electron spectroscopy and molecular structureelectron spectroscopy has been reviewed several times...
TRANSCRIPT
Pure& AppL Chem., Vol. 48, pp. 77—97. Pergamon Press, 1976. Printed in Great Britain.
ELECTRON SPECTROSCOPY AND MOLECULARSTRUCTURE
KAI SIEGBAHN
Institute of Physics, Uppsala University, Box 530, S-75121, Uppsala, Sweden
Abstract—Electron spectroscopy can now be applied to solids, liquids and gases. Some fields of research requireultrahigh vacuum conditions, in particular those directly concerned with surface phenomena on the monolayer level.Liquids have recently been subject to studies and several improvements and extensions of this technique can bedone. Much advance has lately been achieved in the case of gases, where the pressure range presently is 10' — 1 torr.Signal-to-background ratios for core lines can be — 1000: 1 and the resolution has been increased to the extent thatvibrational fine structures of is levels in some small molecules have been observed. These improvements are basedon the monochromatization of the exciting AlK radiation. Under such conditions the background is somuch reduced that shake-up structures are more generally accessible for closer studies. ESCA shifts are also mucheasier to resolve and to measure with higher precision around 0.02 eV. It is interesting to compare the results fromelectron spectroscopy with those recently obtained from X-ray emission spectra of free molecules. By means ofgrating at grazing incidence a resolution has been obtained at a sufficient intensity to record not only transitions fromresolved molecular orbitals to the is levels but also in some cases vibrational fine-structures inside such separatelines.
The photoionization dynamics including atomic and molecular relaxations have been investigated, both experimen-tally and theoretically. In the valence electron region improvements in energy resolution and in the application of theintensity model based on the MO-LCAO approximation greatly facilitate the assignments of the valence orbitals.Accumulation of empirical evidences gathered from series of similar chemical species and also better methods ofcalculation, both ab initio and semi-empirical, have gradually resulted in a much better understanding of themolecular orbital description.
The experience of the latest ESCA instrument with monochromatization has motivated an attempt to design anoptimized apparatus according to the general principles of this prototype. A considerable gain in intensity can bemade at an improved resolution set by the inherent diffraction pattern of the focussing spherical quartz crystals.
INTRODUCTION
Electron spectroscopy has been reviewed several timesduring the last years and more recently by the author inthe conference volume' of the Namur Conference onElectron Spectroscopy (1974). The reader is referred tothis book for a more complete account of the presentsituation in the field and to a comprehensive list ofreferences. In the following I will briefly summarize firstsome features of this spectroscopy and then illustratewith some recent applications, some of which have notyet been published.
Electron spectroscopy was originally started in mylaboratory for precise measurements of electron bindingenergies in atomic and solid state physics. After some tenyears development work it gradually became clearthat this spectroscopy should have interesting applica-tions in chemistry. With this conclusion in mind theacronym ESCA, Electron Spectroscopy for ChemicalAnalysis, was coined around 1965. The word Analysis wasthen taken in a more modern and broader sense than theclassical one. For example, one very important applica-tion of ESCA is surface analysis. One can analyze bothsolids, gases and, more recently, liquids. All elements,including the important light elements carbon, nitrogenand oxygen, can be analysed on the surface with highaccuracy. A particular feature of ESCA is that inaddition to the elemental composition of a compoundfrequently other useful information can be obtained aboutthe chemical bonds and charge distribution. This informa-tion can be deduced from the so-called ESCA shifts of thecore level electron lines. When atoms are brought closetogether to form a molecule the electron orbitals of eachatom are perturbed. Inner orbitals, that is, those with
higher binding energies, may still be regarded as atomicand belonging to specified atoms within the molecule,whereas the outer orbitals combine to form the valencelevel system of the molecule. These orbitals play a moreor less active part in the chemical bonds which are formedbetween the atoms in the molecule and which define thechemical properties. The chemical bonds affect the chargedistribution so that the original neutral atoms can beregarded as charged to various degrees while the neutralmolecule retains a net charge of zero. One may describethe situation by regarding the individual atoms in themolecule as spheres with different charges set up by thetransfer of certain small charges from one atomic sphereto the neighbouring atoms taking part in the chemical bond.Inside each charged sphere the atomic potential is con-stant in accordance with classical electrostatic theory.The result of this atomic potential is to shift the wholesystem of inner levels in a specified atom by a smallamount, the same for each level. Levels belonging todifferent atoms in the molecule are generally shifteddifferently, however, and by measuring these ESCA"chemical shifts" for individual atoms in the molecule amapping can be made of the distribution of charge orpotential in the molecule. This is then a reflection of thechemical bondings between the atoms which in turn canbe described by the orbitals in the valence level system. Itshould be remarked that these chemical shifts are differ-ent from those obtained in NMR and may even go inthe reverse direction.
The chemical shift effect has its theoretical importancein the study of molecular electronic structure. In othercontexts its usefulness lies precisely in its ability tospecify the chemical composition, in particular at sur-
77
78 KAI SIEGBAHN
faces. For example, a metal and its oxide give twodistinctly different lines because of the chemical shifteffect. It is therefore often easy to follow the rate ofoxidation at a surface or the removal of an adsorbed gaslayer from it. The chemical composition of a surface or thechanges due to various chemical or physical treatments areamenable to detailed examination.
MODES OF EXCITATION
Electron spectra can be excited in three different ways(Fig. 1): by means of X-rays (for example A1Ka at1487 eV, or YM at 132 eV) UV-light (usually the Heresonance radiation at 21.22 eV or 40.8 eV) or an electronbeam. The various modes of excitation generally providecomplementary information. From X-rays and UV oneobtains well defined photoelectron lines, whilst electronsproduce Auger lines (and autoionization electron lines).Auger electron lines are also always obtained by means ofX-rays. Figure 2 illustrates how photons ofvarious wavelengths can excite electron spectra fromdifferent levels of an atom within a molecule. The bindingenergy region between 0—50 eV is usually classified as the
(54kev) valence electron region. These orbitals are more or lessdelocalized and can be excited by both X-rays and UV.
(21 eV) The core region inside the valence region is atomic incharacter and is typical for a given element and thereforeconvenient for analytical purposes. This region can bereached by X-rays.
More recently, synchrotron radiation has been used toexcite electron spectra (see Fig. 3). If a variable mono-chromator is inserted between the synchrotron and theelectron spectrometer one can excite the different electronspectral energies in a continuous way. Since thisradiation is linearly polarized in the direction of the planeof the accelerator one can also study the spectra as afunction of the angle between the emitted electrons and
ELECTRON SPECTROMETER
Fig. 1. Excitation of electron spectra by three different modes.
Fig. 2. Excitation of core and valence electron lines at variousphoton energies.
ESCASPECTROMETER
'SYNCHROTRON RADIATION BEAM
FOCUSSING MIRROR
VER11CALLY FOCUSSING MIRROR
MONOCHROMATOR ENTRANCE-SLIT
CONCAVE GRATING
- POLARIZATION DIRECTION
MONOCHROMATOR EXIT-SLIT
ELECTRON MULTI-DETECTOR
Fig. 3. Electron spectroscopy by means of synchrotron radiation as the source of excitation.
Electron spectroscopy and molecular structure 79
the electric vector of the radiation by turning the electronspectrometer in the photon beam.
Usually, it is sufficient to use only one energy ofX-radiation. It is, however, highly desirable to mono-chromatize this radiation to an etent which is betterthan the inherent width of the main X-ray line itself.This can be done in different waysbut common to them allis an X-ray dispersive element such as a bent quartzcrystal. Figure 4 shows the principle of such a system. Anelectron gun yields a well focussed electron beam of highintensity impinging on the periphery of a swiftly rotating,water cooled anode F. Along the Rowland circle one orseveral spherically bent quartz crystals (Q) accept theAlK radiation at a small angle of emittance (--5°). Onlythe central part of the Ka line satisfies the diffractionconditions and this narrow wavelength region is reflectedtowards the slit of the sample chamber G. This chambermay contain a gas, liquid or a solid. The expelledelectrons are partly transmitted through a narrow slitfacing the electron spectrometer, and after passing a lenssystem, which •can retard the electrons to a convenientenergy, they are analyzed in a spherical electrode energyanalyzer. Alternatively, a He lamp or a small electron guncan be used for excitation in the same arrangement. Theelectrons are focussed by the analyzer in its focal plane.Along this plane an extended electron detector is situated.This is a position sensitive device consisting of twoelectron channel plates close to each other, giving a totalmultiplication of 108. These multiplied electrons arefurther accelerated onto a phosphorus screen continu-ously observed and scanned by a TV camera. In this waythe whole system works at maximum efficiency. All data arerecorded and stored into a computer. The apparatus isshown in Fig. 5. An extensive pump system is required toprovide the differential pumping which is necessary forhandling gases and liquid samples.
EXAMPLES OF ELECTRON SPECTRA
Some examples may illustrate the kind of spectra oneobtains in ESCA. We start with the gas phase of a dropletof mercury which is introduced into the sample chamberby means of a vacuum lock device in the instrumentshown in Fig. 5.Already at room temperature the vapour
pressure (10 torr or 10' Pa) is enough to produce thespectrum given in Fig. 6. This spectrum contains all thelevels which canbe excited by A1Ka, namely N1, N11, N111,N1, Nv, N1, N11 and 08, Oii, Oiv, Ov and P1. Noselection rules discriminate against the appearance ofthe complete level systems One can also observe somesatellite structure behind the Nv1N11 lines. Figure 7 is acloser study of this structure in the vapour phase(upper figure) and the solid phase (lOwer figure). Thesespectra disclose some other typical features of ESCA,namely so-called shake-up lines (in the gas) and plasmonloss peaks (for the Hg metal). In addition to the formerlines appearing in the gas phase, one can also observesome extra peaks due to external, discrete energy lossesof the electrons originating from the N1N0 peaks, whichoccur when these electrons are traversing the gas beforereaching the slit. All these electron lines give informationabout excited states in the residual ion or in the neutralmolecule. Plasmon peaks are characteristic of mercurymetal electron plasma.
It is interesting to look at the very last part of the Hgspectrum close to the binding energy zero (see Fig. 8) bothfor the gas (upper figure) and for the metal (lower figure).One clearly observes how the broadening of levels occursat condensation when the free atoms combine to form asolid, the uttermost level P1 (or 6s112) thereby resulting inthe broad conduction band with the Fermi edge. Conduc-tion band spectra of all metals and their alloys can bestudied in this way by ESCA.
Figures 9a and 9b show the conduction bands of goldand platinum. Osmium (Fig. 9d)2 displays a number offeatures which can be compared to theoretical bandcalculations. One observes for the case of gold the s-bandextending to the Fermi edge and furthermore the splittedd-band structure. The platinum has a very high density ofstates at the Fermi edge.
As mentioned, a special feature of ESCA is that bymeasuring the exact position of the electron lines charac-teristic of the various elements in the molecule, one canmove the area of inspection from one atomic species tothe other in the molecular structure. For example, one canobserve one single iron atom and two disulfide bridges ina large biomolecule like cytochrome-c (mol. wt = 12,700),
0
Fig. 4. The prototype ESCA instrument (1972) for gases and solids with monochromatized A1K,. radiation(Ahv 0.2 eV).
8
C')
Erm
m
ECV bQCLfIUJ O H AbOflL OUOCJLOW11C VIIC !1!ou
BII.1DU40 EIIEbOA
2 bP00LbP ot bLooObG E2CV COLqIIJ O L! F EOL I1LWGL qCACobWCu E'
80 KVJ 2IEGBVH4
Si
Electron spectroscopy and molecular structure 81
110 iSo iSo 110 100
BINDING ENERGY (eV)
Fig. 7. ESCA satellite structures following the N1N11 mainelectron lines of Hg (see Fig. 6). Upper figure shows shakeup linesfrom gas phase, lower figure shows plasmon lines from solid
phase.
one Mg atom in chlorophyll, one Co atom in vitamin B12and the three disulfide bridges in insulin etc. ESCA shiftsare related to the chemical bonding.
Figure 10 shows the electron spectrum from the is levelof the carbon in ethyl trifluoroacetate. All four carbonatoms in this molecule are distinguished in the spectruii.The lines appear in the same order from left to right as dothe corresponding carbon atoms in the structure shown inthe figure. If the structure of the molecule is known, thecharge distribution can be estimated in a simple way byusing, for example, the electronegativity concept andassuming certain resonance structures. More sophisti-cated quantum-chemical treatments are preferably ap-plied. Conversely, if by means of ESCA the approximatecharge distribution is known, conclusions concerning theelectronic structure of the molecule can be drawn.
Experimental evidence obtained so far for variouselements in a large number of molecules indicates strongcorrelations between chemical shifts and calculatedatomic charges. Such correlation curves can be obtainedfor many elements including sulfur, carbon, nitrogen,oxygen, phosphorus. Chemical shifts are also observed inthe electron lines due to the Auger effect. Second-orderchemical shifts from groups situated further away in themolecule (inductive effects) are also observed.
A further example of typical ESCA shifts (in acetone isgiven in Fig. 11. For comparison the spectrum observedwhen the A1K0 radiation was not monochromatied isinserted. One observes a substantial improvement ofresolution and also decrease in signal noise. The strongest
Fig. 8. The effect of condensation on electron lines close to zerobinding energies. Upper spectrum from gas phase of Hg, lower
spectrum the corresponding band structure for solid Hg.
V 8 6 4 2 0
PLATINUMZ=78.Fcc
(b) FiE.C B
D....--.......f.f.
.
H.
//FT..—
Fig. 9(a, b). Conduction electron band spectrum of gold (a) andplatinum(b).
110
Hg 14 shake - upand inelastic scatt.
VAPOUR
shake-up ret. scott.._______*____________ —.---
•
A
SOLID
INELASTIC SCATTERINGINTERBAND TRANSITIONS
Pt00mons f Hio
BINDING ENERGY
GOLD
eV 10 8 4 2
Fig. 10. The ESCA shifts of Cis in(monochromatized AlK).
Cis line belongs to the two equal carbons in the CH3 groups,whereas the middle Cls line is the carbon in CO.Theleft Clsline is due to CO2 which is mixed in as a calibration standardgas. This latter procedure has considerably improved theattainable accuracy (around 0.02 eV) in actually measuringthe ESCA shifts and enables one also to get a consistentreference level for various gaseous compounds.
In a recent, not yet published study4 of a number ofbenzene derivatives the above procedure has been used toinvestigate the ESCA shifts at various sites in the ring fordifferent substituents. It turns out that one can obtain inthis way some close correlations between ESCA shiftsand chemical reactivity. First the unsubstituted benzenecarbon is measured against CO2 which is used as calibra-tion for all the benzene derivatives. Figure 12 shows this
82 KAI SIEGBAHN
LU
If)
0>-
Id)zLU0az
a:
>-I-U-)zLU
zz
-JLU
Fig. 9(c, d). Conduction electron band spectrum2 of Os, theoretical(c) and experimental (d). The calculation of the density of states isbased on a recent approach3 called LMTO, 'linear combination ofmuffin tin orbitals'. The result is corrected for finite energyresolution but not for photocross-section dependence of bandsymmetry. The correspondence between the structures in theexperimental spectrum (A, B,,C . . .) and the calculated (1,2,3...)
is remarkably good.
6 4 2CHEM. SHIFT
ethyl trifluoroacetate
5BINDING ENERGY
U,zwI-.z
BINDING ENERGY, eV
Fig. 11; ESCA shifts in acetone with carbon dioxide as calibration gas. Inserted in upper left part is a previousspectrum obtained without monochromatization of A1K.
Fig. 13. ESCA shifts in fiuorobenzene—referred to tet-rafluoromethane(Fls) and carbon dioxide (Cis).
comparison. Figures 13 and 14 are given as representativespectra for two benzene compounds in the series ofmeasured spectra. One can observe also that the ringcarbons which are not directly situated at the substituent,also, exhibit ESCA shifts to various degrees. Some of theresults are summarized in Figs. 15—17. In Figs. 15 and 16 theESCA shifts, referred to Cis in C02, are measured at thestarred carbon atom, whereas in Fig. 17 the ring carbonshifts, also referred to CO2 are given.
It is interesting to correlate these shifts with chemicalparameters such as Hammett's substituent constants (if).Acomplete discussion is beyond the scope of this surveyand the results are only indicated by the relation betweenmeasured ESCA shifts and the corresponding Hammettconstants for the series of substituents as given in Fig. 18.
Molecules like 02 or NO contain unpaired electrons andare therefore paramagnetic. Large classes of solid materi-als have similar properties. In such cases ESCA spectrashow typical features called spin-, multiplet- or exchangesplitting. The phenomenon was observed in oxygen whenair was introduced into the source chamber of the ESCA
Cls eV 293 292 291 290
Fig. 15. CIs ESCA shifts in some benzene derivatives. This figureillustrates the range, '—3.5 eV, as well as the additivity of the
substituent effects.
Fig. 14. ESCA shifts in parafluoroaminobenzene referred to tetrafluoromethane (Fis), nitrogen (Nis) and carbondioxide (C1s).
Electron spectroscopy and molecular structure 83
Cis
C6H6
Co2
eV
—
295 290BINDING ENERGY
Fig. 12. Cis ESCA shifts between benzene and carbon dioxide.
GNH2 j-1o I
OF-NO2 1
N4D 1*4
HN4-F 42.F.ONH2
:_.-.-!--
FtQ 1F NO2 I4—_i-1
I
BINDING ENERGY
Fis Nis Cis
CF F-O-NH2
F—O--NH2
Co2
F—O-NH2
BINDING ENERGY
84 KAI SIEGBAHN
Fig. 18. Cis ESCA shifts in some para substituted benzene seriesversus Hammett substituent constants a illustrating the correla-
tion between ESCA shifts and reactivity parameters.
AIR
Ols Nis A2p- 1200
84510iio$.INETIC ENERGY, eV
Cis eV 29Z 291 290
Fig. 17. Ring carbon shifts of the compounds shown in Figs. 15and 16. The measured carbon atoms carry no substituents. Theshifts are due to the 'relaying' of the primary substituent effect
through the aromatic ring and encompass a range of —1.5 eV.
instrument some years ago. The upper part of Fig. 19shows the spectrum which was obtained when thisphenomenon was first found and the lower part is thesame spectrum as recorded with the new instrument withX-ray monochromatization. The is line of 02 is split in theintensity ratio of 2: 1. This spin splitting is due to theexchange interaction between the remaining is electronand the two unpaired electrons in the 1Tg-2 orbital,responsible for the paramagnetism of this gas. The result-ing spin can be either 1/2 or 3/2. The correspondingelectrostatic exchange energies can be calculated andcorrespond well with the measured splitting of 1.11 eV.Apartfrom oxygen and nitrogen, argon and CO2 canbe seenin spite of the low abundance of the latter gases in air. Astatistical treatment of the data even exhibits the presenceof neon (0.001%).
Recent improvements in resolution now enables one tomeasure the inherent widths of core levels in molecules.
Fig. 19. Electron spectrum of air. The Ols is split into twocomponents due to 'spin' or 'multiplet' splitting. Upper spectrumis the original one when this phenomenon was first observed.Excitation was performed by means of non-monochromatizedMgKa radiation. Lower spectrum is taken more recently withmonochromatized AlK. (Ahv = 0.2 eV) using the prototype
instrument (Fig. 5).
FNH2 iiFO0H -FO I
FOFFOCN
F€NO? I.-I- I I
Cls eV 294 293 292
Fig. 16. Cis ESCA shifts illustrating the substituent effect in aseries of para substituted fluorobenzenes. The shifts are due to thevariation of one substituent in the same position and encompass a
range of —1.2 eV.
/
ClseV
293
292
291
290
H2N4-X
-1.0 ! 0 +1.00'
0-NH? 1211-4
23FØNHo II
3 24FQ Ii
213324
0-NO? I I32
F-GCN -I32
F-Kij-N02 I I
0
zz
BINDING ENERGY, eV
Electron spectroscopy and molecular structure 85
Fig. 20. Cis electron line of CH4 showing vibrational structure.
As an example, Fig. 20 shows the line profile of the Cis inCH4. It turns out that this line can be separated into threecomponents caused by the symmetric vibration whenphotoionization occurs in the is level of the central carbonatom. Figure 21 demonstrates what happens. When theelectron leaves the carbon atom the latter shrinks about0.05 A. The minimum of the new potential curve for theion will consequently be displaced by the correspondingamount and Franck—Condon transitions which take placewill then give rise to the observed vibrational fine struc-ture of the electron line and with the intensities given bythe Franck—Condon factors quoted in the figure. Thisfinding canbe correlatedwitha series ofrecentobservations(Ref. 1, p. 70) in our laboratory, that there existvibrational fine structures in X-ray emission lines. Com-bined, these results show that vibrations occur in these
?vV c1fv5,W dtl
3--.:\,"---2
o
v;
o
-O.O5A
Fig. 21. Potential curves for CH4 and CH4 explaining vibrationalstructure at photoionization of the carbon core level.
molecules during X-ray emission both in the initial and thefinal states.
There are also other causes of line broadenings inmolecules and solids which can now be studied quantita-tively by means of electron spectroscopy. Two spectramay illustrate more generally what one observes concern-ing linewidths in the case of solids, namely NaC1 (Fig. 22)and Cu (Fig. 23). In these spectra all levels with theirinherent widths are recorded, from binding energies1400 eV to the valence and conduction band atzero binding energy. In the case of NaC1, being an ioniccrystal, the ionicity can be measured simply by taking thedistance between the loosest bound levels in the valence
i —---•—To---- O
BINDING ENERGY, eV
Fig. 22. Electron spectrum of NaCI showing the core and valence electron lines with their inherent widths. There is apartial splitting of the Cl 3p level due to a lattice effect.
FRANCK CONDON TRANSITIONS
eV 291.5 291.0 290.5
BINDING ENERGY
NciCL
Nals CL2s Ct2p Na2s
xl25
.58eV
1L 2.I9/1.
1075 070 270
Na2p25.87 eV
:113eV
60
C13p
200
CI 3s
.03 eV
35 30
_L3jeV iii
86 KAI SIEGBAHN
400 V 530 540 550KINETIC ENERGY,
.-.""(570) (560) (550)
910
region of the electron of the cation, Na2p, and anion,C13p.
It was mentioned before that conduction bands can bestudied for metals and their alloys. Figure 24 shows arecording of pure copper (upper spectrum), pure pal-ladium (middle spectrum) and the alloy CuPd (60% Cuand 40% Pd, f.c.c.). There are many interesting features inthese spectra' First, one observes in the case of Cu asymmetric core line (2p3/2) and a conduction band, havingalowdensityofstatesattheFermiedge. ForPdthe contraryis seen, namely quite asymmetric core lines (3d512,312) and ahigh density of states at the Fermi edge of the conductionband. When these metals alloy there is a remarkabledecrease in both linewidth and also asymmetry of the Pdcore lines and the slope at the Fermi edge is less steep whichmeans that the density of states is decreased compared topure Pd. There are also changes in binding energies of thecore levels. The asymmetry of core lines for metals isconnected with the density of states as observed in theconduction bands in the ESCA spectra. Asymmetries ofcore lines for variou meials having high density of states attne Fermi edges are shown in Fig. 25. Itis likely thatthe coreelectrons loose more easily a small part of their energies tocreate electron-positive hole pairs when there is a highdensity of states in the conduction band at the Fermi edge.These electrons are responsible for the electronic specificheat. If the asymmetryof corelinesisplottedas afunctionofthis quantityforthe metals studiedintheprevious figure oneobtains therelation showninFig. 26. These problems shouldbe further studied and any connection between theseproperties and catalytic effects (see below under SurfaceSensitivity) exhibited by some of these metals more closely
560 " 830 840eV
Cu3p
investigated. The effect of alloying on these results alsoneeds investigations.
YM AND UV EXCITED ELECTRON SPECTRA
So far we have been dealing with (monochromatized)X-ray (A1Ka) excitation of electron spectra. This isprobably the excitation mode leading to the widest diver-sity in applicability, since core levels and their shifts,apart from the valtnce levels, are accessible this way.There are also some theoretical reasons why the electronspectra are easier to interpret when the excitation energyis high enough to expel the electrons far away from emptymolecular states. The synchrotron radiation provides agreat flexibility in this respect but due to many obviousreasons it has limitations in its practical applicability.
Before giving some examples of UV excited spectra itis appropriate to discuss a complement to the A1Ka modeof X-ray excitation. Some years ago it was suggested thatultrasoft X-rays from ZrM or YMC, energies of 151.4 eVançl 132.3 eV, respectively, be used for this purpose.5There are some technical difficulties due to the extremelyhigh absorption rate of these soft X-radiations both in anysmall surface contamination of the anode itself and in thewindow to be used. However, good progress has recentlybeen made 1y several groups and in Fig. 27 one ex-ample of a spectrum is shown, as obtained in laborat-ory.8 In this experiment a water cooled, rotating yttriumanode was used with a positive potential withrespect to the source chamber window, a very thincarbon foil (20 pg/cm2). Electrons, scattered from theanode, are retarded to zero energy before reaching this
Cu2p3,
Cu2s
Cu2p-ç x 0 x I
\\41100 1090 1O8 950 940 930
BINDING ENERGY, eV
LM M2, 2,3 4,5.,
x 6
80 390
L3M5M45
x6
/;L2M45M45
Cu3s xIOCu3d4s
1.
, ixIO,:
x20j
BINDING ENERGY, V° 10 "0
920 930 - 940 "i' 1360 1370 " 110 1480KINETIC ENERGY
Fig. 23. Electron spectrum of Cu showing the core lines with their inherent widths and also the conduction band. Seealso the more detailed shape of the band given in Fig. 24.
Electron spectroscopy and molecular structure 87
Fig. 24. Conduction bands of Cu and Pd and the alloy CuPd(60% Cu and 40% Pd, f.c.c.). The asymmetric line shapesof the Pd core lines are strongly influenced by the alloying with Cu x 30 and are related to the changed slope at the
Fermi edge of the alloy.
foil. To keep the yttrium anode constantly clean, there isan oven situated below the rotating anode, which continu-ously evaporates fresh yttrium metal onto the anode,surface. These arrangements turn out to provide thenecessary conditions to enable one to run yttrium Mexcited electron spectra for long periods of time at aconstant X-ray level. A multidetector system in the focalplane of the ESCA instrument ensures a high speed ofaccumulating the data. The spectrum given in Fig. 27shows the valence region of N2. In the upper part the N2spectrum excited with monochromatic A1Ka is given, inthe lower partwith YM. Inserted is a more detailed studyof the three valence orbitals which show line-broadeningsdue to vibrational structure. The width of each line is setby the inherent width of YM, around 0.6 eV. Below thespectrum is indicated the spectrum which is obtained byUV-excitation using Hel radiation at 21.22 eV. Themost interesting feature of this comparison of electronspectra performed at different excitation energies is thewidely different relative intensities of the electron lines inthe three modes. This is due to the dependence of thephotoelectric cross-section on not only the energy of theexciting radiation but also on the symmetry of the molecu-lar level concerned. Consequently, from comparisons ofthis kind orbital symmetry assignments can be madeexperimentally, provided the cross-sections are deter-mined either theoretically9 or semi-empirically from anumber of known cases. The so-called intensity model(Ref. 1, p. 70) based on the MO-LCAO approximation has
been used extensively in ESCA during recent years totreat these problems.
In ourlaboratory another possibility, i.e. using apolarizedHe light source, is being studied. In this case the relativeintensities of the valence electron lines are recorded as afunction of the angle between the electric vector of thephotons and the emitted photoelectrons. Due to thelosses atthe successive reflections of the UV light at the mirrors(these have to be flat to within 50 A) at present to getpolarized light required a substantial reduction in intensityis inevitable and has to be compensated for by a highefficiency in the rest of the system.
Three examples of UV excited spectra are given inFigs. 28—30, showing the valence electron spectra ofwater vapour'0 (Fig. 28), benzene (Fig. 29) and threemetal hexafluorides1' (Fig. 30), namely MoF6, WF6 andUF6. In Fig. 28 the vibrational structure of the electronbands corresponding to the lb1 and 2a, valence orbitalsare well resolved. For the 2A, state a vibrational progres-sion of the P2 bending mode consists of 20 members. Thedifference between H20 and D20 is also striking. Acareful study of the H2160 and H2180 spectra reveals anisotopic effect also in this case. Vibrational structures arepresent in both the benzene and the hexafluoride spectrabut due to the increased complexities introduced by moredegrees of freedom for vibrating modes the variousvibrations are not always possible to resolve. Wherethey can be resolved a high precision in the ionizationenergies is always possible to attain, due to the extremely
JBINDING ENERGY, eV
88 KAI SIEGBAHN
Fig. 25. Asymmetries of core lines of various metals having high density of states at the Fermi edges of theirconduction bands.
I 12 I 4 6 8 110
y, m.VK mol
Fig. 26. Asymmetry of core lines (see Fig. 25) as a function of theelectronic specific heat of the metals.
narrow Hel resonance line at 21.22 eV. Electron linewidthsaround 10 meV can be obtained with an accuracy in the linepositions of 1 meV.
SURFACE SENSITIVITY
The analytical sensitivity of ESCA is related to itsproperty of being a surface spectroscopy. The depth ofinformation, which is the electron escape depth withoutany energy losses, is limited to <50 A, depending some-what on the material and, in particular, on the energy ofthe electrons being emitted. The electron escape depth asa function of electron energy is a parabola-like function
with a minimum at about 100 eV, where the escape depthis some 3 A.
We have seen that bulk properties such as conductionbands are reproduced in the ESCA spectra and manyother examples could be given which show that the analysisof the materials usually corresponds to the composition andproperties of the matter in bulk. One further example ofsome practical interest is given in Fig. 31, showing the Ciselectron spectrum of a plastic material, namely Viton withtwo different compositions, Viton 65 and Viton 80. Thebranching and relative compositions of these two polymersare well reproduced in their ESCA spectra.
The surface sensitivity is, however, likely to be ofspecial importance for the application of ESCA. In par-ticular if the emitted electrons are being looked upon at asmall angle relative to the surface by the electron analyzer,the escape depth may correspond to only one singlemolecular layer, and within this layer, which is by defini-tion truly a surface layer, the composition and chemicalbondings can in principle be analyzed by ESCA. Adsorp-tion of gas molecules onto a surface, corrosion,heterogeneous catalysis, depth diffusion, and surfacereactions are natural research objects. Such studies areknown to be difficult to perform, no matter what techni-que is being used and require clean surfaces (oftenultrahigh vacuum) and well defined and controlled experi-mental conditions. One of the main roads of develop-ments in electron spectroscopy is to explore these pos-sibilities. Already interesting information is being obtainedfrom studies in this promising field. To take a fewexamples,'2 it is possible to distinguish between COstandingup(a form)witheithertheOortheCatomontheW
Asymmetric core lines
Pd 3d512
3d312FWHM.137 cv
3d312 FWHM.09OeVA .116
4-0.420.37
__i ___________) —3d5,2
3/2
• 092 cvA•l.0
E10A6
i •\.•.•.____________i
0eV
Pt '47/2415/2
/ FWHM-1.l2eVA-1$2/ / A.O.70
___J J j043Hg 417/2
415/2FWHM- 0.76 cv
• 4-1.23• : 0.42• . . .
—, .••—/Pb 417/2
415/2FWHM-O.67eV
A1.240.37
2030
0.7
06
> 0.5
0.4
0.3
0.2
0.1
Core Line Asymmetries in ESCAvs. Electronic Specific Heat
>-
I— 10
000
w
F.- z
IlllII
IIIIII
IIlI1
1111
1111
_'11
_iI_
l nV
+
V2
D2
16Q
+
2A
x27
-
x12
- -\-J
ni,2
ni
,1
0000
2000
0
xO.C
2B
1
ni,2
I I
I I
I I
I I
I I
I I
I I
I I
I I
I ii n
+V
2
H2
16Q
+
x0.I
2B1
BIN
DIN
G E
NE
RG
Y,
4000
2000
ni'2
ru
'1
IIIJI
IIIIII
IIIIIL
IIH
flV1 +
Fig.
27.
Val
ence
elec
tron
spec
tra
of N
2 ob
tain
ed by
mon
ochr
omat
-
ized
AlK
a ex
cita
tion
(upp
er s
pect
rum
) an
d by
YM
exc
itatio
n
(low
er sp
ectr
um).
The
thr
ee m
ain
lines
are
plo
tted
in m
ore
deta
il an
d de
conv
olut
ed to
indi
cate
inhe
rent
ele
ctro
n lin
ewid
ths (
0.6 e
V)
and
vibr
atio
nal s
truc
ture
s. I
ndic
ated
at t
he b
otto
m o
f th
e fi
gure
is
also
the
Hel
-exc
ited
vale
nce
elec
tron
spe
ctru
m.
Obs
erve
the
w
idel
y dif
fere
nt in
tens
ity ra
tios i
n the
thre
e ca
ses.
-.
2A
,• .
____
__
x 0.
06.
2B1
x20
- —__
____
_ 6.
0 Jb
.0
4.0
3.0
BIN
DIN
G E
NE
RG
Y,
eV
Fig
. 28
. Par
t of
the
vale
nce
elec
tron
spe
ctru
m o
f H20
exc
ited
at h
v =
21.2
2 eV
(Hel
). O
ne o
bser
ves
isot
ope
effe
cts
betw
een
D21
60, H
2160
and
H21
8O. T
he 2
A1
stat
e ha
s a vi
brat
iona
l pro
gres
sion
of t
he P
2 be
ndin
g mod
e w
hich
con
sist
s of
20
mem
bers
in th
e el
ectr
on sp
ectr
um.
90 KM SIEGBAHN
0000
zUiz
E,, eV
20000
Fig. 29. The valence electron spectrum of benzene, excited at hv = 21.22 eV.
MoF
C,)2LiiI—z M .:
: f/. . . I,.JV\,.,,.f .. /
/,1\,,/ :) LI
9.0 8.0 7.0 16.0
BINDING ENERGY, eV5.0
>-I—
Cl)
LiiH2
BINDING ENERGY, eV
BINDING ENERGY, eV
Fig. 30. The valence electron spectra of MoF6, WF6 and UF6, excited at hi' = 21.22 eV.
Electron spectroscopy and molecular structure 91
[ 1 r 1 CFof exciting radiation. These lines provide information
4F-CFj-CF-CH2- :. 2 about the molecular structure which is complementary toL
m n the photoelectron lines. For example, Auger electron linesI
CHare similarly subject to chemical shifts. Auger electron
VITON 65A 2 spectra are complicated to interpret for the light elements
\ /' \ belonging to the first row (C, N, 0 etc.) since the valence
CF3 CF I -C- orbitals are involved in the transitions.'6"7 From the nextA\ / ' row elements, starting with Na Mg, Al, Si, P S and Cl/\ •\1 \—.,, . - _/ . ,,
. \_________ however, KLL Auger transitions occur between coreVITON 80 A .-. levels and give rise to a characteristic group of lines with
! \ 1' \ one line dominating the other ones in intensity, namely the/ I. \ one which can be described as the K-L11L111 transition
/'\J "'\ J ____\ (2s22p4('D2)).—, ' '- , Figure 32 shows this Auger electron line group for Mg_v 295 290 285 and MgO and at an intermediate stage of oxidation. For
BINDING ENERGY the metal one can also observe series of volume plasmonFig. 31. ESCA spectra of Viton 65 andViton 80 polymers. lines accompanying primary Auger lines as satellites. The
Auger chemical shift between Mg and MgO amounts to5.2eVwhich largely surpasses the chemical shift between
surface or COlyingdown(f3 form). Inthe /3 form CO seems the photoelectron lines, being 1.1 eV.to be dissociated on the surface of Mo. These conclusions In order to study more closely the relationships be-can be drawn from the differences in ESCA chemical shifts tween Auger- and photoelectron line chemical shifts forbetween the free CO molecule and when CO is adsorbed on free molecules and for solids in a systematic way, anthe surface, and is due to the metal surface ability to supply arrangement has been designed which is shown in Fig. 33.electrons to polarize the electron hole left behind during The external coil of a magnetic ESCA instrument hasphotoionization. This influences the "external" relaxation been removed in the photo to show the sample housing withenergy and decreases the binding energy as observed in the various excitation facilities incorporated in thisESCA. These are typical "final state" effects, just as are apparatus. InordertogethighAugerelectronlineintensitiesshake-up electron lines or vibrational line structures. It is an electron gun with quadrupole focussing lenses etc. isalso likely'2 that heterogeneous catalysis is dependent on attached to the bottom of the housing, the electron beamfinal state effects when in the extreme case an electron is being directed upwards and hitting the gas sample volumetransferred from one molecule to another to form an (or the solid sample) through a small hole in the bottom ofactivated complex. Such a transfer is facilitated when this chamber. The sample chamber is introduced andconduction electrons from the catalyzer metal can shield interchanged from above through a vacuum lock device.the resulting electron hole. This will decrease the binding X-ray excitation can simultaneously be done by means ofenergies of the electrons and reduce the potential barrier the strong electron gun inserted into the back wall of theleading to the final reaction products, since an additional housing and the watercooled rotating anode atthe other sidesource for external relaxation is provided. wall. A He lamp can, furthermore be introduced from
There are also "initial state" effects which are influ- above. There is a two stage differential pumping systemenced by surface conditions. For example, molecules around the source chamber to enable the study of gaseouslying on a surface, say benzene, expose their ir-orbitals samples. The Auger and photoelectrons leave the sourcemore strongly than the u-orbitals and are consequently chamber through a narrow slit and pass a smallmore influenced by the surface. This results in an in- four-component electrostaticlens system forming an imagecreased ir-orbital binding energy. The best surface sen- at the entrance slit to the analyzer, after being retarded to asitivities reported with present days technique so far is low energyin order to increase the energy resolution. At the2 . iO atomic layer for a heavy metal evaporated onto a other side of the double focussing analyzer, i.e. after 255°,surface.'3 there isa multidetector system along the focal plane of this
It has been suggested and also tried to combine the magnetic ESCA instrument..surface sensitivity of ESCA with a scavenge technique to The first series of studies concern Auger electronextract various ions in very low concentrations from a spectra of various sulfur compounds as compared tosolution onto a fibreglass surface, which has organic the corresponding photoelectron spectra. Recent theoreti-functional groups attached to it. In this first trial, Hercules'4 cal work by means of so-called "transition potentials"in this way reached sensitivities for lead, thallium and shows that the Auger electron spectra'8 for molecules canmercury in solutions as high as 5 ppb. Quite recently'5 this be quantitatively evaluated this way. Previous calcula-figure has been further improved to —0.1 ppb. tions by this method on binding energy shifts at photoion-
ization (ESCA shifts)'9'2° showed that this approach gaveAUGER ELECTRON SPECTRA results in very good agreement with experiment.
Photoionization giving rise to photoelectron spectra By changing continuously the electron gun voltage andalways accompanied by Auger electron emission keeping the analyzer keeping the Auger line in focus,and whether the excitation is caused by monO- "excitation yield" spectra can be done by means of the newchromatic X-radiation or by a beam of electrons the Auger arrangement. These "spectra" are interesting to followelectron spectra characteristic for that molecule will be around and above the threshold for the Auger process. Therecorded in an ESCA instrument. The emission of the arrangement also allows, in principle, some other experi-Auger electron lines is due to transitions between an ments to be done, such as pulsed double beam excitationinitial state having one core electron vacancy and a final which may concern excited and fragmented molecules etc.state having two vacancies. The energies of the Auger Microwave or UVlaser pulse techniques are also of interestelection lines are independent of the energy and the kind in this context.
EZC!f!OU CSU &J2O pc bcqotwcq pA bnJ2c jccprndcrnaqic aiq;) suq UA (tLow spoAc) jpc SswbIc CSWGL 12 UflOqq jIOW SOAC 4JJL0118J3 jpc ASCnIIW J0CJC qeAloc
SOJIqS 112 S WS1JC41C ECV !U4UTWdUi E1C!iSI!012 CSIJ pc wsqc pA sij c1cCLOU pcsw (Low pGJOM)' x-A (LowE!R fl j4CM SUS1JGWGIJL IOL 2!Wf1J4913c0122 VUCL suq bpoiocrccttou I!Uc CpCW!CSJ 2( UJCU2RLCWCUf2 tOL b2Q2
CXCiSflO1J
J4C KIT VOBCL GJCCILOU 1!1562 SLG !ACU PCl0M S2 LccoLqcq !U We EECV IU2jLIIWCIJ4 pA WCSIJa O CjCC4LOIJ pcswJbccf LOW 12 LLOW 912 1UfCLUJG!9jC Oxiqsj 011 /Off1WG 1)192mW) JIIJC2 sic op2cLAcq LOL cowbsulou we bo2u10u2 o cLLOW s CJCSIJ WC4SJ 211149CC abeGgow DOW we oxiqiscq wcs (mijp onjA s tisce o wcsj) snq pc wiqqjch! 3 N KrF VifeCL CICCILOJI 2bCCfLS Sf qqctcii 2fSC2 O oxiqsuon 92 O49WG !n F2CV flbbCL 2bcqLnW !2
KVI 21909V11L1
VflCEH EMESGA' A
Electron spectroscopy and molecular structure 93
pumped either up into the nozzle (4) - 0.7mm) which thenforms the beam or through a by-pass tube back into thereservoir. Valves can be adjusted to control the speed ofthe liquid beam through the nozzle. The nozzle can be slidon the top of the source housing in order to allow a pre-cise adjustment of the liquid beam in front of the entranceslit of the analyzer of the ESCA instrument. When anelectron spectrum from a liquid is recorded one can, ingeneral, obtain a signal from the vapour as well as fromthe liquid itself. We found that we could actually separatethe two spectra by adjusting the beam position relative tothe slit (shown in Fig. 35). The liquid to be studiedwas formamide (HOCNH2). When the beam is to the leftof the slit only the Cis line from the vapour is recorded.When the beam is gradually moved in front of the slit aline from the liquid itself grows up at a distance of1.6 eV from the gas line. At a certain position of the beamthe gas line is completely suppressed and only the liquidline is recorded. A further movement of the beam restoresthe gas line. The reason for this fortunate behaviour is theshadowing effect of the liquid beam itself on the vapourwhich is excited by the X-rays in this geometry.
Figure 36 is a recording of the Ols, Nis and Cis fromHOCNH2 at a beam position which allows both the liquidand the gas to be recorded simultaneously. A very smalltrace of water in the liquid produces an Ols line (to theleft) in the gas phase.
Figure 37 is a liquid spectrum of a solution of a salt,UQUID namely KI in formamide. One observes the ESCA lines
from both iodine and potassium.Since these experiments were completed we found a
Fig. 35. ESCAlines from the gas phase and the liquid phase as a function of liquid beam position relative to the analyzerslit.
ESCA SPECTRA OF LIQUIDS
So far we have discussed electron spectra of severalkinds of gases and solids. During the last two years wehave been trying to extend the ESCA techniques toinclude also liquids. Our first arrangement consisted informing a "liquid beam" of the sample inside the sourcecompartment of the ESCA instrument. Figure 34 showsthis arrangement. The liquid or solution to be studied iscontinuously circulated by means of a pump. The liquid is
LIQUID
Fig. 34. First arrangement to form a 'liquid beam' in ESCA.
HOCNH2
Cis LIQUID + VAPOUR
JPOULCUID1 j300(
200C
I00C
3000
2000
000
>-
U)zLiiI-z
3000
2000
000
1•-
Ii
535 410 405 295 290
BINDING ENERGY
Fig. 36. Electron spectrum of formamide in beam position to yield both gas and liquid lines.
somewhat better method2' of arranging liquid studies.This arrangement (see Fig. 38) consists of a thin stainlesssteel wire running from one bobbin to another. The wirepasses down into a small beaker of liquid and then, afterbeing covered with a thin skin of the liquid, it proceedsupwards along and in front of the slit of the ESCAanalyzer. The wire at the same time provides a stabiliza-tion of the electric potential of the liquid. This system canrun for some hours without recharging and works quitesatisfactorily. Many core and valence electron spectra forliquids and also Auger electron spectra from dissolvedsalts have now been recorded with this system.
Figure 39 shows the electron spectrum from glycol(CH2(OH))2. The left spectrum is from the liquid. Theextra Cis peak with the upper part of the figure has nocorrespondence at the Ols line and disappears in factafter the wire has cleaned the liquid surface from carboncontaminants at a higher speed (see figure below). Theright figure shows the spectrum when the wire is adjusted
for recording both liquid and gas phase (above) and onlygas(below).
Figure 40 shows22 the valence electron spectrum offormamide for the liquid and gas phases separately. Acharacteristic change of the orbitals at condensation canbe explained in terms of intermolecular interactions in theliquid. The liquid spectrum is consistent with hydrogen-bonded linear chains of molecules in the liquid. Alsomixtures of liquids can be studied.23
Finally, Fig. 41 shows the first Auger electron spectrumfrom a liquid, namely from sodium when Na2(CN)2 isdissolved in (CH2OH)2.
PRESENT DEVELOPMENTS
The field of electron spectroscopy is now a well estab-lished field of research with numerous applications andwith many possible ways to proceed, some of which havebeen indicated in this survey. Already with present days'techniques one can explore large fields and add much new
94 KAI SIEGBAHN
HOCNH2Nis
GAS LIG.
3000 -
(/2I-z2000
C.)
1000•
OlsGAS LIQ.
CisGAS LIQII
eV 540
K12M +HOCNH2
I 3d3,2 5/2
C is2000
1500
>-
U,
1000wz
500
0
K
x2/3
eV 640 630 620 300 290
BINDING ENERGY
Fig. 37. Spectrum of a dissolved salt, KI, in formamide.
information about the electronic structure of moleculesand condensed matter. With further improvements thisspectroscopy is inherently capable of developing in severaldirections, one of the most interesting ones being surfacescience. Ultrahigh vacuum technology and improvedsample handling are necessary prerequisities for progressalong these lines. Higher intensities and better resolutioncan also soon be expected. This means increased sensitivity
(CH2OH)2Cistiquid OlsLiquid
WIRE SPEED:3mm/sec
CISCISEXT
,—
6 mm/sec :
,—
•
95
and should normally result in a better chance of studyphenomena which may now be at the borderline to thefeasible. One could list many examples of this kind, amongthem time studies of surface reactions etc. In all cases,
Fig. 40. Valence electron spectrum of formamide for the liquidand gas. At condensation hydrogen bonded linear chains of
molecules in the liquid are formed.
(CH2OH)2 GAS LIO
OlsCisGAS LID.
\
. :-b0 555 710 715 720
Electron spectroscopy and molecular structure
U)
zD0C.)
LUz
Fig. 38. Wire system for ESCA liquid studies.
BINDING ENERGY
30000
20000
10001
3001
2000)
10001
(1)
z:70
100
5001
101
500
I—z:700
KINETIC ENERGY, eV KINETIC ENERGY, eV
Fig. 39. Electron spectrum from glycol using the wire system shown in Fig. 38. Left figure: liquid with extra line dueto carbon impurity at the surface (upper part) and after wire has.cleaned surface at higher speed (lower part). Right
figure: wire adjusted for recording both liquid and gas phase spectra.
wb pon:JJC LOJU uoq wnU!cLA2! pojqr. pwq pc
pc ptc wb1 puqjru 2?cw 10 ipc lc:ji oqcj ot qc UGt obuwiscq E2CV ! tnuIur 1° W
1UrCIJI1JC MGLC AqpJcDiLqiou nuqt wncp iuJbLoAcq couqipou i ppcFEED if P°"N p bo22pJc jo cxbjoL JJC E2CV1P!CP JJ9 GU ncq GXCUiACjA qnuu LCCCU Acn2 !uGXCLUj pcw CICLOU i qiLCCq ! CL?fj uqbLcAion2jA MJj-JCUOMU CJCLOJJ qJJLcJou MGUpc IU4Lu1wGIJ JJJJ o qJtciou qqccu L0UJ pcic c o qLciou bLu MjJ corijq pc LCCOLq€q iiiuq pc qucciou o cjcciou cwiiow jjn bpduowGuou
qiGLCuj UJCGMCCU1JC XI O }JG CLAJjCACj uq jO VflL CIGCI1OU M!}J buoqicA ieaAiu/•JjIJ wr X-LA2 cwcq bpoocjccpou tow iuonbLoc w ct2j v MCi iuc cLA2j pcu Iuqijcq!UGLUL CJGCI100 qqLci0U OCCflLLiU qnuu U ECVopuq OWC JJGL 0 10 OIIL E2CV !U211WC1U q Occnwnjiou o jLG wonu o u Gxwbjc t1ccouqiiou o 2wqA bpuowGirn M1CjJ LGdfIILG flJ0IAACL !ucLccq !IUGU!G WU WOLG C0IJAG1J!CU
1COI VI IPG OJ COLC JIUG tLOW J?eCOj !2 LGC0LqGqpG ww VL juc iom oqinw iu k1(cM) qJ2oJAcq 113i i E! GJccLou thccpnw Ll.ow pdniq opcti
KiI.IEIIC EI1EbOAarc Sac 350 300
GJ!R2 uq ic 2cppu nnJ-Jo (J,x�YR iUPCL 2 2AG1J201J b j AjWdAi2 J' 2JJ1GL if(T).O l' 0 KP VUqGLGU 'J V 'U°P 1Hq°' flflIb-J3 (J,x
3W 1" V B Uql21J J4 U2øU jç 4?aJJOJW uq i•iq i ACLP1 &IJq LCGLG1JC pCiCIU
(i't) cowc.1Jcc A'-F qq pA jç nquo1K 2!GPPU EJCCILOU bGCLOCObA—VU O1TJOOJC
KEEKE1.4CE2
!wbLoAGq bLoAjiou o1. owbjc pouqpu iJ0LG WCLG0G JO i!UcuiA (>Jyj) i opmucq ouq wncpbGLoLuJoucc J JJGA ObcLoc 0ccoLqiU 0 JJC C0jCflJ0JJ0U 0ouq LcdnlLG mncp coi.cnJ oq1n4wcu obiwnmoi.c JqcuicoJ JjJGG iio !u2jLflmcu2 OLG cowbjGx U qGUcombm.wGuL ponc w OGL LbGc WG Mo !unwGu
!'''t i bjOcGq O JJG LIJU iqc O fG 2OfILCGiu jpc bLcAion coc jpc owbjc pouqJiu 2?w OL pGciu po o p in0Gq MJJGLG pc ouJb jc pouqjiu Acw iiocdncucc o pi qc2u LO0W ouoq ouq G ccciouwcuiA o JJG wouocpiomoic oqmiow 2 0 cou-O' cU WGU I1C WOLG CL 20J (5) ouq JJ11 WCLG0G JJGOPG!GU2 jifouq uooJoJJfpcIwGx-Lo1 pipc wbjcLowpc 21qcrc bcLbcuq!cnJoLI!dI!1 ouq nqocca iu p!2 coc i i WOLG oqoucon o2JIJJIJJjOIJGOfl2JA pciu pn bum LqA wcuqcq OL aopqopq con ojao pG nqicq y ccouq uuimcuwJ.nmcu pocjq buuJouJA pc ncq OL 02G2 oponpo ojjo wouA cxbciiwcujoj &u0uGWGU2 o pc woqc jpicjcctou pcoui qGAiCG jpc 7A4cw ponq p upt cuonp0 H-1m A1-UoqC CJGCJ1013 f1IJ ouq 0 WOIJ0JUUG1CJJGLG OLG tflJ4JJGL 0C1JJJG2 OL OJJGL moqc OJj GXCi0iOU JIJCG
owbIc bLcboLoiou 2CUF yionuq pc 2ombJc bo2lpou2cb0Loc HG ctAobffwb2 ouq pA pio oqqiiouoj ouc In JJC2JJiGJq2 0 t-WGJ0J2 OAGLOJI Ipc iup.nwcnj bnmbcq pALPG !'q AOJDWG O JJG IU4LIIWGIU !2 bLoAiqcq mijp oLc2bouqnJJA JoU WfIJPqGGCOL bjoccq w pc tocoJbJouc2cboLo1ou ot j cm wc 2bpGLG2 COL-
bjJcucoj 0U0JAGL 02 0 W000 Loqin2 O! cm ouq oGIJGLA 0 JJG CJGCLO1J2 (o LG0Lq010IJ 0C0L nb o o) ipUJIUIUJIIUI opcI.LoJioU ouq o bLoAiqc 0 I0IC LGI1CP0IJ 0J jJJGGJGCLO201C jCU2 2A2GW flIJCL couq!Jou2 qGiuGq 0 AGL01U O ouq 0jOIJ JJG 2J1j 0 055 Cm ou toI1L-COWb0IJGJJLoqioJiou MIJj JJ1 JJC 20UJbJC 0 02 OL 0OJ! J.0W 0OAG !'bLoqncc o wouocptowojc v!iC jmc o '1W! 0• T8 GK Jpcompmc o uc o WUJTUJIJIJJ o 2bpcucoj opcLLo1ou ouq( = cm) m.c wornucq 013 0 pjocjc o Luqcq j022 onqcacjc on o2GwpJA ot j 2bpGL!COJJA pcu dnoI4s cLA2ojqciucq OL o abccq o >iyjj LcA\mw On flJG jçojouqtA0jGLCOOjG L00UIJ onoqc JJO2 0 10WGGL 0 o Cm ouqurn o o boMGL JGAGJ o 30 (J()() my oiq so jqY ipcwncp WOLG boPAcqcJJ bcLoLwoucc 1pC GJGCILOU 1IJ 4JJ2JWIJOL O qJG bLoojAbc !U4flmGU 21J0U 1U1 pII O 0LGoc1OU cc JJJG x-Loi wouocJJLomooL ! u bLmcJbjcOL G ombjc JJJ(G cAubOLOpou 2bflUGLIu JJGOIURcouqiIou 2CAGLOI mcpoq o bLcboLopou OLG oioIjopjcuIumu JJG CJGCL01J 2bCCLf1W IIIJqGL flJjL0jJJJ A0CD0WiUCLG02iU AOCflflW nnpj i bn !uo onLCc bo2!!ou tOLpLonp pi Acm lu JJLGG 20p2G&JGIJi 2Gb2 M!P20wb1c bLcboLoiou 2AGm jyc 20wb1G 12 JulloqfrCGq0 W! !UUwGU OUG OGLAG2 ! W OLGLOflU tJCOIIL jopoLoOLA 0L 2OWG WG E!IILC 43120 bpoo o o moqcjE2CV cdnibmu H4JJCIJ 1102 CGU cruqct CO1J2L11C1OIJ !'
LP!2 2IILAGA !ATJJ GIJq M1JJ 0 PLICt qI2cn2aiOu 0i
iiC,0
00\ 5b OJ5(VI 20
(CH5OH)0
V&(CV)0 (J5S)
£ L!1J
KVI BIRORVHM
—.- "
I
Electron spectroscopy and molecular structure 9Y
5M. Krause, Chem. Phys. Lett. 10, 65 (1971); M. Krause and F.Wuilleumier, Phys. Lett. 35A, 341 (1971).
6M. Banna and D. Shirley, LBL-3478 and LBL-3479 (1975).7D. Allison and R. Cavell, ACS April Meeting, 1975.8R. Nilsson, R. Nyhoim, A. Berndtsson, J. Hedman and C.Nordling, UUIP-912 (1975).
9J. W. Rabalais, T. Debies, J. Berkosky, J. Huang and F. Ellison,J. Chem. Phys. 61, 516 (1974).
10L. Karlsson, L. Mattsson, R. Jadrny, R. Albridge, S. Pinchas, T.Bergmark and K. Siegbahn, .1. Chem. Phys. 62, 4745 (1975).
11L. Karlsson, L. Mattsson, R. Jadrny, 1. Bergmark and K.Siegbahn, UUIP-911 (1975).
'2D. Shirley, J. Vac. Sci. Tech. 12, 280 (1975).'3C. Brundle and M. Roberts, Proc. Roy. Soc. Lond. A331, 383
(1972).'4D. Hercules, L. Cox, S. Onisick, G. Nichols and J. Carver,
Analyt. Chem. 45, 1973 (1973).'5D. Hercules, personal communication.
'6H. Siegbahn, L. Asplund and P. Kelfve, Chem. Phys. Lett. 35,330 (1975).
'7H. Agren, S. Svensson and U. I. Wahlgren, Chem. Phys. Lett. 35,336 (1975).
'8H. Siegbahn and 0. Goscinski, UUIP-908 (1975).'G. Howat and 0. Goscinski, Chem. Phys. Lett. 30, 87 (1975).20H. Siegbahn, R. Medeiros and 0. Goscinski, J. Electron Spec-
trosc. 7 (1975).21H. Fellner-Feldegg, H. Siegbahn, L. Asplund, P. Kelfve and K.
Siegbahn, .1. Electron Spectrosc. 7, 421 (1975).22H. Siegbahn, L. Asplund, P. Kelfve, K. Hamrin, L. Karlsson and
K. Siegbahn, J. Electron Spectrosc. 5, 1059 (1974).23H. Siegbahn, L. Asplund, P. Kelfve and K. Siegbahn, J. Electron
Spectrosc. 7, 411 (1975).24K. Siegbahn, U. Gelius, H. Siegbahn and E. Olson, Phys. Lett.
32A, 221 (1970); Physica Scripta 1, 272 (1970).25U. Gelius, H. Fellner-Feldegg, B. Wannberg, A. G. Nilsson, E.
Basilier and K. Siegbahn, UUIP-855 (1974).