328R

A N A L Y T I C A L CHEMISTRY, VOL. 50, NO.

5, APRIL 1978

Electron Spectroscopy: Ultraviolet and X-ray Excitation A. D. Baker Department of Chemistry, City University of New York, Queens College, Flushing. New York 11367

Marion A. Brisk Biomedical Program, City University of New York, City College, New York 10037

D. C. Liotta Department of Chemitry, Emory University, Atlanta, Georgia 30322

A. INTRODUCTORY COMMENTS A N D GENERAL ASPECTS This review encompasses both UV- and x-ray photoelectron spectroscopy (UPS and X P S , respectively) for the period late 19'75 to late 1977, thereby replacing the separate reviews on the two branches that have appeared in previous review issues of ANALYTICAL CHEMISTRY. Although we authored the UPS review in the 1976 review issue ( A I ) , it was D. Hercules and his associates who prepared all the previous XPS reviews (A2. A 3 ) . His name also comes readily to mind in describing our task this time. which has often seemed Herculean indeed. Both branches of t h e technique have continued to grow. especially in terms of the number of people using photoelectron spectroscopy in a remarkably wide range of areas. This of course has made our task as reviewers even more difficult than before; our literature searches have often led us into the most unexpected corners of scientific research, sometimes far removed from our own areas of endeavor. Adding to our difficulties as reviewers have been an appreciable number of uses of photoelectron spectroscopy as an adjunct technique for solving various chemical problems. This has meant t h a t many important and/or interesting results appear in papers that do not contain the words "photoelectron spectroscopy" or some similar identifying phrase in their titles. Inevitably we will have missed some applications of this type, and additionally, our coverage is somewhat selective anyway owing to constraints of space. T h u s , as in the 1976 review, we apologize in advance for our omissions, noting only that our coverage is geared mostly to our own interests. \Ye would, however, like to appeal to authors to place us on their mailing lists for reprints of papers, so as to help us prepare for the 1980 review. We thank those who provided this service for this review. In preparing this review, we have not attempted to cover literature pertaining to various highly specialized areas, such as adsorption of gases on surfaces, other aspects of surface physics, and polymer chemistry. These have really become rather large subdisciplines and merit separate consideration. Our intention here is to deal with aspects of general chemical importance or of basic importance along with new types of applications and studies that may have widespead significance for practitioners of electron spectroscopy. Over the review period, some new books dealing with photoelectron spectroscopy have appeared. Rabalais iA.7) has written a text dealing with the principles of I.:PS in terms of modern chemical physics, in which he presents discussion of photoionization transition probabilities, interpretations of spectra, and their applications. Carlson has also authored a text (A6) which deals with both photoelectron and Auger spectroscopies. Briggs has edited a collection of chapters on Photoelectron Spectroscopy (A49). The first volume of a series dealing with all aspects of photoelectron spectroscopy has been published (A8);it contains a general review of electron spectroscopy plus individual chapters that are directed a t 0003-2700/78/0350-328R$05.00/0

theoretical calculations of relevance in photoemission studies, the X P S of inorganic substances. IJPS as applied to small molecules, inorganic compounds, and organic compounds, high temperature U P S studies, and the use of two-parameter coincidence experiments. Three volumes of the series "Structure and Bonding" contain information on photoelectron spectroscopy. Volume 24 deals principally with the spectra of nonmetallic solids and their consequences regarding quantum chemistry, the fractional percentage method for ionization of open shells of d and f electrons, and the applications of X P S to metals and alloys (A9). Volume 30 contains two chapters on photoelectron spectroscopy (AIO); the first, by Campagna et al.. deals with the application of X P S to homogeneous mixed valence rare earth compounds, the second (Jorgensen) describes work on the photoelectron spectra of inorganic solids, particularly regarding deep lying valence orbitals, the crossing of shell ionization energies, the Gelius effect, and the observation of satellite peak structure. Finally, in Volume 31 ( A I I ) ,Ferreira discusses Koopmans' theorem, with special references to the 3d transition metals and the lanthanides. There have been a number of review articles dealing with aspects of photoelectron spectroscopy. Among these are the following: a number of general reviews on various aspects of photoelectron spectroscopy and allied techniques which were presented a t a NATO sponsored meeting (AI9), a survey of the application of X P S to industrial catalysis (A12),a historical survey of the development of XPS (A13), several reviews on solid state characterization (AI4-AI7), a discussion of the molecular structure information available from XPS ( A 1 8 ) ,three reviews on X P S (A20, A40, .460), and a review on methods of' measurement of ionization potentials and appearance potentials (A41). T h e role of photoelectron spectroscopy and related techniques in probing the gas metal interface has been reviewed (Ad),so has the role of these techniques in studying gas phase molecular ions ( A 4 8 ) . Also of interest is a review by Wittel and McGlqmn which deals with the orbital concept in chemistry and spectroscopy, and focuses particularly on the different orbital models one might use depending upon a desired application. T h e discussion is illustrated by the consideration of some photoelectron spectroscopic results (A21). Among applications of photoelectron spectroscopy that we have noted: but which we do not take up elsewhere in this review, are the following: characterization of archaelogical artifacts (A22, A D ) . pigment surfaces (A21), aerosol particles ( A X ) , glasses and silica derivatives (A27- A32). zeolites (A32 -A34),fluorinated graphite ( A X ) , limestone samples (.435), and as an aid in resolving structures by x-ray diffraction methods ( 4 3 7 ) . Grunthaner ( A M ) has described the application of X P S to metal insulator semiconductor devices (MISdevices). Hammond and LVinograd have described an application of XPS and Auger spectroscopy to the underc 1978 American Chemical Society

A N A L Y T I C A L CHEMISTRY, VOL. 50, NO. 5, APRIL 1978 A. D. Baker is currently Associate Professor of Chemistry at Queens College of the City University of New York He was born in England, and received his B Sc (Special Honors Chemistry) from Imperial College of the University of London From 1965 to 1968 he worked as a graduate student of D W Turner first in London and then at the Universty of Oxford obtaining the Ph D degree in 1968 His Ph D work involved studies with the original retarding field type of photoelectron spectrometer and on the first high resolution focusing type ultraviolet spectrometer Between 1968 and 1971 Dr Baker held an appointment as a Research Fellow at the Swansea campus of the University of Wales, working with D. Betteridge on analytical aspects of photoelectron spectroscopy. In 1971. Dr. Baker moved to the United States, first to be Assistant Professor, and later (1975) Associate Professor of Chemistry He has co-authored two books on photoelectron spectroscopy, and numerous research papers and review articles. He is also co-editor of the series ' Electron Spectroscopy-Theory, Techniques, and Applications", and a member of the editorial board of the Journal of Electron Spectroscopy and Related Phenomena, I n addition to electron spectroscopy. his research interests lie in other types of instrumentation for chemists. and in aspects of organic nitrogen chemistry. He has published a number of papers in this area, and enjoys the opportunity of participating in a broad spectrum of chemical reseach.

Marion A. Brisk received a B.A. and M.A. from Queens College in 1970 and 1972, respectively. She earned her Ph D from the City University of New York in 1975. Her doctoral work involved both the UPS and XPS techniques. and was under the supervision of A. D. Baker. She is currently an Assistant Professor in the Sophie Davis Center for Biomedical Education located at City Collage.

Dennis Liotta received his undergraduate and graduate training at Queens College of the City University of New York. Afer two years of postdoctoral study at the Ohio State University. he became an Assistant Professor of Chemistry at Emory University. His research interests include new synthetic methods singlet oxygen chemistry, and the application of photoelectron spectroscopy to chemical reactivity probiems

potential deposition of silver and copper on platinum electrodes (A39). Petrovic et al. have studied feldspar grains as part of an effort aimed a t the investigation of rate control in dissolution of alkali feldspars (A42). Fluck and LZ'eber (A43) have authored a useful survey of the use of photoelectron spectroscopy for studying aspects of phosphorus chemistry. X P S has been used to study the dissolution and passivation of nickel ( A M )and to probe passive layers on stainless steel (A-25).It has also been used to measure the chlorine content in iridium supported catalysts (A.16).and to study vanadium bearing aegirines ( A 4 i ) .

B. ADVANCES IN INSTRUMENTATION AND INTERPRETATION An important experimental advance in photoelectron spectroscopy has been the continuing development of capabilities for examining liquid samples. Two laboratories have been particularly active in this area, those of Siegbahn and Delahay. Siegbahn's group has concentrated on X P S aspects and has used a wire system for transporting the liquid to the x-ray beam (see reference A40 for a description of this work), while Delahay and his co-workers have so far worked with

329R

vacuum ultraviolet excitation sources and ha\re utilized an ingenious rotating disk device to continually renew a liquid surface exposed to the illuminating radiation. They have worked at photon energies of 21.2 eV and lower, and the curves they obtain are similar to gas-phase spectra, but are broadened and shifted owing to the electronic polarization of the liquid medium (BIGB3). I t will be interesting to compare results of this type with results obtained from what are. in effect, photoionization efficiency curves on liquids, obtained by measuring the electrical response in a cell containing a liquid and two platinum electrodes, the cell being illuminated with Nakato and colight from a monochromator (H-27-B49). workers have done considerable work of this type, and recently have compared their results on the ionization potentials of tetraphenylporphyrin derivatives (B47)with gas phase values on the same molecules, measured previously by Khandelwal and Roebber (B50)using high temperature UPS. Considerable differences in the ordering of the lowest ionization potentials for the compounds studied were apparent, and merit further study; Nakato e t al. pointed out the possibility of thermal decomposition in the high temperature U P S studies. An advantageous approach would be to examine t h e same compounds in a Delahay-type apparatus. Miksche et al. have described a technique for quickly freezing a solution prior to X P S examination (on a SlcPherson ESCA 36 instrument). Their method (B7)apparently follows an initial suggestion by Burger and Gutmann. Burger e t a]. have used such a freezing technique for studying nitrogen containing ions in contact with varying counterions ( B 7 a ) . Hercules noted in his 1976 review (43) that sample charging remained the "bCte noir" of x-ray photoelectron spectroscopy. We now reporot that the beast is still with us, black as ever. In a recent interesting but rather alarming paper, Madey. Wagner, and Joshi (B33)compiled results from different X P S laboratories on powdered, insulating materials typically used as catalysts or catalyst supports. T h e results obtained indicated a standard deviation in absolute X P S line positions that is far greater than the precision of any one instrument. Furthermore, there was a large spread in reported intensity ratios for instruments nominally having the same transmission characteristics, and even of the same manufacture. Many of us have long suspected this type of problem, SO it is to be hoped that more comparisons of this type can be arranged. and better "standard" procedures developed as a result. On t h e same theme of calibration procedures have been the following papers: .4sami ( B I 3 ) has proposed a consistent energy calibration method for XPS. Kinoshita et al. (B29)have also commented on energy calibration, Nefedov et al. (R36) have compared different x-ray photoelectron spectrometers and the methods used to correct for charging effects, and Sitte (B46)has commented on reference levels, work functions, and contact potentials and their significance in photoelectron spectroscopy. Interestingly, the comparison of spectrometers and charging effect corrections reported by Nefedov did not indicate such a worrysome spread of data as was found by Madey et al. (see above); clearly more comparisons need to be done. Hirowaka et al. (B4)have discussed the application of X P S to quantitative analysis without standards. Wagner has discussed the factors affecting quantitative determinations by X P S (€5).He concludes that the technique of dividing photoelectron line areas by elemental sensitivity factors or standard intensities can be applied (a) in single-phase but not multiple-phase systems and (b) only in the case of nontransition elements. because of the propensity of transition elements to undergo multielectron excitation/ionization processes. CZ'oodruff, Torop. and West have discussed the quantitative interpretation of peaks in photoelectron spectra obtained with a dispersive analyzer (R14). They point out that if height is used as an intensity measure, one must consider the energy spread of the photon beam and the analyzer bandpass. Such measurements in general require a complete knowledge of the variation of the analyzer efficiency with electron energy. Bancroft. Brown, and Fyfe (B9, B9a) have carried out calibration studies for quantitative x-ray photoelectron spectroscopy of ions. They obtained linear and reproducible calibration plots for monolayer and sub-monolayer amounts (less than 6 X lo-' g/cm?) of Ba2+and Pb2+ on cleaved calcite surfaces. Their plots were of the area ratios of the X P S peaks P b 4f Ca 2pl,? or Ba 3d5;?: Ca 2p,,,, vs.

330R

A N A L Y T I C A L CHEMISTRY, V O L 50, NO 5, APRIL 1978

the amount of Pb2+or Ba2+syringed into the calcite surface. Schultes and Ebel (B47) have carried out an absolute determination of the work function of palladium with an XPS instrument (B5.5). Ebel and co-workes (6’54)have described t h e use of an XPS instrument as a soft x-ray spectrometer. There has been considerable interest in novel sources of excitation for electron spectroscopy and in de\ eloping existing sources. Pinetta and Lindau (B8)have described the source characteristics and optical systems involved in X P S with a synchrotron source. Laser sources continue to find use in obtaining the ionization potentials of negative ions and, hence, the electron affinities of the corresponding neutral species (e.g., see Ref. B41). Electron-Electron coincidence measurements are becoming more common. A recent paper b) Hood, Hamnett, and Brion (R39) contains references to much of the existing work, and summarizes the state of the art. Shirley et al. (B35)have worked a glow discharge lamp on a conventional UPS instrument at a very low pressure. and find that under these conditions t h e lamp generates enough electrons for the operator to carry out electron impact experiments (autoionization electron spectroscopy and Auger electron spectroscopy) in the U P S instrument. Castle and co-workers have described a Si KO source, 1739.5 ‘e! ( B I I ) . Yttrium hl{ sources (B12, D5) and zirconium MC sources (BI 7-B19) producing, respectively. 132.3 and 151.4 eV are extremely promising prospects for aiding the interpretation of spectra, but the great reactivity of the metals makes for difficulties. Nilsson et al. overcame some of the reactivity problems by continuous evaporation of yttrium onto a rotating anode (B12). They noted that the use of the yttrium source offers great advantage in studying photoemission from adsorbed species, since the surface sensitivity is much greaater than a t 1253.6 e\.‘, and the results obtained are not plagued by the final-state distortion effects found when very low energy (UPS) sources are used. Allison and Cave11 (B42) used the zirconium source to study neon, argon, methane, nitrogen, and water. They noted that orbitals of chiefly 2p character dominate in the spectra obtained with the Zr source. while orbitals of 2s character are dominant in a spectrum obtained with Mg K n radiation. A book on electrostatic lenses has been published (BIG), which fills a need for a comprehensive source of information on these devices. Wannberg and Skollermo (R37) have published a paper dealing with computer optimization of retarding lens systems used in XPS. Prutton has made a comparison (theoretical) of analyzers degraded to obtain high sensitivity (R38). Several papers have dealt with mathematical methods for resolution enhancement (B21-B24). Clark and his associates have worked out a method for reducing hydrocarbon contamination in a n AEI ES200B spectrometer (B34). Apparently most of the contamination “boils off’ the x-ray cap as a result of heating; it can be greatly minimized by cooling with water or liquid nitrogen. This belies the popular belief t h a t hydrocarbon contamination is mainly a result of residual gases from diffusion pump oil. Van Attekum and Trooster (B31) and Beatham and Orchard (R32) have devised methods for eliminating peaks in XPS spectra caused by the presence of satellite lines in the light source. Gadruk (B30)has discussed plasmon satellites in XPS. Nornes et al. (B20) have described a miniature electron bombardment sample evaporation source for X P S Leeuw and co-workers (RIO) have de\eloped a double modulation technique for studying transient species by UPS. They modulate a t the frequency of the discharge which produces t h e transients, thereby limiting interference by “parent“ species. Brion and Crowley (B26) and Jonathon et al. (B27) have devised methods for carrying out high temperatures UPS-above 1500 K. Berkowitz has reviewed high tempeature U P S studies (A8). Frost et al. (B28) have constructed a fast pumping U P S system for handling transient and unstable species. Connor and associates @ I S ) have measured the He I and He I1 spectra of solid films of silver, gold. anthracene, the alkali halides, sodium hydroxide. and lithium sulfate. Problems in assigning work functions or binding energies are discussed. A new technique for measuring high resolution threshold photoelectron spectra has been devised (B53). It involves the following: (a) photoionization; (b) attachment of the photoejected electron to a n electron “trap”, e.g., sulfur hexafluoride; (c) detection of the product negative ion, e.g. SF6

by mass spectrometry. A GC-UPS combination developed a few years ago has been further investigated (B6). It has been used to obtain spectra of pure N-nitrosamines, and to investigate the possibility of using a photoelectron analysis system as a selective GC-detector. It works quite well; there are really only two drawbacks. T h e cost is high and, because there is such an overlap in the range of photoelectron bands observed for different functionalities, its selectivity is not always guaranteed. Gray and Hercules have examined correlations involving X P S shifts and Sanderson electronegativities (B25). Ballard has tried t o compare electrode potentials with UPS derived ionization energies (B45). Briggs et al. (R40)have determined trace elements in water by adding 10-20 pL of the water to an etched aluminum plate, and then evaporating the solvent by exposure to a UV lamp, followed by an X P S scan of the now coated plate. They were able to detect g P b in a 10 pL sample. Asbrink et al. have developed a new semiempirical method for calculating ionization energies (HAM/3, “HAM“ = hydrogenic atoms in molecules) (B43). Brujn (R44) has raised some questions regarding the application of this method to certain systems. Manson (B52) has presented a general theoretical framework for treating satellite line intensities. Penn ( 8 5 1 ) has presented a compilation of electron mean-free path data for inelastic scattering as a function of energy for all elemental solids, and has shown how this information makes it possible to deduce from X P S measurements the relative concentrations of atoms or molecules distributed in the surface region of a material. Finally in this section we note a n interesting attempt to develop a unified treatment of X P S chemical shifts, dipole moments and polarizabilities (B56). Othe aspects will be taken u p in the remainder of this review.

C. PHOTOIONIZATION CROSS SECTIONS AND ANGULAR DISTRIBUTIONS: ANGLE-RESOLVED MEASUREMENTS Angular-resolved X P S studies of solids have recently been reviewed by Fadley (C2), who has made several important contributions in this relatively new and rapidly developing area (C2), C3, C6, and references therein). H e has pointed out that x-ray photoelectron angular distribution measurements are of importance for a number of types of study of solid samples. Baird and Fadley picked out the following specific areas in a recent paper (C6),and gave leading references to each type of application: (1) the selective enhancement of surface-atom intensities for low (grazing) exit angles or low x-ray incidence angles; (2) the quantitative analysis of XPS angular distribution data to yield parameters such as electron attenuation lengths, surface-layer thickness, surface concentration profiles, and surface roughness contours, and in measurements of single-crystal specimens; (3) fine structure in angular distributions and total peak intensities changes in valence-peak relative intensities with angle, both of which can be related to atomic and orbital symmetry factors. Representative papers on the use of angular-dependence studies are given in references C4 and C2O-C22. There has been continued debate on the merits or otherwise of calculations involving plane-wave approximations. Ritchie, in discussing this issue, has noted that results obtained using the plane wave (PW) method, length form, are in disagreement with results from both coulomb wave (length form) and PN‘ (velocity form) methods a t all energies, and that the P W (velocity form) and coulombic (length form) approaches give different results for all except 1s photoejection (C7). Shirley and Williams (C13) have calculated differential photoemission cross sections for the Is, 2s, and 2p shells of neon using several different methods. They note that methods based on the P W or orthogonalized P R approximation “have very limited applicability” to cross-section calculations, both generating spurious local minima and incorrect angular distributions. On the other hand, calculations involving the use of Hartree-Fock functions were found to be in good agreement with experiment. T h e sources of the failures of the PW and OPW methods were analyzed, and limits set on values of n , 1, and z for which the PW model gives qualitatively correct total cross sections. Shirley and Williams concluded that reliance on P\V type calculations should be discouraged. However, this method still has i t q advocates. Beerlage and Feil (C8)claim to have improved the P W method for U P S

A N A L Y T I C A L CHEMISTRY, VOL. 50, NO. 5, APRIL 1978

studies by making the assumption that Ephotoelectron - Ephotun. This is believed to correct for one of the major deficiencies in the Pb'model, viz., the neglect of electrostatic interaction between cation and photoelectron. In principle this effect should become insignificant for high energy photoionization but for U P S studies it becomes serious, since the outgoing electron classically moves away from its origin with decreasing velocity. Thus Beerlage and Feil contend that a plane wave description of the U P S process should use a plane wave with a wavenumber corresponding to the velocity of the electron near the molecule. This modification results in considerable improvement in calculated cross sections for a number of molecules as compared to the results from a standard PLY treatment. Hilton et al. ( C I I ) approach the problem by using a spherical square wave to represent the potential of the electron. They conclude that their method will experience the same difficulties as the standard PLV method when applied to large atoms, and will be unable to predict angular distributions. The Pit' approach has been applied to calculations of photoionization cross sections of diamond and silicon (C9). It has shown that consideration of orthogonality terms greatly influences the calculated ratio between cross-sections of s- and p-electrons Scofield (C10)has calculated subshell photoionization cross sections for all Z values u p to 96 for the energies of the Mg K a and A1 KCI lines. T h e calculations were done using transition matrix elements with the electrons in the initial and final states treated as moving in the same Hartree-Slater potential. T h e potential was determined self-consistently for t h e neutral atom occupations of the sub-shells with the potential introduced by Slater to approximate the effect of exchange. Dill ( C I 4 has calculated fixed molecule photoelectron angular distributions. His analysis is not dependent on any particular dbmamical description of the photoionization process, and will serve as a framework for studying processes such as molecules oriented a t surfaces, photoionization in molecular beam experiments, etc. Fixed molecule angular distributions for CO and N 2 were determined by Dill. Siegel, and Dehmer ( C 1 5 ) for the case of electrons ejected from K-shell orbitals of these molecules by electric-dipole interaction. T h e results correspond to the gases adsorbed onto a surface, and it is predicted that the distribution function will he rich in structure. It will he of great interest to test this prediction by experiment since the experimental results will give a sensitive and detailed insight into the dynamics of molecular photoionization. Davenport (C19j has reported calculations of a similar tbpe for oriented CO. his method being based on the Xci method. Hancock and Samson (C16) have described a new method for measuring angular distributions of photoelectrons using He1 radiation. This new method appears to circumvent some of the problems previously encountered with this type of experiment. They fix the position of their analyzer at 90"from the beam direction and then polarize and rotate the plane of polarization of the photon beam. T h e polarizer used is of a reflection type (no suitable transmission type exists for such short wave length radiation). Data on 3 were found for Ar. Xe, N2, 02,CO, C O Y ,and NH3. Kalman ( C 1 2 ) has discussed the variation in @ for vibrational states of nitrogen and carbon monoxide. He criticizes previous claims that a breakdown in the Born-Oppenheimer approximation may be the cause of experimentally observed angular dependences. Previous papers to this effect ignored an orientational averaging that allows for a randomly oriented ensemble of molecules. Hirokawa et al. ( C I ) have discussed applications of calculations of photoionization cross sections and mean free paths to quantitative surface analysis by XPS.Nefedov et al. have followed up previous work on relative intensities in XPS (C5). XPS studies of alkali metal halides in the solid phase have been examined using various sources (C17). The ratios of the 2s to 3p and 3s to 3p peaks were found from the peak intensities. Samson et al. have measured cross sections for O2 in the region 100-800 A (C18).

D. STUDIES OF ATOMS AND SMALL MOLECULES There continue to be a very large number of investigations of atomic and simple molecular species. Some of these are new studies, made possible by advances in instrument capabilities (e.g., high temperature studies, angularly resolved

331 R

studies. studies with different light sources. work with metastable or transient species), others are repeats of earlier studies on the same types of substances. with particular emphasis on some special point of interest to the investigators. T h e borderline between "small molecules" and other types of molecules is, of course, arbitrary, and there may be some overlap in our coverage in this section with coverage in later sections of this review. Williams and Potts have investigated complexities in the U P S spectra of alkali metal vapors ( 0 6 ) . "Anomalies" in the spectra of K. Rb, and Cs are caused in part by inelastic collisions of the outgoing photoelectrons with the metal atoms in the ionization chamber, and in part by autoionization initiated by the weak 51.56 n m He1 line. For example, a peak a t an apparent I P of 1.51 eV in the spectrum of potassium was assigned to (4s)-'ionization produced by the 51.56 nm line, but enhanced in intensity by a factor of more than 50 by autoionization. Hush and Suzer (040)studied satellite peak structure in the spectra of Zn. Cd, and Hg using He 58.4 nm and 30.4 nm radiation. They found no dependence of the satellite structure on the wavelength of the incident light. and postulated an initial state C.I. mechanism to account for the observed result. Suzer, Lee, and Shirley obtained UPS results for Bi nd Bi2 ( 0 3 4 ) ;the spectrum of the former was found to be consistent with a J-J coupling description of the ground state. Satellite peaks were observed. and autoionization seemed to be prevalent as judged by the unusual intensity of the He17 spectrum. Gardner and Samson ( D 1 ) measured the intensity distribution of the vibrational peaks in the U P S of H 2 and compared their results with calculated values. An I P of 13.01 eV has been found for the hydroxyl radical OH (D46). Chong and Takahata (02.5) discussed UPS results, especially regarding vibrational structure, for CS, P N , SiO: and ps. P2 and P N are considered particularly interesting because they afford the possibility of a comparison of the family Ne, P N , P2. Earlier studies on P N had been carried out by R u and Fehlner (D29a), and new experimental studies have been provided by Bulgin, Dyke.and Morris (048). Two major changes are noted in comparing the spectra of the family referred to above. T h e relative intensity of the third band in the spectra decreases in intensity on going from Nzthrough PK to P1. and the ordering of the first two hands is reversed in P2 as compared to N2 and P N . Calculations on the P N ionization energies have been provided by Bulgin et al. ( 0 4 8 ) to complement their experimental study spoken of above, and by Domcke et al. ( 0 2 9 ) . T h e latter group also calculates vibrational structure in fairly good agreement with experiment. and point out that there is probably a systematic error implicit in an earlier calculation of the vibrational structure associated with the 2 r - l band (025). Natalis e t al. have embarked on a series of investigations of vibrational structure as revealed in U P S spectra and Franck-Condon factors (018. 022). Basically, they set out to answer the question "can U P S peak intensities be used to estimate Franck-Condon factors for direct ionization processes?". Their approach is to pool together experimental information from literature sources, and compare the vibrational intensities with calculated Franck-Condon factors. The answer they obtain to the question just posed is generally in the negative apart from a few cases, the main reasons apparently lying in autoionization related contributions to peak intensities and in instrumental factors such as detection angle. The family of compounds 02.S,. Se?,Te2 has been subjected to a U P S study by Shirley et al. ( 0 3 3 ) . "Anomalous" X211, - 2 r I l L2r:I12 , intensity ratios were shown to arise from spin-orbit coupling in the ground states. T h e diatomic molecule KS has an ionization potential of 8.87 e\' ( 0 4 7 ) . Dyke et al. generated this species by heating N3S3CI3at 400 K thereby producing NSCI, then subjecting it to a microwave discharge. T h e photoelectron spectrum of H 2 0 has been reinvestigated along with those of HDO and D 2 0 ( 0 1 1 ) . The Franck-Condon factors obtained were used to generate a geometry, rOH= 1.00 A, i H O H = 110'. Formaldehyde has been studied using the electron-electron coincidence technique that is the electron impact counterpart of photoelectron spectroscopy (D32);angular variation studies demonstrated that the third level is 5al and the fourth level lh,. The formaldehyde derivatives HCOCI and HCOF have

A N A L Y T I C A L CHEMISTRY. VOL. 50, NO. 5, APRIL 1978

332R

Table I. UPS Studies of Some Small Polyatomic Molecules Compounds studied NOCl Nitrosyl and nitryl halides, "0, NSF FCN

co.

Seo, As$,, p,s,, P&3 AS406 pa, p,s, S,NZ H,C=S=O H-N= S= 0 NCN,, NCNCO /

co-co

\

X X (X = halo-

CH,NH NZO, HF, DF, F , AsBr,, AsI, Thallous halides

so3

Table 11. XPS Studies of Elements or Simple Compounds Compound examined or type of study

Type of study

Gaseous and frozen Hg; linewidths, relativistic effects Relaxation and Final State Structure in XPS of Atoms, Molecules, and Metals Tellurium, tellurium compounds Sulfur vapor, S, Iron Thin gold films Silver Formaldehyde, orbital relaxation accompanying core ionization OF* Amorphous ice Y Mt spectra for N2, CO, C,H,,

Ref

UPS, He I UPS, He I

D42 D81

He I, He 11, Ne I He I, He I1 Threshold PES He I He I

D43 D51 D7 3 D 28 D26

UPS and XPS He I He I

D80 D58 D74

He I He I, variable temp. conformational study

D27 D30

He I, d-substituted compounds also studied He I He I, eqm of N,O,/NO, system Threshold He I, emphasis on spin-orbit coupling He 11, solid samples

D12

Ref D3 D4 5 D23, D33 D54 D61 D67 D68 D9 D5 D21 D7 6

and 0 ,

He I. Reassignments of bands: vib. structure

D13 D16, D17, D55 D36 D4 D2 D82 ~~

also been investigated ( 0 1 4 , 015). Table I contains reference t o U P S studies of other polyatomic small molecules. Of particular note is the report by Frost et al. (012)that the n+,nsplitting in N2H2is as large as 5.01 eV, the largest splitting of this type known to the authors. Although organic compounds are to be dealt with in a later section of this review, we note here the following studies on small organic compounds: a study of methanol and deuterated methanols, aided by a deconvolution procedure to improve resolution ( 0 3 2 ) ,a U P S study of alcohols, interpreted with t h e help of SPINDO calculations (050),a U P S study of CH,CHO, with special reference to deuteration effects (0491, an angular distribution/He I/Ne I UPS study of cyclopropane ( 0 3 2 ) , a comparison of MO calculations on the UPS of nalkenes (D8), a study on the hyperconjugative effect of the methyl group as a substituent (057), a study of the intramolecular hydrogen bond in malonaldehyde by XPS (D70), a n investigation of electron-donor/acceptor complexes involving bromine and alkylamines @%), an X P S investigation of t h e C 2s valence orbitals in cyclic alkanes (0711, a U P S investigation of some simple push-pull alkenes (075), and a reinvestigation of p-difluorobenzene ( 0 4 1 ) . Photoelectron spectra of negative ions are becoming increasingly popular; a laser source is used to generate the spectrum, which provides electron affinities of the corresponding neutral species. Representative studies are those on PO-, PH-, PH,, CH,- (035,0 3 5 a ) , NH- ( 0 3 8 ) ,and C,H< (072). There have been countless studies of gases adsorbed onto various substances; however, we shall not deal with this area in this review. X P S studies of elements and simple compounds continue to proliferate; we indicate some areas of interest in Table 11.

E. TRANSITION, LANTHANIDE, A N D ACTINIDE METAL COMPLEXES T h e data accumulated over this past two-year period is encouraging as well as disconcerting t o photoelectron spectroscopists in this area. T h e technique has been applied to

Gaseous alkenes: satellite structure HZ02 0,:XPS and Xol calculations HCONH, : satellite lines CO: Theoretical study of satellite structure H,CO: Role of spin in shake-up spectra Shake-up events in XPS of small molecules HF: F l s shake up GeS, GeSe, GeTe amorphous; UPS, XPS study of valence levels U0,:UPS and XPS. Nature of U valence electrons 5f vs. 6d UF,, K,UF, Single crystal titanium carbide Carbides of Zr, Hf, and Ta Metal fluorides Transition metal difluorides Chalcogenides of transition metals and alkaline earth metals Cerium nitride Lead chloride, lead bromide, and similar materials

D77 D60 D69 D37 D59 D19 D24 D 20 D10 D44 D39 D62 D66 D63 D64 D65 D78 D79

many new systems under varying conditions: new areas of application have been revealed; limitations of the technique in some cases have been ascertained. Sample decomposition under x-ray irradiation is becoming a formidable problem confronting photoelectron spectroscopists as evidenced by several studies on different systems (E38-E42). New information on the structure and bonding in lanthanide and actinide complexes in particular has been obtained via application of XPS and U P S to these systems. Koopmans' Theorem in PS studies has been used frequently with the contention that relativistic effects, electron correlation, and electron reorientation in the positive ion remain fairly constant for similar ionizations from similar molecules. Transition metal carbonyl complexes continue to be favorite subjects for photoelectron investigations. Avanzino and Jolly ( E l ) reported X P S data on CH,COMn(CO),, [n-C5H5Fe( C O ) L ] 2and , Co,(C0)12in the gas phase. 0 Is spectra of these complexes indicate that 0 1s binding energies can he used to distinguish between terminal and bridging carbonyl ligands. Bridging carbonyls displayed lower b e.'s due perhaps to increased back-bonding. The results of this study imply that the 0 1s b.e. value of CO adsorbed on metal surfaces can be used to elucidate the mode of bonding: a low 0 1s value suggests a CO carbon coordinated to two or more metal atoms while a high b.e. indicates only one M-CO bond. Jolly et al. ( E 2 ) correlated 0 1s binding energies in transition metal carbonyl complexes with multiplicity-weighted C-0 stretching frequencies, effective metal nuclear charge, and the number of C 0 ligands in the complex. All three correlations supported the qualitative ideas concerning back-bonding by CO ligands. A decrease in either the effective nuclear charge on the metal atom, or in the number of CO ligands in the complex, will increase back-bonding; increased back-bonding leads to a

A N A L Y T I C A L CHEMISTRY, VOL. 50, NO. 5, APRIL 1978

decrease in both the 0 Is binding energy and the multiplicity-weighted average of the C-0 stretching frequency for the complex. Willemen et al. ( E 3 also correlated IR and X P S data: substitution of C1 for CH:< in t h e series (CH,,), ,,C1,Sn\~(CO):,s-CsHi.n = 0.1-3. is accompanied by an increase in CO stretching frequency and an increase in the carbonyl C Is, 0 Is and \V 4f; binding energies. It was noted in this study that the spectrum of (CH,I),ISnI\/lo(COI,;i-CiH;, continuously changed during the experiment: sample decomposition under x-ray irradiation was thought to be the cause. Van de Vondel et al. (El) examined chemical shifts of t h e h)ln 2p, C Is, 0 1s. and other core electron signals of a series of manganese carbonyl complexes. Electronegativity and structural considerations were used primarily to account for observed shifts, especially for bromine derivatives; a n exceptionally high Br 3 p , , b.e. measured for [Mn(CO),Br], was attributed to the presence of two bridging Br atoms between the two M n atoms in the complex. Lichtenberger and Fenske ( E 5 ) reported U P S spectra of Cr(C0):CS. SI’(C0IXS. and CnMn(CO),: CD = cvclopentadienyl. T h e a”and u ionizations associated with the thiocarbonyl ligand appeared well separated from the other peaks in the spectra; this fact, coupled with molecular orbital calculations provided interesting insights into the nature of the thiocarbonyl ligand. T h e CS ligand, for example, was found to be a better A electron acceptor than the CO ligand. An important point emanated from this study in regards to investigations of complexes: a consistent description of electronic structure cannot always be established by consideration of bonding capabilities of isolated ligands alone. Block and Fenske (E6)employed C P S data and approximate M O calculations in their studies on a series of Dentacarbonvl chromium carhene complexes: Cr(COjiC(Xifi; S = O C H 3 , SCH3.N(Me),, NH]; T = CHI: S = OCH,, NH?, Y = C,H,O; and S = OCH,, N(hle) h” Y = C,H,. The carhene ligands uere shown to be poorer 7r acceptorsihan the carbonyl Ggand as commonly thought. However, the charge on the carbene C i i less positive than on the carbonyl carbon atom in contrast t o the popular contention uhich depicts the carbene C as an electron deficient center. Lichtenberger et al. ( E 7 )employed similar techniques to elucidate the binding of the din&rogen (N2)and n”, ligands in Cphln(CO)zN,and CpMn(CO),NH,. The He(I) PS spectra of these two complexes were compared with that of the Darent comDound: CaMn(C0L. The sDectra differed primarcly in the iohzations ‘associated with k e t a l d levels. T h e I P shifts measured in the dinitrogen complex are accurately predicted by the Fenske approximate MO method although experimental and theoretical data are not as well correlated for the amine complex. Theoretical and experimental IF’‘. were correlated by deriving scale factors from a study on the valence electronic structure of CpMn(CO), (E12). It u a s ascertained that the r\J2 ligand destabilizes predominantly metal orbitals and ring e”l orbitals because of its lower s-acceptor ability leading to an increase in the electron density on the metal center. T h e amine complex also exhibited a destabilization of these molecular orbitals when compared to the parent molecule: in fact, the first I P appears at 6.63 eV which is perhaps the lowest binding energy ever observed for a closed shell molecule. The bonding interactions of the PF, ligand with some transition metals :vas elucidated by Head et al. ( E 8 ) . T h e He(1) PS of MnHL5, FeH2L4.and CoHL4 (L = CO or P F J were obtained by these authors; the IP‘s associated with the M-H and the Dredominantlv M d molecular orbitals are shown to be slightiy higher in the PF, complex as compared to the CO analogue. In both the CO and PF3 derivatives, however, a steady increase in X)I-H u bonding IP‘s in the order Mn < Fe < Co was reported. I t is interesting to note that the band, corresponding to ionization of the FeeH bonding electrons in FeH,(PF,J,, exhibits the same anomalous high intensity as previously observed in the carbonyl derivat ive. Green and .Jackson ( E 9 ) reported the He(1) PS of some mixed transition metal carbonyl complexes: MeMCp(CO),, 11 = hlo and LV: MeMCp(CO),. M = Fe and Ru; and (CH2CNjFeCp(CO)2.IP’s associated with M-C bonding electrons were provided from spectral analysis and comparisons with P S data on other transition metal alkyl complexes. A ”molecules in molecules“ approach (EIO) was applied to facilitate the interpretation of the spectra. Green et al. ( E l l ) also emploh-ed such an approach in elucidating ),

).

333 R

the bonding in several tricarbonyl Fe and Ku complexes. He(1) P S spectra of MLM(CO)3;L = cyclohexa-l-3-diene,M = Fe or Ru; L = cycloocta-1,3-diene, cyclohepta-l,3,5triene and cyclooctatetraene, M = Fe, were presented. Band assignments were also aided by ab initio SCF MO calculations and relative band intensity data. Lichtenberger and Fenske ( E l 2 ) also applied the UPS technique to study the valence electronic structure of several mixed metal carbonyl complexes: CpM(CO),, M = Mn, Re: and CpFe(CO),X, X = CH3, C1. B. I. T h e bands appearing in the 7-13 eV region demonstrated a sensitivity to metal and ligand substitutions. T h e spectra were interpreted primarily by results of a b initio calculations on the Cp ion and approximate calculations on transition metal complexes. T h e peaks assigned to ionization from molecular orbitals of mainly metal character in the Re spectrum exhibit spin-orbit splitting. and q-mesiHe(1) U P S spectra of q-cyclohepta-1,3,5-triene tylenetricarbonyl metal complexes ( M = Cr, hlo, \V) have been reported by Gower et al. (E13). T h e first hand in the latter series was attributed to ionization from the a] and e type molecular orbitals in C3\-symmetry. These MO‘s are largely of M d character although their IP‘s appear to be independent of the metal in the complex. T h e 2nd band was identified with the degenerate H orbitals localized on the arene ligand; these MO‘s are raised by -1.7 eY upon complexation exhibiting a substantial interaction between metal and arene. T h e 3rd and 4th peaks of the M(C,H8)(C0)3spectra were assigned to H MO’s of the cycloheptatriene ring; interestingly, the M-L interaction is greater for the more stahle ligand s M O (a”), although, the energy separation between the appropriate metal d orbital and this ligand orbital is larger. The M d orbital IP‘s are similar for all three metals but the differences are greater than those observed for the MICO),. M(PF,), or M(CsH,Me,-1,3,j)(CO),~series. Koepke et al. (E14) investigated the charge distributions in trimethylenemethaneiron tricarbonyl ((CH,),CFe(COj,i by measuring the core electron binding energies in this compound and comparing t h e values with those of busadieneiron tricarbonyl. T h e C Is spectra of these complexes implied that the charge on the central carbon atom of the (CH,),C moiety is between that of the carbonyl and the CH, carbon atoms; published C N D 0 / 2 calculations disagree with such an ordering. This discrepancy was explained by consideration of relaxation effects: core ionization of a ligand atom causes an increase in back-bonding if negative charge shifts to the core-ionized state. Such an increase in back-bonding lowers the b.e. because of a significant associated relaxation energy ( E l 5 ) . It is proposed that formal negative charge is shifted from Fe to the central C upon core ionization of this C atom; therefore, a higher positive charge is attributed to the central carbon atom than is indicated by the position of its C Is peak. Back-bonding was also used to account for the observed core level binding energies in a n iron tricarbonyl complex of a thiophene sulfoxide derivative. Since the low S 2p b.e. resembles that of thiophene it is proposed that the S takes an electron pair from the Fe(CO), moiety to form an aromatic system. Eekhof et al. (E16i also reported the core level binding energies of several other iron tricarbonyl complexes of thiophene and dihydrothiophene derivatives; the results indicate that the Su and Faller ( E l 7 ) criteria for using b.e.‘s to establish whether rvl-S or M-0 coordination occurs in sulfoxide complexes does not appear to be valid for these iron tricarbonyl complexes. C. Batich (E181 employed simple Huckel calculations, relative band intensity data, and methyl substitution effects to provide evidence which supports the validity of applying Koopmans’ Theorem to photoelectron studies. Careful analysis of the He(1) spectra of bis(a-methallyl)nickel, his(s-crotyl)nickel,and bis(x-1,3-dimethallyl)nickel prompted a reassignment of the spectrum of bis(7willyl)nickel;previous analysis of this spectrum and MO calculations had indicated a gross failure of K T (E19). In this study however, the a,(s) I P is placed within 0.4 e\’ of the 1st I P , bringing the experimental value closer to the theoretically calculated one; the error introduced by application of K T is therefore minimized, and its use as a first-order approximate technique is defended. Cauletti and Furlani (E20) examined the He(1) PS of some transition metal diethyldithiocarbamates: M(S,CNEt,),. M(IIj = Ni, Cu, Zn; and M(S,CNEt,i,, M(II1) = Cr, Fe, Co. All six

334R

A N A L Y T I C A L CHEMISTRY, VOL. 50, NO. 5, APRIL 1978

spectra exhibit three band regions, the first (6-8 eV) of which corresponds to MO‘s of predominantly M d character, and consequently varies for each metal. It is interesting to note that ionization of the presumably low IP antibonding Cu(I1) d electron (b2g) is not observed below ‘7.5 eV, while the first band observed in the Ni(I1) complex appears at 6.95 eV. Maier and Sweigart (E21)reported the effect of changing the metal on the He(1) spectrum of a diethyldithiophosphate metal complex (M[S,P(OEt),],, $f(II) = Ni, P d , Pt). T h e Ni and P t complexes gave rise to similar spectra while the spectrum of the P d complex differs substantially. In particular the d-n, separation was shown to be smallest for the P d complex in agreement with known trends from C-T data for d6 planar systems: n, consists largely of S 3p orbitals perpendicular to the S P S plane T h e He(1) PS of M[S(SihleJ211 complexes (M = Sc, Ti, Cr, Fe, Ga, In) were examined by Lappert et al. (E22). The bands corresponding to M d levels appear at higher b.e.’s than analogous levels in the spectra of the dialkylamide analogues; the :N-M (p-d)ir bonding known to characterize the electronic structure in the dialkylamides is suppressed in the disilylamides due to a strong :N-Si (p-d)n interaction. T h e I P associated with the Ti-C bond was determined by Basso-Bert e t al. (E23) in their U P S studies on TiR4 .X,complexes: R = Me; X = C1, OR, NR2. and Cp. In the case of TiMeC1, the 1st band corresponds to the Ti-C bond orbital: the alkoxy and dialkylamino derivatives however. show considerable mixing of Ti-C bond orbitals with 0 and N orbitals and therefore. a t least t a o MO‘s must be considered for the assignment of Ti-C energy levels. Extended CNDO ’ 7 calculations aided in the spectral analysis described in thys work. U P S is currently being applied to many lanthanide and actinide complexes. Clark and Green (E24)investigated the valence electronic structure of M(V-CBH& by analysis of both He(1) and He(I1) P S spectra. T h e U complex spectrum displayed a first band a t 6.2 eV which was assigned to f electron ionization; the band exhibited a significant increase in intensity relative to C MO‘s upon changing the exciting radiation to He(I1). It was also noted that the predominantly ligand MO increased significantly in intensity relative to the e2 ( F )band implying a metal 5f contribution to the e?, MO. T i e major stabilizing interaction in these molecules however is proposed to exist between the ring e2(n)orbitals and M 6d orbitals. T h e same workers (E25)also elucidated t h e bonding in uranocene and thorocene; their He(1) PS spectra were found to be remarkably similar to those of the transition metal sandwich compounds. T h e 5f? electrons ionized at 6.2 eV compared with a reported 6.08 eV I P for the U atom; an electron rich uranium ion is therefore indicated in this comdex. Frapala et al. (E26)reDorted IP‘s between 6.35-7.10 e‘V for U f“ electrons in (Cpj,MCl, M = T h , U; (CH,CjH,),MCl, M = T h , U; (CH3CjH4)3UBrand (CH3C5H4),UBH4 complexes. T h e He(1) spectra were qualitatively described using a simple MO scheme. The proposed 5f electron band appeared to be sensitive to the nature of the ligands in the complex. Brant et al. (E27)studied core electron binding of P d and Pt complexes of carbenoid and related ligands; P t 4f; and P d 3d5,2b.e.‘s indicate that the carbenoid ligands are superior electron donors than methyl isocyanide, while the C Is b.e.’s imply that the carbenoid C atoms are less positively charged than the C atom of coordinated methyl isocyanide. Send and Tsuchiva (E28. E29) reDorted X P S data for a series of novel Pt(I1) “and Pd(I1) yfide complexes: [PdC12L2]. L = RR’2T\J+N-COR”;and [PtCI,(ILle,S+(O)C HCOPh)(SEt)] and [PtClo(Me2S+(0)C-HCOPh)].IR, ‘H and NMR,and XPS were employed to probe the bonding interactions between metal and ligand. By comparison of the multipeak N 1s spectra for the ylides, the salts, and the appropriate complexes, P d exists. it is apparent that a net electron transfer from N Studies on the S ylide complexes rendered-similar results; however, the P t 4f b.e.‘s of [PtC1,(hle2S+(0)CHCOPh)]were surprisingly high while the S 2p b.e. appeared to be low. Back donation from Pt L presumably accounts for these results. R . A. Walton (E30) reviewed studies which focused upon measurement of ligand associated core binding energies: emphasis is usually placed on acquisition of metal core b.e.’s in P S investigations of metal complexes. It is made apparent that studies on N Is, S 2p, and C1 2p electron spectra of coordination complexes have yielded important structural

-

-

information, and it is suggested that by monitoring h e . and intensity changes of ligand signals much insight into the M-L bond can be acquired. Ebner et al. (E31)and Best et al. (E32) reported core binding energies of ligands in a series of complexes. The former workers measured the C12p b.e.’s for three different structural types of chlorine complexes: [M2CISn-](M-M may be present); [ML,Cl,]Cl, [ML,CI]C12and [ML,]Cl,; and double salt compounds containing a chloro anion and a free chloride ion. It was shown that bridging chlorine atoms in dinuclear chloro anions have higher associated C1 2p b.e.‘s than terminal C1 atoms; the separation varies between 0.5-1.2 eV depending on the anion and cation charges. Best et al. (E321 observed S 2p b.e.’s in a series of transition metal complexes of S-containing ligands. S 2p b.e.‘s of Ni(II), Pd(II), and Pt(I1) complexes of 1.2-ethanedithiol and benzenethiol occurred between 163.5-161.5 eV. S 2p b.e.’s associated with bridging S atoms in the square planar polymers investigated appear to be higher than analogous b.e.‘s of terminal S atoms in the square planar monomers. T h e S 2p b.e. of the methionine S was shown to undergo a 1.2-eV shift to higher energy upon complexation t o Pt(I1) due to the formation of a S-Pt coordination bond. T h e magnitudes of the S 2p b.e.‘s measured in this study and C1 2p b.e.’s measured in previous studies, prompted the authors to question the assignment of a high binding energy S 2p peak (-169 eV) reported in the S 2p electron spectrum of cytochrome C (H13) and bean plastocyanine ( H 9 ) ;the -169 e\’ peak in both works were attributed to a S bound to the metal in the metalloprotein. Yatsimirskii et al. (E33)acquired XPS and Mossbauer data to study the influence of N-containing ligands upon the properties of Fe(II)(CN)5;K2, N 2 0 , NO2,N3, KH3, N2H4,and H20 were the ligands involved in the study. Correlations between Fe 2p3$2b.e.’s and quadrupole splitting values, as well as between Fe i s b.e.‘s and isomer shifts were discovered. The dinitrogen ligand was shown to be a weak electron donor consistent with the known poor stability of its transition metal complexes. Brant and Feltham (E34) examined the N 1s binding energy region for several aryldiazo derivatives of some transition metals. T h e data indicates that the R N 2 ligand is highly reduced in these complexes, and that binding energy trends can generally be accounted for by a structural knowledge of the complex. Bakke et al. (E35) reported the C 1s b.e.‘s of gas phase bis(cyclopentadieny1) complexes of the first row transition metals and magnesium, and of the bis(fulvalene) diiron complex with an experimental uncertainty of h0.05 e\’; previous studies of the hl(Cp), complexes in the solid state were accompanied by a binding energy uncertainty of f0.3 eV. By application of the point-charge potential equation, a reversal of the usual b.e. vs. charge relationship was demonstrated. Results of a XPS and Mossbauer study on ylidic compounds of gold in the +1, +2, and +3 oxidation states by Schmidbauer et al. (E36),indicate that the Au 4fi,,Zsignal is sufficiently sensitive to distinguish between oxidation states. Oku and Hirokawa (E37)studied the metal core level spectra of some mixed valence spinels: Co304, Fey04,Mn304, and related compounds. T h e Co 2p photoelectron spectrum was considered to be the sum of spectra associated with magnetic cobaltous and low spin cobaltic ions; oxidation states in Fe304 and Mn304 were not clearly delineated in the core metal electron spectra, presumably because of the high spin state of both divalent and trivalent ions. Angelis et al. (E38) explained the appearance of the LV 4f electron spectrum of Na,W03 bronzes by assuming the presence of W(VI), LV(V), and W(1V) oxidation states; partial reduction of the two higher oxidation state ions was noted during the recording of the spectrum. Chatt et al. (E39)studied Mo in formal oxidation states of 111-VI. It was demonstrated that Mo 3d b.e.‘s could not be used to distinguish oxidation states in the compounds examined: [MoC1,02L2],[MoC1,0L2],and [MoCI,L,]: L = PPh,, PMePh?, PMe,Ph, PEtPh,, MeCN. EtCN. thf, pg-ridine, 2,2’-bypyridine, 1,lO-phenanthroline, and 1,2--bis(diphenylphosphino)ethane. Binding energies appeared to be more a function of proportions of anionic to neutral ligands rather than of oxidation states. In addition, the authors noted that complexes which contained tertiary phosphines, potential reducing agents, gave rise to a variet>-of spectra; for example, a consistent spectrum of CoC1, ( N 0 ) 2 ( P P h 3 ) 2could not be obtained. Residual pressure in the spectrometer, x-ray flux,

A N A L Y T I C A L CHEMISTRY, VOL. 50, NO.

and irradiation time appeared to affect the appearance of the signals. The effect of x-ray irradiation on the surface of CrO,, K2Cr04, and silica supported CrOj was monitored by De Angelis (E40);the extent of reduction in these compounds was determined by examining the Cr 2p electron energy region and noting the growth of the band corresponding to Cr?+. The extent of reduction appeared to be primarily dependent upon irradiation time. T h e silica-supported CrO, sample was reduced most readily and it was suggested that the presence of water of hydration may favor the reduction process. De Angelis attributed the x-ray induced reactions to the presence of secondary electrons: those electrons which are produced inside t h e sample by a slowing down of t h e primary photoelectrons, and those falling onto the sample from the surroundings within the spectrometer. Both Lazarus et al. ( E 4 I )and Burness et al. (E42) reported some very disconcerting results from their XPS investigations. Lazarus et al. measured metal 2p photoelectron spectra of Fe(I1) spin equilibrium complexes and their Co(I1) and Ni(I1) analogues. A temperature-dependent study of the 2p spectrum of one of the Fe(I1) complexes- -90% low spin a t 40 OC-showed a dependence of the appearance of a broad satellite band upon temperature; the extra peak was present in the spectrum measured a t 150 "C and not in the spectrum of a fresh sample obtained at 40 "C. This temperature-related change, however, was not reversible, prompting the workers to suspect sample decomposition. The paramagnetic Fe(I1) complexes appeared to be particularly prone to the decomposition; complete decomposition was assumed to occur even in the short times needed to collect sufficient data for a meaningful spectrum in two of the Fe(I1) complexes at 40 "C. T h e x-ray beam was thought to be the primary source of decomposition. The results of this study clearly warn against attributing these extra peaks commonly observed in metal core electron spectra of paramagnetic complexes to a secondary process involving the original compound, unless evidence exits which precludes decomposition. Burness et al. reported another example of sample decomposition under x-ray irradiation, but in this study the ligand underwent a chemical change. T h e N I s photoelectron spectrum of a tetradentate Schiff bas e 1ig a n d , N.N '-ethylene bis (p y r r olid e n i mi ne ) (H,PRLEN), and its copper complex, CuPRLEN, were obtained using different sample preparations and at different irradiation times. It was demonstrated that decomposition of the free ligand was enhanced by greater irradiation periods and by dusting the sample onto cellophane tape admixed with graphite. Complexation appeared to inhibit the x-ray promoted chemical change. An interesting explanation was proposed: the imine linkage in the ligand is thought to hydrolyze on the surface of the sample; water is either being provided b5- the graphite or cellophane tape which accounts for an enhanced reaction when these substances are present, or by the water-vapor contamination associated with the intrument. U 4f binding energies and valence band PS spectra were obtained by Pireaux et al. (E-23)for UF, and K,UF,. T h e U 4f b.e.'s for these fluorides were shown to be higher than those of the uranium oxides, assigning a greater ionic character to the U-F bond. T h e U 4f electron spectra for these two compounds were analogous, both exhibiting a shake-up satellite peak separated by 7.1 eV from the main peak. The valence band spectra revealed information regarding bonding in CF4;the U 6s, 6p, 5f, and F 2s levels retain their atomic core-like character, while the C 6d, is, and F 2p levels form a bonding band approximately 8 eV below the Fermi level. T h e corresponding spectrum of K2UF6 appears to be a combination of the KF and UF, PS spectra. T h e ionization of metal 4f electrons of the rare earth oxides and of correspondin tris(dipivaloy1methide) chelate complexes was Orchard and Thornton ( E 4 4 ) . T h e measured studied separations between peaks in the 4f spectra were shown to be in general agreement with electronic absorption data and an appropriate coupling model. Castro et al. (E45) studied the XPS spectra of some representative compounds for polynuclear complexes of two f block elements with metal Schiff base complexes. Core electron binding energies of two trinuclear derivatkes of HoC1, and the corresponding binuclear ones of UO2CI2, and of HoC13.H20:U0?.3H20 and the neutral ligand complexes were presented in this uork. T h e XPS data offered some quali-

\>

5 , APRIL 1978

335 R

tative information regarding bonding in these polynuclear complexes. Satellite Lines. Many studies have been directed at the observation and analysis of those "extra peaks" which commonly appear a t the high binding energy side of metal core electron signals of transition. lanthanide. and actinide metal complexes. Various theories have been promulgated in an attempt to account for observed satellite peak trends: the most widely accepted however, attributes satellite peak intensity t o a secondary electron process, in which a charge-transfer transition of either a M L or L M nature occurs along with ionization of a core electron (E46: E47). S.Larsson (E481 conducted a theoretical investigation on the satellite structure observed in the Cu 2p photoelectron spectra of Cu(I1) complexes. Satellite peak intensity for Cu(I1) complexes is known to be greatest for complexes with highly electronegative ligands. In addition. it has been observed that in some spectra one satellite peak is associated with the 2p,,2 electron signal while two appear near the 2p3, main peak. Larsson explained these experimental observations by assigning a u u* C-T transition of a M L character to the shake-up process. T h e complex satellite structure is thought to be due to multiplet splitting associated with a pad9 final state configuration. The same author (E49)calculated satellite peak intensities associated with the core electron ionization of some do first row transition metals in their complexes: the Multiple Scattering method was used in this theoretical study. I t was determined that intensity increases with relaxation M), and decreases with 3d electron charge-transfer (L occupancy in the neutral ground state. T h e Multiple Scattering MO method was also applied to some diamagnetic octahedral transition metal cyano complexes (E50).Absence of satellite structure in the inner shell electron spectra of these complexes is attributed to small charge relaxations associated with ionization. J. A. Tossell (E.51) applied the SCF-Xct Scattered b-ave MO method and Larsson's formalism to investigate the origin of reported satellite bands in the h l n 2p electron spectrum of hlnF, and MnI, (E51). Calculated satellite intensities and hole state energy separations for both eg eg* and t2g t2g* (L M) transitions compared poorly with experimental data. Observed satellite splittings resembled most clearly the hole state energy separations calculated for a M n 3d conduction band transition; such a transition is forbidden by the monopole selection rules derived from the sudden approximation (E52). Tossell reported similar results on some transition metal oxides and consequently proposed a valence conduction band transition as well (E53, E54). Vernon et al. (E%) observed satellite peak splittings and relative intensities in a multitude of 2p photoelectron spectra of transition metals in their complexes. Several trends were noted: satellite structure due to electron shakeup does not commonly occur with photoionization in core levels of second and third row transition metals; intense satellite structure appears to be correlated with paramagnetic compounds; and in going across the periodic table to higher 2 for the first row transition metal compounds, one generally notes an increase in satellite band intensity. On the basis of experimentally and theoretically observed trends, Vernon et al. suggested that a charge-transfer transition of L M character is responsible for satellite peak intensity as originally proposed by Kim (E46). In addition, on the basis of these trends, a eg eg" transition was proposed to be exclusively responsible for the satellite structure observed in this work although monopole selection rules permit a t2g t2g* excitation as well. Results of M O calculations on ground state molecules were used in making this assignment. Sen et al. (E56) reported satellite peaks in the Ti4+2s and 2p photoelectron spectra of Ti02not previously observed. Specifically. the 2s, 2 p , , and 2pn photoelectron main peaks were each shown to be accompanied by three extra signals: the satellites a t 4.0 e\' and 5.0 e\' to higher b.e. of the 2s and 2p main peaks, respectively, which have not been previously noted were attributed to a t2g + t2g* ( L M ) shake-up transition; the satellite appearing at 13 e\' from the main peaks was associated with an eg --• eg* (L Mj transition; and the third satellite peak (1E 26 eVj is thought to be due to an energy loss process. The existence of these additional satellite peaks was predicted b y Sen et al. in a previous study (E57). Frost et al. ( E % ) also examined metal core electron spectra of some 3d" transition

- -

-

-

-

-

-

-

-

-

-

-

-

-

-

~

-

336R

A N A L Y T I C A L CHEMISTRY, VOL. 50, NO. 5, APRIL 1978

metal complexes; satellite structure was observed at -8 and 1 2 e\' to high b.e. from metal 2p main peaks for Sc'" and Ti4+ compounds. T h e 8-eV satellite bands were attributed to a t2g - + t2g* (I, --- M)shake-up excitation while the 12-eV satellite peaks were associated with an eg eg* transition (L Mj. Since the relative intensities of these shake-up transitions are dependent upon the extent of L-M orbital mixing, the latter transition is expected to be favored in t2g* compounds with strong (i interactions, while the t2g transition is predicted to occur in complexes with significant K interactions. T h e Sc3+ 2p photoelectron energy region of S C ~ ( S Oexhibited ~)~ a small satellite peak at -8 e\.' from the t2g* L M C-T main peak which was attributed to a t2g shake-up excitation. Borodko et al. (E591 correlated the intensity of the satellite structure accompanying the Co 2p3 L' main peak of several CoiII) complexes with magnetic moments. Ioffe and Rorodko (E601 reported a correlation between satellite intensity in the Cu 21) photoelectron spectra of some Cu(I1) complexes and spin density on the Cu atom as calculated from ESR data. Both studies proposed a mechanism based on spin exchange between photoelectron and unpaired valency electrons during the photoionization process. Wood and Urch ( E 6 l ) compared x-rab- emission spectra (XES) and the XPS of several manganese and nickel compounds. It was shown that such a comparison permits an assignment of extra peaks to either multiplet splitting or multielectron effects; bands due to the former would be common to both types of spectra. Orchard et al. (E621 reexamined exchange splitting in the Cr 3s electron spectra of K,3Cr(CN)6and K,Cr(NCS),. Two peaks appeared in the Cr 3 s spectra of fresh samples of these compounds separated by 3.6 eV. Such a splitting is in line with exchange splittings reported for chromium complexes ranging from 3-4 e\'. It was also noted that after -5 h two new peaks appeared in t h e core electron spectrum of K3Cr(CK)6due to surface decomposition of an unknown nature. Satellite structure is being observed in core electron spectra of lanthanide and actinide metals in their complexes. Bancroft et al. (E631 conducted a detailed study of core level spectra of some diamagnetic actinide complexes; extra peaks due to both electron shakeup and configuration interactions were observed. It was shown for example that the intense satellite peaks appearing in the T h 5p3 spectra of Th(IV) complexes originte from configuration interaction of the 5p hole state with a 5d2 hole state. Shake-up satellites cited were explained by I, M C-T excitations. The L M C-T mechanism was supported by X P S data presented by Allen and Tucher iE64. This process appeared to be responsible for satellite peaks obseriTed in the T h 4f spectra of several Th(I\.') compounds. It was noted that the positions of the satellite peaks depended on the nature of the ligand in the complex; also, satellite---main peak separations were indicated to be generally greater in actinide as compared to lanthanide spectra: while satellite peak intensities displayed an opposite trend. T h e 5f group complexes niay show low satellite intensities due to a small degree of overlap between the appropriate metal and ligand orbitals. Berthou et al. (E65)examined the 3d photoelectron region of the first four lanthanide metals in their trivalent oxidation states. T h e satellite structure appearing in the LalIIIi spectra was assigned to a L M 4f electron transfer as previously proposed. Satellite intensities for these compounds and the nephelauxetic effect are shown t o be better indicators of the covalent character of the M-L bond than satellite-main peak separations and electron transfer energies in the ground state. T h e cerium. praseodymium, and neodymium spectra exhibited a different satellite structure which was attributed t o the presence of 4f electrons in the ground state giving rise to an excited 3d9 4fB (q = 1, 2.3) configuration. X I, M (0 Pp * M 4f) C--T excitation was assigned by Hurroughs et al. (E66)to the satellites appearing in the core level spectra of La(II1) and Ce(1V) in some binary and mixed oxides. A shake-down transition was assigned to several extra peaks observed: for example, the high KE component in the 1,;i :id3,? and 3d;.? spectra of La(tmhd):j (tmhd = 2.2,6.6tetramethylheptane-3,5-dionato) is of lower intensity and was attrihuted to a shake-down transition. Weber et al. iE67i studied the inner shell electron spectra of lanthanide trii'liiorides by application of the MS Xn method and Imson's formalism (E-18).It was demonstrated that satellite intensities

-

-

~-

-

--+

-

-

+

-

are stronglb- dependent on both the amount of charge transferred to the 4f shell at ionization and 4f occupancy in the neutral ground state. Calculations imply that 3d satellites should shift from the higher b.e. to the lower b.e. side of the main peak with increasing 4f occupancy in lanthanoid fluorides. T h e intensities of shake-up satellites in the 0 1s and C 1s core electron spectra of some transition metal carbonyls were observed to vary relative to the main peak according to the energy of the exciting radiation. Photon energies of 21.2,40.8, and 1253.7 eV were utilized in this study leading to the conclusion that relaxation effects are time dependent in the carbonyls examined. Rajoria e t al. (E68) also demonstrated, however. that satellite intensities are constant for photoelectron kinetic energies greater than -20 eV. Oku and Hirokawa ( E 6 9 ) ,by obtaining the Co, Xi, and Cu 2p photoelectron spectra for samples of M,hIg, ,O. CuO, NiO, and CuO, showed that next nearest neighbor metal ions may influence final states after photoionization; shake-up satellite main peak intensity ratios and F W H M values of main peaks, were observed to change when the oxides were immersed in an MgO solution. These results question the validity of employing ML, cluster models when calculating satellite intensities.

F. SOME ORGANOMETALLIC COMPOUNDS Flamini et al. i F I ) investigated bonding interactions in MhTe,,,XlM = C , Si, Ge. Sn. Ph; X = CI. Rr) compounds by He(1) PS and CNDO MO calculations. Halogen lone-pairs in these compounds are shown to participate in bonding by their mixing with appropriate orbitals of the MMe3 moiety. T h e He(1) PS of bivalent homoleptic alkyls and amides of Group 4a elements were interpreted by Harris et al. iF2). The first I P for the alkyl series is close to that of the free metal. In the amide series. the second band corresponds to the metal lone-pair and increases as the metal increases in atomic number, while the first I P is identified with the N lone-pair orbitals. It should also be noted that evidence for a N-Si (p-djT interaction is presented. A UPS study on several phenyl derivatives of Group .la elements (HMPh,, H M (mesityl):,,and H2MPh2;M = Si, Ge, Sn) by Distefano et al. ( F 3 ) .showed a lack of significant through-space interaction among K orbitals of the rings, although such an interaction is apparent in the C analogues. T h e different behavior is attributed to the short central atom-ring distance characteristic of the C analogues. Bernardi et al. (F4obtained He(1) PS spectra for two series of compounds containing the SX(CH3)?-Sgroup (X= C, Si and Sn), namely. dithiolanes and the related open-chain species. MO calculations coupled with P S data lead to several informative conclusions: metal d orbital participation in bonding becomes significant only when X = Si: when X = Sn, the S lone-pair combinations are destablized by the inductive effect which appears to be masked by other effects in the other compounds. T h e [IPS of tetrathiofulvalene ( T T F ) , diselenodithiofulvalene: and tetraselenofulvalene were examined by Schweig et al. (F5) in order to investigate a previous interpretation of the TTF spectrum in which the ordering of the 5th and 6th IP's as suggested by CIiDO i s calculations was reversed. Results of this study, hotvever. supported the original ordering presented by the calculations. T h e valence electronic structure of methyl-, dimethyl-, trimethyl-. and ethyl-silane are described by Szepes et al. (F6) as a result of their UPS studies and CNDO/2 calculations on these compounds. Experimental and theoretical trends are in agreement in this work. MO's which are indicated to have the greatest amount of d orbital character are localized on the Si-Cand Si-H bonds, while the 7i MO's localized mainly on the alkyl groups are almost devoid of d orbital participation. A high resolution He(1) PS spectrum of SiF, was reported by ,Jadrny et al. ( F 7 ) ;vibrational structure is evident in three out of four accessible electron hands. Ionization potentials of silanes (Si,,H,,,+,) were reported by Rock et al. (F8).He(1) PS spectra were interpreted largely by spectral comparisons and modified CKDO calculations. Rock et al. iF9) obtained the UPS spectra of four azo compounds including bisitrimethylsily1)diimine and its carbon analogue: only the strongly substituent first bands could he assigned unambiguously in this study. Ah initio SCF-310 calculations and HeiI) PS of halo-

A N A L Y T I C A L CHEMISTRY, VOL. 50, NO. 5, APRIL 1978

gen-hridged dimeric Group 3a metal halides and methylmetal halides were described by Lappert et al. ( F 1 0 ) . T h e predominantly bridging halogen orbitals exhibited in all dimeric species studied a higher associated I P than in the corresponding monomers; such a result implies a loss of electron density at the bridging halogen atom. Several series of silyl and germy1 compounds have been investigated by monitoring shifts in core level binding energies. Drake et al. (FII)determined b.e.'s for all atoms in the following series: MenMC14-nand Me,MX; n = 0 4, M = Si or Ge, X = F, C1, Br, I: M = Ge, X = CY, N3! NCS. Measured chemical shifts were correlated with estimated atomic charges derived from an electronegativity-equalization procedure. T h e same procedure was successfully applied in a study on bromo(methy1)- and iodo(methy1)-silanes and germanes ( F 1 2 ) : hle, MBr,.,, M = Si, Ge, n = 0 3; and Me,MI,.,: M = Si, Ge. n = 2. 3. T h e C Is, Ge 3d, and Si 2p b.e.'s appear to be fairly similar for each value of n regardless of the nature of the halogen implying that in these compounds the halogen electronegativities are almost the same: such an observation is consistent with a steady increase in halogen s-orbital participation along the series C1 > Br > I as indicated b y the model applied to this study. The same workers (F13) measured core level b.e.'s for methyl substituted disilyl and digermyl chalcogenides: (Me,MH,_,),E; M = Si, Ge: E = 0, S, Se, Te; n = 0 3. Binding energies were compared to those of the Me2E and H,E series. For the chalcogenide compounds, it was noted that binding energies varied only slightly. (p-d)s bonding, varying s orbital participation. and inductive effects were used to account for experimentally observed trends. Gray et al. ( F 1 4 ) reported Si 2p b.e.'s for 33 organosilicon compounds in response to the paucity of X P S data on Si compounds. Partial atomic charges were calculated by various electronegativity models and CNDO calculations, and were correlated to measured b.e.'s. Group shifts were determined for various groups bonded to Si and were found to correlate with known group shifts for C and P compounds. Evidence for strong Si-N bonding in some pentacoordinate Si compounds was presented by Hercules et al. (Fl.5). T h e Sn 3d photoelectron line was observed by Avanzino and Jolly ( F 2 6 ) for 15 S n compounds in the gaseous state; measured S n b.e.'s are shown to span a range of 4.4 eV and to correlate well with b.e.'s calculated using the point-charge potential model. By studying N Is and S 2p b.e.'s, Limouzin and Llopiz (FI 7) determined that a free thiol group is required if a complex is to form between Bu3SnC1 and amino acids. Bancroft et al. (FIR) studied solid state broadening in an X P S investigation on some S n organometallic compounds; the S n 3d line widths obtained by subliming thin films onto P t metal were within 1 0 7 ~of gas-phase values. The correlation between S n 3d linewidths and Miissbauer quadrupole splittings demonstrated that the observed S n 3d line broadening is due to the splitting associated with C20 term (quadrupole term) in the crystal-field expansion. T h e C2' term is shown to also be responsible for the observed splitting of the Cd i d signals of gaseous Me,Cd, and for apparent S n 4d line broadening in various solid organotin compounds (8'19). Such S n 4d line broadening was also reported in a study employing Synchrotron radiation as the photon source (F20). Both valence band and outermost d core levels spectra of some solid Sn, In, Sb, and P b compounds were obtained in this work. Spin orbit splittings of 4d and 5d core levels are shown to be independent of the chemical environment of the metal. In addition, the advantages of this technique are reviewed in this paper as well.

-

-

-

G. STUDIES O N SOME INORGANIC SYSTEMS Hargis and Worley ( G I )obtained and interpreted the UPS o f some tris(dialky1 aminophosphines) in an effort to resolve a controversy regarding the He(1) PS spectrum of P[N(CH,),],. T h e data presented appear to contradict previous interpretations of this spectrum, and consequently a different band assignment is offered. T h e U P S of some caged phosphorus compounds and acyclic species were reported by Cowley et al. ( G 2 ) ;it was ascertained that the I-' lone-pair IP's of the caged species differ appreciably from those of the acyclic analogues. T h e spectra of these caged compounds are important since they convey information regarding the effect of stereochemistry on the nature of lone pair-lone pair interactions. Such knowledge can be utilized to determine

337R

molecular geometries from ionization potentials of acyclic compounds. He(1) P S spectra of a series of phosphorus halides (R,P(X)Y and RP(X)Y,: R = Me, F; X = O! S!Se: and Y = CI, Br) were completely analyzed bb- Elbel and Dieck (G3); interpretation was facilitated by comparison of these spectra with the spectra of R3PX, Y3PX, R2PY, and RPY, ( G 4 , and of similar molecules. It was determined that all original R2PY and RPY, orbital energies are significantly stabilized upon coordination as a consequence of the strong electron-withdrawing effect of the acceptors and somewhat by hyperconjugation. Furlani and Andreocci (G5) also assigned bands largely by comparisons between spectra, and by C N D 0 / 2 calculations; the spectra of the methyl dihalogenophates: PX,O(OMe), PCl,O(OEt), and PC12 ( 0 M e ) S (X= F, C1, Br) were compared with the spectra of the parent compounds: POX3 and PSX3. Several trends were noted upon substituting a OMe group for an X atom: IP's were shifted to lower values; band contours appeared less resolved, probably due to the lowering of symmetry; and several new bands appeared corresponding to the OMe group. U P S data and CNDO results were shown to be in agreement regarding the electronic structure of some phosphorus ylides ( G 6 ) . For the parent system, (CHJ,PCH,, a P+-C- charge distribution was supported. T h e first band (6.81 eV) was assigned to carbanion ionization and shifted to lower I P in most derivatives investigated. Core electron b.e.'s of some cyclophosphazines were reported in the gas phase by Avanzino et al. (G7).namely of (NPX&, X = F, C1; and (NPX?)4,X = F, C1, CH3. T h e small change in b.e.'s between analogous tetramers and trimers prompted the authors to conclude that the degree of K bonding is also similar in these analogues. Clark and Rizkalla ( G 8 ) ,and Carriere et al. (G9) studied some silicon-oxygen compounds by XPS. In the former work, the Si 2s and 0 Is b.e.'s of some silicates were examined and correlated with calculated charges obtained from an iterative Pauling method. Deconvolution of the 0 Is signals of the metasilicates revealed two peaks. A study of these peaks prompted the authors to propose (M2Si03.2H20), (M = monovalent metal ion) as a more appropriate formula for these compounds. Carriere et al. (G9)applied X P S to more than 50 silicon-oxygen compounds. The Si 2p lines were analyzed in this study; three different oxidized states for Si were identified, and associated chemical shifts relative to unoxidized Si were determined for these states. Adams e t al. (GI01 analyzed some aluminosilicates quantitatively through standard techniques and by application of X P S in order to evaluate X P S as a quantitative analytical tool. It was determined that XPS can provide bulk quantitative analysis of homogeneous solids with an accuracy of - 5 7 ~ . The study also demonstrated the potential value of the technique to geochemists interested in the surface composition of mineral specimens. Brant and Feltham ( G I I ) showed that X P S in some cases can be used to obtain the stoichiometric ratios of S/Mo and N IMo in uncharacterized molecular compounds. Briggs et al. 612) demonstrated the ability of X P S to identify trace cationic and anionic species in water. It is interesting anion appeared to undergo a rapid to note that the 103~ reduction to 1- during examination. Orchard and Thornton (G13) revealed X P S data which attributes mixed valency character to tr-Sb,O, and BaBi03. In addition. the study indicated that unless special precautions are taken in sample treatment. a mixed valency compound may have a surface phase which contains only the most stable oxidation state of the metal and. therefore, is unrepresentative of the bulk material. X P S was applied to neutral and anionic indium halide species ( G I 4 ) . It was shown that because of lattice effects, no correlations can exist between In b.e.'s and oxidation state, ligand electronegativity and coordination number. T h e He(1) P S spectra of some electron donor-acceptor complexes between Br2 and alkylamines were obtained by Utsunomiya et al. (G15). The essentially lone-pair N orbital of the alkylamine shifted towards higher IP, while the bromine K 4 P and (T 4p peaks display lower IP's upon complex formation. Such trends agree with the notion that complex formation stabilizes the lone-pair because of its new bonding character, while the T 4p and u 4p MO's are destablilized because of a n increase in electron density on bromine. Ikemoto et al. (GI61 analyzed the shapes of S 2p peaks to determine the amount of charge-transfer in the tetrathio-

338R

A N A L Y T I C A L CHEMISTRY, VOL. 50, NO. 5. APRIL 1978

fulvalene-tetracyanoquinodimethane(TTF-TCNQ) complex.

Measured S 21, binding energies lead Monroe and Swinele ( G I 7) to conciude that-there-is a greater charge-transfer"in TTF-TCNQF, than in the TTF-TCNQ complex, implying that a significant charge-transfer mav be detrimental to high conductivity. Shake-6p satellites appearing in core electrin spectra of TCYQF, were also investigated in this study. Copperthwaite and Lloyd (G18) conducted a novel experiment in which the kinetics of the photoinduced decomposition of solid KaC104 and NaCIOg were determined from XPS data. Sodium perchlorate is reduced in this reaction to the chloride ion via sodium chlorate in a consecutive first-order process. C1 2p and 0 Is signals were monitored as a function of exposure time.

H . STUDIES ON BIOMOLECULES Some novel and interesting applications of X P S and U P S t o biochemical problems have been recently reported. Such studies strongly support the contention that photoelectron spectroscopy is developing into a useful tool for the biochemist in many diverse areas. Several workers studied the surface and subsurface compositions of various cells by application of X P S and etching techniques. Millard and Bartholomew ( H I ) studied the surface elemental composition of mouse fibroblast cells and malignant transformed cells. The line intensities coupled with appropriate elemental sensitivity factors were used to determine approximate atom ratios. The 0 to Tz atom ratio was reported to be generally lower for transformed cells. In addition X P S and oxygen plasma etching revealed that P concentration appeared to change in a linear fashion with time to a depth of -30 nm, and appeared to be similar for cells before and after transformation. Millard et al. ( H 2 )applied XPS. electron microscopy. and oxygen etching to investigate the surface and subsurface composition of several types of bacterial cells including E. coli cells. Distribution of teichoic acids were determined throughout the cell wall of two Bacillus species by examining the P core level signal. Na exhibited strong signals on the surface of all cells and diminished as subsurface levels were exposed. while the K line intensity increased with oxygen etching. The surfaces of the bacterial cells investigated in this work displayed some substantial differences in atomic ratios (C/O/N) which agreed with known surface composition of biopolymers. Meisenheimer et al. ( H 3 ) determined in depth profiles of vacuum dried human red cells by application of X P S and argon etching. Thallium signals (4fi!2, 3f,,,) appeared after -100 8, of surface had been etched away. T h e data accumulated in this study implies that TI+ is concentrated in and just below the phospholipid region of the red cell membrane. In addition, the Fe 2 p , , and Fe 2p,,, signals appeared in the XPS spectrum after etching to a depth of -1000 A attributing a thickness of >500 8, to the red cell membrane. White et al. ( H 4 ) employed XPS to ascertain the oxidation states of Os in erythryocyte ghosts and related systems treated with osmium tetroxide. OsO, has been used extensively as a fixative and stain for electron microscopy studies of biological samples. Os 4fTI2b.e.'s for compounds containing osmium in known oxidation states were determined, and compared to those of various lipids treated with osmium tetroxide. The spectum of er5Throcyte ghost cells was reproduced by combining the spectrum of phospholipid and of cholesterol all treated with OsO,. suggesting the preference of this stain for lipids in the fixation process. Lack of reaction with the saturated compounds studied in this work prompted t h e researchers to support the previous contention that unsaturated lipids are the predominant substrates for OsO, in membranes. Winograd et al. ( H 5 ) obtained the C IS and 0 Is electron spectra of chlorophyll a at various temperatures. Binding energy data and relative peak intensity measurements via computer deconvolution verified a monohydrate stoichiometry of Chl a.H,O. At temperatures above 120 "C the shoulder in the 0 I s spectrum diminishes and vanished completely a t 250 "C. This photoelectron study marks the first direct spectroscopic observation of the water of hydration in chlorophyll a. XPS continues to be a useful technique in providing information concerning bonding and structure of metalloproteins and of their model compounds. Rupp and Weser ( H 6 ) concentrated on using X P S to determine the oxidation state of copper in some complexes of biochemical significance. Their

work agreed with previous studies ( H 7 ) in that the Cu 2p photoelectron signals of Cu(I1) complexes were accompanied by shake-up satellite peak(s) in all cases, while the analogous spectra associated with Cu(I) complexes were devoid of these extra peak(s). In addition, satellite structure was dependent on the nature of the ligands and the geometry of the complex. It is emphasized that the oxidation state of copper in antiferromagnetically coupled Cu(I1) complexes can be determined by XPS. For example, the Cu 2p spectrum of the diamagnetic Cu2.(1.3-diphenyltriazene), complex exhibited satellite peaks in agreement with the known divalent oxidation state of copper. X P S of several complexes which contain S ligated to copper were reported in this work. T h e copper cysteine complex is shown to contain Cu(I) only, while the copper cystine complex is characterized by Cu(I1); both of these complexes are prepared with a Cu(I1) salt. In addition, reaction of Cu2+ with n-penicillamine gave rise to a product which contained both Cu(1) and Cu(I1) according to X P S data in agreement with Birker and Freeman ( H 8 ) who recently showed this compound to be an anionic Cu(I), Cu(I1) cluster complex. The S 2p binding energies corresponding to S ligated to Cu all fell within a 162.7-163.7 e\' range. Such values indicate small increases in S 2p binding energies upon complexation. Because of these low values, Rupp and Weser questioned a previous X P S study on plastocyanin ( H 9 ) ,a blue copper protein, in which a binding energy of 169 eV was assigned to the 2p electrons of a cysteine S ligated to copper. Such high S 2p binding energies have been associated with highly oxidized sulfur e.g. RSOT or S O:.. Peeling et al. ( H I 0 ) also challenged the spectral interpretations of Solomon et al. ( H 9 ) as a result of their studies on hemocyanin. A high binding energy S 2p signal was observed due to a dialyzable oxidized sulfur impurity. Such contamination was therefore attributed to the samples examined by Solomon et al. (H9). This impurity theory was criticized by Solomon et al. ( H I 1 ) who emphasized the fact that the bean plastocyanin, apoprotein, and Co(I1) reconstituted proteins were all prepared from the same protein sample. Consequently, an impurity could not explain the absence of a high binding energy S band in the S 2p spectrum of the apoprotein, and the presence of such a peak in both the native and Co(1I) substituted protein spectra. In addition, a high binding energy(HBE) S 2p signal at 169 e\: was reported by these workers in their spectra of two other blue copper proteins: Rhus cernicifera stellacyanin and Pseudomonas aeruginosa azurin ( H l l ) . Studies on the corresponding apoproteins again reveal an absence of H B E sulfur while the Co(I1) reconstituted stellacyanin exhibits the anticipated HBE S 2p peak. Solomon et al. also investigated Limulus oxyhemocyanin and detected a small H B E S 2p signal implying an oxidized-sulfur impurity present in the samples examined by Peeling e t al. T h e absence of a H B E S 2p peak of significant intensity is reasonable in light of the fact that no evidence indicating S-Cu bonding in hemocyanins has been presented a t this time. An important aspect of these protein studies which should be mentioned involves the oxidation state of copper. Solomon et al. observed both Cu(1) and Cu(I1) 2p photoelectron peaks in the native plastocyanin protein a t low temperatures. T h e former peak is attributed to partial photoreduction of Cu(II), a process which is becoming a severe problem to photoelectron spectroscopists. Because of the increasing number of reports citing x-ray beam induced reactions, it is difficult to rule out the possibility of a photocatalyzed reaction in which the cysteine S bound to Cu(I1) is oxidized: such oxidation has been documented in copper proteins where a Cu-S bond is present. Prinz and LVeser ( H I 2 ) reported the presence of oxidized S in aged preparations of Cu-thionein while fresh samples were shown to possess cysteine S exclusively. Erythrocuprein exhibited a similar oxidative decomposition. The binding of S to Cu(I1) may enhance its susceptibility to oxidation. T h e feasibility of this mechanism is presently being investigated in regards to the H B E S 2p signal detected in cytochrome c (H13). LVeser et al. ( H I 4 studied the reaction of selenite with biochemically active thiols including mercapto-ethanol, dodecane-thiol. N-acetylcysteine, D-penicillamine, and thionein. The Se 3d: