nearest neigh-

2 h or AO's and can be estimated from the bonding-antibonding orbital energy separation. y is the covalent

2c 2Z + V + 6 UI - U1 (6) interaction between nearest neighbor BO's and can be esti-

0.I U1 1 - U12 2Z + V + 61 mated from the s and p atomic orbital energy separation. ris the covalent interaction between neighboring GO's; for

where U,, (f Iritf[) and U1 j - <fl fl r1' I./ f). example, the GO's are the planer arrangement of three N-OClearly if V < U:. - U1 , very little mixing occurs and the BO's about a single N atom. (In metals, the extended bandhole state M.O.' r 0, and 0. properly describe the locaiuza- width is determined primarily by V, but more commonly thetion of the two holes. In this instance the Auger line shape symbol r is used to refer to the extended bandwidth. Con-is molecular-like. If V > U1 - U1, the mixing of the sistent with this, r is used in the previous section to refer toconfigurations is complete and the linear combination the bandwidths in metals.) U., Ub, and U, are the effective0, ± .- ff orf' properly describe the localization of the Coulomb interactions between holes localized on a singleholes. As such. the line shape is atomic-like and gives Auger AO, BO or GO, respectively, and are schematscally definedintensities reflecting the DOS on the atom with the core- in Fig. 1.hole. If V Z U1a - U,2, of course intermediate mixing To intuitively understand this intermediate level locahza-occurs giving both contributions. tion phenomena consider Fig. 2. Figure 2, utilizing a sche-

For metallic behavior. U11, will be drastically reduced by matic one-electron DOS. ilustrates the covalent parameterselectron screening, so that AU U,,. For extended V, 'j, and r and the application of eq. (1) with increasinglysystems, the parameter 6 goes to zero. This is the essence of larger values of U. p*p in Fig. 2(b) is representative of thethe Cini-Sa%atzky expression. eq. (2). Auger line shape provided the a, a, x., and no bands are all

Dunlap et al. have generalized these concepts to multi- filled and all localization effects are negligible, ie. AU = 0.element covalent systems by providing criteria, Figures 2(c)-,r show clearly that distortion effects on each

AO: V <AU_.. subband are reasonably independent of the other regions ofBO: V > AU, < AUt, the spectrium until U is stfficiently large to encompass these

GO' 7 > AUt, r < AU,, other regions. For U = 0.1-.SeV, no sigmficant distortionEBO- r > AU,,, (7) effects occur indicating the EBO's best describe the final

\, state holes. For U = 1.5-2.5eV, the various I' (Ir = s. sp,for assessing the nature of localization in covalent systems etc.) subbands are distorted into narrower resonance-likewhere intermediate levels of localization can occur [28]. In features indicative of localization onto single GO's.these systems the localization can occur onto atomic, bond. U = 2.5--eV causes mixing of bands with bonding-rouu ,r -xenced band orbitals (AC BO GO. or EBOI antibonding character and demonstrates localization onto

- - I - "".'T t~~I - , Z.., I ',r--- -. --nn, -rrrk I A0,6 if 'hs ., ,,,,.

D..E.Ramaker

a a " " bands can be treated independently (e.g. the ssiS) andt-1-pp(IS) multiplets dcauple]. Furthermore for snmall Uis. theT t -v-_ different Is L multiplets arising-,from the same band

L [] apparently can be averaged together to give an effective ss.b , , as o P" e ro"* r €,," sp. and pp contribution (e.g. the 'S and ID multiplets from

, .- t p he pp contribution can b-- summed before applying the Ciniv V 4 4 expression).

We can generalize this picture for the hydrocarbons.e Examination of the wavefunctions for most hydrocarbonsU--1 show that the MO's can best be considered as linear com-

binations of CH. GO's (22]. It follows that for most of theC based ,vrzems GO localization is the best qualitative

d picture for characterizing the localization.13=2 Analysis of the Auger profiles and bonding in covalent

systems indicates that a correlation exists between the ionic,, . , bonding character and the nature of the localization. As the

bond orbitals polarize from say atom N to 0 in an NO3U=3 type cluster in Fig. 1, Ubb increases, and 7 decreases. The

increase of Ubb relative to y raises the extent of BO localiza.& "tion. We conclude that increasing BO ionicity increases the

U=n10 character of the localization from GO to EQ. Thus the moreionic BeO, (30) BN anti B20 3 (Q solids can best be char.acterized as exhibiting BO localization. The silicon tetra-

94 78 62 4 3halide molecules, SiX,, exhibit BO localization, consistent94 780 14 with the large Si-X electronegativity difference, in contrast

BINDING ENERGY (eV) with CF,,, which does not, consistent with the smaller elec.

Fig. 2. (a) Sehemaic one-electron DOS. p. and an illustraion of r. -. and tronegativity difference here (32]. The oxyanions NO; (33],Vas defined in Fig. I. Here r - 2eV, 7- SeV nd V - 30eV. (b) Corn. PO3 and SO-, regardless of cation, exhibit GO locahza-panion of p'p (dashed tine) and the hbet iranoit, lIE) (solid Iane. tion (2a]. In this latter case, it is expected that the holes(04) Compaion of p*p (dashed line) aith AE) (sohd line) obtained from canr. get off of the oxyanions where 'W.e A ere created byurahngleq.12)ih theU'sandicated '. " L3 VV process, so that GO ,calization is a at

._complL The Si L23 VV profile for S.02 exhibits structureare empty, as in most covalent systems, localization onto similar to that for the SiX, molecules, so that it also appearsthe atomic orbitals will not occur because the tnixing in of to exhibit some BO localization character (27].empty orbitals means the final state holes are being screened The application of the Cini expression to the empiricalby other electrons and hence the effective U is decreased. DOS for molecules is another approximation requirmng

Now consider diamond. It is well known that the best further justification. This ,3 problematic since the empiricalstarting point for considering the occupied DOS in diamond DOS contain the vibrati, tal broadeung which has nothingis to consider a linear combination of bond orbitals, since to do with the correlation problem. However, vibrationalthe bonding and antibonding a bands are so far removed in broadening is also present in the empirical DOS for solids;energy from one another (they are separated by a large band it just is not as evident since banding effects dominate.gap). Thus. we can say at the outset that AU,. < V (the Extended covalent solids contain banding effects which alsoAO-AO interaction parameter), and we will not have do not dictate the correlation. To illustrate this, let us con-atomic localization such as that seen in the metals discussed sider the DOS for benze with that for graphite The localabove. What about BO localization? The one-electron DOS .sp2 bonding around each C atom is essentially inc same andfor diamond has the s and p dominated peaks scparated by this is reflected in the similar gross features in the DOS. Themore than 10eV (29). These s and p DOS arise from the small differences arise because vibrational effects dominateclustering of rour bond orbitals about each C atom having s in benzene while banding effects (i.e, second, third etc.and p tymmetry (actually ai and t5 symmetry in the point nearest neighbor interactions) domnate for graphite.group for tetrahedral symmetry, Td). If the s and p bands However, the principal correlation effects are determined bycan be treated as separate bands, i.e. AUbt, < 7 (the BO.BO the local bonding. Thus neither the extended banding effectscovalent interaction parameter), then we do not have EQ in graphite nor vibrational effects in benzene (both broadenlocalization. This leaves us with GO localization, which is the major DOS features) determine the correlation distor-proposed for diamond. tion. It has been shown that the Cini expression numics the

We can state this in another %ay. When the zsiL multi- -ffects of a configuration interaction for molecules. Thus. weplet contributions arising from the GO's are resolved suffi- intuitively expect that the Cini expression has comparableczently so that separate features are visible, we can speak of validity for molecules and covalent solids. This is true pro.BO localization, since these separate features then corre- vide, the separation. AE. between the molecular levels isspond to configurations of holes on the same or different small compared with the vibrational broadening. AV This3O's When the multiplcts are not sufficiently resolved we does appear to be true for the molecules such as benzene.

,ocak of GO localization. In this case. the same -"'SL hexane. and larger molecules, but may not be appropriate,,aiets arism irom different bands decounle. anu the 'or ethylene. etc. involving three or less caroon atoms [n

Auger Line Shapes as a Probe of Electronic Structuire in Covalent SYstems 5

BENZENE C KVV

" C2H,(gas) ;I (a)

Exp. Ex(a).

in

W . Theo ITheory i -vVv -'v

~' - ,.. Akv-vvv ke-vve

Iv "j.

keVYC '~ ke-v ~ C21'l/NI(100) Tek.-s - 100 K

.5- -(b)

80 70 60 80 40 30 20 10 Excp., VvBIN0N ENERGY (*VI :/

Fig. 3. (a) Comaniato of the C KVV experimtental (34] and theoretical .f ''

Auger lines shapes for besesse (from Ref, (6]). (b) Ile total theoretical lineshape and each of the coniponents as indicated. The various contributions.* .

(kvv, kx.vv. k-vvv. ke-v. lie-et) were obtained as described in the text, * .- -,

3nd having iU'. 4s and relative intensuueasindicted inTables and kr ..I... I it.

70 s0 S0 40 30 20 10 Ev

thee sallr mlecles mutilet ar alo mre mpotan.BINDING ENERGY Wo)sthst hsaproaechleincertinlyer r cormplestiat Fig. 4. (s) Sane as for Fg. 3 but for ethylene gas (34] (b) Comparison of

so tat hisappoac is ertinl no copletly atifacory the expesiraeail and theoetal Asger line sihr for ethylene clientsfor If:.-= small molecules (6]. soebed on NXIOO) at WOOK Weibaded ethylense).~b Thtre component (VV.

Figures 3 arid 4 compare the optimal total theoretical line ve. xn'a) line shaWi wie obtained an descrbed in the test and in Refshape with the experimental ine shapes for ethylene and C63] 'The relative iialstte were obtained by [caat squares fit to theenxerine [34]. In general, the theoretical line shapes gener. experimental damL(FromnRA 631).ed by the formalism described agree nicely with the.penmental line shapes. Similarly good agreement is

obtained for other gas phase molecules riot shown. : . :or stood from the definition of AU = VUI - U12 . For verymethane. ethane arid cyclohexane A6. Table I suomr~i,"- short screening leng.,his, one might expect both U 1 and U 12the AU and i

5 parameters for the principal kvv components. to be reduced substantially, so that AU would be deceasedComparison of the AUl's for molecules and extended -CI10]. For long screening lengths, one might expect Us to

nolids in Table I indicates nomething about the nature of the be decreased more than U1 1, having the effrz- of increasingscreeeung processes in these covalent systems. Note that the AU. Apparently the latter is occurring in .hese materials.AU for the CC-CC contribution incsasi in the order The longer chain length in polyethylene and full threec~clohexane < potyethylene < diamord. Thit can be under. dimensional covalency in diamond suggests that the extent

Tasble 1. SuntmarY of A U and ci paranmeters ob~atned empirically for the theoretical kin, line shape

Moteeste AVt. leV) 6 e

Athones CII-CH CPCC CC-~CC CHt-CH CH-CC CC-CC

Methane 0 12Ethane I 1 0 i2 to) 10Cyctohenasne 3 3 1.25 9 99Polyethylente 3 3 1.25 0 00Diamond 2 0

Aikeies do on irs go as Ir

Effiviree 2 i 0 9 111Benn 2 1 0 8 6 6Graniite 2 1 U, 0 0 0"Suckyostr- 1 0 0 0 0

vroet Ret 161

D. E. Ramakr

of polarinzation should increase in the order cyclo- Some theoretical work has been reported for initiallyhexane < polyethylene < diamond. This increased polanza- unfilled bands by Treglia et aL. [38], Cini et al. (39-4i] andton then has the effect of increasing AU. For the alkenes, Liebsch [42]. We briefly summarize some of this work.the AU's are all the same. This suggests that the screening According to Treglia et al- the Cini equation is still applic-length is much shorter so that "full" screening already able for nearly filled bands (i.e. low hole density); however.occurs in ethylene. This is consistent with the more delocal- (E) must now include thi self-energy effects. In this pre-ized 7 electrons in the alkenes. scription, p(E) might be determined from the measured pho-

The 6 parameters in eqs (1) and (6) are interpreted as the toemlssion spectrum which includes static and dynamicdelocalized hole-hole repulsion (6, 28]. As the size of the screening effects, etc. (the latter introduces satellitemoiecule increases. Table I shows that 6 decreases. reflecting contrbutions) (43]. Cubiottt et al. (44] applied the model ofthe ability of the two final state holes to stay apart from Treglia et al. to consider the XVV spectra of several alloys;..ach other in the delocalized molecular orbitals. Note also unfortunately, they did not compare their results withthat for similar sized molecules. the 6's for the alkenes are experiment to verify their findings. However, Cini et al.smaller than for the alkanes. This may reflect the increased suggest that within the ladder approximation, the undressedscreening due to the i electrons. Note that the 6's are zero or bare one-hole propagator is more appropriate (this hasfor the extended solids, as expected, although a controversy become known as the bare ladder approximation. BLA)has existed for polyethylene (the problem has arisen from (41]. Cini has concluded for Pd (40] and graphite [37],energy calibration problems which will not be discussed both possessing unfilled bands, that the Cini expression canhere (35, 36]. satisfactorily be utilized with the undressed DOS, provided

Gri has recently applied the BLA theory (to be discussed AU is treated as a parameter to give an optimum fit tobriefly in tne next section) along with the full multiplet and experiment, and the correlation rffects are small. Appar-intermediate spin-orbit coupling theory, without further ently, the two holes tend to interact more as bare holessimplifying assumptions, to the C KVV line shape of graph- because the size of their screening clouds is larger than theite (37]. The theoretical line shape he obtained was amaz- range of hole-hole repulsion. Bonnet et al. [43] have irtro-ingly similar to that obtained empirically by us (10], when duced a procedure which removes any photoemission satel-we ignored the multiplet decomposition and applied the lite contributions so that the undressed DOS might teCmi expression directly to the full ss to the ss, sp and pp approximated empirically assuming the remaining staticcontributions (10]. It seems clear that the approximations screening effects are relatively small. We believe that Cini'sdiscussed above are indeed valid for these carbon based bare ladder approximation is, at last qualitatively, able tosyqems when the correlation effects are not too dramatic. handle all situations. Treglia's approximation may also beAltough the theoretical line shapes obtained by Cini and adequate provided the satellite contributions are removed.us were nearly the same, the resultant U's were very differ- Which of these approaches is better is however not yet clear.ent. We initially found U's of 0.6, 1.5 and 2.2eV for the nr, To intuitively understand some of the additional com-arn, and au components (10] (in a slightly different pro- plications which arise for half-flled bands, and how negativeccdure we (6] obtained the values in TableH namely 0, 1, U's may enter, we consider again the simple diatomic mol-and 2eV), while Cini found values of 5.5, 11 and IleV. ecule as in eq. (6) above, and compare the case when theWhat can account for these dramatically different U's but band is initially full on the left and when the band is half fullwith the same line shape? The AU's reported in Table I are on the right:interpreted as U, - Us,, i.e. appropriate for group orcluster orbitals as discussed above, and these values are ra- Full Halffullsonable in this context. I suspect the U's reported by Cini Iitial state 0? O 02

represent the single center Uti on a single carbon atom or Final state (A emptycluster orbital, and thus are also reasonable. Nevertheless Matrix element <lslk]ra[(4liii> <4,s:]r12 [lskl>more theoretical work needs to be done here to more funda- (8)mentally understand the U parameters obtained for highly )covalent systems. The Is and kl refer to the core and continuum orbitals

involved. The Auger matrix can be written in either the holepicture or the electron picture; i.e. we can enumerate the

3. The problem of unfilled bands electrons or the holes. Obviously, on the left, it is better toWihi The poxim andised aconsider the holes, while on the right better to consider theWithin the approximation discussed above. ovalently electrons, because in either case we then have an emptybonded systems generally have initially filled bands. This is band: empty of holes in the initial state on the left, andparticularly "rue for the a bands, which are generally far empty of electrons in the final state on the right. Therefore,removed from the a* bands. On the other hand. the separa- we need to worry only about final state hole-hole inter-tion between the a and n* bands is much smaller, so that action on the left. and initial sate electron-electron inter-the x and n' bands together may have to be treated as a action on the right. This initial state electron interaction onsingle band. In this case the a band is half-filled in the initial the right will increase the initial state energy, thus increasingstate and many addtional complcations enter. Sawatzky the Auger kinetic energy, and make the convetonal Uhas presented a simple illustration for nearly half-filled parameter come out negave. For U > W the two electrons,ands snowing that the usual theory is totally inadeauate will remain on separate atoms (antiferromagnetic) and the'r ieinhv corre!ate systems in this case. and that even u.uger ntensity goes to zero for less than one-half filled

Auger Line Shapes as a Probe of Electronic Structure in Covalent Systems *"7

both the intensity and the profile. This is to be expected, pounds, such as CrSe 2 and TsSe2 (47). Different suggestonssince by the nature ol the sum rules, the initial state must have been iven to account for these negative U's. DeBoeralways determine the total intensity (12]. et al. [47] suggested that the negative U's were caused by a

We have previously applied these intuitive ideas by uti- dynamic bipolaron effect involving the conduction electrons.lizng the Cini expression as depicted in Fig. 5 where the Others have proposed that these are caused by the potentialfinal state and the initial state rule prescriptions am com- of the core-hole in the initial state, i e. due to edge or non-pared (45] In the initial state prescription. appropriate for orthogonality effects [48, 49]. Ramaker et aL. have arguedless than or equal to half-filled bands, the Cini expression is that for the low electron density limit, i.e. nearly emptyapplied to the entire DOS, occupied and unoccupied por- bands, the negative U values can be interpreted as arisingtions. As U increases the occupied DOS self-fold'is distorted naturally from correlation effects in the initial state as dis-upward by a negative U and the total Auger intensity cussed above "45,46).(shaded area) decreases. This is in contrast to the final state We can summarize these intuitive discussions as followscase. whe:e the Jh:!ortion is downward, and the total inten- (I):sity is conserved becau,: ;t is applied only to the occupied (a) The final state %Le. ground state for CW processes)

part. The upward distortion in the less than half-filled case DOS is the most appropriate one in nearly all cases.occurs because the Auger process tends to stlt6 those lec- (b) It is best to use the uncorrelated DOS, or the corre-irons which exist on the same atom in the initial state. lated DOS but without the satellite contributions. Agree-

Evidence for negative U effects have indeed appeared in ment on which of these is best has not yet been obtained.the Auger lineshape of covalent systems involving the a (c) For nearly filled bands, the Auger profile apparentlybands. The n Auger contribution for the benzene molecule reflects the multiplets from hole-hole coupling in the finalshows an upward distortion (45]. It is believed this arises state that gives the usual positive U's.because the total n band (bonding plus antibonding) is half For significanty less than half-filled bands, the Augerfilled in benzene and hence a negative U is appropriate profile reflects the multiplets of the electrons in the initialwithin the initial state rule. Ramaker also has shown that state, that gives negative U's in the usual sense.the na contribution arising from the reconstructed clian (d) The initial state determines the total intensity in alldiamond surface within Pandey's a-bonded chain model can cases. For lesa than half filled bands, electron-electron coup-be explainrd by a negative U, which strongly reduces the nn . ling is critical to determining this intensity.contribution vs. the an coming from the surface (11). It seems clear that much more rigorous work is requiredFinally, some very interesting correlation effects have been to sort out these complicated correlation and screeningobserved in core EELS spectra for Ti and V metals which effects for initially unfilled bands. We anticipate much moresuggest similar effects (46]. work in this critical area in the near future.

The existence of negative U's has also been observed forthe transitio, metals on the left side of the periodic table, L.e.with less th.-. half-filled 3d bands. Indeed, the U becomesincreasing negative as the band becomes more unfilled (47]. The presence of satellite contributions in the Auger spectraNegative U's have also been found for some transition com- of molecules and free atoms has been known for some time.

but their importance (at least in solids) has been recognizedonly relatively recently (6, 50-54 . It has been shown thatthe pro es producing these satellite contributions are re

FS Rule IS Rule onatt excitauon. initial-state shakeoff and shakeup, anc

tand > 112 fled) teand112 ned) final-state shakeoff and shakeup (6].The process creating each satellite in the lineshape is illus-

iE) trated in Fig. 6. The notation indicates the particles in the

initial and final states before and after the hyphen (6]. Here,the "k" refers to the instial Is core hole, the "e" to the reson.antly excited bound electron, and v or n to a general valence

NlN (#or more specifically a-valence) hole created either by the"shakeup" or "shakeoff" process or by the Auger decay. The

AQ X principal Auger process is indicated without the hyphen

(kvv rather than k-vv) consistent with that used historically.

cMN.) We use kvv to indicate this principal or normal Auger con-tribution to differentiate it from the tota. KW experimentalline shape.

In Fig. 6 ke-vve refers to the resonant Auger satellite. Itarises when Auger decay occurs in the presence of a local.

u>O U COtzed electron, which was created by resonant excitation intoan excitonic or bound state upon creation of the core hole.

Fi, 3 Schemauc ilustraion of the final (WS) and Initial lis) state rules The ke-v contribution arises when the resonantly excitedapphed to a singie-band rectangular DOS. NI(. with peater and less than electron participates in the Auger decay. The kire-vvne and',i fi vaience 6ant reipctiheiv The DOS self-foid N.* and the Cm k-vi. satellites arise from initial state shake-up followed by

8 D.E. Ramaker --

kw ke-a ku -. Y k..n, k- e5-vrW k--

i0. Su-at~n sO ii *1 0 *0 1 lo. - ISM

0-pht" -45 -40 -35 *30 -25 -20 -15 -10 -50 5.0/.1n.. 0 . Y* 7 -~ n... 81ASHG EINGY(*V re. HOWM)

Fi..CcpsofhCVepnetintersclsgrn50 shapes for C,. (Fromn Ref. (56].) The hoe-vx (is. SU/partl and k-vvoe ifs. '

Fig 6 Summary of the various processes giving rie to the total Auger hone SUi) satellites were shifted by +53S and -s 535V respectively consistentshape. Core VB. ec' and CB indicate the core level. vale=se band (or filled with that indicated by the C ts XPS data F60] The relative intensities areoritalsi. excitotnic' (usually nt) electron orbital and condtsctiont band (or given in Table Itempty orbitalsl respectively. Spec (spectator) and Part (Participant) Indi-cate the subsequent fate of the resonantly excited electron durng the Augerprocess. Us and U~s indicate tmsts-tate and fintal-state and refer to the hoe.Tekvlnsap inctrtreetsdrtytestate to which the shakeoff event occurs relative to the Auger decay. NON hl.Tek- ieh ncnrs elcsdrcl hand N (N - P in eq. (0I] refer to the approximsate hoe shape, Ite. either a DOS; however, shifted by the binding energy of the excitedDOS self-fold, or jsst the DOS. with the relative size of AUJ to the Cmi electron (6]. The resonantly excited satellites appear onlyexpression (eq. (2)] indicated. The resonant satellites oneur Only under e1ec- with electron excitation, and are clearly present for ethylenetron excitation. At the bottom, the presence (yes) or absence [either defi- and beuzene (Figs 3 and 4). They do not appear in graphiteottely not present (no) or not evident lnAe)] of each satellite as tndicsted forbenzene. graphite. or C60. The presence of a small h-vvae satelhite for or C.0 because the resonantly excited electron propagatesgraphite ts indicated but not ctear away from the core hole before the Auger decay. This is

evident from comparison of the electron excited and x-rayshake-up upon filling the core hole during the Auger excited Auger lineshapes which are identical for graphiteprocess. The kv-vvv satellite arises when Auger decay occurs [10] aad C60 (56-58], but not for benzene (6).in the presence of a localized valence hole, which was The initial and final state shakeup satellites enter with,:reated via the shakeoff process during the initial ionization. large intensity for Cio. They are not evident in benzene nor'I w:s shakeup or shakeoff of a valence electron is an intrinssc the aikaies, because the shakeup probability .s significantlyphenomena resulting from the "sudden" change of the core smaller, as indicated by the C Is XPS spectra, and becausehole ?otenttal upon ionszatson. The k-vvv tern denotes the at least for benzene, the resonantly exctted contributions arefinal state shake Auger satellite, which arises when Auger larger and appear in the same energy region. The kneevvndecay occurs simultaneously with shakeoff of a valence hole. waellite appears in graphite, but only with small intensity,These latter five terms arise as a direct result of core hole and only from shakeup right near the Fermi evel (te. excita-

scenn.The ke-vve and ke-v terms arise because the tion of electron-hole pairs across the negs..Iy smtall bandAucrn estng eealyecte.y lcro xitto gap) (59]. The higher CL.-rgy xt to n* excitations are appar.Auger prloces ise eneally excitedbyelctooeciato ently not sufficiently localized in graphite to cause a satel-wniht alos the resonat le shtat eapaetlonitso lite. but they do appear quantitatively in C60 . C IS XPS

the sum of several intensities, namely that of the principal daarvlsahkepatit oabu15±0%whakvv contribution plus several satellites& The relative inten- shakeup energy around 5.5eV for C,0 (60]. Simple calcu-

si~es f te stelits ae gnerllyobtind b lest quaes lations reveal that thfei'4 on should parttcipate in thest'es f te stelits ae gnerllyobtind b lest quaes Auger decay about 47% of the time giving a theojeta st

fit of the several line shapes to the experimental spectra (6, eiaes-521. The results for benzene. ethylcee, and C60 are showniin mate for this satellite intesity of 046 in .im;e agreementFigs 3. 4 and 7. respecttvely. The cesults in Fig. 4 for ethyl.ine and mne concluston.' concerning the importance of satel- Table II. Swnmury of satellfie intensties in percent'lite contributions utilizing the semji-emptincal method tototally consistent with recent ab-initio Green's functions hvv

5 he-v he-we hoe-vx h.vvxe hyvoyv h-vvv

results reported by Ohrendorf et 41. (55]. Methane S1 0 12 0 0 20 17Figure 6 schematically indicates how we generate the line- Edisas 52 0 12 0 0 21 IS

shape for each satellite and compares the presence of these Cycilhesne 54 0 8 0 0 19 19satellites for benzene, graphite, and C60. Table 11 sum- Polyethylene 67 3 1i 0 0 17-21 0marizes the intensities of cach satellite for all the carbon Dtimontd 100 0 0 0 0 0 0

sytm tdid nFLL6 Nidiae h nra O thylene 50 2 13 0 0 210 issytesstdid I Fg 6 J' adcaeste ora DS Benzenei 56 1 0 0 211 16

self-fold weighted wtth atomic matrix elements and with the Grapltite 93 0 0 7- 0 0 0optimal U's and/or 6's includc-d (i.e. eq. (1) with the U's and lluckyball 67 0 0 16 Is 0 0

sn Table 1). 'From Ref. (6]The kc-vve lineshape also reflects a DOS self-fold but

er Line Shapes as a Probe ofElectronc SucaeinCovalntSsem '9

with that found experimentally (Table II). Furthermore, the imated in Fig. 4(b) by the Ni L5 VV Auger line shape.final state shakeup k-vv7e satellite should then appear with Although the latter two components are facilitated through0.13 intensity (0.5"(l - 0.15)], slightly larger than the 0.8 an intra-atomic Vx* and x*r Auger process, respectively,found experimentally, they ultimately appear inter-atomic in character because

We note that the relative intensities of the kv-vvv satel- one or both holes ultimately end up on the substrate (63].htes for the 6 molecules listed in Table II are essentially all The relative intensities of the three components can bearound 20% to within experimental error (6]. Methane is uniderstood within the final state rule. The electronic con-isoelectronic with the neon atom. The shakeoff probability figuration per carbon atom in the ground state of chem-for neon has been both measured and calculated to be sorbed ethylene, assuming charge neutrality, is nominallyaround 21%. This is in excellent agreement with that found a x

1'~xr ', where the x indicates the 7n bonding and s*

for all of the carbon systems [61, 62]. Table II shows that back-bonding charge transfer involved in the interactionthe empirically determined intensity for the k-vvv satellite is with the metal substrate. Upon creation of the core hole, wequite constant around 17% in close agreement with that expect that the valence electronic configuration assumingexpected theoretically (i.e. 0.21"(1 - 0.21) = 0.17]. charge neutrality, becomes a

3rtn,, or V'n*,, where y is the

The kv-vvv satellite does not appear in graphite, net charge transfer in the presence of the initial core hole.diamond, and C6 , because the shakeoff valence hole is 'not The relative intensities of the components, VV: Vt*: n'n*sufficiently localized to witness the Auger decay [10, 11, 563. should then be 16: 8y: ?, or upon including the Pill matrixIn the presence of a core hole, the occupied valence band elements 13: 7y: y. Best agreement with the resultsDOS of diamond indeed does not exhibit any bound states obtained from the fit to the experimental line shapes(9]. on the other hand, the DOS for polyethylene in the (56: 34: 10, total normalized to 100) is obtained whenpresence of a core hole does exhibit narrow peaks indicative y - 1.3, which gives relative intensities of 55: 38 : 7 (63].of bound-like states, consistent with the the kv-vvv satellite The 1.3 total electron transfer is consistent with the 1.3 coreobserved for polyethylene. The k-vvv final state shakeoff hole screening elections found in benzene as determinedsatellites are not evident (although surely present) for the from ab-initio theoretical calculations [64].solids because these are removed along with the extrinsic Similar interpretations of the Auger line shapes forloss contributions in the deconvolution process utilized to ethylene/Ni at different temperatures have been reported bygenerate the lineshape. Ramaker et 41. (63, 651. Most interesting of these results is

In summary the correlation effects in benzene, C60 , and the evidence that the line shape at 600K is not representa-graphite are the same. The 1s'in

* resonant and Is'Iv-1 tive of a trte carbide, since considerable C-C bonding

shakeoffexcitations are delocalized in times short relative to character is reflected in the line shape. Similar C KVVthe core level width in both graphite and C6o; however, the Auger data reported by Caputi et al. reveal further changesIs-It In shakeup exci:ation is localized in Cio, but not in the line shape around 620K [66]. Quanutative interpre-graphite. Finally no evidence exist for local spin ordering of tations of these data suggest that the amount of C-Cthe it electrons in C60, as seen on the surface of diamond bonding character decreases at this temperature (65].(see Section 3). We might have expected this because of the Recently, CEELS data has shown that the C-C bondingreduced non-planar p-p interaction in the it band of C60, below 600K corresponds to C, (n mostly equal to 2) hon-but apparently this reduction is not significant (56). zontally bonded to the surface, and that above 620K a sig-

nificant fraction of the C, dissociatet; however, at high

5. Molecular adsorbates coverages, some of the C2 s flip up vertical to the surface(67]. Comparison with theoretical calculations [68] and

Once a molecule is placed on a metal surface, localization of additional experimental data indicate that these vertical C2 'sthe hole on the molecule is no ;onger a fait accompli. The serve as precursors to the graphite nucleation sites (67].hole can escape via electron transfer from the substrate, i.e.charge transfer. It is therefore interesting to compare the

Auger spectrum for the free molecule with that for themoecular adsorbate. In summary, the chemical effects seen in the various exper-

We consider in detail here the results for ethylene/Ni at mental C KVV line shapes do not arise from one-electron100 K. Figure 4 compares the Auger line shape for gas phase effects, but rather from many-body correlation and screen-CIH, with CzH,/Ni at 100K (i.e. for i-bonded ethylene) ing effects. This is apparent because the DOS self-folds are(63]. The spectrum in Fig. 4(b) was excited by x-rays, so very similar to each other, in contrast to the experimentalthat no resonant satellites appt at. Charge transfer from the line shapes which reveal signsficant differences. Thus the dif-substrate into the n* orbit.. oc .urs to screen the holes, in ferences seen between graphite and diamond P',r exampleboth the core-hole inital olate ,nd the two- or three-hole result because diamond has just the o orbitals with a singleAuger final state. This charSe transfer has the affect of AU, graphite has both c and i orbitals with different AU'sdecreasing the AU and 6 parameters; the transferred charge for the a. ax and tn holes. The differences betweenplaying the role of the resonantly excited electron in the gas benzene, diamond, and C60 arise primarily from the differ-phase (63] Thus the kvv and kv-vvv contributions which ent satellite magnitudes present. On the other hand, thecomprise the intramolecular component (i.e. tencied the VV hole-hole correlation and repulsion eff ect are much dimin-componenti for the chemisorbed state are similar to the ished for chermsorbed systems because of metallic screeningke-vve and kvv in the gas. The Vn* component is similar to from the substrate. In the cheusorbed case. and only in this

,,,e. llimen i.s, a, .., im., -t~, . ., - m .......-r-& , the KV /.=__ 16"111 DOS%" Wif * LhOW-.f~~

10\ D. E. Ramacker

The review above of several areaa where Auger line shape Referencesinterpretation has provided considerable insight into local-ization, charge transfer, polarization, and screening pro. 1. 11makee. D. E, to be published in: "Critical Revmew in Solid Stateceases should convince m~any readers that the results and Materials Sciences" (CRC Press. Boca Raton. Fl. 199)

obtinale rom (hi aproch re ort th efor . Raiker. D. I pl.VSurTnfA. iCL21l1(989)obtanabe frm tis aproch re wrththe ffot. 2. Raneaker. D. 9.. Jp. Vc Sc )h. A114(19 9)

However. Auger line shape analysis has not realized its full I Itam&acr. D. E.. Scanning Microscopy Suppl. 4. 207 (1990)potential because a quantitative interpretation is still a for- 5. Ramaker. D 2.. in: TChenuistry and Physics or Solid Surfaces IV-midlaoie task in view of the many simultaneous processes (Edited by It. Vanselow and R. Howe) (Springer Series in Chemicaland the many-body effects reflected in the total line shape. 6Physics) (praget.Vedag Heidelberg (982). Vol 20. p. 19.

Indedoftn jst he xtrctin o th lie sape(i 6 Huuson, F. L and Ramaker. D.j. J. Chemts Phys. 87.6824 (198n)Inded otenjut he xtacinelastic lssesa (is e no .a Ciii. M. Solid Stale Common. 20. 603 (1976): Solid Slate Commas

removal of the background and ieatclss)snoan 29 691 (1977); Phys. Rev. B17.2788 (i979).easy task. 8. Sawazy, G. A. Pbs's. Rev. Letn. 39. 504 (1977); Sawaisky. G. A. and t

One way to simplify thesm two tasks is to limit the Lenelink. A. Pbs's. Rev. B21. 1790 (1980.number of variables or to ilmit the number of simultaneous 9. Huuson. F. L and Rumaker. D. 9.. Phys. Rev. W&5.9799 (198n)

proesss.I bliee hisis hedirection of the future for I& Houston. J. EL Rogers J. W., Rye. Rt. It. Hutson, F. L and Ramnaker.proese. n believ tuhipoes wi th e maeithsrain 1 D.LEPys. Rev. B34.121tS(1986). Comn63 3

AES.and hat uch rogrss wll b mad in his rea n ii Ramaker. D. E. and Hutson. F. L- Solid State Cma 3 3the coming years. The background and magnitude can be (1987)5controlled by positron induced AES, PAES (69-713 and 12. Rnimiker. D. E.. Pbs's. Rev. 3125.,734i10982).

coincident techniques (Auger-photoelectron coincidence 13. Jenios. D R,. Madden, H. H. and Zehner. . M. Pby. Rev. B21, J430(199%)spectroscopy, APECS (72]) The latter technique also con- 14 Almbtadli. C. 0. Morales. A. L and Grossmnin. G. Pbs's Rev. B39.

trol the initial state which ontrols the magnitude of the 349(999.initial state satellites. A second way of controlling the initial IS MeGune E. J. Phys. Rev. IN&. 1(1969); Phys. Rev A. 273 (1970),state is by utilizing a tunable photon source, such as a syi. Phs's Rev. A3. 1801(11971).chrotron. to resonantly excite a core electron into a bound 16- wallen. D-. Land Ehala. C. P., Pbs's. Rev A3.,1919 (197 1). Al. Data

(i~e vi dc-xciatin elctrn spctrscoy, DS (3)) 3.3010(970v; Phns. Rev. A4. 2164 0I97111slate (e.vad-xiaineetospcrsoyDE 7). 1.Chen. M. H. Laiman. E.. Crasemans. B,. Aoyai. M, and Mark. H.By measuring the spin polaizatson of the Auger electrons Phys. Rev. A19. 2253 (1979).one gains additional information about the possible process is. Chris M. H. and Cravema. B. Phys. Rev. AS. 7 (1973); Atomicwhich caused them (74). Spin polasized AES (SPAES) thus D., t and Nuclear Data Tables 24. 13 (197 9): Cheui M. H.. Crate.has two aspects; namely it assists in interpreting the normal mann. B. and Murk. H. Phys. Rev. A21, 436 (1980); Phys. Rev. A24.

N(E spctr an ithasthepotntil t prvid usfulinfr- 177 (1"91)N(E)spetraandit as te ptenialto rovie uefu inor- 19. Bambysek. W.. Crassaun. B, Fiash. It. W. Freund. H. U.. Murk. H.

mation in surface magnetism. Strong angular variations in Swift. C D. Price. It. E. and Rao P. V. Rev. Mod. Pbs's. 44. 716the Auger line shapes of adsorbed molecules have also been (1972.observed (75, 76). These angular variations are of tigifi. 20a Ramaker. D. E.. Hutsott. F L. Turner. N H. and Ma. W N. Phys.

cant help in determining the correct assignment of some of Rev. 3BU 257411986)~he inestrctue inthelin shpe.21. Lander. L.1. Pbs's. Rev. 911.13112 '53).!he inestrctue inthelin shpe.22 Jorgensen. W. L and Salem. L. -Meo Organic Chtemists Book ofWe know tht the Cini expression will continue to be OrtmWli (Academic Press. New York 1973).used on a wide variety of systems; even in the event of Some 23. Jennison. D It, Kerber J. A. and Rye. It. It, Pbs Rev. 25 1384

progress in developing a theory for unfilled and degcnerate (1992).bands. But much further work is needed to learn of its lim- 24. Macday. J. S. Dunlap. B. 1. Huon. F. L and Celbafen. P. Phys.

Rev. B24.476401981).tations and validity for a wider variety of materials. 23. Matthew. J. A. D_ Prutton. M. El Gomau. M. M. and Peacock, .

Most Auger line shape interpretations to date have been D. SIA. Surf. Iater. Anal. 11, 173 (19881.on homogeneous solids, molecules, or at least on well char- 26 Dunlap. B 1. Mills. r'. A. and Ramaker. D 9.. J. Chem Phys 75.300acleriztgi surfaces. And most interpretation schemes in the (1991)past rr mire a one-electron DOS. which is then utilized to 27. Itamaker. D. .. Pbs's. Rev. 321,.460911980),

genrat asel-fod or nsetin ito heCini expression. 28. Dunlap & L HtnsF.LIand Rm ker. D.E-J. Va. ScLTechol.geneate sel-fol forinsetionintotheIS. 36(1991).

The DOS are normally obtained empirically as described 29 Painter. P. S.. Elks. D. E. and Labiniky. A. It. Phys. Rev U4 3610above. But the future will demand a study of iuore complex (1971).practical systems. As the technology moves to more X0 Madden. H. H.. Sue) Sc. 126.80 (1983.complex "engineered" inhomogeneous materials such as 31. Haiike.0. and Mueller I-J. Vac Sei.Teciiol. A2.964 (94

mult-elmenal aloy, cmpostes marice, hgh e- 32. Rye. It. It. and Houston. J.9.. J. Chesm. Phy. 78, 4321 (1983).mult-eimenal aloy, cmpostes marice, hgh em- 33. Huuson. F. L. Ramaker. D. L- Dunlap. B. 1. Ganjer. J. D andperature superconductors, and interfaces, AES is going to be MurdyJ. S. J. Chem. Pbs's. 76. 2191 (1992).

increasingly useful because of its ability to sample a site or 34 Siegbahn K. Noedhing. C.. Johanison. 0. Hedmnan. J.. Heden. P F..element specific local DOS. Thus, the future of AES is Hairsm. K.. Gehas. U. Bersmark. T. Werme, L 0.. Manne. It. andbright, but the theory must advance to be able to handle Baer. Y. 'ESCA Applied to Free Molecules (North Holland Pub.

thee icrcsinlycomlexsysems Mch orkreminsto Co. New York 1969). p. 103; Rye. It. It. Mailey. T. 9.. Houston. J.9E.thes inceasiglycompex sstem. Mch wrk rmain to and Holloway. P. H..J. CThemi. Pbs's 69. 150411978). Ind. Eng. Chris.

be done. Prod. Res. Div. iM 2 (979L.3$ Rtye. R. . Iltys. Rev. 839. 10319(199)36 Turmer. N. H. Rimaker. D.2.- and ! lutson. F Liiio be published

Acknwlegemnts37 Cii. X. and Andrea. A. D. in: "Auger Spectroscopy sod ElrctronicAcknwlegemntsStructure" (Edited by G Cubiotu. G Mondto sod K Wandrltt I

Auge ie S;a4es4s a Probe c Eleerro,,i Svmu'tr in Cocaleff Sysum 11

59 CL-L.N -L Sm $7.41*1 119L n 5&. Rm~. M E dTmr-H..!be pabim4. C=r M.. mnd Verdean C. =: -Arlm Sn=s* =d Evnmnor 57. Tom& W. 3'L 0kbag At A. A-. Yoi. H8. K. V.mR iS A.r I

Sum=-r lEdnen by G. Ccbsomn G. M.onr ad K. Wxand) L. Airmn SC.34. V%cn 3 L- Rube: Y. =nd D= Lt X- L.ispc,-Vcaa. Borba 395) P. In2 Pkh Qin.S3.40 1993

41. C'm M.. 2=d Vcvcozr. C.. SoF.d S:=~ Ccnn. 57. 6F# (19W.5 5L Weaw.L H.. pa c nu r cuo" C m.D9. I 191M N5. Hommig .LRkee. D. E- 3m L. R- R*;s. Xr J. ....

£.Lensmr A-. Phmr ftc Lem. 43. 1431 1197,9. PhYL ftc,. 323.5203 HneneEF.k1-.3ev.LJim m6.3L 302 (195ft(13:1.f 6M. WunmI E M.1 L5 LK5ad. T- Cben Y- 0-n4xT- R

43 Bcnnet P. A. Feane. J C.. Hd'khbecL F.. I - -. A. 2zd Saw. KroCl.. H arTrO~k 54.. Y rv.- Ln6&.174111991Laixky. G. A.. Pb-v R",. Sr.:194119S31. 61. Cxbo. T. A-. RadiaL flea 60L -94197-9: Cadsoe T. A- Nc=. J--

44 Cubabru G- Guffliano S. Grnaupo. B. 2nd Szaaze. A.. C W- Tde. T. C md MA F- -P IL. 169.27 . 965Nuovo Ci=r 1- 53311M). 611 Ramakce. F. mad hduday.J.S-i. V=Sd-.Teekeol.16. SI01197M)

45 Ramaker. D. E. and eLson. F. L- J. V=n S=. Teeboo.L AS. 630 'I Huus. F. L- famakm . E K od. B. E. 2nd Gebb&nd. S. C.. SeL1198'P Sd..24S 11941991.

46 Lercksoe.. N. E- Pon-d). C J. anM Ramakm. D E- Ph)rL Rev Le:L. S8 64. Bz P.. HnAmm. .. A.- Hm-ronwc. E- =d Ho=cnn. V- HA.,.W, (1987) Chia Ann 52. 1745 (1969t.

47 d-Boer. D. K. G-. Haa C and Sanmuzky. G. A.. .. PbrvL F: eL 65. Huo& F.kfL- inkm .O.LKod B.ESeLSLZU.I 10: (19911Phys. 14.2769 (19241 66. Cwaumi L S Qiao.GQaad K'gagaI- LSad Sd 259 (1935L

48 Hodepzd. P. a=d HdkebredsL F. U. PbyL R"c. B334.3045 (IM6) 67. Ramker. D. ELApl. Sad. S&. AWM9 351(1991t.49 Jeamson. 0. ft. Hillebeeebt. F U. a=d Feeg3. L. C. . V=r. So 63. Darft& G. R-. Peadry. 3. &. and JoyI C. I. W. SarL S&d. 169

Tcchnei. A2. 1049 (19M4 (19).5Q. Cim. M.. Man=. F. a=d Phu=ma f-t. n. -Auare Specuropy and 69. Wei= A. Mqcr. RL likly. M.. L4) C.. Mold. C and Lymn K- G-

Elecuonc Struoue (Endted by G. Cobmoa. G. Mondwo and K. PhyL Rev.61. 2245S(193n.Wanddt(Spangee.Ve~as. Bcdan 3988). p. ISO. M0 Wean. A. SMtL D.. ioyanan A. R- Lme K. H. and Id. C- J. V=e So.

S[ AksejL. H. and Aksela S-. L P3.ys Colloque C9 (Suppie 124L -65 TecknoL A& 2517(1990L(1997). 71. Lea. C-. MekI D.. Kommn A. ft. Gooweld. F.. Jiba. M. and Wean.

52. Aksela. S.. PtkkaLa. T.. Aksda. H-. WajILe M. and Had..M A.. ARev. Sm Iowea= GIL365 (199nPhys. Rev. A35 3426 (198n. Aksda. h-. AIMa S.. Pekkala. T. and 72. Hank. HL W.. Sawnanky. G. A. and Tboaas T- D, Pkhys Re". Lems 41Waflernus.M.. Phys. R"v A35.1S2211987L 1825 (3973)

53 Aksela. H-. Akseia. S-. Polkksne H aod Yaeasltaa. A. PhYs. R"v 73. Clan. C T. OD~i. ft. A. Ford. W- K.. Pluanae. E. 91. and Ebee.A40. 6275(1989). bana W1. Phyr. Rev B32.93434 (19115).

54 Akieja. H.. Akscla. S-. PuLcnen H-. Kwamik A. and Sajeanen 0. P.. 74. Ladoa , .. Hand.L D.. Plys. ?mv Leame 49. 1' 33(195PlayL.Se. 43.425(3990). 7S. 'Jmaback.F. and Hosan.Z..PlayL.I v.Leae 5.7(3984.

55 Oheendoaf. E.. Koeppel. H.. Ctdeebaun.. L IS. T=Eanels. F. and Sga- 76. Uanbach. E-. Comosenas Aaeausc and.Mole. Ph),& 1& 23(1986.meflot. A.. J Elema Spelcosc. Related Pheanua 6-3.2113(I90)

Accession For

DTIC TAB 0Unannounced 13Justifi.cation

By

_R 6tz~ but ion/

IAvailability Codes

v~ jail and/or01st Sp~ca

TECN.ICAL REPORT DISTRIBUTION LIST - GNERAL

Office of Naval Research (2)" Dr. Richard W. Drisko (1)Chemistry Division, Code 1113 Naval Civil Ei.ginecringS00 North Quincy Street LaboratoryArlington, Virginia 22217-50C0 Code L52

Port Hueneme, CA S3043

Dr. James S. Muruay (1) Dr. Harold H. Singerman (1)Chemistry Division, Code 6100 David Taylor Research CenterNaval Research Laboratory Code 283Washington, D.C. 20375-5000 Annapolis, MD 21402-5067

Dr. Robert Green, Director (1) Chief of Naval Research (1)Chemistry Division, Code 385 Special Assistant for MarineNaval Weapons Center Corps MattersChina Lake, CA 93555-6001 Code 00MC

800 North Quincy StreetArlington, VA 22217-5000

Dr. Eugene C. Fischer (1)Code 2840David Taylor Research Center Defense Technical InformationAnnapolis, MD 21402-5067 Center (2)

Building 5, Cameron StationAlexandria, VA 22314

Dr. Elek Lindner (1)Naval Ocean Systems CenterCode 52San Diego, CA 92152-5000

Commanding Officer (1)Naval Weapons Support CanterDr. Bernard E. DoudaCrane, Indiana 47522-5050

* Number of copies to forward

11