Esca Studies of Core and Valence Electrons in Gases and Solids C

ESCA STUDIES OF CORE AND VALENCE ELECTRONS IN GASES AND SOLIDS C. Nordling

To cite this version:

C. Nordling. ESCA STUDIES OF CORE AND VALENCE ELECTRONS IN GASES AND SOLIDS. Journal de Physique Colloques, 1971, 32 (C4), pp.C4-254-C4-263. 10.1051/jphyscol:1971447. jpa- 00214648

HAL Id: jpa-00214648 https://hal.archives-ouvertes.fr/jpa-00214648 Submitted on 1 Jan 1971

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés. JOURNAL DE PHYSIQUE Colloque C4, supplkment au no 10, Tome 32, Octobre 1971, page C4-254

ESCA STUDIES OF CORE AND VALENCE ELECTRONS IN GASES AND SOLIDS

C. NORDLING (*) Institute of Physics, Box 530, S-751 21 Uppsala 1, Sweden

Rbsumb. - Les processus electroniques dans les systkmes atomiques sont habituellement associes a l'kmission ou l'absorption de photons ou a remission d'electrons. L'ktude spectrosco- pique peut par consequent s'effectuer par l'analyse du rayonnement electromagnetique ou par la mesure de 1'6nergie cinetique des klectrons. Alors que la spectroscopie electromagn8tique dans le domaine optique se pratique depuis des siecles et depuis plusieurs decades pour les rayons X, l'electron lui-m&men'a pas ete tres utilise pour explorer la structure Blectronique et les processus electroniques. Cependant en raison du developpement ces dernikres annees de moyens experimen- taux pour I'analyse precise des spectres d'klectrons, ce type de spectroscopie a produit des rksultats tres encourageants qui montrent que la spectroscopie basee sur l'observation directe des electrons est une methode efficace d'etude des systkmes atomiques et mol6culaires. La spectroscopie des electrons fournit egalement des renseignements qu'on ne peut obtenir par d'autres types de mesure et il y aun grand nombre d'applications de ce nouveau type de spectroscopie. Noys exposerons brikvement le travail effectue par notre groupe k Uppsala dans le domaine de la spectroscopie des electrons pour les atomes et les mol6cules. Un expose plus etendu peut 6tre trouv6 dans les rkfkrences [I] et [2]. Abstract. - Electronic processes in atomic systems are usually associated with theemission or absorption of photons or the emission of electrons. The spectroscopic study of these processes can therefole be made by the analysis of the electromagnetic radiation or by a measurement of the kinetic energies of electrons. While electromagnetic spectroscopy in the optical region has been made for centuries and in the X-ray region for many decades the electron itself has not been used very much to probe the electronic structure and the electronic processes. However, following the development in recent years of experimental devices for the exact analysis of electron spectra this type of spectroscopy has now produced some very encouraging results which indicate that the spectroscopy based on the direct observation of the electrons is a powerful method for the study of atomic and molecular systems. Electron spectroscopy also produces information which cannot be obtained by other types of measurement and there is a multiplicity of applications for this new type of spectroscopy. A brief account will be given of the work which has been done by our group at Uppsala in the field of electron spectroscopy for atoms and molecules. A more comprehensive account until the present year is given in references [I] and [2].

1. EIectron binding energies and photoionization dynamics in the noble gases. - Different modes have been used>to excite the electron spectra, viz X-rays, UV-radiation, and electron impact, and the energy (or momentum) analysis of the spectra is made in double focussing electron spectrometers of electrostatic or magnetic type, see figure 1. When photons

are used for the excitation the kinetic energies of the SPECTROMETER expelled electron are

where Eb is the electron binding energy (ionization FIG. 1. - Different modes of excitation of electron spectra. energy). With X-ray quanta one can liberate electrons from all parts of the electronic structure, i. e. one can widths can be obtained but the technique is limited study both the atomic core and the valence electrons to the outermost parts of the electronic structure. in molecules and solids. This is the mode of excitation X-ray induced electron spectra from the noble that we have used in most cases. UV excitation has the gases are shown in figure 2. The spectra map out in advantage that electron lines with smaller inherent some detail the electronic structure of the noble gas

Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphyscol:1971447 ESCA STUDIES OF CORE AND VALENCE ELECTRONS IN GASES AND SOLIDS C4-255 c/120~, atoms, from the outer shells and as far into the atomic - 400- He Mg Ka core as attainable with the quantum energy of the magnesium Ka radiation, 1 253.6 eV. Electron emission from deeper lying shells was induced by harder radia- loo- tion, for example Cu Kcc. The electron binding energies

30 20 ' ' ev obtained for the noble gases are given in Table I (see 1s 2m0' Ne also ref. [2]). One can conveniently study the entire electronic structure by use of one and the same ins- trument and at a resolution close to the limit set by 1000 inherent width of the atomic levels. For example, the observed width at half maximum intensity of the neon o- 1 s line is 0.80 eV which is near the natural width 1. I 880k 870 860 " &I ' Lo ' 30 ' 20 ' 'ev of the exciting X-radiation. The spectrometer window was in this case approximately 0.2 eV. This high resolution is valuable for a more detailed study of e. g. C I>Z the dynamics of the K photo emission. Figure 3 8 100-

100 8 1

KINETIC ENERGY 100 FIG. 3. - Neon 1 s electron spectrum excited by Mg K radia- 690 680 670 " 210" 140" 70 tion at a pressure of 0.5 torr. The main peak (0)is the Ne l s line BINDING ENERGY excited by Mg Kal,z. The MgKX-ray satellites and KB radiation give the peaks with higher kinetic energy. The peaks with lower FIG. 2. - ESCA spectra from the noble gases excited energies (numbers 1-12) are due to shakeup, shake-off and by Mg Ka X-radiation. inelastic scattering.

Binding energies for the noble gases (eV)

(*) Reference value (from optical spectra). C4-256 C. NOR shows the electron spectrum from neon over an energy is an order of magnitude smaller, i. e. E,,,, E 1.8 eV, range of 130 eV around the Ne 1 s line. On both sides which is a reasonable numerical value. of the main line a number of satellite lines are observed. The intensities of these lines are less than 10 per cent 2. Autoionization and Auger electron spectra. - of the main line which in the figure has been reduced Electrons are used to excite autoionization and Auger in intensity by a factor of 20. electron spectra from gases. (Auger electron spectra The satellite lines have essentially three different are also excited by photons.) The electron energies are origins. The high energy lines are due to the high then independent of the energy of the bombarding energy satellite lines in the incoming X-radiation. particles : Photoelectrons from the neon 1 s shell induced by more energetic X-radiation will consequently have higher kinetic energy. The low energy part of the spectrum contains a number of fairly sharp peaks E" denotes the energy (above the first ionization energy) superposed on a broad continuum starting sharply of the excited atom in the autoionization process. E+ is around 362 eV and extending some 50 eV toward the energy of an atomic or molecular ion with an lower electron kinetic energies. The intensity of the inner (Auger) or outer (autoionization) shell vacancy, continuum and the distinct features 1 to 4 in figure 3, EC+ is the energy of the doubly ionized atom or measured relative to the main line, is found to be molecule in the Auger process. E,(i) denotes the bin- pressure dependent. These features are therefore ding energy of electron i (i = 1 is initial state vacancy interpreted as due to secondary collisions between in the Auger process, i = 2, 3 are the final state vacan- ejected photoelectrons and neutral atoms. The remai- cies). Thus from autoionization and Auger electron ning features 5 to 12in the spectrum are independent of spectra one obtains complementary information to pressure and are due to ionization processes at which that obtained from photoelectron spectra : from the a valence electron is simultaneously ejected or excited. autoionization spectra one obtains information on The former process is usually called shake-off and for highly excited states of the neutral atom or molecule the latter we have used the term shake-up. It turns out and from the Auger spectra one obtains information that lines 7-1 1 are due to shake-up processes to states on doubly ionized atoms and molecules. of the type 1 s 2 s2 2 p5 np2S and line 12 to a shake- As an example of an autoionization electron spec- up process of type 1 s 2 s 2 p6 ns2S. trum of a noble gas figure 4 shows a high resolution Lines 5 and 6 are interpreted as the Ka,,, satellite lines of 7 and 8 + 9. The complete shake-off of a 2 p 410. ARGON electron, i. e. the ionization limit of the first term series, lwo- I , -4,. - occurs close to line 11 at an excitation energy of 47 eV. The shake-up states have three unpaired electrons. 1 I Using a multiconfigurational SCF procedure the term splitting could be calculated for n = 3,4 and 5. Thus, lines 7 and 8 could be identified as a lower and upper doublet state, respectively, in the term lines 9 and 11 as a lower and upper state in FIG.4. - Autoionization electron spectrum from argon showing four identified series of lines. and line 10 as the lower state in Ne' 1 s 2 s2 2 p5 5 p. Compared to the experimental spectrum the calculated study of argon. Several Rydberg series can be identified energies reIative to the main Ne 1 s lines are consis- in this spectrum with series limits corresponding to tently displaced toward lower energies by about 1.8 eV. ionization of a 3 s electron in the initiaI state. The The spacing between the states is very close to that appearance of two Rydberg series for each of the experimentally observed, however. These findings initial state configurations 3 s 3 p6 nd and 3 s 3 p6 np may be explained in the following way : The main line can be assigned to the spin-orbit splitting of the 3 p represents a state with one 2 p electron pair more than shell. The splitting is 0.18 eV and the intensity ratio he shake up states. There will thus be a larger electron between the two Rydberg series of each configuration correlation energy for the main 1 s hole state than for is 2 : 1. It is interesting to note that these findings are the shake up states. If the SCF calculations, which in close agreement with what we previously have omit the correlation energy, are accurate enough, the observed by means of photoionization of argon difference found between the observed state energies using the He resonance radiation at 21 eV for excitation and the calculated ones should yield an approximate This doublet is shown in figure 5. The resolution in value of the electron correlation energy for a 2 p eIec- this experiment, defined as the full width at half tron in Ne' 1 s 2 s2 2 p6 since the relativistic correction maximum height of the electron line, is I3 meV. ESCA STUDIES OF CORE AND VALENCE ELECTRONS IN GASES AND SOLIDS C4-257

Molecular Auger and autoionization electron spectra have been little studied so far. Evidently, the electron source for excitation need not be monokinetic (as is the case for studies of discrete energy loss spectra of electrons). In the experiments which we have perfor- med, the electron source has been a beam from an electron gun with an energy of a few keV traversing the gas target chamber perpendicular to the emission angle of theelectrons to be studied. In this way we have recorded at high resolution Auger electron spectra from all the noble gases. These spectra are very line rich and can in part be compared to transitions in UV spectra. An interesting field from a chemical point of view is the study of such spectra emitted from molecules. We have been able to observe chemical shifts in such spectra both for solids and gases, although in general these spectra are much more complex and more difficult to interpret than the photo electron spectra. We have also found evidence for a vibrational structure in Auger electron lines as well as in autoionization electron lines. An example of this is the carbon Auger electron spectrum of CO, figure 6. The left part of

co CARBON AUGER

25000

OLL L I KINETIC ENERGY M t60 no eV KlNErlC ENERGY I I eV- 16.00 15.75 FIG. 6. - Part of the carbon Auger and autoionization BINDING ENERGY electron spectrum from CO excited by electron impact. The insert figure shows the vibrational structure in some of the autoionization lines.

the spectrum is the Auger part showing a closely spaced vibrational structure at the right side of a strongly excited single Auger line. Further out to the XENON higher energy side of the spectrum there are a few autoionization electron groups with clearly resolved vibrational components.

3. Core and valence electron spectra in molecules. - As mentioned previously one can study both the core and the valence electron structure when X-rays are used to excite the electron spectra. Figure 7 shows an electron spectrum from carbon tetrafluoride, excited by Mg Ka. Between 1 200 eV and 1 250 eV kinetic energy we find the valence molecular orbitals which in

KINETIC ENERGY this molecule derive mainly from the atomic 2 s and

1325 1100 1271 llS0 1225 2 p orbitals of carbon and fluorine. The two lines at - ENDING EtLERCI kinetic energy 952 eV and 558 eV are the core electron lines C 1 s and F 1 s. One interesting feature of the FIG. 5. - Electron spectra of argon and xenon showing the spin-orbit splitting of the 3 ps(2P) and 5 p5(2P) term, respectively. core electron lines is that the width of F 1 s is consi- The spectra were excited by helium resonance radiation. derably larger than that of the C I s line, and much C4-258 C. NORDLING

~140s AI I?

01s Nls A2P

I I FIG. 7. - ESCA spectrum from carbon tetrafluoride excited 710 715' 840 845 " 1000 1005 eV by Mg Ka radiation. The atomic-like core orbitals F 1 s and C 1 s, KINETIC ENERGY and the valence molecular orbitals are seen in this ESCA spec- FIG. 8. - ESCA spectrum from air excited by Mg Ka radia- trum as well as the fluorine K Auger electrons. tion. The oxygen l s line shows spin splitting (paramagnetic molecule). Argon is detected through its 2 p electrons and the larger than the K level width in this region of the spin-orbit splitting is well resolved. periodic system. The spectrum in figure 7 also shows the fluorine K Auger spectrum in the energy interval oxygen and nitrogen and also the 2 p spin-doublet in between 630 eV and 660 eV. argon which in this case had a partial pressure of The 32 valence electrons in CF, are distributed torr. The oxygen 1 s line is split into two components, among 16 molecular orbitals but due to the high 1.1 eV apart and with the intensity ratio 2 : 1. The two symmetry of the molecule several of these orbitals lines correspond to the quartet and doublet states are degenerate and only 7 different ionization energies respectively in which the oxygen molecule can be left are observable in the valence electron spectrum. Three upon emission of a 1 s electron. Due to the exchange of these states have previously been studied in UV interaction between the 1 s core electron (spin s = +) excited spectra and are attributed to mainly non- and the two unpaired outer electrons (resulting spin bonding molecular orbitals. For an understanding S = 1) the two states have different energies and give of the chemical bonding it is of interest to study also rise to two lines with intensities proportional to the the bonding orbitals. In figure 7 these deeper orbital weights of the states (4 :2). In the case of 0, one can states can now be seen. The deepest lying valence calculate this t( spin splitting )) by use of the vector orbitals are 1 a, and 1 t, which are mainly of F 2 s coupling method. The energy difference between character. There is a small bonding C 2 s contribution the (S + 3) and (S - 3) states is obtained as to the 1 a, orbital and a similar C 2p contribution to the triply degenerate 1 t, orbitals. The large width of the deepest valence orbitals can be explained by an where K,, is the exchange integral defined by increased inherent width due to radiationless transitions of Coster-Kronig type and, to a lesser extent, to vibrational structure of these strongly bonding orbitals. In this expression for the exchange integral s(i) and The 2 t, has mainly C 2 p-F 2 pa bonding character p(i) denote the 1 s and n, 2 p wave functions. The spin while 2 a, has C 2 s-F(2 s + 2 p,) a bonding character. splitting of core electron lines has also been studied in The non-bonding orbitals 1 e, I t,, and 3 t, are seen to transition metals with unfilled d-bands [3]. the right in the spectrum. The spin splitting disappears when oxygen is bound For the interpretation of the molecular orbital chemically to other atoms in a diamagnetic molecule. spectra we have employed semiempirical (CNDO) This is illustrated by the 0, and H20 spectra in and ab initio caIculations of the valence electron figure 9. However, chemical binding introduces a new configuration and the parentage of the molecular feature in the core electron spectra which is of great orbitals. A particularly useful feature of the X-ray importance, viz. a change in energy of the lines, mode of excitation in this context is the strong depen- characteristic of the chemical bonds with other atoms dence of the photoemission cross-section on the in the molecule. For the oxygen 1 s line this chemical symmetry of the orbitals. shift >) is 3.5 eV between (the quartet state of) molecular An interesting line splitting is observed in the core oxygen and oxygen in H,O, see figure 9. The chemical electron spectra from paramagnetic molecules. This shift effect and its application to molecular spectro- splitting is shown for molecular oxygen in figure 8. scopy will be discussed in the following section. The spectrum was obtained by letting air into the In conclusion the valence electron spectra contain collision chamber at a pressure of 0.1 torr and irra- information on the electrons that take part in the diating the gas with Mg Ka. It shows the 1 s lines of chemical binding and the chemical shift in the core ESCA STUDIES OF CORE AND VALENCE ELECTRONS IN GASES AND SOLIDS C4-259

figure 10, showing carbon in ethyl trifluoroacetate. All four carbon atoms in this molecule are distinguished in the spectrum. The lines appear in the same order from left to right as do the corresponding carbon atoms in the structure that has been drawn in the figure.

F 0 H H I I1 I I F-C -C-0-C-C-H I I I F H H

L I I J eV 545 540 BINDING ENERGY FIG. 9. - Oxygen 1 s electron lines from a mixture of oxygen gas and water vapour. The water line is shifted towards lower dinding energy in the water molecule and the spin splitting is remov-d.

electron spectra, vide infra, contain information on w chemical binding and molecular structure. Moreover, since in the electron spectrum each element of the J ' sample makes its characteristic contribution, it is possible to identify the different atomic species contained CARBON 1s in the sample and the intensities of the core electron I I I J lines are measures of the number of atoms of the res- 1190 1195 eV pective elements and valence states. Because of all KINETIC ENERGY - I I I these chemical implications the acronym ESCA (Elec- CV 295 290 285 tron Spectroscopy for Chemical Analysis) is used for BIND lNG ENERGY the spectroscopy which is based on X-ray induced FIG. 10. - Electron spectrum from carbon in ethyl trifluo- emission of electrons, roacetate. All four carbon atoms in this molecule are distinguished in the spectrum. The lines appear in the same order'from left to right as do the corresponding carbon aton~sin the structure 4. Chemical shifts. - When an atom is bound that has been drawn in the figure. chemically to other atoms there is a,change of wave function for the valence electrons. ConsequentIy there In the application of ESCA to chemical problems, is a change in the interaction between the valence it is desirable to be able to explain the chemical electrons and the core electrons and this induces a shifts by means of a simplified chemical language. slight change in the binding energies of the core (and By making use of the electronegativity concept in a valence) electrons. Therefore a change in chemical quantitative manner this has turned out to be possible. environment of an atom is relayed to its core electrons Correlations between chemical shifts and an atomic and can be observed as a line shift in the ESCA spec- charge parameter q, have been empirically established. trum. These chemical shifts of the inner electron The parameter q, is the sum of the partial ionic charac- energies can now be measured and are likely to assist ters of the bonds formed by the atom. The partial in the solution of many problems in chemistry. A ionic characters are obtained from the electronegativity particular feature of ESCA is that one obtains infor- differences between the atoms forming the bonds, by mation on chemical and molecular dynamics by use of the relationship by Pauling as described and measuring a quantity that remains essentially atomic discussed in ref. [I]. in character. Thus one can move the area of inspection In figure 11 the chemical shift AE of the carbon 1 s from one atomic species to the other in the molecular line has been plotted against q, for a selection of solid structure. carbon compounds with sp3 and sp2 types of hybri- A conspicuous chemical shift spectrum is given in dizations. Compounds with as simple substituents as C4-260 C. NORDLING

small carbon molecules [4]. Extensive basis sets of Gaussian type functions were used. For hydrogen six s-type functions contracted to two, and for the first row atoms eleven s-type functions contracted to five and seven p-type functions contracted to three were used. All points fall very close to the straight line which has a slope of 1.09. The closer calculations approach the Hartree-Fock limit the more does the orbital shift (- A&) approach the experimental shift (BE) (slope 1.00).

FIG. 11. - AE plotted against qpfor carbon of tetragonal and trigonal types of hybridization in molecules with small inductive effects. Open circles represent chlorine compounds for which qp has been cak ulated using an uncorrected value for the electronegativity of chlorine. possible were chosen in order to avoid secondary effects on the shifts of the binding atoms [4]. The electronegativities used are those given by Pauling except for chlorine and bromine. The points for compounds containing chlorine indicate that the electronegativity for this element, when bound to carbon, is not well described by the Pauling value. Inner electron binding energies of free molecules I I can be determined to within a few electron volts from 0 5 10 eV ab initio quantum mechanical calculations within the EXPERIMENTAL CHEMICAL SHIFT Hartree-Fock (SCF) approximation. Separate calculations are required for the neutral state and the ionized FIG. 12. - Comparison between carbon 1 s energy shifts state, and the electron binding energy is taken is the measured in the gaseous state of some small carbon molecules and the shifts obtained from ab initio MO-LCAO-SCF calcu- difference in total energy between the two states. lations. Inner electron shifts obtained this way so far are accurate to within a few tenths of an electron volt. A simplification of the theoretical calculation of The high accuracy depends on a cancellation of errors binding energy shifts can be made with the help of an due to relativistic and electron correlation effects electrostatic potential model [I], [2]. Through this which for inner electrons are unaffected by changes model the binding energy shifts are related to the elec- in the chemical environment. tron distribution of the neutral molecule. The model Electron binding energies can be obtained also from is purely classical although it can be described and Hartree-Fock calculations on the neutral systems used in terms of quantum mechanics. through Koopmans theorem. The additional assump- In the potential model the chemical shift is determi- tion over separate calculation on neutral and ionized ned by a change in potential for the core electron. This systems is that the remaining electron of the ion can potential can be considered as a superposition of two be described by the same orbital wavefunctions as in potentials. The first, which generally is the dominating, the initial state. The binding energies obtained in this originates from the change in electronic distribution way for inner levels of light element are systematically around the particular nucleus being studied within the 10-20 eV larger than those found experimentally. molecule. The second potential, which we may call Moreover, they are very sensitive to the size and the molecular potential, is set up by the charge dis- optimization of the basis set used. For this reasofi tribution from the rest of the molecule. The molecular comparisons of calculations with different basis sets potential is easily estimated having condensed the become meaningful only if the calculations are close charges to point charges. The first potential obviously to the Hartree-Fock limit. Systematic series of inner is not well described by a point charge at the position electron binding energy shifts from Koopmans theorem of the nucleus. However, from comparisons with free are now available and have been compared with atoms and ions it can be expected to be approximately experiments in our recent monograph 121. proportional to the charge of the atom. The expression Figure 12 shows the correlation between experimen- for the core level shift then becomes tal chemical shifts and orbital energy shifts obtained from ab initio MO LCAO SCF calculations on some ESCA STUDIES OF CORE AND VALENCE ELECTRONS IN GASES AND SOLIDS C4-261 where mental Cls shifts with shifts calculated in this way from the ab initio wavefunctions and CND0/2 wavefunctions, respectively. The charges from the ab initio wavefunction are Mulliken gross atomic charges. The The first term in (4) represents the potential from the constants k and 1 were in both cases determined from charge at the atom considered while the second term, a least squares fit of AE - V to kgi + 1. Thus values the molecular potential, accounts for the potential for k, 18.3 eV from the ab initio wavefunctions and from the rest of the molecule. The third term is a 23.5 eV from the CND012 calculations agree reaso- constant related to the choice of reference level. The nably well with calculated 1 s-2 p electrostatic repul- constant k is approximately equal to the electrostatic sion integrals. With an atomic Hartree-Fock wave- interaction integral between the considered core orbital function the value 21.2 eV is obtained while simple and a valence atomic orbital in the same atom. This Slater orbitals give 22.0 eV. The change of the Cls integral is close to the expectation value < llr > for binding energy upon the removal of a 2 p electron, a valence electron [I], [2]. obtained from independentcal culations on atom free Figures 13 and 14 show the correlation of experi- the four and ion states involved is 18.8 eV.

5. Some applications. - The oxidation of cystine provides a simple example of how the chemical shifts in ESCA spectra can be utilized to solve chemical structure problems [5]. The cystine molecule contains two equivalent sulfur atoms : HOOC-CH-CH2-S-S-CH2-CH-COOH I I

CYSTINE S-DIOXIDE SZp(AIKa) "0° t

CALCULATED SHIFT (18.3q +V+3.0) FIG. 13. - Comparison between measured shifts and shifts calculated with the potential model using charges obtained from ab initio calculations.

CYSTINE S2p(AIKu)

1310 1315 1320 1325 eV KINETIC ENERGY I I I 1 I 10 ev eV 170 165 160 CALCULATED SHIFT (23.5q+V+0.22) BINDING ENERGY FIG.14. -. Comparison between measured shifts and shifts FIG. 15. - Electron spectrum from the 2p shell of sulfur in calculated with the potential model using charges obtained from cystine S-dioride and cystine. The valence states of the sulfur CNDOJ2 calculations. atoms can be determined from the spectra. C4-262 C. NORDLING

If the cystine dioxide is synthetized two different investigation confirmed that coordination of Iigands structure may be formulated. If one oxygen is attached to the metal results in a significant charge transfer to each of the two sulfur atoms, the resulting compound from metal to ligand, see figure 17. with equivalent sulfur atoms would give rise to a single line in the electron spectrum. If both oxygens are attached to one of the sulfur atoms, i. e. if the disulfide dioxide has a thiolsulfonate structure, the two sulfur atoms, having non-equivalent structural positions, would give rise to two lines at different energies in the electron spectrum. According to figure 15 this is actually the case. Instead of one single line as in the symmetrical cystine, two lines are obtained from the 2p subshell in sulfur, one unshifted and the other shifted by 4.0 eV. The electron spectrum of cystine S-dioxide therefore gives conclusive evidence for the thiolsulfonate structure : 0 t HOOC-CH-CH2-S-S-CH2-CH-COOH I 1 I NH2 0 NH2 A number of other structure problems more compli- cated than the above quoted example have already been solved by means'of the ESCA technique. It is likely that with the improved resolution now under BINDING ENERGY (eV) development still more detailed informations can be FIG. 17. - Platinum 4 f7/2 energies in a series of metal-organic obtained on structure problems for practical use. complexes. The question of electron transfer between the metal and the carbon atoms in transition metal carbides has been a matter of discussion over the years. Experimen- 0 K mission EszSZa Eels 02s tal data have been lacking and there is serious disa- 'ma greement between the different theories that have been proposed. ESCA measurements on the core level shifts in various transition metal carbides and related compounds have zhown that electrons are transferred from the metal to the carbon 161, see figure 16.

BINDING ENERGY FIG. 18. - Electron spectrum from Be0 excited with Mg EKE radiation. The energy distributions obtained from K-emission MTi metal spectra are shown at the top of the figure.

BINDING ENERGY FIG. 16. - Chemical shifts for titanium and carbon in Tic. The shift of the Ti 2 P3/2 level indicates that titanium is more positive in the carbide than in the metal ; the carbon 1 s shift indicates that carbon is more negative than in the hydrocarbon reference.

In the study of catalytic reactions much interest is focussed on the binding of the metal in metal-organic complexes. We have recently investigated by ESCA a number of complexes of platinum, in which the metal is in a formally low oxidation state [7]. Relative oxidation states of platinum in the complexes, as determined from the binding energy data, were ordered and the FIG. 19. - Valence band spectra from transition metals. ESCA STUDIES OF CORE AND VALENCE ELECTRONS IN GASES AND SOLIDS C4-263

We have also recently applied the ESCA technique in transition metals and other solids have been inves- to some crystal and solid state phenomena. For exam- tigated [I], [ll]. Figure 19 shows the valence band ple, the angular variation in intensity of elastically spectra obtained from the transition metals. This scattered electrons expelled by Mg Kcr from various work will be discussed in more detail in another contri- shells in a sodium chloride crystal have been studied [8]. bution to this conference [12]. The escape depth of electrons photoemitted from a metal by X-rays has been measured [9]. Core lines Acknowledgement. - It is a pleasure for me to and valence bands of LiF, BeO, BN and graphite acknowledge the cooperation of my colleagues at have been studied and compared with X-ray spectros- Uppsala in the research work described in this copic data [lo], see figure 18, and the valence bands review.

References

[I] SIEGBAHN(K.), NORDLING(C.), FAHLMAN(A.), LING (C.) and LINDBERG(B. J.), Spectrochim. NORDBERG(R.), HAMRIN(K.), HEDMAN(J.), Acta, 1967, 23, 2015. JOHANSSON(G.), BERGMARK(T.), KARLSSON(S.-E.), [6] RAMQVIST(L.), HAMRIN(K.), JOHANSSON(G.), FAHL- LINDGREN(I.), LINDBERG(B.), ESCA, Atomic, MAN (A.) and NORDLING(C.), J. Phys. Chem. Molecular and State Structure studied by ESCA Solids, 1969, 30, 1835. Nova Acta Regiae Soc. Sci. Upsaliensis, Ser. IV, [7] COOK(C. D.), WAN(K. Y.), GELIUS(U.), HAMRIN(K.), Vol. 20, 1967. JOHANSSON(G.), OLSON(E.), SIEGBAHN(H.), [2] SIEGBAHN(K.), NORDLING(C.), JOHANSSON(G.), NORDLING(C.) and SIEGBAHN(K.). J. Am. Cchem. HEDMAN(J.), HEDBN (P. F.), HAMRIN(K.), SOC.1971,93,1904. GELIUS(U.), BERGMARK(T.), WERME(L. O), [S] SIEGBAHN(K.), GELIUS(U.), SIEGBAHN(H.) and MANNE(R.), BAER(Y.), ESCA applied to free OLSON(E.). Physica Scripta, 1970, 1, 272. molecules. North-Holland Publ. Co., Amsterdam- [9] BAER(Y.), HEDBN(P. F.), HEDMAN(J.), KLASSON(M.) London, 1969. and NORDLING(C.). in Solid State Comm., 1970, [3] FADLEY(C. S.) and SHIRLEY(D. A.), FREEMAN(A. J.), 8, 1479. BAGUS(P. S.) and MALLOW(J. V.), Phys. Rev. [lo] HAMRIN(K.), JOHANSSON(G.), GELIUS(U.), NORDLING Letters, 1969, 23, 1397. (C.) and SIEGBAHN(K.). Physica Scripta., 1970, [4] GELIUS(U.), HED~N(P. F.), HEDMAN(J.), LINDBERG 1, 277. (B. J.), MANNE(R.), NORDBERG(R.), NORDLING [I11 BAER(Y.), HED~N(P. F.), HEDMAN(J.), KLASSON(M.), (C.) and SIEGBAHN(K.). Physica Scripta., 1970, NORDLING(C.) and SIEGBAHN(K.), Physica 2, 70. Scripta, 1970, 1, 55. [5] AXELSON(G.), HAMRIN(K.), FAHLMAN(A.), NORD- [12] BAER(Y.), This conference.

DISCUSSION

Mr. DAS GUPTA.- Why Calbon behaves as an Reply to question from Nagel : We are presently acceptor rather than a donor ? On alloying carbon measuring valence bands and core electron shifts in with iron group alloys carbon usually behaves as a some palladium alloys. To my knowledge there have donor rather than an acceptor. not yet been any measurements reported on core Reply to questionfrom Das Gupta : I have not seen the electron shifts in other metal alloys. evidence you refer to that carbon usually behaves as a C. Nordling. donor in iron group carbides. The electronegativity of Fe, Co and-Ni-(X = 1.8) is lower than that if Mr. WIECH.- In one of your slides you showed carbon (X = 2.5) and would rather suggest that carbon is an acceptor. For the transition metal carbides that the spectrum of the valence electrons of 12 metals. we have studied (group IV b and V b) there had been In some of the curves the intensity at the high and the great controversy as to the direction of the charge low energy side is of the nearly same magnitude, in transfer. Our ESCA results confirmed the prediction others the intensity on the low energy side is high that can be made from the electronegativities that than on the high energy side. Is this high intensity due there is a transfer of charge from metal to carbon. to inelastic scatte~edelectrons or does the valence band C. Nordling. extend to low energies ? Is it possible to determine the botton of the balence band by your method ? DAVIDJ. NAGEL.- Models exist for non-transition metal alloys in which an electronegativity parametel is Reply to questionfrom Wiech : Inelastically scattered used to calculate electron redistribution upon alloying. electrons contribute to the intensity on the low energy The conduction electron concentration around, for side of both core level and valence band spectra. The example, aluminum a~omsin alloys should be reflected low energy tails of core level peaks which are close in in core electron level shifts because the conduction kinetic energy to the valence band spectra can be electron density partially determines the potential used as a measure of this contribution. We have not experienced by core electrons. yet applied this and other corrections that would be Have core electron shifts been measured in alloys necessary to determine the bottom of the band. in order to determine conduction electron distribution ? C. Nordling.