XPS, UPS and XAES studies of the adsorption of nitrogen, oxygen, and nitrogen oxides on W(110) at 300 and 100 K

Surface Science 0 North-Holland

79 (1979) l-25 Publishing Company

XPS, UPS AND XAES STUDIES OF THE ADSORPTION OF NITROGEN, OXYGEN, AND NITROGEN OXIDES ON W( 110) AT 300 AND 100 K I. Adsorption

of N,, N,O and N02/N,04

J.C. FUGGLE * and D. MENZEL Institut fiir FestkBrperphysik, Physik Department D-8046 Garching bei Miinchen, W. Germany Received

7 July

1978; manuscript

received

E 20, Technische

in final form 30 August

Universitri‘t Miinchen

1978

The adsorption of Nz, NsO, and NOa on W(110) at 300 and 100 K and the behaviour of the resulting adlayers upon heating have been investigated using XPS, UPS, and XAES. For 02 adsorption, some results of importance to the present investigation (sticking coefficients, XPS peaks and satellites, UPS, and XAES spectra) are also reported. The determinations of absolute coverages of 0 and N are based on these data. All nitrogen oxide layers are found to be deficient in N, as compared to the gaseous species. The occurrence of dissociative or molecular adsorption under the various conditions is determined. It is found that at 300 K, NaO and NO* adsorption are completely dissociative, while at 100 K initial dissociation is followed by the formation of condensed layers of NaO and Ns04, respectively, with different growth characteristics. Adsorbate line widths in UPS and line shapes in XPS are discussed. The usefulness of the various methods for fingerprinting is compared.

1. Introduction Whilst a great deal of effort has been expended to achieve some quite detailed understanding of the adsorption of CO and 0 on W surfaces [l-4], there have been few investigations of the adsorption of nitrogen and nitrogen oxides on W. Indeed the only photoelectron spectroscopic investigations reported are those of Madey et al. [5] who described XPS investigations of the adsorption of N2 and NO on polycrystalline W, and of Feuerbacher [6] who studied the UPS spectrum of nitrogen on W(100) at 300 K. As to other metals, the most important paper is by Brunclle [7], whose systematic study of the adsorption of nitrogen and nitrogen oxides on polycrystalline nickel will be discussed later in the text. Here we will report on XPS, UPS and XAES (X-ray excited Auger electron spectroscopy) studies of N2, NzO, and NO2 (or N204) adsorption on W(110) at -300 and -100 K, where * Present address: Institut Jiilich, W. Germany.

fur

Festkorperforschung,

Kernforschungsanlage

Jtilich,

D-5 170

possible comparing these results with those obtained for the same gases on other clean metals. Studies of NO adsorption are reported in part 2 of this work. The W( 1 10)/N-oxide system was chosen for this study because we have some experience of oxygen and CO adsorption on this face and can use results from those systems to aid interpretation of the present results; no other photoelectron studies of these systems have been reported. Nitrogen may adsorb molecularly ar atomically on transition metal surfaces. Species which appear to he molecularly adsorbed N2 on polyc~st~~e Ni fS19 Ru(100) [9], and Ir field emission tips [IO] are found to have desorption energies between 30 and 60 kJ mol-’ and are thus less strongly bound to metal surfaces than molecular CO. We know of no report of molecular nitrogen coverages being stable in UHV on clean metals at room temperature. Forms of adsorbed nitrogen which are almost certainly dissociatively adsorbed N are more strongly bound to most transition metal surfaces; adsorption/desorption energies of -240 kJ mol-r for polycrystalline Re and 330-390 kJ mol-’ for polycrystalline W [I 1,X2] are given in the literature (i.e. desorption temperatures are 1000-1450 K). The sticking coefficient for dissociative adsorption of Ns is variable, being small (-0.01) on polycrysta&ne Re [f1,13] and Fe 1143 but of the order of 0.2-0.4 on polycrystalline Nb, Ta, MO, and W 1151. It varies very strongly with crystal face on tungsten f16-I?]. ft is generalfy accepted to be very low on W(ll0) and is certainfy increased drasticahy by the presence of steps [ 19]_ Thus the only well characterised state of nitrogen on flat W(ll0) is molecular and desorbs around 150 K with an activation energy of -25 kJ mol-’ 1201. This state has characteristic XPS, UPS and XAES spectra 1211 which we reproduce here in order to show that this state is not formed during reaction of nitrogen oxides with W(110). The adsorption of nitrogen oxides on clean transition metals has been even less studied than adsorption of nitrogen. At low temperatures molecular adsorption of NO is observed on Ni [7,22], and Fe [14,23], and XPS gives some evidence for molecular NO on polycrystalline W at low temperatures [5]. Even at -80 K, dissociation of NzO and NOz is reported on poIyc~st~~~e Ni [7] but the evidence is far from un~bi~ous. When atomicaliy adsorbed nitrogen is formed by reaction of nitrogen oxides with metal the nitrogen can be desorbed as nitrogen molecules at 500-SOOK from Ru(IO0) [9], and at 700-900 K from NitIll) 1221 and Nifl IO} [24] r The desorption energies of 130-220 kJ mot-’ cannot be compared with data for desorption of nitrogen from atomic states in the absence of adsorbed oxygen. However, Madey et al. found that the desorption temperature of nitrogen from polycrystalline W is similar with [S] and without [l I] the presence of coadsorbed oxygen. In a series of papers Gasser et al. [25] have shown that the uptake of N20 by Rh, Re, and W was 5-8 X JO”” atoms rnm2. !*$nst this background information the main tasks we set ourselves in this work are (1) to establish the electron spectroscopic fmgerprints of adsorbed Nz and N on metals;

J.C. Fuggle, D. Menzel /Adsorption

of N2, 02 and nitrogen oxides on W

3

(2) to establish estimates for the amounts of nitrogen and nitrogen oxides which can be adsorbed on W(110); (3) to investigate the possibility of molecular adsorption of nitrogen oxides on W( 110) at low temperatures. The present paper presents results for N,, N,O, and NOz adsorption, together with some results for O2 adsorption which are important for comparison and calibration. The subsequent part II [58] will present results for adsorption and reaction of NO on W(110).

2. Experimental The apparatus used for this work is a custom-built Vacuum Generators instrument which has been described elsewhere [26]. It has a base pressure of about 3 to 6 X 10e9 Pa. Digital counting methods are used to accumulate data. All binding energies given below are referred to the Fermi level. The W(110) crystal discs used for this work were cut by spark erosion from boules produced by Metals Research Corporation. The crystals were oriented by Laue photographs to approximately 30’, cut by spark erosion, and polished mechanically to about 3 pm, then chemically and mechanically with 1 and 0.05 pm A1203 and a solution of CrOs in water. N20 (99%) and Nz (99.99%) were purchased from Messer Griesheim. NO, was made by heating lead nitrate. The first fraction obtained was pumped away as it contained a high proportion of impurities. The fraction used was purified by repeated condensation and re-evaporation at 100 K to remove components volatile at this temperature (in particular oxygen). In the storage bulb this gas must have partially dimerized to Nz04, as expected from equilibrium data and evidenced by the light colour; the condensed solid was white. Unfortunately, the molecular state of the gas under dosing conditions is not known. Thermodynamically, the expansion to low pressures favours dissociation to NOz, but we do not know the rate of dissociation, especially under the possible catalytic action of the metal walls. Mass spectrometrically, no Nz04 was detected under dosing conditions, but this could be due to efficient decomposition in the mass spectrometer. Because of this uncertainty we use the label “NOz” for this dosing gas in a generic sense. Mass spectral studies were also made for the other gases in the vacuum system under conditions similar to those used for exposure. This does not give an absolute check on the gas purity because Nz and NO form part of the normal cracking pattern of N*O and NO*. However, we can at least say that the oxygen content of all gases used was below l%, and the observed cracking patterns were similar to those given in the literature. The CO content was below 2-3s for all gases (shown mass spectrometritally for the nitrogen oxides and by XPS of the adsorbed layer for N,). Instead of exposures, we here give doses (or, more accurately, values offlfluence), in the unit 1 Ex = 10” molecules~m-2 striking the surface [27]. At 300 K, 1 Ex

4

J.C. Fuggk, D. Menzef /Adsorption

of N,

02 and nitrogen oxides on W

NzO is equivalent to 0.33 L, or 4.4 X lo-’ Pa+ NzO. The figures for NOz are very similar (referring to the monomeric form). The ionization gauge used for pressure measurements in our instrument is situated -50 cm from the sample. In order to calibrate the gas pressure at the sample we plotted the variation of the 0 1s XPS peak from adsorbed oxygen on W( 110) at 100 K versus the nominal gas dose as calcufated from the ionization gauge. The resulting curve was compared with the known kinetics of oxygen adsorption on W(ll0) at 100 K [28] and the correction factor between real and nominal gas dose was thus found. A dif~culty arises in studies of N20 and NOz in our equipment because exposures are made by flooding the whole experimental chamber. N20 and NOz become trapped on the cold areas around the titanium sublimation pump and the cooled sample, so that the pressure sinks slowly after a gas dose. This leads to difficulties in estimating exact gas doses and to slow condensation of NzO and NOz from the ambient onto the cooled sample. To reduce this effect we periodically cleaned and re-exposed the sample and added up the scans from the short counting periods.

3. Results and Interpretation

In order to calculate the absolute coverage of W(110) by the gases used in the present study we again use the oxygen/W(l 10) system as a reference point. The oxygen coverage of a W(110) surface at 100 K is shown as a function of oxygen dose in fig. 1. In agreement with Wang and Comer [28], we find that the curve

o-100K 0 -3OOK ,

0

5

15

10 ~

I

I

20

,,.._

I

100 DOSElEx) -

I

loll

I

Hxxx)

Fig. ;. Plots of the relative integrated 0 1s XPS peak intensities from oxygen on W(110) surfaces at 100 and 300 K as a function of oxygen dose. The coverage may be assumed to be proportional to peak intensity in this case. Note the change of scale from linear to logarithmic at 16 Ex.

J.C. Fuggle, D. Menzel /Adsorption

of N2, 02 and nitrogen oxideson W

5

reaches a plateau at a dose of about 1.5 Ex. We refer to the coverage at such plateaus as “saturation” coverages although it is clear that very high oxygen doses would lead to some further adsorption. (Notice the logarithmic scale of part of the abscissa in fig. 1.) Wang and Gomer [28] found that at “saturation” at -100 K (actually they measured at 87 K) the oxygen coverage on W(110) was 7.1 X 10” atoms m-* and we use this as our reference coverage. As shown in fig. 1 the curve for adsorption of oxygen at 300 K does not show such a clear plateau which could be used as a reference point. We can, however, readily calibrate the coverage after any given oxygen dose with respect to the reference point at 100 K. This was done to produce a secondary reference point for NO2 and N20 measurements at 300 K. In passing we note that information on the sticking coefficients, s, of oxygen can be derived from these measurements. For a gas temperature of 300 K, Wang and Gomer [28] have given s = 0.56 for the surface temperature 7’, = 87 K, and s = 0.47 for T, = 300 K. Engel et al. [2] and Butz and Wagner [29] have measured s r 0.4 for T, = 300 K. Our corresponding values are 0.68 (T, = 100 K) and 0.45 (T, = 300 K), in acceptable agreement. On the other hand, Besocke [30] has found s to be increased by surface imperfections like steps, and has concluded from the relative values that for flat W(110) surfaces s must be below 0.28 at T, = 300 K. If this argument is correct, our crystal faces as well as those of refs. [2] and [28] were rather faulty, and to about the same degree. However, imperfections also lead to dissociative adsorption of N2 [19] which we do not observe (see below). This suggests that our (110) faces did not contain a large concentration of faults. In order to calculate the quantity of adsorbed nitrogen we also need to know the relative sensitivity of our spectrometer to oxygen and nitrogen core peaks. This depends on the relative photoionization cross sections for the electron levels used and on various machine-specific factors. We made a calibration with condensed NOz and found that the sensitivity of our experiment for oxygen (using the 0 1s XPS peak) was 1.6 times that for nitrogen (using the N 1s XPS peak). We recognise the shortcomings of the XPS technique for measuring concentrations of surface species. A correction for shake-up features must be made where these are important; formation of thick oxide-nitride layers or diffusion of adsorbate into the bulk will falsify the results; and XPS peak intensities are also dependent on the precise sample position. However, in the absence of other measurements of adsorbate coverages the results of XPS measurements are useful, at least as a guide to the true coverages of both N and 0, and we always give the possible errors alongside our coverage estimates. Such results are certainly more reliable than absolute values derived from, say, flash desorption or LEED. 3.2. Adsorption of N2 Fig. 2 shows the UPS, XPS N Is, and N KLL XAES spectra from N2 on W(110). As explained elsewhere [21] we believe that only one state of N2 is present on W(110). The two peaks at 11.7 and 7.0 eV in the UPS spectra of adsorbed N2 can

1

G

12

1

10

+---

I

8

6

I

4

BE (eV1 _-

I

2

I E,

-

’

350 -

K E (eV1

-

2

Pig. 2. Valence band difference spectra fram Nz on W(l10) at 100 K (left), and N 1s XPS and N KLL Auger spectra fright). Part of the N Is XPS spectrum has been run at higher resolution to show that the peaks from the two N atoms can be clearly separated, N Is XPS and N KLL XAES difference spectra (covered minus clean) have been plotted to reduce the effect of peaks from &heW substrate at BE > 420 eV and KE < 350 eV.

thus be associated with the 4a and 50 + ln levels of N2 by anatogy with the isoelectronic UJ/W(llO) system (see e.g. ref. [3] and references therein), Malecular nitrogen is thought to sit vertically or nearly vertically on the W(l10) surface so that the two nitrogen atoms in each molecule are inequivalent. This gives rise to the two N Is peaks at 399.1 and 400.4 eV. The broad XPSN 1s peak at 405.5 eV is most probably due to a rather special type of shake-up which has been discussed before in connection with theoretical treatments [3 l], According to these treatments, the screening process of the core hole created is due to the transfer of electrons from the metal, and the “main” peaks around 400 eV are due to the fully screened core holes on the two different N atoms, while the peak around 406 eV is due to tr~sitions to final states in which the created core hole is screened by the image charge which is less effective. The high intensity of the more weakly screened peak is due to the weak surface bond which leads to weak hybridization of metal and adsorbate orbitals (or weak hopping between them). The KLL Auger spectrum of nitrogen gas [32,33] is completely different from that of adsorbed nitrogen molecules. This may also be related to the special form of shake-up shown by adsorbed Nz, ~thou~ further work on this question is necessary. Kunimori et al. [34] have published AES spectra of Nz an MO at - 190” which they b&eve to have been influenced by the presence of mofecular Nz which is thought to adsorb on MO at low temperatures [35]. Their spectrum bears littie resemblence to ours which may be due to the presence of di~oc~at~ve~y adsorbed

J.C. Fuggfe, D. Memel /Adsorption

of Nz, O2 and nirrogen oxides on

W

7

nitrogen or electron beam effects in their experiment. Chesters et al. 1361 have published an electron induced AES spectrum of N2 on W(ll0) at SO K, but it has insufficient resolution for detailed comparison with our work. Their peaks appear to be shifted by about 20 eV relative to ours due to difficulties in calibration of their energy scale. 3.2. Adsorption

ofNzO and NO, at room temperature

After exposure of the W(110) surface at room temperature to 30 Ex NzO or NOz the intensity of the 0 Is peak reached its maximum and no further growth was observed. In fig. 3 the 0 Is XPS peak from a W(l10) surface “saturated” with Oa, NZO and NOz is shown together with the N Is peak for the surfaces saturated with NzO and NOz. The N 1s peaks from Nz adsorbed at 100 K, together with the

N

ts+Shake-up

I

I 16OOc/s

I

‘.‘:

x

I 560

550

540

630

5x) -

BE

120

410

ml

(@VI

Fig. 3. XPS spectra and shake up features from N2, NO2 and NQO on W(110) at 300 K together with similar data from a condensed layer of N2 and N2O4 at 100 K, and of adsorbed oxygen at 300 K. For chemisorbed layers N 1s difference spectra (covered-clean) are plotted to reduce the effect of satellites from the WN3 peak at -425 eV.

412.4 405.1 - 405.8 a 404.2 397.6 + 0.3 396.8 - 398.0 a 400.8

UN0 408.7 401.8 t 0.2 -400.6 f 0.4 402.0 -396.8 r 0.4 396.2 - 396.8 a 396.2 + 0.3

Gas [40,41] Molec. at saturation on W(110) 100 K Molec. at tow coverages on W(~lO), 100 K Moiec. poly. Ni “77 K” ]7] Diss. W(110) at 300 K Diss. W(110) at 100 K Diss. ads. W(110) at 100 K, and heated to 300-900 K

Gas Molec. condensed on W(110) at 100 K Molec. condensed on poly. Ni at “77 K” [ 71 Diss. W(110) at 300 K Diss. W(I 10) at 100 K ?, poly. Ni at 100 K [7], low coverages

409.93 t 0.10 409.93 f 0.05 399.1 -’ 0.2, 400.2 400.6 -398.5 397.3 -397

N 1s BE (eV)

Gas [ 37) Gas [ 381 Molec. on W( 110) at 100 K Molec. poly. Fe at -100 K [ 141 Molec. poly. Ni at -100 K [7) in W-N2 complexes [ 391 Diss, poly W [5 1 Diss. poly. Fe [ 14 1

State~substrate

a Increases with coverage

NO2

@2O

Starting molecule

NE0 412.6 405.9 _+0.2 -404.4 _+0.4 406.0

400.4 f 0.2 405.3 405.7 -400.3

-405.5

f 0.5

+ 0.2 - 530.4 i 0.2

i 0.3 + 0.4

541.3 533.3 - 533.8 a 533.0 530.5 + 0.2 530.4 * 0.2 531.9 _-...-_

541.2 534.3 533.6 534.6 530.3 530.1 530.3

0 IS BE (eV)

Table 1 XPS binding energies relating to adsorbed nitrogen and nitrogen oxides (referred to the Fermi Ievei for condensed phases, and to the vacuum level for gases)

9 Q “?”

i;

to

J.C. Fuggle, D. Menzel/

Adsorption

of N2, 0,

and nitrogen oxides on W

9

0 1s and N 1s peaks of condensed N204 are also shown. A number of satellites are seen in fig. 3 alongside the main XPS peaks. In some cases, as indicated by the question marks, the signal to noise ratios are poor and background effects produce too many uncertainties to be sure that a satellite is really present. Most of the satellites less than 20 eV from the main peaks are probably due to shake-up, i.e. the peaks appear because there is a finite probability that the final state is an excited state of the molecular ion. The broad peaks labelled 4 in the spectra of condensed NOz (N,04) are probably due to extrinsic losses. It is seen from fig. 3 and table 1 that the 0 1s binding energies and 0 Is satellites from the surface covered with oxygen, N20 and NO2 at 300 K are all similar. This suggests that the environment of the oxygen is similar in all these cases, i.e. that NOz and NzO are dissociated at room temperature. Also the strong satellites present in molecular N204 (no data for molecular NzO could be obtained because this could not be conc!ensed in thick layers) are absent in the spectra obtained after adsorption at 300 K which again suggests that adsorption is not molecular. This conclusion is further supported by the N 1s results from the same surfaces. The N 1s binding energy is in the same region as that found for nitrogen thought to be dissociated on other metal surfaces (see table 1); as in the case of oxygen [42] the small shifts between different dissociated species on the same surface can be attributed to small differences in the weak adsorbate-adsorbate interactions. It is quite clear from the N 1s XPS results that adsorbed molecular N, is not formed during the reaction of NzO and NO* with these surfaces. As described in the experimental part of this paper the 0 1s and N 1s peak intensities were used to estimate the quantity of N and 0 on the metal surfaces. The results are presented in table 5 and discussed in section 4. The similarity of the 0 KLL XAES spectra (fig. 4, table 2) from adsorbed O2 and NzO or NO* adsorbed at -300 K shows that all these spectra stem from atomically adsorbed oxygen and therefore supports the conclusion that NzO and NOz are dissociatively adsorbed on W( 110). The observed N KLL XAES spectra from NzO and NO* adsorbed at 300 K do not resemble those of adsorbed Nz which shows that no molecular Nz is present under these circumstances. These spectra must then arise from nitrogen atoms on the surface. The N KLL Auger spectrum from NO* dissociatively adsorbed on W(110) has much broader peaks than the corresponding spectrum from NzO. This must be related to the different local stoichiometries at the surface. The peak in the NOz spectrum at -345 eV is probably spurious. because W has an Auger peak in this region and the background subtraction can produce a misleading result here. UPS spectra from clean W(110) and W(110) covered with NzO, NOz, and oxygen are given in fig. 5. The changes on adsorption which concern us most here are the development of broad peaks at 5-7 eV due to the adsorbate 2p levels. Adsorbed nitrogen on W(100) gives rise to a nitrogen 2p peak at -6 eV [6]. UPS spectra of oxygen adsorbed on W have been extensively investigated [6,45,46], and in all cases a peak or group of peaks due to the adsorbate oxygen is found at 5-7 eV. We have made fairly extensive studies of the 0 2p resonance in He I and

J.C. Fuggle, D. Menzel /Adsorption

10

of NZ, O2 and nitrogen oxides on W

N KLL Auger I

\_

t

yO.3OOK

NZ.lOOK

.

.

350

400 -

L60

460

500

520

5L0

KE(eV)-

Fig. 4. N KLL and 0 KLL XAES spectrafrom chemisorbed NzO, NOz, and oxygen on W(110) at 300 K. For comparison, spectra of adsorbed Nz and condensed N204 are also given. For chemisorbed layers N KLL difference spectra (i.c. covered minus clean) are plotted to reduce the effect of substrate Auger peaks at KE < 350 eV.

He II spectra, from oxygen adsorbed on W(l lo), as a function of oxygen coverage, sample temperature, and take-off angle [47]. We found that: (1)The 0 2p peak was often split into more than one peak with intensities not necessarily equal (see fig. 5B). (2) In general the main 0 2p peak in He II spectra was at -0.3 eV higher BE than in He I spectra. (3) In general the 0 2p peak intensity was not linearly related to coverage (see e.g. fig. 5B). (4) The shape of the 0 2p peak group changed as a function of take-off angle. (5) If the sample was heated after adsorption of oxygen then the shape of the 0 2p peaks changed drastically at temperatures well below the desorption temperature of oxygen, i.e. the shape of the 0 2p peaks is dependent on the structure of the adsorbate complex and not just the substrate or the concentration of oxygen (fig. 5B). In view of the complexities sketched, we should not try to draw conclusions and line shapes found when different gases are from the different intensities adsorbed. However, the UPS spectra observed after adsorption of NzO and NOz on W(110) are certainly consistent with dissociative adsorption at 300 K. If we compare the spectra in fig. SA with those obtained after molecular adsorption at

Ru(001) W{110) WjflO)

W(110)

W(110) MozN MO, 300 K

02 N20 FQ

W&O

NO2

a Taken from figure.

N2

W(110)

Pm-

Substrate

02

Adsorbate

--

495.4 t 0.3 493.9 t 0.8 493.4 i a,4

493.1 f 0.5

2

3 Sol,6 1+0.2 500.4 t I.0 499.1 ?r 0.6

---499.9 * 1

-.“---

358.6 ” 3 358.9 362

368.0 + 2 372.0 372

?374.6 1: 3 375.0 378

N KLL peak KE (ev) WXI resped to Ef _.---_ ..“.“I_l__.“I1”.““” ?36O‘O t 3 369.9 + 2 ?313.& *- 2

480.8 + 0.2 480.L k 2.0 478.8 + 2.0

479.4 i 0.S

I

0 KLL peak KE (eV) with respect to Ep l_*l_ --_.a

382.9 * 0.8 385 .o 383

384.0 i 0.5

515.0 t 0.2 513.6 t OS 512.9 f 0.2

513.2 f 0.5

4

Table 2 Comparison of 0 KLL and N KLL Auger peaks from dissociatively adsorbed nitrogen and oxygen

392.0

?389.O c 2

---

523.4 J: 1.6 521.1. * 2.5 520.7 f 2‘0

,._-,-- 521.0 + 1.5

5

.,- ..ll_^l__T

_x-_

--*I-

This work [341 [441 a

This work

i43f This work This work

[31

Ref.

----

B 2 0 2. Ez B %

a3. P % 2 c; (s h) 3 R, g.

B 0

2

$ $_

a?-

P

$

12

J.C. Fuggle, D. Menzel /Adsorption

-

, 10

.

. I

of N2, 0, and nitrogen oxides on W

BE kV)

.-Clean

.

5

-

a)

EF -

BE

(eV)-

,..,.I.... 10

5

EF

Fig. 5. He1 valence spectra of W(110) before and after coverage with NzO and NO2 (A) and oxygen (B), at 300 K. Left: direct spectra; right: difference spectra. The electron take-off angle was always about

16’ to the surface,

but was not identical

for the three experiments.

low temperature (figs. 7,8, 10 and 11) we can categorically state that the UPS data are nat consistent with molecular adsorption of NzO or NO2 at 300 K. 3.4. Adsorption of N20 at -100

K

Qur results suggest that the initial adsorption of N20 on W(l IO) at this temperature also leads to dissociation. However, we find that additional N,O can be adsorbed at temperatures below -130 K. Evidence for this can be seen in figure 6 where XPS spectra in the region of the 0 and N 1s peak are given. After low doses

13

535 +------

BE kv]

-

Fig. 6. XPS N 1s and 0 1s spectra from W(l10) at -100 K after the pretreatments given in the figure (the temporal sequence runs from bottom to top).

with NzO

of N20 (
Rm in MO+ Pa

N20

‘-3UEx

NzO

Heated to 160 K -30ExN10

-Clean I5

10

5

f,

e) df

c)

/

a)

‘15

E, A

B E

-

Y---y

IO

5

-.->,

b-0)

E,

ieV1

7. lieI (A) and lfeII (Bf UPS spectra and difference spectra after the pretreatments N&l given in the figure ftcmporat sequence a -+e, r).

I:&.

with

a difference between N20 molecules in contact with the metal, and condensed N20, Le. the small quantities of ‘*anomalous” N&I may be chemisorbed. molecular N&I can be recondensed on top of the dissociated N20 layer produced by heating to 160 K, as shown in figs. 6, 7.4 and 7B. The quantity of N20 which can be adsorbed is critically dependent on the sample temperature but is not very dependent on the N20 pressure in the system, and even with an ambient pressure of 5 X 10s5 Pa the observed attenuation of the XPS peaks from the substrate and from the dissociated N20 suggest that the heat of conderlsat~on of this thin layer represeat

J.C. Fuggle, D. Memel/

Adsorption

15

of N,, 02 and nitrogen oxides on W 2n ,... :

I

He I N20 gas 16

13

12

11 10

+-

9

6

7

6

BE leV) -

was taken from ref. [48] of the different reference levels in spectra from gaseous and condensed phases. The background, peak resolutions, and lines used to estimate the FWHM given in table 4 are shown. Fig. 8. UPS spectra

of gaseous

and adsorbed

N20.

The gas spectrum

and has been shifted by 6.8 eV to lower BE to allow for the effects

of molecular N20 on W(110) is larger than thenormal heat ofcondensation ofN20. We did attempt to observe the 0 KLL and N KLL Auger spectra during adsorption of N20 on W(110) at low temperature. During the initial adsorption the spectra were very similar to those of N20 adsorbed at room temperature. At high coverages no well resolved spectra could be obtained due to peak overlap.

Table 3 NzO valence

level binding

energies

Sample

(eV) from UPS (referred

to EF, unless indicated

otherwise)

Photons

2n

40

In

30

(He I, He II average)

6.1

9.6s

11.35

13.3

N,O ads. on W(110) surface pretreated and heated to 160 K

(He I, He II, He III average)

5.9

9.45

11.2

13.1

NzO gas (-6.9 was referenced

(He I)

6.0

9.5

11.3

13.2

(He I)

6.3

9.1

11.5

13.4

NzO ads. on W(110)

at 100 K

eV; original to VL (481)

NzO ads. on Ni (poly)[7]

16

J.C. Fuggie, D. Menzel /Adsorption

of N, 02 and nitrogen oxides on W

Table 4 Linewidths (FWHM) in gaseous and condensed NaO (given in eV) -----. Level ..._ _.._ 2n

40

In

30 --..-.~

NzO gas

He I instruments contribution NaO/W(llO) He1 Instrumenta contribution N~O/W(110) He Ii Instrumental contribution Approximate broadening due to condensation (assuming arithmetic addition of contribution)

0.04 _a

0.03 _

0.8 -0.15 1.05 -0.25 -0.6

0.7s -0.15 0.9s -0.25 -0.6

-0.6 _

-0.03 -

1.3 -0.15 1.4 -0.25 -0.4s

0.8 -0.15 1.0 -0.25 -0.6

---.-a Negligible in this experiment.

3.5. Adsorption ojf‘AJ0, at IO0 K XPS studies (fig. 9, table 1) showed that initial adsorption of NOz on W(ll0) at -100 IS gave rise to an 0 1s peak at 530.4 eV and an N 1s peak at about 397 eV which can be associated with NO* dissociated into N and 0 atoms (see also table 1). With increasing coverage up to a total dose of 15 Ex the N Is peak shifts by more than I eV to higher B.E. At the same time it broadens which suggests that there is more than one state of nitrogen on the surface. After total doses of NOz above 20 Ex, new 0 1s and N 1s peaks appear at 533.6 and 405.5 eV. These peaks are attributed to condensed NO*, probably in the form of Nz04. The 0 1s peaks from condensed, and from dissociated NOs are too close together to investigate the possibility of any intermediate state, This is not the case for the N 1s peaks. Here a very broad shoulder centred at 402 eV is observed after NOz doses of 30-40 Ex. This shoulder may be due to the presence of molecularly chemisorbed NOz but we could obtain no other evidence to support this conclusion. UPS spectra (fig. 10) are also consistent with the initial dissociative adsorption of NOz on W(110) at 100 K followed by condensation of NO* or Nz04. At doses up to -10 Ex NOz a broad peak appears at 5-7 eV which we attribute to the 0 and N 2p levels of N and 0 on the surface. At higher doses the UPS emission from the W levels slowly disappears and a complex collection of new peaks appears. There are some shifts in the peaks from the condensed layer as a function of thickness and at very high doses (-300 Ex) much of the structure is lost as the broadening increases. We cannot rule out the possibility that these effects are due to charging of the condensed layer in the photon beam. As shown in fig. 11 the UPS spectra at intermediate coverages, where broadening is not yet severe, show marked resemblance to the UPS spectra from Nz04 gas [49,50]. The resemblance to the spectrum of NO* gas [49--5 l] is less marked.

N Is

I

i

800 c/s

Heated to -9OOK

hated

to

-300K

-

LOEx

- 20Ex

- 15Ew * 1OEx

- IOEX

~ 5Er

-

5Ex

Clean 540

531,

528 -

LOB BE leV1

401

39L

-

Fig. 9. XPS N Is and 0 Is spectra from W(ll@ at -- 100 K after the pretreatments indicated in the figure, and after stepwise heating of the fully covered surface.

with NOz

Fig. 11 ako contains plots of valence band spectra from the condensed layer obtained with 2 1.2,40.8, and 1253.6 eV radiation, above each other so that they may be compared. We should remember that the total instrumental contribution to the resolution in XPS is -1.3 eV and thus far greater than the -0.2 eV instrumental broadening of the UPS spectra. Also the highest binding energy valence levels cannot be excited by 21.2 eV radiation. However, we can see that the intensity distribution in the photoelectron spectra is very dependent on photon energy, A detailed con~gura~o~a~ analysis of the various levels would be needed for further SIdySiS of theSe changes.

J.C. Fuggle, D. Menzel /Adsorption

oflv,,

02 and nitrogen

oxides on W

Fig. 10. UPS spectra run during the adsorption and condensation of NOa on W(110) at -100 K.

Table 5 Coverages of N and 0 at “saturation” on some metal surfaces (for nitrogen oxide reactions only Substrate

Adsorbate/adparticle

NZ N2

W(llO), 100 K W(1 lo), 300 K W(llO), reacted at 100 K and warmed to 300 K Poly. W, 300 K [25a] Poly. W, 300 K [S] Poly. Re, 300 K [ 25b] Poly. Rh, 300 K [ZSc] Poly. Ni, “77 K” [7] Poly. Ni, 300 K [ 71 Poly. Ni, reacted at “77 K” and warmed to 300 K [7]

15.4*3 -0 -0 2.6 _ 2.4 _ -0

N2O

N -0 -0 0 2.6 5.6 -

-0

N

0

Total

5.3 + 1.8 2.3 t 0.7 7.7 t 2.8

3.2 + 0.6 2.4 f 0.2 4.9 f 0.5

8.5 + 2.4 4.7 4 0.9 12.6 i 3.3

1.8 -0 -0 2.8 a 0 0

5.9 _

7.7

5.7 5 2.9 = 1.5 2.5

5.7 5 5.7 a 1.5 2.5

a Brundle does not distinguish between molecular and dissociative adsorption here. b This state may be molecularly chemisorbed.

19

15

5

10

+---

EF

BEleVI

-

Fig. Z1. Valence band spectra from a condensed layer of Na04, compared to a HeI spectrum of N&)4 gas (from ref. 1491, shifted by 7.2 eV to lower BE for alignment with the spectra of the solid). Ref. [49] does not give any comment on the relative intensities of parts (a) and (b) of the spectrum.

the coverage of dissociated particles are given); the units are 1O’a adatoms/m2

NO

NQ2

N

0

Total

N

0

Total

2.4 * 1.2 1.8 f 0.7 1.8 + 0.7

7.0 * I.4 8.4 t 1.5 12.6 _+x.3

9.4 r 2.6 10.2 f 2-2 14.4 f 2.0

3.4 f 1s 4.0 + 1.5 5.2 i 2

5.6%2 5.9 t I.5 8.8 * 2

9 -I 3.5 lo* 3 14 $4

5.6 -

8.4 _

1.4

4.0 b

4.6 b

8.6 b

4.2

4.6

8.8

-

-

_

2.9

4.2

7.1

3.1

6.9

10.0

20

J.C. Fuggle, D. Menzel /Adsorption

of N,, 02 and nitrogen oxides on W

3.6. Thermal desorption and conversion Most thermal desorption studies are done by heating the sample and observing the desorbed particles in the gas phase [52]. This method is very sensitive but has the disadvantage that the desorbed particles are often different from those which were on the surface. In our studies we can observe the species left on the surface after heating to a chosen temperature and an example of the results was given in fig. 9. There it can be seen that after adsorption of NOz at 100 K, heating to 300 K caused the XPS peaks due to molecular species to disappear. At the same time the 0 1s peak has increased by -80% (see also table 5) whilst the nitrogen peak at -397 eV has narrowed and lost -20% of its intensity. This indicates that on warming not all of the condensed Nz04 is desorbed; a complex reaction leading to increased concentration of atomic 0 species and resulting enrichment of oxygen in the surface layer takes place, while the nitrogen of the dissociating molecules leaves the surface. The narrowing of the atomic N peak is produced by the elimination of all species with BE above 398 eV, in particular the “broad shoulder” tentatively identified above as molecularly chemisorbed NOz. On heating the surface reacted with NO* from 300 to 900 K very little changes in either N 1s or 0 Is XPS peaks, but between -1000 and -1400 K the N 1s peak disappears indicating that the atomic nitrogen has desorbed. Our temperature control in this temperature region was poor which may have been the reason why we could not desorb all the nitrogen without any desorption of O2 oxygen. However, as illustrated in fig. 9, the decrease in the oxygen peak when all the nitrogen had desorbed was only -lo-15% and we believe that the nitrogen and oxygen desorption are mutually independent. The changes in XPS N 1s and 0 1s spectra of N,O covered surfaces on heating from 100 to 160 K were shown in fig. 6. There was no further change in the spectra on heating from 160 to 300 K or even 900 K. As seen in fig. 6 (see also table 5) the N and 0 1s peaks from dissociated species both increase by about 50% when the surface is freed from molecular NzO indicating that NzO decomposition is competing with desorption. Contrary to NOz, no marked preferential loss of Nz occurs here. The desorption behaviour of the N,O covered surface at high temperature is essentially the same as that of NO*, i.e. N and 0 desorb independently.

4. Discussion 4.1. Molecular versus dissociative adsorption The results reported above show clearly that all nitrogen oxides dissociate completely upon adsorption at 300 K on W(110). At 100 K the initial adsorption is also dissociative; at higher exposures condensed phases are formed. In the case of NO*/

f.C. Fuggle, D. Menzel/ Adsorption

ofiv,, O2 and nitrogen oxides on W

21

N204, a molecularly chemisorbed species may exist at lower temperatures. Nitrogen does not adsorb at all at 300 K; at 100 K a molecular layer is chemisorbed on W(l lo), in agreement with former conclusions [20]. The only other substrate on which similar work has been done over a comparable range of gases and temperatures is polycrystalline Ni. In the following we compare our findings to those of Brundle [7]. He showed that adsorption of NzO 2nd NOz on Ni at 300 K is dissociative, but N20 adsorption does not lead to any atomically adsorbed N. The situation at low temperature (80 K) was less clear. For NzO he shows a spectrum with N 1s peaks at 402 and 406 eV but none at 396-398 eV where atomic N should show up. This could be interpreted as due to either molecular N, or N,O; his UPS data show that N,_O is present. The 0 Is spectra indicate the presence of atomic 0 as well as condensed N20. For NOz he shows XPS spectra with N 1s and 0 1s peaks at only 400.8 and 531.9 eV, respectively, which we believe may be due to molecularly chemisorbed NOz, although UPS spectra do not show the expected structures. NO,, can be condensed on Ni at 80 K, as on W(110). N2 does not adsorb at all at 300 K on Ni, and only molecularly at 100 K. The main differences between W(110) and Ni (poly) are seen to be the probable absence of NOz dissociation on Ni at 80 K and the nonexistence of atomic N from N20 dissociation at 300 and 80 K. Both can possibly be explained by the weaker affinity of atomic N to Ni as compared to W. 4.2. Quantitative aspects of adsorption In table 5 we have given the coverages of dissociated nitrogen oxides at which the sticking coefficient drops dramatically (“saturation”), or condensation of molecular species becomes the dominant process. We have also given data for NO [53], for nitrogen, and published data for nitrogen oxides on other metals. Several points may be made here. (1) The coverage of nitrogen molecules on W(110) at 100 K was found to be (15 +_ 3) X 10” N-atoms rn-’ in Nz. Thus, the surface density of Nz molecules is seen to be about half that of surface W atoms (14.1 X 10” m-* j. (2) In no case does the number of adsorbed particles exceed the number of surface metal atoms by more than the experimental error. For all three nitrogen oxides the total number of ad-particles usually approaches the number of metal atoms when the sample is saturated at 100 K and warmed to 300 K. An exception to this is the adsorption of NzO on Ni [7] which gives very low coverages for dissociative adsorption. (3) In general dissociative adsorption of nitrogen oxides at 100 or 300 K gives “saturation” coverages of between l/2 and 3/4 adatoms/surface metal atom. From our measurements of peak intensities as a function of gas dose we can say that the initial sticking coefficient for dissociative adsorption of NzO, NOz, or NO on W( 1 IO) at 100 or 300 K is always between 0.5 and 1. (4) After dissociation of NzO, NOz, and NO on W(llO~, and N20 on Ni [7], the

22

J.C. Fuggle, D, Menzei /Adsorption

of N2, O2 and nitrogen

oxides on W

ratio of N : 0 in the surface region is lower than that in the reactant molecule. This suggests that some N leaves the surface as N2 during N after dissociation. The situation for NO2 reactions with Ni is less clear: even for NO2 condensed at 77 K Brundle gives an N : 0 ratio of 1 : 1.4 (and the same for reacted layers), although this should really be 1 : 2. 4.3. Line widths in conu’ensed NT0 Gadzuk has discussed the possible appearance of vibrational structure in UPS spectra from chemisorbed molecules [54] and concluded that such structure should be observed unless hole-dipole interactions lead to multiphonon excitation. This effect should depend on the size of the dipole within the neutral molecule. In UPS studies of condensed CO and Nz, Norton et al. [55] showed that the difference in linewidths for ah equivalent levels in CO and Nz was 0.1 to 0.2 eV which could be understood in terms of such a mechanism. However, they also found that all the spectral features of the condensed Nz and CO were broadened by 0.6 eV on condensation and could find no really satisfactory explanation for this. They could only suggest that a combination ofincreased contributions to the linewidth by interatomic Auger transitions, electron hole pair shake-up, and effects of site heterogeneity might explain this broadening. It is of interest to note that the broaden~g of all the N20 valence levels on condensation is also about 0.6 eV. Further, by collating XPS data from gas phase N20 [40] with data we have obtained in computer studies of line shapes in adsorbates [56] and with the present results we find that the broadening of the XPS lines from N,O on condensation must be about 1 eV, i.e. of the same order of magnitude as that found in the valence levels. This fact rules out the possibiIity that interatomic Auger effects of the type suggested by Matthew and Kominos (571 play a significant role in the broadening. We also wish to draw attention to the greater linewidth of the NzO In peak in both gas and condensed phase. In the gas phase this peak is observed as a band of vibrational levels. From the fact that the difference between In and 3u,40 or 2rr linewidths is SO similar in gas and condensed NzO we conclude that the effective contribution to total linew~dth from the NzO vibrations is similar in both cases. 4.4. Chemical shifts arId fingerprinting In previous studies we have found X-ray excited 0 KLL Auger spectra to give the most characteristic and unalnbiguous fingerpril~ts for adsorbates [3,43]. These, together with the N KLL spectra, were again useful here for identifying dissociatively adsorbed NO, and N,O, but for conditions where molecular adsorption was thought possible two or more species produced overlapping spectra in which the details could not readily be distinguished. This will be a general disadvantage of XAES; it will only be really useful in identifying majority adsorbate species. UPS is somewhat more helpful in this respect because the intensity of the 0 and

J.C. Fuggle, D. Menzel /Adsorption of N,, O2 and nitrogen oxides on W

23

N 2p adsorbate resonances is rather low by comparison with molecular species, especially for NzO which has a very characteristic UPS spectrum. The UPS spectra from condensed molecular NO1 or Nz04 are rather complicated and do not show sharp, characteristic peaks. This may be partly responsible for the fact that molecularly chemisorbed NO,? has not yet been observed. XPS core peaks provide the best possibility for tagging different adsorbate species present in comparable amounts and following their interconversions. The separability of core peaks is strongly influenced by the range of chemical shifts of the core peak concerned. In this respect the N 1s peaks are clearly superior to the 0 1s peaks. The range of C 1s shifts in adsorbates also appears to be larger than that of 0 1s [3], so that 0 Is seems to be particularly unsuitable. However, the photoelectric cross sections vary as 0 1s > N 1s > C 1s which partly offsets the effect of energy spreads. The disadvantage of XPS for fingerprinting is the possible ambiguity of core peaks; therefore it should always be coupled with UPS and/or XAES wherever possible 133.

5. Summary The combination of XPS, UPS, and XAES has been used to investigate the adsorption of N2, NzO, and NOz/Nz04 on W(110) at 300 and 100 K, and the behaviour of the adsorbate layers with heating. Some new results for O2 adsorption on W(l lo), which are relevant here for comparison, have also been obtained. The main conclusions are: (1) At 300 K, NzO and NOz adsorption is completely dissociative, with partial loss of N. Nz does not adsorb at all. (2) At 100 K, the initial adsorption of N,O and NOz/N,04 is dissociative. At higher exposures condensed layers of NzO and N204 are formed; a molecularly adsorbed NOz species may also be formed. The NzO condensation layer is restricted to a few monolayers which are more strongly bound than in the normal condensed phase. The condensed NO* layer grows infinitely at 100 K. Nz forms a molecularly adsorbed monolayer (about 0.5-0.6 N*/W) at 100 K. (3) Heating of condensed NzO and NO* layers leads to further dissociation, as well as desorption. For NzO, the N and 0 increases have about the same proportion as in the dissociated starting layer; for NO?, only the oxygen content of the surface layer is increased, while no significant increase of N is found as a result of additional NO* adsorption and heating.

Acknowledgements We thank E. Umbach for help and discussions, and W. Back for technical assistance. This work was supported by Deutsche Forschungsgemeinschaft within SFB 128.

24

J.C. Fuggle, D. Menzel /Adsorption

ofN,,

0,

and nitrogen oxides on W

References [l] R. Gomer, Japan. J. Appl. Phys., Suppl. 2, Part 2 (1974) 213. [2] T. Engel, H. Niehus and E. Bauer, Surface Sci. 52 (1975) 237. [3] See, e.g., E. Umbach, J.C. Fuggle and D. Menzel, J. Electron Spectrosc. 10 (1977) 15, and references therein. [4] See, e.g., C. Steinbrtichel and R. Gomer, Surface Sci. 67 (1977) 21, and references therein. [5] T.E. Madey, J.T. Yates and NE. Erickson, Surface Sci. 43 (1974) 602. (6) B. Feuerbacher, Surface Sci. 47 (1975) 115. [7] C.R. Brundle, J. Vacuum Sci. Technol. 13 (1976) 301. [8] D.A. King, Surface Sci. 9 (1968) 375. [9] R. Klein and A. Shih, Surface Sci. 69 (1977) 403. [lo] B.E. Nieuwenhuys, D.Th. Meyer and W.M.H. Sachtler, Surface Sci. 40 (1973) 125. [ll] J.T. Yates and T.E. Madey, J. Chem. Phys. 51 (1969) 334. [12] R.W. Joyner, J. Rickmann and M.W. Roberts, J. Chem. Sot. Faraday 170 (1974) 1825. [13] M. Scheer and J.D. Mc.Kinley, Surface Sci. 5 (1966) 332. [ 141 K. Kishi and M.W. Roberts, Surface Sci. 62 (1977) 252. [15] S.M. Ko and L.D. Schmidt, Surface Sci. 42 (1974) 508. [ 161 D.L. Adams and L.H. Germer, Surface Sci. 27 (1971) 21, and references therein. [ 171 A. van Oostrom, J. Chem. Phys. 47 (1967) 761. [ 181 S.P. Sinph-Boparai, M. Bowker and D.A. King, Surface Sci. 53 (1975) 55, and references therein. [19] K. Besocke, Verhandl. Deut. Physik. Ges. VI, 13 (1978) 579, and private communication. 1201 J.T. Yates, Jr., R. Klein and T.E. Madey, Surface Sci. 58 (1976) 469. [21] J.C. Fuggle and D. Menzel, Vakuum-Technik 27 (1978) 130; J.C. Fuggle and D. Menzel, in: Proc. 7th IVC and 3rd ICSS (Vienna, 1977) p. 1003. [22] H. Conrad, G. Ertl, J. Ktippers and E.E. Latta, Surface Sci. 50 (1975) 296. [23] K. Kishi and M.W. Roberts, Proc. Roy. Sot. (London) A352 (1976) 289. [24] G.L. Price, B.A. Sexton and B.C. Baker, Surface Sci. 60 (1976) 506. [25] (a) R.P.11. Gasscr and C.P. Lawrence, Surface Sci. 10 (1968) 91; (b) R.P.H. Gasser and D.E. Holt, Surface Sci. 52 (1975) 475; (c) R.P.H. Gasscr and C.J. Marsay, Surface Sci. 20 (1970) 116. [26] A.M. Bradshaw and D. Menzel, Vakuum Technik 24 (1975) 15, and references therein. [27] 1). Menzel and J.C. I:uggle, Surface Sci. 74 (1978) 321. [28] C. Wang and R. Gomer, Surface Sci. 74 (1978) 389; C. Wang and R. Gomer, in: Proc. 7th IVC and 3rd ICSS (Vienna, 1977) p. 1155. [29] R. Butz and H. Wagner, Surface Sci. 63 (1977) 448. [30] K. Besocke and S. Berger, in: Proc. 7th IVC and 3rd ICSS (Vienna, 1977) p. 893. [3l] J.C. Fuggle, E. Umbach, D. Menzel, K. Wandelt and C.R. Brundle, Solid State Commun. 27 (1978) 65, and references therein. [32] See e.g. W.E. Moddeman, T.A. Carlson, M.O. Krause, BP. Pullen, W.E. Bull and G.K. Schweitzer, J. Chem. Phys. 55 (1971) 2317. 1331 K. Sicgbahn, J. Electron Spectrosc. 5 (1974) 42. 1341 K. Kunimori, T. Kawai, T. Konow, T. Onishi and K. Tamaru, Surface Sci. 54 (1976) 526. [35] TX. Madcy and J.T. Yates, Jr., J. Chem. Phys. 44 (1976) 1675. [36] M.A. Chcstcrs, B.J. Hopkins and R.I. Winton, Surface Sci. 59 (1976) 46. [37] G. Johansson, J. fledman, A. Berndtsson, M. Klasson and R. Nilsson, J. Electron Spectrosc. 2 (1973) 295. [ 38) T.D. Thomas and R.W. Sham,1 J. Electron. Spcctrosc. 5 (1974) 1081. [39] See, c.g., H. Binder and P. Sellmann, Angcw. Chem. 85 (1973) 1720; B. I‘olkcsson, Acta Chem. Stand. 27 (1973) 287; J. Chatt et al., J.C.S. Dalton (1975) 2392.

J.C. Fuggle, D. Menzel /Adsorption

of N,, O2 and nitrogen oxides on W

25

[40] K. Siegbahn, J. Electron Spectrosc. 5 (1974) 42. [41] Handbook of Spectroscopy, Ed. J.W. Robinson (Chem. Rubber Co., 1974), cited by C.R. Brundle in ref. [ 71. [42] J.C. Fuggle and D. Menzel, Surface Sci. 53 (1975) 21. [43] J.C. Fuggle, E. Umbach, P. Feulner and D. Menzel, Surface Sci. 64 (1977) 69. [44] T. Kawai et al., Phys. Rev. Letters 33 (1974) 533. [45] A.M. Bradshaw, M. Steinkilberg and D. Menzel, Japan. J. Appl. Phys., Suppl. 2, Pt. 2 (1974) 841. [46] E.W. Plummer, B.J. Waclawski and T.V. Vorburger, in: Progress in Surface Membrane Science, Ed. S.G. Davison (Pergamon, 1976) p. 149; B.J. Waclawski, T.V. Vorburger and R.J. Stein, J. Vacuum Sci. Technol. 12 (1975) 301. [47] J.C. Fuggle and D. Menzel, unpublished results. [48] D.W. Turner, C. Baker, A.P. Baker and C.R. Brundle, Molecular Photoelectron Spectroscopy (Wiley, London, 1971) p. 89 ff. [49] D.A. Ames and D.W. Turner, Proc. Roy. Sot. (London) A348 (1976) 1975. [SO] D.C. Frost, C.A. Mc.Dowell and N.P.C. Westwood, J. Electron Spectrosc. 10 (1977) 293. [Sl] C.R. Brundle, D. Neumann, W.C. Price, D. Evans, A.W. Potts and D.G. Streets, J. Chem. Phys. 53 (1970) 705. [52] See, e.g., D. Menzel, in: Interaction on Metal Surfaces, Ed. R. Gamer (Springer, Berlin, 1975), p. 101. [53] We thank E. Umbach and R. Masel for the permission to publish these figures in this context. [54] J.W. Gadzuk, Phys. Rev. B14 (1976) 5458. [55] P.R. Norton, R.L. Tapping, H.P. Broida, J.W. Gadzuk and B.J. Waclawski, Chem. Phys. Letters 53 (1978) 465. [56] J.C. Fuggle, P. Steiner and D. Menzel, unpublished. [57] J.A.D. Matthew and Y. Komninos, Surface Sci. 53 (1975) 716. [58] R.I. Masel, E. Umbach, J.C. Fuggle and D. Menzel, Surface Sci. 79 (1979) 26.