Adsorption of carbon monoxide carbon dioxide on clean and cesium

Publication Date: June 1989. ACS Legacy Archive. Cite this:J. Phys. Chem. ... ACS Catalysis 2011 1 (4), 365-384. Abstract | Full Text HTML | PDF | PDF...
0 downloads 0 Views 1MB Size
5238

J . Phys. Chem. 1989, 93, 5238-5248

Adsorption of CO and CO, on Clean and Cesium-Covered C u ( l l 0 ) Jose A. Rodriguez, William D. Clendening, and Charles T. Campbell*,t Chemistry Department, Indiana University, Bloomington, Indiana 47405 (Received: October 20, 1988; In Final Form: January 30, 1989)

The interactions of Cs with Cu( 1 IO) and the coadsorption of CO and C 0 2 with Cs on Cu(ll0) have been studied with X-ray photoelectron spectroscopy, Auger electron spectroscopy, work function measurements, and thermal desorption mass spectroscopy (TDS). Cesium adsorbs in relatively cationic form at low coverages, depolarizing to relatively neutral, bulklike Cs by one close-packed monolayer (ecs* = 1.0). At low &,*, indirect interactions stabilize CO, by -15% relative to its desorption temperature (-220 K) from clean Cu(l10). For eCs* > 0.5, a C0,-Cs, complex appears, which decomposes at 400-500 K with simultaneous Cs and CO evolution in TDS. For a fixed CO dose, the concentration of this complex increases with Bcs*, reaches a maximum at Bcs* 1 . I , and decreases rapidly to zero at BCs* = 2.0. No CO will adsorb on Cs multilayers at 1 I O K . No C 0 2 adsorption could be detected on clean Cu( 110) at 110 K. A cesium/carbonate surface complex is formed already from the lowest Cs coverages and all higher Cs coverages below 180 K via 2C02, + Cs, CO, = Cs.CO,,, (perhaps through an 0, or oxalate intermediate). Weakly adsorbed C 0 2 is also seen when Cs, is present, desorbing at 130 K. The Cs.C03,, complex decomposes in a complex TDS spectrum at 5OC-700 K via CPCO,,~ Cs, + COzg + 0,. Cs, then desorbs at high coverages simultaneous to C 0 2 evolution. The implications of these results to Cs promotion of catalytic water-gas shift and methanol synthesis over Cu are discussed.

-

I. Introduction The objective of the present work is to investigate the effects of adsorbed Cs on the adsorption and chemistry of CO and C 0 2 on Cu( 1 IO). This study is motivated by the fact that both C 0 2 and C O are thought to play major mechanistic roles in catalytic C H 3 0 H + higher alcohols) and alcohol synthesis (CO + H, in the water-gas shift reaction ( C O H 2 0 C 0 2 + H2) over Cu and Cu/ZnO catalysts.1,2 In both these reactions, Cs is known to be a promoter of activity or ~ e l e c t i v i t y . ~In, ~addition, there is some controversy in the literature concerning the rate of CO, dissociation at Cu surfaces and its role in the catalytic reaction mechanh" The present study on C 0 2 / C s / C u ( 110) adds to the few studies (C02/K/Pd(100)8a,band CO,/K/Rh(lI l)8c*d)in which the coadsorption of alkali-metal atoms and C 0 2 on metal surfaces has been investigated with the modern techniques of surface science. Quantum chemical calculations for adsorption of C 0 6 and CO? on Cu surfaces suggest that the heat of adsorption of these molecules should increase when they are coadsorbed with alkali metals. To our knowledge, no experimental work has appeared studying the coadsorption of C O and Cs or of C 0 2 and Cs on Cu surfaces. The adsorption of C O on Cu( 110) has been investigated by means of thermal desorption mass spectroscopy (TDS),9 lowenergy electron diffraction (LEED)?bJo high-resolution electron energy loss spectroscopy (HREELS)," work function measurements,1° electron energy loss spectroscopy (EELS),I2 inverse photoemission spectroscopy (IPS),], and infrared reflectance absorption spectroscopy (IRAS).98 Carbon monoxide is adsorbed on Cu( 1 10) molecularly, with its C end toward the surface. It desorbs without any dissociation in the temperature range between 200 and 220 K.839LEED,14 ultraviolet photoelectron spectroscopy (UPS),15and work function measurement^'^^'^ have been employed to study Cs adsorption on Cu( 1 10) surfaces. On these surfaces Cs is adsorbed with substantial repulsive lateral interactions, forming quasi-hexagonal structures at 80 K and causing reconstruction of the surface at temperatures above 150 K.14 The interaction between C 0 2 and Cu surfaces has been studied by Auger TDS,16 UPS,I7X-ray photoelectron spectroscopy (XPS),17,18 electron spectroscopy (AES),I9 and ellip~ometry.'~ Carbon dioxide is thought to be adsorbed on polycrystalline Cu both in a physisorbed form (AH,,, = 4.3 kcal/molI6) and in an anionic (C02-) chemisorbed state (AHad,< 14.3 kcal/molI8). In the present work, we investigate the effect of added Cs upon the adsorption and chemistry of CO and C 0 2 on Cu(l10) using TDS, XPS, and AES. The paper is organized as follows: The

-

+

-

-

-

next section summarizes the experimental procedures. The third section presents the results obtained for the Cs/Cu( 1 lo), C O / Cs/Cu( 1lo), and C02/Cs/Cu( 110) systems, in that order. The fourth section contains a discussion of these results and, in their light, the nature of the Cs promoter and its role in catalytic reactions involving C O and C 0 2 over Cu. 11. Experimental Section

The ultrahigh-vacuum apparatus and the procedure employed for the preparation and cleaning of the Cu( 110) crystal are described elsewhere.20 The apparatus contained capabilities for TDS, XPS, AES, work function measurements, and sputter cleaning with Ar' ions.

Newsome, D. S. Cural. Rev.-Sci. Eng. 1980, 21, 275. Klier, K. Ado. Caral. 1982, 31, 243. Campbell, C. T.; Koel, B. E. Surf. Sci. 1987, 186, 393. (a) Bogdan, C. E.; Nunan, J. G.; Santiesteban, J. G.; Herman, R. G.; Klier, K. Srud. Surf. Sci. Carol. 1987, 38, 745. (b) Nunan, J.; Klier, K.; Young, C.-W.; Himmelfarb, P. B.; Herman, R. G. J . Chem. Soc., Chem. Commun. 1986, 193. (5) (a) Campbell, C. T. Appl. Catal. 1987.32.367. (b) Chinchen, G. C.; Spencer, M. S.; Waugh, K. C.; Whan, D. A. Appl. Caral. 1987, 32, 371. (6) Rodriguez, J. A.; Campbell, C. T. J. Phys. Chem. 1987, 91, 2161. (7) Rodriguez, J. A. Lungmuir 1988, 4 , 1006. (8) (a) Solymosi, F.; Berkb, A. J . Cutal. 1986, 101, 458. (b) Berkb, A,; Solymosi, F. Surf. Sci. 1986, 171, L498. (c) Solymosi, F.; Bugyi, L. J. Chem. Soc., Faruduy Truns. 1 1987, 83, 2015. (d) Kiss, J.; REvEsz, K.; Solymosi, F. Surf. Sci. 1988, 207, 36. (9) (a) Woodruff, D. P.; Hayden, B. E.; Prince, K.; Bradshaw, A. M. Surf. Sci. 1982, 123, 397. (b) Harendt, C.; Goschnick, J.; Hirschwald, W. Surf. Sci. 1985, 152J153, 453. (IO) Horn, K.; Hussain, M.; Pritchard, J. Surf. Sci. 1977, 63, 244. ( I 1) Lackey, D.; Surman, M.; Jacobs, S.; Grider, D.; King, D. A. Surf. Sci. 1985, 152/153, 513. (12) Spitzer, A.; Liith, H. Surf. Sci. 1981, 102, 29. (13) (a) Rogozik, J.; Scheidt, H.; Dose, V.; Prince, K. C.; Bradshaw, A. M. Surf.Sci. 1984, 145, L481. (b) Gumhalter, B. Surf. Sci. 1985, 157, L355. (14) Fan, W. C.; Ignatiev, A. Phys. Reu. E 1988, 38(1), 366. ( 1 5) Woratschek, B.; Sesselmann, W.; Kuppers, J.; Ertl, G.; Haberland, H . J . Chem. Phys. 1987, 86, 2411. (16) Hadden, R. A,; Vandervell, H. D.; Waugh, K. C.; Webb, G. Caral. Lett. 1988, 1 , 27. (17) Norton, P. R.; Tapping, R. L. Chem. Phys. Lett. 1976, 38, 207. (18) Copperthwaite, R. G.; Davies, P. R.; Morris, M. A,; Roberts, M. W.; Ryder, R. A. Catal. Lett. 1988, 1, 1 1 . (19) Habraken, F. H. P. M.; Kieffer, E.; Bootsma, G. A. Surf. Sci. 1979, (1) (2) (3) (4)

83, 45.

'Alfred P Sloan Research Fellow

0022-3654/89/2093-5238$0 1 .SO/O

(20) Clendening, W. D.; Campbell, C. T. J. Chem. Phys., in press.

0 1989 American Chemical Society

The Journal of Physical Chemistry, Vol. 93, No. 13, 1989 5239

Adsorption of C O and CO, on Cs/Cu( 1 IO) 0 ; s

0.25

0.75

0.50

1.0

1.25

TDS

Cs

from

Cs/Cu(llO)

2

z

. z

m

U c

4

.

0 +I

5

5 w

3

z

0

2 Y)

4

B ”

az CI

250

i 3 3

350

450

550

650

750

050

TEMPERATURE/ K

U

I

I

0.02

0.04

Cs(MNN)/Cu(LVV)

AES

0.06

0.08

Figure 2. Representative thermal desorption spectra showing the evolution of cesium at various Cs coverage. Cesium was vapor deposited on the surface at 110 K.

-

0.10

PEAK-TO-PEAK R A T I O

-

Figure 1. Change in the work function of the Cs/Cu(llO) surface (at 1 IO K) as a function of Cs/Cu AES ratio. The upper coverage scale was determined from the Cs(MNN)/Cu(LVV) A.ES ratio, such that a ratio of 0.08 reflects a close-packed monolayer of Cs on the surface (see

text). The carbon dioxide employed in the experiments was from Air Products (minimum purity: 99.995%) and the C O from Linde (minimum purity: 99.97%). High purity of the gases was proven in situ by mass spectroscopy. C O and CO, were dosed onto the surfaces at a sample temperature of 110 K. For dosing C O and C 0 2 ,we used a cosine-emitting pinhole doser such as described previously.2’ This gave enhancement factors of -3.5 for C O and 15 for CO,, measured by comparison to dosing only from the background pressure rise. These enhancement factors were taken into consideration in reporting the exposures here. In reporting exposures here, we used the ion gauge reading directly without correction for any possible sensitivity factor differences for C O and CO, relative to N,. Cesium was dosed to the front surface (sample temperature 110 K) by using a vapor deposition source similar to that described previously,21where now a SAES Cs zeolite holder was resistively heated to generate Cs vapor. After each adsorptiondesorption experiment, the sample was routinely cleaned by sputtering at -750 K followed by annealing briefly to -825 K. The cleanliness of the surface was checked by AES and XPS. The TDS experiments were performed using a feedback circuit in which the heating rate was held constant at 0.169 mV/s on the chromel-alumel thermocouple. This leads to a temperature rate of 6.5 to 4.25 K / s between 120 and 250 K and to a constant value of 4.25 K / s above 250 K . Detection in TDS was with a lineof-sight mass spectrometer set at -45O from the surface normal. The Cs(3d) and Cu(2p) XPS spectra of section 111.1 were recorded with Mg K a radiation and a pass energy of 100 eV, which gave a Cu(2p3/,) peak for clean C u ( l l 0 ) with a 1.7-eV fwhm (full width at half-maximum). The Cu(2p), Cs(3d), O(ls), and C( Is) spectra of sections 111.2 and 111.3 were recorded with AI K a radiation and a pass energy of 100 eV, which gave a Cu(2p3,,) peak for clean Cu(1 I O ) of 1.8-eV fwhm. (These Cu fwhm can be taken as the overall instrumental resolutions.) Our XPS binding energy scale was calibrated by using the Cu(2p3,,) peak and the Cu(L3VV) XAES transition for clean Cu(1 IO), which were set at binding energies of 932.422and 335.0 eV,22

-

-

-

(21) Campbell, J. M.; Seimanides, S. G.;Campbell, C. T. J . Phys. Chem. 1989, 93, 815. (22) Wagner, C. D.; Riggs, W. M.; Davis, L. E.; Moulder, J. F.; Muilenberg, G.E. Handbook of X-ray Photoelectron Spectroscopy; Perkin-Elmer: Eden Prairie, MN, 1978.

respectively. Detection was normal to the surface in AES and XPS. For AES, a defocused electron beam of 2650-eV primary energy incident at -45O from surface normal was used. Its current was -2 X IO” A, and an effective modulation voltage of 4.5 eV was used. To monitor the changes in work function for the Cs/Cu( 1 10) surfaces, we followed the variations in the onset of secondary electrons in the Auger spectra, at -60 times better resolution and 20-fold less current. A bias voltage of -35.00 V was applied to the sample during the work function measurements. Absolute coverages (8) are reported here relative to the Cu( 1 10) surface atom density (Le., 1.085 X I O l 5 cm-, corresponds to 0 = 1 .O).

111. Results III.1. Cs Adsorption on Cu(220). Figure 1 shows the work function variation (A4) of the Cs/Cu( 110) surface (at 1 IO K) as a function of Cs coverage. In the present work the relative Cs coverage was assumed to be proportional to the Cs(565 eV)/ Cu(920 eV) AES peak-to-peak ratio. (This assumption is proven for submonolayer coverages below.) This Auger ratio was scaled to absolute Cs coverage by inspection of the work function data in Figure 1. In this figure, the shape of the curve is identical with those reported for Cs adsorption on Cu( and Cu( 11 For those surfaces an initial large decrease in A 4 to a minimum at about one-half monolayer, reversing to a relatively rapid increase until one full monolayer of Cs is adsorbed, above which the work function is relatively constant. On the basis of the point where the A 4 curve first reaches a saturation level, we assign a Cs(565 eV)/Cu(920 eV) AES ratio of 0.08 to a full monolayer of Cs on Cu( 110). The Cs coverages (e,,*) reported here will be given in relative units, such that Bo* = 1.O corresponds to one monolayer. The close-packed monolayer of Cs on Cu( 1 IO) has an absolute coverage of 0.48 Cs atom per Cu surface atom for the unreconstructed surface at 80 K based on LEED observations.14 From Figure 1, the maximum reduction in work function is 3.07 eV. This value agrees well with those reported for Cs adsorbed on C u ( l l l ) (3.4 eV24), Cu(100) (2.92 eV23), and other metal surfaces (-3 eV25326). Since the work function for clean Cu(1 IO) has been reported to be 4.48 eV,,’ the minimum work function for the Cs/Cu( 110) surface is then 1.41 eV, whereas the value for a close-packed Cs layer is 2.1 1 eV. The last value is very close to the work function of bulk polycrystalline Cs (2.14 eV2’). The large decrease in work function at low Cs coverages observed in

-

(23) Papageorgopoulos, C. A. Phys. Reo. B 1982, 25, 3740. (24) Lindgren, S. A.; Wallden, L. Phys. Rev. B 1980, 22, 5967. (25) Gerlach, R L.; Rhodin, T. N. Surf. Sci. 1970, 19, 403. (26) Bonzel, H . P. Surf. Sci. Rep. 1988, 8, 43. (27) Kittel, C. Introduction to Solid State Physics, 6th ed.; Wiley: New York, 1986; p 537

5240

1

The Journal of Physical Chemistry, Vol. 93, No. 13, 1989 Cs(3d) XPS

:s/cu(llo) h

BINDING

ENERGY/ eV

Figure 3. Cs(3d) XPS spectra for different cesium coverages on Cu(ll0). The spectra were taken after vapor depositing Cs on the clean surface at -I10 K.

Figure 1 is indicative of a large transfer of electron density from cesium to the copper s u r f a ~ e As . ~ the ~ ~Cs ~ ~coverage increases, there is depolarization between the cesium dipoles, and the Cs adlayer is gradually neutralized. At low coverages the adsorbed Cs can be regarded as quite ionic, whereas at high coverage its behavior is more like that of neutral metal atom^.^^^*^ The Cs thermal desorption spectra of Figure 2 show that Cs desorbs in a very broad spectrum, beginning at -250 K. The Cs was not completely desorbed by 950 K, the maximum heating temperature (as judged from work function and AES data). The behavior in Figure 2 is reminiscent of Cs desorption from Ni25 and Ag2*surfaces, where desorption of the first monolayer extends from 320 to 1100 K in a structured spectrum. The desorption of submonolayer alkali-metal coverages from transition metals is generally very broad, reflecting a strong decrease in the heat of adsorption with coverage due to repulsive lateral interactions in the a d l a ~ e r . ~It ~should . ~ ~ be mentioned at this point that adsorbed Cs induces a surface reconstruction of Cu( 1 IO) at temperatures 2150 K.14 LEED experiment^'^ showed that the local reconstructing phase transforms to either a (1 X3) or a (1 X2) phase at BCs > 0.08 as temperature is increased above 240 K. In the spectra of Figure 2, a peak at 300 K first appeared at 8,,* E 0.76. It then grew in intensity without saturation, but its leading edge was independent of coverage beyond approximately two monolayers. It is attributed to desorption of multilayer Cs atoms, which were bonded on the surface only to other Cs atoms. A leading-edge analysis of this peak (assuming zero-order kinetics for multilayer desorption) yields an activation energy for desorption of 18.5 kcal/mol, which is close to the sublimation energy of bulk Cs (18.7 k ~ a l / m o l ~ ~The ) . submonolayer TDS features (above I . 1. The -320 K) continued to grow in intensity until Ocs* fact that the multilayer state starts to be populated at considerably lower coverages shows that the first monolayer is not perfectly completed before some atoms begin to build in the second layer. 1.6 and then flashing this surface to 310 By depositing &,* K to remove the multilayer, we found an average Cs/Cu AES ratio of 0.084 for the saturated monolayer. This is within 5% of the value used in our coverage calibration above. Figure 3 displays the Cs(3d) XPS spectra of the C u ( l l 0 ) surface following various Cs doses. The spectra were taken immediately after dosing Cs onto the clean &face at 1 I O K. The coverages in Figure 3 were determined by using the corresponding

-

-

(28) Campbell, C. T.J . Phys. Chem. 1985, 89, 5789. (29) CRC Handbook of Chemistry and Physics, 49th ed.; CRC Press: Boca Raton, FL. 1968-1969.

Rodriguez et al. Cs(565 eV)/Cu(920 eV) AES peak-to-peak ratios (see above). At submonolayer Cs coverages, the integrated Cs(3d):Cu(2p3,,) XPS intensity ratio was found to be nearly proportional to this Cs/Cu AES ratio, so that both ratios have been used in the paper for coverage calibration. We found that a single monolayer of Cs (Bcs* = 1.0) attenuates the Cu(2p) signal by -19% with respect to its value from a clean Cu( 1 I O ) surface and gives a Cs(3ds/,):Cu(2p3/,) ratio of 0.1 5 with Mg K a radiation here. (This value was 0.13 when using AI K a radiation.) The Cs(3dsj2) peaks in Figure 3 always appear in a narrow range between 725.4 (Oo* = 0.9-2.1) and 725.7 eV (eo* 50.15). These binding energies (BE’S) are much larger than the typical values reported for bulk compounds with Cs in a formal oxidation state of + I : -723.5 to 724 eV.” This result is consistent with experimental observations for cesium coadsorbed with oxygen on metal surface^,^' which show that the Cs(3d) XPS peaks appear surprisingly at larger binding energy for pure Cs than after oxygen dosing to the Cs adlayer. (A similar shift can be observed comparing the K(2p) XPS spectra of metallic and oxidized potassium films3’ and pure and oxidized potassium on Fe( 1 Comparison of the curves in Figure 3 shows that there is a decrease of -0.3 eV in the Cs(3d) binding energy of adsorbed Cs as Ocs* increases in the first monolayer. A similar fact has been observed33 for the K(2p3l2) peak of potassium adsorbed on Fe(ll0). The direction of these shifts is consistent with the notion of a gradual neutralization of the (cationic) adsorbed alkali-metal atoms with increasing ~overage.*s~*~ For large coverages of Cs the spectra of Figure 3 are characterized by the presence of unresolved satellites that extend to 2-6-eV higher binding energy (lower kinetic energy) than the main peaks. These satellites can be attributed at least partially to simple and multiple plasmon losses associated with the adsorbed Cs.24,34 In experiments with thick Cs films,34first-order bulk (volume) and surface plasmons have been identified with loss energies of 2.96 and 2.1 1 eV, respectively. A plasmon loss of 1.5 eV has been observed in energy loss experiments for a monolayer of Cs adsorbed on Cu( 1 1 Strong plasmon losses are usually detected in energy loss experiments for adsorption of alkali-metal monolayers on transition metal^.^^^^^ In those cases the energy of the losses varies as a function of alkali-metal coverages.24 The results of energy loss experiments on Na/Ni(100)35 and Cs/Cu( 1 I show that the relative intensity of the plasmon loss increases dramatically when the alkali-metal coverage is increased through one monolayer. A similar behavior is observed for the satellites in Figure 3. Other contributions to the skewed line shapes of alkali-metal XPS spectra have also been discussed.36 111.2. Coadsorption of CO and Cs on C u ( l 1 0 ) . Here we investigate the effects of adsorbed Cs on the adsorption and chemistry of C O on Cu(l IO). We first vapor-deposited Cs on the surface at 1IO K, then dosed C O at the same temperature, and finally observed the changes in the TDS, XPS, and AES spectra as the concentration of the cesium overlayer was increased. A series of thermal desorption spectra for C O adsorbed on Cs/Cu(l I O ) are shown in Figure 4. This figure displays the evolution of C O ( m / e = 28) and Cs ( m / e = 133) for different Cs precoverages and a fixed C O exposure of 1.3 langmuirs at 1 I O K. With no cesium present, C O desorbs from Cu( I I O ) In a narrow peak ( a )centered at -220 K. This temperature is in good agreement with results presented in the literature for the ~ can be seen in Figure 4, the width of CO/Cu( 1 IO) ~ y s t e m .As the a C O desorption peak at 220 K increases with increasing Cs coverage. A shoulder between 240 and 300 K is clearly seen for 0.90 IOCs* 5 1.28, although intensity in this region starts to grow from the lowest Cs coverage. For Bcr* > 0.53 a weak, hightemperature CO desorption feature (@)appears near 425 K. The

-

-

(30) Ayyoob, M.; Hegde, M. S. Surf. Sci. 1983, 133, 516. (31) Petersson, L. G.;Karlsson, S. E. Phys. Scr. 1977, 16, 425. (32) Pirug, G.: BrodCn, G.; Bonzel, H. PI Surf. Sci. 1980, 94, 323 (33) Broden, G.; Bonzel, H. P. Surf. Sci. 1979, 84, 106. (34) Hartley, B. M. Phys. Left. A 1967, 24, 396. ( 3 5 ) Andersson, S.; lostell, U. Surf. Sci. 1974, 46, 625. (36) Doniach, S.; Sunjic, M. J . Phys. C 1980, 3, 285.

The Journal of Physical Chemistry, Vol. 93, No. 13, 1989 5241

Adsorption of C O and COz on Cs/Cu(l 10) I

t

.

TDS: CO from

I

I

ec',

,

1 . 3 L CO/Cs/Cu(llO)

9:, -

1.3L o f

co

0.03-

1.00

c;

E

0.75

0.02-

y)

ec',

Y 4

0.50

2

.

0.01. 0.25

U

F

5

0.00

0.00 Cs(MNN)/Cu(LVV)

v)

TDS: Cs

from

5

PEAK-TO-PEAK

RATIO

s:8

1.3L CO/C+/Cu(llO)

n

r-1

AES

E

l

0.25

0.5

0.75

1

1.25.,2.5

I

CO ADSORPTION 1.3L o f CO

2.75

i

1.2

zt 150

250

350 450 TEMPERATURE/ K

550

650

Figure 4. Representative thermal desorption spectra showing the evolu-

tion of CO (top) and Cs (bottom) for different Cs precoverages and a fixed CO exposure of 1.3 langmuirs at -1 10 K. intensity of this CO peak increases with increasing Ocs* until Ocs* = 0.90 and then remains approximately constant through Ocs* = 1.1 6, when it starts to decrease to zero. Another small, hightemperature CO peak (near 450 K) is seen for 1.01 IOcs* I1.78. For all Cs coverages below 1.3 monolayers, the presence of cesium on the surface leads to new C O adsorption states which have desorption temperatures (activation energies for desorption) larger than the value of 220 K (10-1 3 kcal/mol) reported for CO on clean Cu( 1 For larger precoverages of cesium, no adsorption of C O was observed. Measurements with XPS and AES for a C O dose of 1.3 langmuirs at 1 10 K and Cs precoverages lower than half a monolayer showed that neither carbon nor oxygen remained on the surface after heating to 300 K. A similar type of analysis for 0.63 IOcs* 5 1.28 (where the high-temperature CO TDS peaks are present) proved that much carbon and oxygen remained at 300 K but that it had all desorbed by 525 K. Comparison of the Cs TDS spectra for Cs coadsorbed with CO (bottom of Figure 4) with those for Cs adsorbed alone (Figure 2) on Cu( 110) shows that the only differences are a relative increase in desorption intensity at -440 K and small decreases in the relative intensities of the peaks at 290 and 350 K. For Ocs* I0.63 there is simultaneous evolution of Cs and C O at -440 K. This is probably due to thermal cracking of Cs-CO adsorption complexes to liberate Cs and CO gas. Comparing Figure 2 and Figure 4 for similar Cs precoverages, one can conclude that the presence of C O on the surface stabilizes a small number of Cs adatoms up to 440 K which on clean Cu( 110) desorb at -290 and -350 K. Figure 5a displays the effect of Cs precoverage on the total amount of adsorption of C O following a fixed C O exposure of 1.3 langmuirs at -1 10 K. On a clean C u ( l l 0 ) surface this exposure leads to adsorption of -50% of a saturated C O monolayer. I n the present work the relative C O coverage, Oca*, was determined by measuring the O(KVV)/Cu(LVV) AES peakto-peak ratio. (Beam damage effects were meticulously avoided by short and defocused beam exposures.) This Auger ratio was normalized to the value obtained for saturation (7 langmuirs) of C O on clean Cu(1 I O ) (O(KVV)/Cu(LVV) z 0.029) and as-

-

-

Cs(MNN)/Cu(LVV)

AES

PEAK-TO-PEAK

RATIO

-

Figure 5. Effect of Cs precoverage on the amount of adsorption of CO, following a fixed CO exposure of 1.3 langmuirs at 110 K, as determined by (a) AES (top) and (b) CO TDS (bottom). The Cs(MNN)/

Cu(LVV) AES ratios reported here were always measured before dosing CO.

suming a linear relation with C O coverage. (This coverage calibration was also checked at a few points using integrated O( 1s) XPS intensities.) The C O coverage, Oca*, reported in Figure 5a is given in relative units, such that Oca* = 1.0 corresponds to saturation. This saturation (monolayer) coverage of CO on clean Cu( 110) has been reported to be 0.8 molecule per Cu surface atomgb Figure 5a shows that the total amount of adsorption of C O is enhanced as Cs is added, until a maximum is reached at 0.5 I Os,* I0.75. This increase can be attributed to a kinetic effect (increase in the sticking probability), since for the CO exposure used here (1.3 langmuirs) there is not yet saturation on a clean C u ( l l 0 ) surface (see above). For Ocs* > 0.75 the amount of adsorbed CO decreases monotonically and has almost disappeared by e,,* = 1.3. For the C O exposure used here (1.3 langmuirs) we did not observe C O chemisorption on pure Cs (ecs* > 2.60). This fact is consistent with results which show that C O has a very small sticking coefficient on metal surfaces saturated with alkali ,,ta1.37.58,75 In Figure 5b we summarize the effect of Cs precoverage on the various CO TDS peak areas. These areas were scaled to &* by using the value obtained for saturation (7 langmuirs) of CO on clean Cu( 110). A comparison between the curve for total desorption of C O (a @)in Figure 5b with that for adsorption of C O in Figure 5a shows that there is a reasonable agreement between the relative C O coverages obtained from TDS and AES data. As can be seen in Figure 5b, the (3 CO TDS peak only starts to grow beyond Ocs* = 0.55, and it maximizes at Ocs* = 1.15. Figures 6 and 7 display the C( 1s) and O( 1s) XPS spectra after dosing 1.3 langmuirs of C O onto the clean Cu( 110) surface at 1 10 K. The spectra are characterized by the presence of broad features in a large range of binding energies: 284-296 eV for carbon and 531-542 eV for oxygen. A very similar broadening

+

-

(37) (a) Broden, G.; Gafner, G.; Bonzel, H. P. Surf. Sci. 1979, 84, 295. (b) Crowell, J. E.; Garfunkel, E. L.; Somorjai, G. A. Surf. Sci. 1982, 121, 303.

The Journal of Physical Chemistry, Vol. 93, No. 13, 1989

5242

r1 I C ( 1 s ) XPS

1

2

1 31 C O / C s / C u ( l l G )

c

cc.

on CS/C"(llO)

L

-

A

.

-

'

3

L,rj

c l e & l r>(!:3) I

1

I

1

292

296

BINDING

L

288 ENERGY/ eV

A

254

Figure 6. C ( l s ) X P S spectra following C O adsorption on clean and cesium-coveredCu( I IO): (a) clean Cu( 1 IO) surface; (b) after dosing 1.3 langmuirs of C O onto clean Cu( 110) at 1 I O K; (c) after dosing 1.3 Iangmuirs of CO onto a cesium-covered (ec,* = 0.97) Cu( 110) surface at 1 I O K; (d) after flashing surface (c) to 325 K to remove the a C O

-

-

state.

,

-- - -- - - -- '

- -._ 0

L

'

542

540

538

BINDING

536 534 ENERGY/ eV

532

53'

Figure 7. O(ls) XPS spectra following C O adsorption on clean and cesium-coveredCu(1 IO): (a) clean Cu( 110) surface; (b) after dosing 1.3 langmuirs of C O onto clean Cu( 110) at 110 K; (c) after dosing I , 3 langmuirs of CO onto a cesium-covered (ecJ* = 0.97) C u ( l I O ) surface at 1 I O K; (d) after flashing surface (c) to 325 K to remove the a CO state. In spectra b, c, and d the dashed line represents the (smoothed) background for clean C u ( l 1 0 ) .

-

-

was observed in the C( Is) and O( Is) XPS spectra of C O adsorbed on C ~ ( l 0 0 ) . I~n ~that case the features toward high binding energy in the spectra have been attributed to strong shake-up transition^.^^ The same phenomenon is probably occurring in the O(ls) and C(1s) spectra of CO on C u ( l l 0 ) . Curves c of Figures 6 and 7 were taken immediately after dosing 1.3 langmuirs of CO onto a cesium-covered (e,* = 0.97) Cu( 1 IO) surface at 1 I O K. Note that this Cs coverage is beyond the maximum in Bco* versus Bcs* (Figure 5), so that the C O coverage is about the same as on the clean surface. Curves d show C ( 1s) and O( Is) XPS spectra obtained by briefly flashing this Cs + CO overlayer to 325 K. At this temperature the only species present on the surface are the Cs-CO complexes (p) that decompose at -425 K and Cs itself (see Figure 5). A comparison between

-

(38) Norton, P. R.; Tapping, R. L.; Gocdale, J. W. Sur/. Sci 1978, 72, 33. (39) Messmer, R. P.; Lamson, S.H.; Salahub, D. R. Solid Sfafe Commun. 1980. 36. 265.

IEE

:o

Y l i

c :z

PCIK-'i-"TPI

o

1813213::

__L-L-~L,

:1

o :6

c

PP'!:

Figure 8. Effect of Cs precoverage on the total amount of surface oxygen as measured by the O/Cu AES ratio, obtained from fixed C 0 2 exposures of 2.3 and 11 langmuirs at 1 I O K. The absolute oxygen coverage scale, Bo, was determined by relating AES to X P S intensities, and XPS intensities to a known 0, standard (see text). The Cs(MNN)/Cu(LVV) AES ratios reported here were always measured before dosing C 0 2 .

-

curves b, c, and d of Figures 6 and 7 shows that coadsorption with Cs strongly decreases the average binding energy of the O( 1 s) and C(ls) features of CO. We observed that for Bcs* < 1 the O(1s) and C ( l s ) binding energies decrease with increasing Cs precoverage (maximum shift -2 eV). An analysis of the Cs(3d) XPS spectra for Cs coadsorbed with C O (Bcs* L 0.7) showed the following differences with respect to the spectra for Cs adsorbed alone (comparing only spectra with the same Cs precoverage): (1) a small (-0.1 eV) shift of the main Cs peaks to lower BE, (2) decrease in the intensity of the satellite loss tails (plasmons, etc.), and (3) a resulting increase in the intensities of the two main Cs peaks. Similar effects have been observed upon dosing O2to alkali-metal monolayers on metal surfaces24~30,32,40 and have been interpreted as evidence that exposure of the gas either increases the oxidation state of alkali metal or at least removes Cs-Cs interactions which lead to band formation and metallic plasmon^.^^^^^ In any case, this result shows that the C O is interacting very intimately with the Cs adatoms when BCs* > 0.7. 1iI.3. Coadsorption of C02 and Cs on C u ( l l 0 ) . In these studies, we first vapor-deposited Cs onto the surface at 110 K, then dosed CO, at the same temperature, and finally observed the variations in the TDS, XPS, and AES spectra as the precoverage of Cs increased. Using exposures up to 350 langmuirs, we were unable to react any measurable amount of carbon dioxide with the clean Cu( 1 IO) surface at 110 or 250 K, as evidenced by the complete absence of oxygen signal in XPS and AES after exposure. The presence of preadsorbed cesium dramatically affects the adsorption behavior and chemistry of C 0 2 on Cu( 1 IO). Our results to be presented below indicate that when C 0 2 is dosed onto cesium-covered Cu(l10) at I 10 K, it reacts to yield CO,, CO,,,, and C03,, on the surface. Figure 8 displays the effect of Cs precoverage on the total amount of surface oxygen obtained from fixed C 0 2exposures of 2.3 and 1 1 langmuirs at 110 K. As will be shown below, this oxygen Auger signal has contributions from at least three different species produced upon C 0 2 adsorption: CO,, CO,,,, and C03,,. The relative coverages of these species varied depending on C 0 2 exposure and Cs precoverage. In any case, the total amount of surface oxygen can be taken as qualitatively representative of the total amount of CO, that interacts with the surface to give adsorbed CO, or to produce CO, and C03, upon adsorption at 1 10 K. The surface oxygen concentration was determined by measuring the O(KVV)/Cu(LVV) AES peak-to-peak ratio. This Auger ratio was calibrated approximately to absolute oxygen coverage units, Bo, by comparing its values from the experiments

-

-

-

-

(40) Lindgren. S. A.; Walldtn, L. Surf. Sci. 1979, 80, 620.

The Journal of Physical Chemistry, Vol. 93, No. 13, 1989 5243

Adsorption of C O and C 0 2 on Cs/Cu( 110) TDS: CO,

from

2.3L C O z / C s / C u ( l l O )

r66OK

0,: 0.5

1.5

1.0

2.0

1.24

in

k 3

i2 z

CI 0

N (0

150 200 250 300 350 TEMPERATURE/ K

C s (YNN) /CU (LVV) AES PEAK-TO-PW

3 RATIO

Figure 10. Effect of Cs precoverage on the line shapes (A) and integrated peak areas (B) of CO TDS spectra obtained after dosing 2.3 langmuirs of C02onto different Cs/Cu(l10)surfaces at 110 K (same conditions as Figure 9).

-

250

350

450

550

650

750

TEMPERATURE/ K

Figure 9. Representative thermal desorption spectra showing the evolu-

-

tion of C02(top) and Cs (bottom) for different Cs precoverages and a fixed C02exposure of 2.3 langmuirs at 110 K. in Figure 8 at Ocs* N 0.4 to the integrated 0(ls)/Cu(2p3,,) XPS intensity ratios in the corresponding XPS spectra. The O/Cu XPS ratios in turn were scaled to absolute coverage units by using the ratio (6.2 X measured in this same instrument for 0, saturation exposure at 300 K on clean Cu( 1lo), which is known to give an absolute oxygen coverage of 0.5 atom per Cu surface atom (i.e., Bo = The results of Figure 8 indicate that the amount and probability for adsorption of CO, are enhanced as Cs is added to the Cu( 1 10) surface. For the conditions investigated in the present work, no saturation limit on the amount of reacting C 0 2 was reached as the cesium coverage was increased to almost three monolayers (although there must eventually be a limit at unit sticking probability). There was, however, a break in the slope of oxygen coverage versus Ocs* at a cesium precoverage of about one-half monolayer. The slope changes by a factor of -2.2 here. This change in slope is in part caused by the fact that at high Cs precoverage most of the C O formed on the surface upon dissociative CO, adsorption desorbs due to the lack of empty Cu sites. (Note that C O does not adsorb on pure Cs; see section 111.2.) There was also a noticeable change in the O(KVV) line shape at this point, so that line shape influences on the peak-to-peak height also contribute to this change in slope. A series of thermal desorption spectra obtained after dosing CO, onto Cs/Cu( 110) are shown in Figures 9 and 10. Figure 9 displays the evolution of C 0 2 ( m / e = 44) and Cs ( m / e = 133) for different Cs precoverages and a fixed CO, exposure of 2.3 langmuirs at 110 K. For this exposure a small peak for C 0 2 desorption (not shown in Figure 9) was found at 130 K already for BCs* > 0.2. The intensity of this peak grew with CO, exposure

-

-

(41) Grazalski, G. R.; Zehner, D. M.; Wandelken, J . F. Surf. Sri. 1985. 159, 353. (The O(1s) binding energies taken from this work have been

corrected here by -0.3 eV due to differences in the energy calibration used here.)

well beyond 5 langmuirs. We assign it to the desorption of a very weakly adsorbed CO, species. This species was observed only in the presence of Cs,. The higher temperature features in the CO, TDS spectra of Figure 9 are a strong function of Cs precoverage. Many of these involve simultaneous evolution of Cs. On the basis of the XPS and AES results below, all the peaks for C 0 2 evolution above 300 K will be assigned to thermal cracking of adsorbed Cs, n0, nCOzg. cesium carbonate complexes via Cs.nC03, These carbonate species appear to be formed on the surface upon CO, adsorption at 110 K (see below). Measurements with XPS and AES showed the presence of small amounts of oxygen and cesium on the surface after the thermal desorption experiments (see below), consistent with this mechanism. The results of Figure 9 for low Cs coverages show a desorption state of CO, at -550 K which, in contrast to the peaks seen at higher Ocs*, does not involve simultaneous evolution of Cs. At these low Cs coverages, Cs adatoms are stable on the surface to above 650 K even in the absence of stabilizing carbonate (see Figure 2). Thus, at these low Cs coverages, cesium evolution during the thermal cracking of the C S . ~ C Ocomplex ~,~ at -500 K is not expected. This is because the resulting Cs adatoms are stable on the surface even in the absence of carbonate. These eventually desorb only at higher temperatures. The CO, TDS spectra for Bcs* = 0.54-0.71 are characterized by the presence of three desorption peaks at -550, -620, and -685 K. Only the peak at highest temperature is associated with the simultaneous desorption of cesium. By comparison to Figure 2 for the same Bcs*, it is clear that the carbonate stabilizes surface Cs relative to the clean Cu( 110) surface, slowing the onset of desorption only with the removal of the last of the CO, from the carbonate. For e,,* I0.83 there is simultaneous evolution of Cs and CO, in peaks at -500 and 620-700 K. These desorption peaks correspond to thermal cracking of Cs.nC03,, complexes. At Bcs* 1 1.24 a sharp peak for multilayer Cs desorption appears at -300 K (compare to Figure 2). A small amount of CO, is also liberated here. The CO,/Cs intensity ratio for this feature is much, much smaller than those for the features at -550 and 620-700 K . This fact suggests that the CO, at -300 K is not the product of thermal cracking of a Cs.nCO, complex but perhaps a consequence of rearrange-

-

-

+

+

5244

The Journal of Physical Chemistry, Vol. 93, No. 13, 1989

ments in the adlayer induced by the morphological and surface area changes accompanying multilayer Cs desorption. It may also be an artifact related to the interactions of the large amounts of desorbing Cs with the chamber walls. Comparison of Figures 2 and 9 shows that adsorption of C 0 2 leads to a strong stabilization of Cs on the Cu(l10) surface, presumably by forming Cs-nCO,,, complexes. For cesium precoverages lower than a monolayer and a CO, exposure of 1 1 langmuirs, we found that there was no desorption of Cs from the surface until -620 K. At this temperature Cs desorption starts (with simultaneous evolution of CO,). Thermal desorption spectra for C O evolution from CO,/Cs/ Cu( 110) are displayed in Figure 10. Here we show only the m / e = 28 peaks associated with actual C O desorption. We of course also saw higher temperature peaks at m / e = 28 mimicking those for CO, at m / e = 44 and with intensities appropriate to the mass spectral cracking pattern of CO,. This figure shows both the effects of Cs precoverage on the line shapes and integrated peak areas of C O TDS spectra obtained after dosing 2.3 langmuirs of C 0 2 onto different Cs/Cu(llO) surfaces at -110 K. In the spectra of Figure 10 the C O TDS temperatures and line shapes are essentially identical with those observed for the low-temperature desorption state of CO after simply dosing C O to Cs/ Cu( 1 10) (see Figure 5 ) . Thus, these are desorption-limited CO evolution peaks. These C O desorption peaks are therefore likely a consequence of the following reactions to produce carbonate on the surface occurring below the onset of C O desorption (Le., at T < 175 K):

-

co, co, + 0, COl,, 0, + CO, and/or

2c0,

-

- co, + coj,,

(la) (lb)

(2)

where CO,, here can refer to a surface Cs.nC03, complex as well. These reactions may occur from gas-phase CO, or from a weakly adsorbed C 0 2 precursor. Reaction l a has been observed for and polycrystallineCuI6 high-pressure C 0 2 exposures to Cu( 1 and for CO, adsorbed on Re(0001),43Ni( 1 Fe(l1 1),45and Fe films.46 Reaction 1b has been reported on silver surface^.^' (We were, however, unable to make this reaction go on O/Cu( 110) surfaces in the absence of Cs,.) Finally, reaction 2 is supported by experiments in argon and nitrogen matrices which show that reaction of Cs2C02with a molecule of CO, leads to formation of cesium carbonate, C S ~ C O The ~ . ~fl ~C O TDS feature at -425 K produced by dosing C O to Cs/Cu( 1 10) (see Figure 4) was not produced from the adsorbed C O created via C 0 2 dissociation. We assume that this is because CO, competed more effectively for the available cesium required for making the fl CO/Cs complex. Figure 10 shows that the amount of C O evolved during TDS after a fixed C 0 2 exposure increases as Cs is added, until a = 0.70. This increase in CO evolution maximum is reached at reflects the increase in the amount of CO, that reacts on the surface to produce carbonate (see Figure 8 ) . For eCs* > 0.70 the amount of desorbing C O decreases with increasing eCs* and has almost disappeared by eCs* = 2.25, in spite of the fact that large amounts of carbonate are still produced as shown by Figure 9. This decrease in the amount of C O desorption in TDS is very (42) Nakamura, J.; Rodriguez, J. A.; Campbell, C. T. J . Condem. Mutter, submitted for publication. (43) Asscher. M.; Kao, C.-T.; Somorjai, G. A. J . Phys. Chem. 1988, 92, 2711.

(44) Bartos, B.; Freund, H. J.; Kuhlenbeck, H.; Neumann, M.; Linder, H.; Muller, K. SurJ Sei. 1987, 179, 59. (45) Freund, H. J.; Behner, H.; Bartos, B.; Wedler, G.; Kuhlenbeck, H.; Neumann, M. Surf. Sci. 1987, 180, 550. (46) Pirner, M.; Bauer, R.; Borgmann, D.; Wedler, G.Surf. Sci. 1987, 189/190, 147. (47) (a) Bowker, M.; Barteau, M. A.; Madix, R. J. Surf. Sci. 1980, 92, 5 2 8 . (b) Campbell, C. T.Surf. Sei. 1985, 157, 43. (48) Kafafi, 2. H.; Hauge, R. H.; Billups, W. E.; Margrave, J. L. Inorg. Chem. 1984. 23, 177.

Rodriguez et al. similar to the decrease in the amount of adsorbed CO produced from C O dosing presented in Figure 4. A difference is that the amount of CO decreases for eCs* > 1 much more slowly for CO, dosing than for C O dosing. Nevertheless, the decreases in both cases can qualitatively be attributed to a decrease in the number of free Cu sites on the surface, which are necessary to form adsorbed CO states that are stable above 110 K. (Remember that C O does not adsorb easily on pure Cs at 110 K.) Thus, for Cs coverages much in excess of one monolayer, even if CO is produced from C 0 2 at 110 K according to reactions 1 or 2, it is unlikely to be stable and will desorb during the exposure rather than during the subsequent TDS. These data in fact show that the reactions to produce carbonate already occur during dosing at 110 K for these coverages. Curves a of Figure 11 display the C( Is) and O( 1s) XPS spectra taken immediately after dosing 22 langmuirs of CO, onto a cesium-covered (ecs* = 1.0) C u ( l l 0 ) surface at -110 K. A large exposure was used here in order to produce a sizable concentration of the weakly adsorbed CO, state which desorbs at 130 K and is associated with the presence of Cs,. (The TDS spectra presented above were for lower exposures than used here because the C 0 2 background pressure was a problem during TDS after such large exposures.) Under the conditions of Figure 11 the total coverage of oxygen on the surface was 2.9 atoms per Cu surface atom, or -6 oxygen atoms per Cs atom. Both C(1s) and O(ls) spectra are characterized by the presence of at least two peaks extended over a range of -9 eV. We found that the positions of the two major peaks in each spectrum varied by as much as 1.5 eV depending on BC,* and on the size of the CO, dose. For large values of eCs* ( 2 1 ) and small C 0 2 exposures ( 5 2 langmuirs) the peak at higher binding energy was present in neither the C( 1s) nor O( 1s) spectra of the CO,/Cs/Cu( 1 IO) surfaces. These peaks, when present after dosing CO,, disappear upon heating to only I80 K as shown in curves b of Figure 1 1, with simultaneous CO, desorption. An analysis of the integrated C( Is):O( 1s) intensity ratios in the XPS spectra (taking into account the factor of 3.1 difference between the surface XPS sensitivity factors of 0 and C49showed an O / C stoichiometric ratio of 1.8:l for this intensity which is lost upon heating to 180 K in Figure 11. Given expected small differences in the sensitivity factors between spectrometers, this stoichiometry is consistent with the simple desorption of weakly adsorbed CO,. The oxygen and carbon that remained on the surface after heating to 180 K showed only the lower binding energy peaks, now shifted to slightly lower binding energy. Notice in Figure 10a that CO desorption will already be largely completed since the sample was held at 180 K during spectra acquisition. Also, the additional CO produced by the larger CO, dose of Figure 11 compared to Figure 10 will have desorbed already by 1 10 K, since the Cs coverage used here is beyond the maximum in Figure 1Ob. Further heating the surface to 310 K (curves c) causes both the C(1s) and O(1s) peaks to narrow due to the loss of the remaining CO, (Figures 6 and 7) and also perhaps due to changes in the adlayer. At this point the C(1s) peak is centered at 289.2 eV. Its width is comparable to the instrumental resolution (- 1.7 eV), suggesting only a single carbon atom environment. When the weakly adsorbed CO, state above is used to obtain a precise value for the relative XPS sensitivity factors for O:C, their XPS intensity ratio here indicates that the O / C stoichiometry at this point is -2.95: 1, consistent with a carbonate (CO,) species. The O(ls) peak here is noticeably broader than the instrumental resolution (fwhm = 2.4 eV compared to 1.7 eV), suggesting different environments for the oxygen atoms in this species. This is consistent with either monodentate or bidentate forms of this species. The O( 1s) peak is centered at 53 1.3 eV. The integrated XPS peak areas for curves c show that the carbonate coverage here is -0.4 C 0 3 unit per Cu surface atom, or 1 CO, unit per Cs atom (calibrated using O(1s) intensities for 0, on Cu(l10) at Bo = 0.5). Thus, at least for these exposures, the value of n in the formula C S ~ O ,is, approximately ~ unity. (This does not

-

-

-

(49) Wagner, C. D. J . Electron Spectrosc. Relat. Phenom. 1983, 32, 99.

The Journal of Physical Chemistry, Vol. 93, No. 13, 1989 5245

Adsorption of C O and C 0 2 on Cs/Cu( 1 I O )

j [ [ A

BF O(1s)

22L co*/cs/cu(11o)

d , e . CO, o n C s / C u ( l l O ) ,

I

I

I

I

292

296

BINDING

XPS

22L co2/cs/cu(110)

T=730K and 810K.

I

I

I

I

284

288 ENERGY/ eV

540

536

532

528

BINDING ENERGY/ eV Figure 11. (A) C(ls) and (B) O(ls) X P S spectra following C 0 2 adsorption on cesium-covered Cu(l IO): (a) after dosing 22 langmuirs of C 0 2 onto a cesium-covered (Be,* = 1) Cu(l IO) surface at -110 K; (b) after warming to 180 K to remove the weakly adsorbed C 0 2 and some CO,; (c) after warming to 310 K to remove the remaining adsorbed CO (note that there was no change after further heating to 450 K); (d, e) after warming to 730 K (to decompose the Cs.nCO,,a complex) and to 810 K, respectively.

rule out other stoichiometries for much different conditions.) Further heating the surface to 450 K produced no changes in the XPS spectra. By 730 K, all observable C ( 1s) intensity is gone, and at least two-thirds of the O( 1s) intensity is removed. The remaining O( 1s) intensity appears in a new peak at -528.6 eV. The only other element on the surface is Cs. Both the O(1s) and Cs(3d) peak positions at this point are the same as those produced by simply dosing O2 to Cs/Cu( 1IO) and heating to this same t e m p e r a t ~ r e . ~ ~ That treatment was thought to produce a coadsorbed Cs, 0, layer, where the Cs and oxygen adatoms are both bonded to Cu but strongly stabilize one another.50 Heating from 450 to 730 K also causes a large (-65%) loss in Cs(3d) intensity here (not shown). All of these XPS results are consistent with a mechanism whereby carbonate species decompose between 450 and 730 K according to Cs.CO,,, C02,g+ 0, + Cs,, where much of the Cs, then desorbs immediately due to the fact that it finds itself above its desorption temperature. Carbonate decomposition to liberate C 0 2and produce 0, also occurs on Ag( 110) in the absence of Cs, at -475 K.47 The C 0 2 and Cs evolution between 450 and 730 K in the TDS spectra of Figure 9 is also consistent with this mechanism. Between 730 and 8 10 K, there is only a small loss of O( Is) intensity (perhaps due to diffusion of 0, into the bulk), and significant loss of Cs(3d) intensity, due to desorption of Cs, as reported previously for 02/Cs/Cu( 1 An analysis of Cs(3d) XPS spectra for Ocs* N 0.5-1.5 before and after dosing C 0 2at 110 K showed the following changes: (1) a small (10.25 eV) shift of the two main peaks toward smaller binding energy, (2) an increase (120%) in the intensity at the two main peak maxima, and (3) a corresponding decrease in the intensity of the satellites (plasmons, etc.) extending to lower kinetic energy. Similar changes have been observed for adsorbed Cs after

+

-

(50) Clendening, W. D.; Rodriguez, J. A,; Campbell, J . M.; Campbell, C. T. Surf.Sci., in press.

dosing CO (see section 111.2), oxygen,50 and formic acids’ on Cu( 110). They are indicative of Cs oxidation, in this case by reaction with CO,. Whether this “oxidation” of Cs involves direct bonding of the adsorbate to Cs, or simply screening of Cs-Cs depolarization and band-forming interactions by locating the adsorbate between the Cs atoms (ions) is still in question.50 Representative O(KVV) and C s ( M N N ) AES spectra for C02/Cs/Cu( 110) surfaces are shown in Figure 12. Curve a displays the Cs(MNN) Auger spectrum of a Cu( 110) surface with a cesium precoverage of 0.61 monolayer. Curve b was taken after dosing 2.3 langmuirs of CO, onto this cesium-covered Cu( 110) surface at 110 K. Note that the C s ( M N N ) AES line shape changes significantly upon C 0 2 adsorption, so that its peak-to-peak height increases by a factor of 1.5 and the peak minima are shifted slightly toward lower kinetic energy. A similar phenomenon was observed in the present work for adsorption of CO on Cs/Cu(l 10) surfaces and has been reported for adsorption of oxygen on cesium-covered surfaces.50 The line shape change indicates a change in the chemical nature of the Cs. After CO, adsorption, the C s ( M N N ) AES line shape is similar to those reported for species with ionically bonded Cs: bulk CSI,~,O,/ Cs/Ag( 1 and O,/Cs/Cu( 1 The narrowing of the transitions upon oxidation, which leads to increase in peak-to-peak height, is probably related to the same decrease in plasmon loss intensity seen in the Cs(3d) XPS spectra. Curves c-e of Figure 12 show AES spectra obtained by briefly flashing the surface overlayer to the indicated upper temperature limits to desorb different adsorption states. Note that the C 0 2 exposure is smaller than in Figure 11, so that much less of the weakly adsorbed C 0 2 desorbs here upon heating to 180 K (curve

-

-

(51) Henn, F. C.; Rodriguez, J. A.; Campbell, C. T., manuscript in p r e p aration. (52) Davis, L. E.; MacDonald, N . C.; Palmberg, P. W.; Riach, G. E.; Weber, R. E. Handbook of Auger Electron Spectroscopy, 2nd ed.; PerkinElmer: Eden Prairie, MN, 1978; p 173.

5246

The Journal of Physical Chemistry, Vol. 93, No. 13, 1989

1

1

' AES '

I

ce2/cs/cu(110)

C s (MNN)

I-

r-

Cs (MNN)

1

460 480

500

1

520 540 KINETIC

I

I

1

560

580

600

ENERGY/ eV

Figure 12. Cs(MNN) and O(KVV) AES spectra following adsorption of Cs and CO, on Cu( 110): (a) after vapor depositing 0.61 monolayer of Cs onto Cu( 110) at I I O K; (b) after dosing 2.3 langmuirs of CO, onto the cesium-covered Cu(110) surface at 110 K; (c) after flashing to 180 K to remove the weakly adsorbed C 0 2 ;(d) after flashing to 300 K to remove the adsorbed CO; (e) after flashing to 735 K to decompose the Cs.CO,,, complex.

-

-

c). Note, however, a narrowing of the main O(KVV) peak and disappearance of the minima at -515 eV, due to desorption of CO,,, and CO,. Upon heating to -300 K, the remaining CO, (produced by CO, decomposition upon carbonate formation) desorbs (see above). The amount, however, is so small that changes in the O(KVV) line shape are hard to see in curve d . Further heating to -735 K to decompose the Cs-nC03,,complex (see Figure 9) causes a decrease in the O(KVV) intensity and leaves an O(KVV) line shape (curve e) that is similar to those reported for 0, in Cu( 1 10) and O/Cs/Cu( 110) surfaces.jO Measurements with AES, like XPS, showed that no carbon was present on the surface after thermal desorption to -735 K. I t should be noted that extreme caution was exercised in the acquisition of the spectra of Figure 12 to prevent beam damage effects. hone were observable within the sampling time required for the portion of the spectra which is shown (