J . Phys. Chem. 1988, 92, 2569-2513
2569
Interaction of Aluminum with the Ru(0001) Surface and Its Influence upon CO Chemlsorption Charles T. Campbell* Chemistry Department, Indiana University, Bloomington, Indiana 47405
and D. W. Goodman Sandia National Laboratory, Albuquerque, New Mexico 871 15 (Received: August 18, 1987; In Final Form: November 6,1987)
The interaction of vapor-deposited A1 with the Ru(0001) surface and its influence upon CO chemisorption has been studied in ultrahigh vacuum by use of Auger electron spectroscopy, thermal desorption mass spectroscopy, and low-energy electron diffraction (LEED). Below 670 K, the AI grows as an overlayer in (1 11) oriented islands. By 1170 K, A1 penetrates the surface to form intermetallic compounds with the Ru, and a (d3Xd3)R30° LEED pattern dominates the surface structure. This is interpreted in terms of a stoichiometric monolayer of the compound A12Ru. Thermal desorption data indicate that this monolayer alloy has a heat of formation of about -24 kcal/mol, compared to -14 kcal/mol for the multilayer alloy which nucleates at higher coverages. The CO/Ru interaction at the surface of these intermetallic compounds is not very different from that on the clean Ru(0001) surface, except for the expected dilution of Ru sites by the added Al. In contrast, when the A1 overlayer has not yet been annealed and the A1 adatoms still sit on top of the Ru(0001) surface, they have a very strong influence upon CO desorption spectra. A new CO desorption state at 850-1050 K appears at submonolayer AI coverages, which is attributed to strong, direct, interaction between adsorbed CO and AI probably involving C-0 bond cleavage.
I. Introduction Chemisorption and reactions on bimetallic or alloy surfaces are of fundamental interest in catalysis, corrosion prevention, and numerous other technological areas. In addition, understanding bimetallic surface chemistry is germane to the study of interactions between a metal catalyst (e.g., Ru) and its supporting metal oxide (e.g., Al20,). Presently, a great deal is known of the interactions of simple molecules with single-element metal surfaces. In contrast, relatively little of a systematic nature is known of chemisorption on bimetallic surfaces. In this study, we investigate the interaction of CO with Al/Ru bimetallic surfaces generated by vapor-depositing A1 on the Ru(0001) surface, in some cases followed by annealing to generate the intermetallic compound A12Ru at the surface. In the paper that follows, Chen, Crowell, Ng, Basu, and Yates' explore the interactions of C O with Ni/Al bimetallic surfaces. The interaction of C O with clean Ru(0001) has been well studied by a number of authors (ref 2-13 and references cited therein). It adsorbs molecularly2 and saturates at room tem3 ~ ' (per Ru surface perature with a coverage Oc0 of 0 ~ 8 ~ or~ 0.6713 atom). It desorbs intact in two distinct peaks (& and 03)at -400 and 450 K, which reflect strong, short-range lateral CO-CO repulsion^.^^^^ On Al, CO also adsorbs m o l e c ~ l a r l y ,but ' ~ with a much weaker bond than on Ru. It desorbs intact already at 125 K.14 In this study, we find that A1 on Ru(0001) induces a much stronger interaction with C O than for either clean metal surface. On the other hand, when A1 is actually imbedded in the topmost Ru layer (Le., in AlzRu alloy surfaces), its major effect on the CO/Ru interaction is simply to dilute the Ru sites on the surface. In the former case, we show evidence which suggests that A1 directly participates in C O dissociation, perhaps by attacking the oxygen end of the molecule. A similar interaction is also suggested for C O on Ni/Al surfaces in the accompanying paper by Chen and co-workers,' who interpret vibrational data in terms of a Nix-C-0-AI, species which is a precursor to CO dissociation.
-
11. Experimental Section
The apparatus used for these experiments has been described in detail previo~sly.'~ Aluminum was vapor-deposited from a source described previo~sly,'~ where now a small AI wire was *Alfred P. Sloan Research Fellow and author to whom correspondence should be addressed.
0022-3654/88/2092-2569$01.50/0
wound about a resistively heated tungsten wire. The cleaning procedures for the Ru(0001) surface where the same as we have used in our previous studies with this surface.16 Coverages here are normalized to the Ru(0001) surface atom density. (e = 1 corresponds to 1.64 X 1015cm-2.) A gas doser was used for CO dosing, and exposures are reported here in units which have been approximately calibrated with C O uptake on clean Ru(0001).
111. Results III.1. Interactions of A1 with Ru(0001). Figure 1 shows thermal desorption mass spectra (TDS) of A1 at m / e 27 following increasing exposures of Al to Ru(0001) at room temperature. At low A1 coverages, a desorption peak at 1570 K appears, which broadens to lower temperature with increasing coverage. The peak maximum decreases to 1530 K, and then a second desorption peak at 1480 K appears, which also shifts to lower temperature with increasing coverage until it reaches a minimum temperature of 1420 K. The peak then grows continuously in zero-order fashion, with a low-temperature edge that is independent of increasing A1 coverage. This is characteristic of some type of multilayer A1 or Al/Ru alloy that decomposes to liberate A1 gas.
-
-
-
-
(1) Chen, J. G.; Crowell, J. E.; Ng, L.; Basu, P.; Yates, J. T., Jr. J. Phys. Chem., following paper in this issue. (2) Madey, T. E.; Menzel, D. Jpn. J . Appl. Phys. 1974, Suppl. 2, Port 2, 229. (3) Madey, T. E. Surf.Sci. 1979, 79, 575. (4) Fuggle, J. C.; Madey, T. E.; Steinkilberg, M.; Menzel, D. Surf.Sci. 1975, 52, 521. (5) Fuggle, J. C.; Umbach, E.; Feulner, P.; Menzel, D. Surf.Sci. 1977, 64. 69. (6) Pfnur, H.; Feulner, P.; Engelhardt, H. A.; Menzel, D. Chem. Phys. Lett. 1978, 59, 481. (7) Menzel, D. In Chemistry ond Physics of Solid Surfoces IV; Springer-Verlag: New York, 1982, pp 397-405. (8) Pfnur, H.; Menzel, D.; Hoffman, F. M.; Ortega, A,; Bradshaw, A. M. Surf. Sci. 1980, 93, 431. (9) Hoffmann, F. M.; Persson, B. N. J. Phys. Reu. B: Condens. Matter 1986, 34, 4354. (10) Williams, E. D.; Weinberg, W. H. Surf.Sci. 1979, 82, 93. (11) Thomas, G. E.; Weinberg, W. H. J . Chem. Phys. 1979, 70, 1437. (12) Leuthausser, U. Z . Phys. B 1980, 37, 65. (13) Biberian, J. P.; Van Hove, M. A. Surf.Sci. 1984, 138, 361. (14) Paul, J.; Hoffmann, F. M. Chem. Phys. Lett. 1986, 130, 160. (15) Goodman, D. W.; Yates, J. T., Jr.; Peden, C. H. F. Surf.Sci. 1985, 164. . . 417. (16j Houston, J. E.; Peden, C. H. F.; Blair, D. S.;Gocdnlan, D. W. Surf. Sei. 1986, 167, 427.
0 1988 American Chemical Society
2570 The Journal of Physical Chemistry, Vol. 92, No. 9, 1988
Campbell and Goodman
? W
I-
< K
z
0 I-
n K 0
t,S E C O N D
v)
W
-
n
0
U
-
P E A K A P P E A R S IN A I - T O S
I
I
I
I
I
1
2
3
4
5
eAl
0 4 0.
,
0.17 0.05, 1300
6
7
-
Figure 2. AI/Ru AES peak-to-peak ratio for 68- and 273-eV transitions, respectively, as a function of the AI coverage after dosing AI at 300 K (0)and after a brief anneal to 1170 K (A). Also indicated are coverage regions where LEED and TDS features were seen.
\
A
,
,
1400
1500
I'""'""'""""""'I
1
1600
1700
TEMPERATURE/K
Figure 1. Aluminum ( m / e 27) thermal desorption spectra for increasing doses of AI to Ru(0001) at 300 K. (Absolute coverage calibration described in text.) For the highest coverage, only the leading edge of the curve is shown.
This leading edge was analyzed for the highest coverages (OAl > 3) in zero-order Arrhenius form and gave a straight line fit with a desorption activation energy of 88 f 4 kcal/mol. Since this is significantly larger than the heat of vaporization of pure A1 at 1300 K (-71 kcal/mol17), we can conclude that A1 desorption in the low-temperature (- 1420 K) peak is not from a multilayer of pure Al. (Note that pure A1 melts at -933 K.I7) Instead, it must represent the decomposition of some Al/Ru alloy or intermetallic compound. Mixtures of AI and Ru form a whole series of immiscible intermetallic compounds from AlRu to A112R~,18*19 any of which could be responsible for the low-temperature (multilayer) desorption peak. The high-temperature TDS peak then reflects desorption from the first monolayer of A1 at (in) the Ru surface. Its desorption temperature of 1570-1530 K then reflects an activation energy for desorption of 100-97 kcal/mol, assuming a typical preexponential factor for desorption of lOI3 s-I in first-order Redhead analysis.20 Following desorption to 1640 K, no A1 was observable on the Ru surface by Auger electron spectroscopy (AES), and the lowenergy electron diffraction (LEED) pattern was identical with the p(lX1) pattern of the clean Ru(0001). Furthermore, subsequent CO TDS spectra were identical with that from freshly cleaned Ru(0001). These results indicate that A1 removal by desorption was quantitative by 1640 K. Since our A1 dose rate was somewhat unstable, we have used the integrated area under the A1 TDS spectra as proportional to eAl here. Absolute coverage calibration will be described below. The variation in the Al(68 eV)/Ru(273 eV) AES peak-to-peak ratio with A1 coverage at 300K is shown in Figure 2. Also shown are the results after briefly ( 5 s) annealing the overlayers to 1170 K. Both the annealed and unannealed AES ratios are very similar below eAl = 0.7, indicating a roughly linear increase in the ratio with A1 coverage. The AES signal growth shows a break in slope, and for the annealed surface it becomes almost constant when eAl exceeds 0.7. This is just the coverage when the multilayer peak begins to appear in the A1 TDS. The signal decrease after an-
t
F
CLEAN
300K
ANNEAL
AES: Al/Ru(OOOl)
-
-
(17) The Handbook of Chemistry and Physics, 49th 4.; CRC Press: Cleveland, OH, 1968-9. (Temperature interpolation used on data here.) (18) Obrowski, W. Metall (Berlin) 1963, 17, 108. (19) Pearson, W. B. Handbook of Lattice Spacings and Structures of Metals and Alloys: Pergamon: New York, 1967; Vol. 2, p 591. (20) Redhead, P. A. Vacuum 1962, 12, 203.
KINETIC ENERGYIeV Figure 3. AES spectra for several AI coverages showing irreversible changes near 60 eV in the AI(LVV) line shape upon annealing briefly from 300 to 1170 K. Dashed line indicates region of change.
nealing indicates some loss of A1 from the surface region. Even below 0, = 0.7, the anneal to 1170 K resulted in a change in the Al(LVV) AES line shape in the region from 56 to 64 eV, as can be seen in Figure 3. This indicates a change in the valence energy levels of the AI atoms. That is, the chemical bonding of A1 at the Ru interface changes distinctly upon annealing to 1170 K. A survey of LEED observations for the annealed and unannealed Al/Ru(0001) overlayers is summarized in Figure 2. At 300 K, diffuse extra spots appear just inside the (01) spots of the Ru lattice at a coverage of -0.4. These increase in sharpness and intensity with A1 coverage. They can be easily explained on the basis of (islands of) an Al( 111) film growing expitaxially and rotationally commensurate with the Ru(0001) surface. Annealing to 670 K only sharpens these extra spots and, in the range 0.67 IOAl I1.9, leads to the appearance of extra satellite spots due to the 13th-order coincidence lattice formed by the Al(111) overlayer ( a = 2.862 A) on the Ru(0001) surface ( a = 2.649 A). All Al( 111) structure disappears upon annealing to 1170 K, and a ( d 3 X d 3 ) R 3 0 ° pattern is apparent for all A1 coverages studied above 0.3. This pattern is sharpest and most intense when BAI 0.7 (Le., when the monolayer peak saturates in A1 TDS). When 0,, > 1.2, additional spots also appear in the LEED patterns of the annealed overlayers. These are complex and generally consist partially of doublets in the vicinity of the spot positions expected
Interaction of CO with AI/Ru(0001) Surfaces
The Journal of Physical Chemistry. Vol. 92, No. 9, 1988 2571
00 A1
R"
a=8.;15H
Figure 4. Structural model for the ( 4 3 x 4 3 ) R 3 O 0 LEED pattern of
annealed AI/Ru(0001). This model involves a slight distortion of a auasi-hexaeonal ~. olane of the bulk comwund ALRu. . . whose dimensions are shown here. For comparison, the Ru-Ru lattice parameter in the pure Ru(0001) plane is 2.649 A. for a (2 X 2) structure. W e will refer to these collectively as '(2 X 2)" below. The ( 4 3 X 4 3 ) R 3 0 ° pattern will be interpreted in terms of a hexagonal-close-packed plane of the intermetallic compound A12Ru, which is shown in Figure 4. This compound has the orthorhombic TiSi2(C54) type structure, with a = 8.015, b = 4.715, and c = 8.780 The lattice consists of (nearly) hexagonal layers such as shown in Figure 4, stacked one on another in an unusual packing arrangement.2' In the plane shown, the AI-Ru nearest-neighbor distances are 2.67 and 2.72 A. This can be compared to the Ru-Ru distance of 2.649 A in the Ru(0001) plane. Thus, this A12Ru plane needs only to be compressed by 2.6%in one direction and 0.8%in the other direction in order for this structure to match perfectly the observed hexagonal (v'3X 4 3 ) R 3 0 ° LEED pattern. In this model, a single A12Ru layer then corresponds to a coverage d,, = '/, The first appearance of the (v'3Xd3)R3O0 LEED pattern for the annealed surface already at a coverage of 0, = 0.3 is consistent with islands of this structure coexisting with relatively clean patches of Ru(000I). Since this pattern becomes most intense at the same coverage (0.67) that saturates the high-temperature AI desorption peak, we assume that this high-temperature peak reflects the decomposition of a single monolayer of this A1,Ru compound. With increasing coverage, the low-temperature desorption peak appears, and eventually additional "2 X 2" structure shows up in LEED along with the v'3 structure. These results indicate that thicker islands of some AI/Ru intermetallic compound begin to nucleate and grow, but that large regions of the A12Ru structure (Figure 4) still exist. The structure of the thicker compound is not yet known. The AES uptake curves to Figure 2 are consistent with this model as well. At 300 K, a true AI film grows above the Ru, and the AI/Ru AFS ratio grows continuously. For all coverages above -0.7, however, annealing to 1170 K results in a nearly constant AFS ratio. This would indicate a constant monolayer of the A12Ru structure over most of the surface, with penetration of thicker alloy islands a t isolated nucleation sites which cover only a small part of the surface. 111.2. Influence of AI upon CO/Ru(OOOl) Interactions. In this section we will compare thermal desorption spectra for saturation exposures of CO to AI adlayers predeposited a t 300 K, without (Figure 5 ) and with (Figure 6) prior anneal to 1170 K. In all these experiments, the CO exposure was 12 langmuirs (1 langmuir = 10-6 T0rr.s) at I50 K,which is beyond that required 0.584.67) on clean Rufor saturation coverage (Oco (0001)?44.7.'0 Figure 5 shows typical TDS results for the unannealed AI/Ru(0001) interface. The well-known 02/&doublet characteristic of clean Ru(OOOI)~".~ is seen in the absence of AI.
-
(21) Structure Reports V I , 1939; Hcrrmann, K.,Ed.; Edward Bra.: Ann Arbor. MI. 1943; p 12.
300
500
100
000
1100
TEMPERATUREIK
Figure 5. Thermal desorption spectra for CO at mle 28 following 12langmuir exposures of CO to Ru(0001) containing various doses of adsorbed AI. Here, AI was first dosed at 300 K, and then the sample was cooled to 150 K for the CO exposure. No CO desorbed below 300 K, so this portion of the spectra is omitted.
300
son
roo
goo
1100
TEMPERATUREIK
Figure 6. Thermal desorption spectra for CO at m / e 28 following 12langmuir exposures of CO to Ru(0001) containing various coverages of annealed AI. In this case, the AI dose at 300 K was followed by a brief anneal to 1170 K, and then the surface was recmled to 150 K for the CO exposure. No CO dcsorbed below 300 K. so this portion ofthe spectra is omitted
With increasing AI coverage, the 02/&doublet narrows to a singlet and decays in intensity, Simultanously, a new high-temperature peak (y) grows and decays with increasing AI coverage. The y peak shifts from 830 to 1050 K. Surface analysis by AES after desorption to 1170 K showed no oxygen or carbon, indicating that CO removal via desorption was quantitative by 1170 K. The integrated areas under the ((3, + 02)doublet and the y peak are shown in Figure 7 as a function of AI coverage. The total (02 y) CO desorption amount is shown in Figure 8 as a function of dAp The (0, + 02)area drops precipitously with OA1, and the peak is essentially gone by O, = 0.7. The y peak does not grow until a coverage of 0.1, a t which point it rises rapidly to a maximum a t,,e = 0.7, and then drops to nearly zero by 8,1 = 1.4. The y-area maximum corresponds to rougly 60% of the saturation CO coverage that will adsorb a t 300 K on clean Ru-
+ os +
2512
The Journal of Physical Chemistry, Vol. 92, No. 9, 1988
1
7
0
5
2w
4
E a
i
CO/AP/Ru (00011 UNANNEALEO
6
a u)
n 3
c;
0 O 2
1
0
c @Al
Figure 7. Areas of the (p3 + p2)-C0TDS doublet and the y-CO TDS peak as a function of AI precoverage for 12-langmuirCO exposures to unannealed A1 overlayers on Ru(0001) at 150 K.
i i
CO/AP/Ru (0001 )
I
I
\ a \
I
\
'\ 1 -
O
0.0
0.5
h
1.0
\
+ 1.5 2
3 4
Figure 8. Areas of the ( p 3 + pz)-CO TDS doublet as a function of AI coverage for 12-langmuir CO exposures at 150 K to AI overlayers which had been annealed briefly to 1170 K. Also shown is the total area of all CO TDS peaks (p3 + p2 + y ) for saturation CO exposures to unannealed AI overlayers on Ru(0001) at 150 K. In both cases, the curves essentially represent the total amount of CO adsorption for a 12-langmuir exposure. (Owl),which is between 0.5810*'7and 0.67.13 That is, the y peak reaches a maximum coverage of OCo = and this occurs at an AI coverage 8, = 2/3. The total CO coverage for the unannealed A1 overlayer (Figure 8) decreases rapidly until OM 0.2 and then stays relatively constant as the decay in (p3+ p2) area is balanced by the increase in y. Finally, it drops to nearly zero as the y peak is blocked. The behavior of CO TDS from the Al/Ru adlayer that has been annealed to 1170 K is far simpler. As seen in Figure 6 , no y peak appears (except perhaps a very small peak at very high coverages which could be due to an incompletely annealed A1 overlayer). The (& p2) doublet narrows and then broadens to a slightly higher temperature with increasing A1 coverage. As seen in Figure 8, the (p3 p2) peak area decays to about 2 / 3 of its original intensity (or Oco = 0.39-0.45) and thereafter remains relatively constant with increasing A1 coverage. These results indicate that, even for initially thick AI films which completely block CO adsorption at room temperature, many Ru atoms become available
-
+
at the surface for bonding to CO after annealing. This would indeed be consistent with the growth of alloys of structure such as that shown in Figure 4, where the topmost surface layer offers both A1 and Ru atoms in large concentrations. If, for the annealed surfaces, the C O TDS experiment for a given A1 coverage was repeated by terminating the first TDS at 1170 K, cooling to room temperature and again adsorbing CO, the same CO TDS spectrum resulted. Furthermore, no species other than A1 and Ru were observable by AES after desorbing the CO to 1170 K. These results also indicate that, for the annealed surface, the removal of CO via desorption is quantitative by 1170 K.
IV. Discussion Let us first discuss the Al/Ru interface structures for the annealed and unannealed surfaces. The Al( 11 1) LEED features and the AES uptake curves both indicate that, at room temperature, the A1 grows on top of the Ru substrate. Thus, at low coverage, the AI goes down as adatoms above the undisturbed Ru(0001) surface. The appearance of Al( 111) features already a t OAl 2 0.4 indicates a tendency for these adatoms to coalesce into two- or even three-dimensional Al( 111) islands. After a 670 K anneal, the clear observation of coincidence satellites in the range 0.7 C 8,, C 1.9 shows that indeed a two-dimensional (one- or two-layer-thick) Al(111) film grows epitaxially on the Ru(0001) substrate. On further annealing to 1170 K, however, a marked transition to alloy formation occurs. This is evidence by the AES uptake curves, the A1 AES line shape change, the multilayer desorption kinetics, and the (v'3Xv'3)R30° LEED pattern, which is stable over a wide coverage range. This pattern is clearly interpretable in terms of the plane of the A12Ru intermetallic compound shown in Figure 4. The formation of this alloy requires that the A1 atoms move into the Ru lattice. It is therefore understandably a highly activated process. It is interesting to consider our results in light of the bulk phase diagram for the Al/Ru system.'* Since our system involves a small amount of A1 and a large amount of Ru, the thermodynamically preferred structure would be a mixture of a nearly pure Ru phase and the compound AIRu. The bulk compound A12Ru is only thermodynamically preferred at concentrations exceeding 50 atom % AI, and less than -75%. At higher A1 concentrations, A12Ru, A16Ru, and higher compounds appear (in mixtures of segregated phases).I8 Thus, when we observe the A12Ru compound after annealing to 1170 K, the system may not have yet reached its equilibrium structure. The fact that A1,Ru appears over such a broad coverage range may be due to the fact that it is kinetically prevented from converting to AlRu, or that the surface A1,Ru alloy is more stable than bulk A12Ru. The melting point of these compounds increases with decreasing A1 content from -950 K for the AI richest phase, to -2300 K for A1Ru.18 Since a substantial fraction ( -2/3) of the melting temperature is normally required for alloy growth in a reasonable time scale, we should not be surprised that AlRu is not formed here at 1170 K. On the other hand, A13Ru2 and A1,Ru melt as a eutectic at only 1570 K, and A12Ru was often observed to form rapidly from solid A13Ru2upon cooling below 1270 K.18 This testifies to the ease of formation of Al,Ru, which would explain why it is seen here. Our A1 thermal desorption data show the decomposition of a multilayer alloy (AlRu,) with an activation energy of -88 kcal/mol at 1300 K. Since there is typically no excess activation energy beyond the thermodynamic barrier for such atomic desorption steps, this energy approximates the reaction enthalpy for Al(gas) + xRu(so1id). Since the the process: AlRu,(solid) sublimation energy of A1 is -74 kcal/mol at this t e m p e r a t ~ r e , ' ~ we can approximate the heat of formation of the alloy as -(88 - 74) = -14 kcal/mol of Al. This value probably refers to either bulk AlRu or A1,Ru. The first A12Ru monolayer has an activation energy for desorption that is 10 kcal/mol larger (see above), reflecting a heat of formation of about -24 kcal/mol. To our knowledge, no other quantitative thermodynamic data exist for these alloys.
-
'---, I
OAl
+
Campbell and Goodman
-
-
-
-
-
Interaction of C O with Al/Ru(0001) Surfaces
The Journal of Physical Chemistry, Vol. 92, No. 9, 1988 2573
It is certainly clear that the structure of the Al/Ru interface greatly influences the way in which A1 modifies the chemisorption of CO on Ru(0001). For the annealed surface where A12Ru alloy formation occurs, the A1 sits within the topmost Ru layer. In this situation, the interaction of CO with Ru is only weakly effected, as seen in Figure 6. At BAl 2 0.67, the absolute C O coverage is reduced to 0.39-0.45, which is close to the value of expected if each Ru atom in the structural model of Figure 4 adsorbs one CO molecule. These Ru atoms are surrounded by A1 neighbors in the topmost plane and therefore might experience significant electronic influences. These, if they exist, are not strongly manifested in the C O desorption behavior. The well-known CO-CO lateral repulsion^'^^^ would be weak for C O adsorbed on Ru atoms in the geometry of Figure 4 but very strong at C O coverages above which exist on the clean surface. This coverage dependence appears to be as important in C O TDS as any directly Al-related electronic influence in the alloy. Thus, A1 seems fairly close to an ideal diluting agent in this alloy, replacing Ru atoms in the topmost atomic layer, without strongly affecting the nature of the Ru-CO chemisorption bond. The lack of electronic influence of A1 upon its neighboring Ru atoms, at least as reflected in the CO-TDS here, is rather surprising. On the other hand, on the unannealed interface when A1 adatoms sit on top of the Ru(0001) surface, they induce a very marked change in the C O chemisorption behavior. At high coverages, they eventually mask the surface completely from chemisorbing CO. This is due to the growth of an Al( 111) film. N o C O will adsorb on A1 at 300 K.I4 In addition, a new, very strongly adsorbed state of CO (y) appears at low coverages of Al. This state desorbs at -900 K, compared to -450 K from clean Ru(0001). This difference reflects a doubling in the activation energy for CO desorption. This state reaches a maximum coverage Bco N 0.33 at BAl = 0.67 (Figure 7). This would suggest that two AI atoms (and a single Ru surface atom) are needed for each C O molecule in the y state. The nonlinear increase in the coverage of this y state with A1 coverage for BAI < 0.2 would also be consistent with such a picture. We should p i n t out that the CO exposure of 12 langmuirs used for Figures 5-8 was proven to be sufficient for saturation on the clean Ru(0001) surface. It is possible that the C O sticking probability could be reduced by Al, so that these data do not necessarily reflect saturation coverages of C O for the Al-dosed surfaces. However, a C O exposure that was 2.5 times larger than those reported here gave no significant increase in the C O TDS peak area for the one case of an unannealed A1 overlayer of coverage BA1 N 0.7. This indicates that many if not all of these experiments were for saturation C O coverages. Very similar high-temperature C O desorption states are seen for Mn overlayers on Ru(0001).22 Clear isotopic scrambling in the case of Mn addition suggested that this C O desorption peak may be due to the recombination of dissociated CO: C, + 0, CO. However, UPS data suggested that the C O was still molecular.22 The C 0 recombination reaction on clean Ru surfaces typically gives C O desorption peaks in the range 550-650
-
+
K,5323
(22) Hrbek, J. J . Vac. Sci. Technol., A 1987, 5, 864; to be published.
unless the carbon is in a graphitic form.23 At this point, we can only speculate on the nature of this y state. The A1-0 bond can be very strong, as evidenced by the high stability of A120,. One can therefore imagine that the oxygen end of the C O molecule is somehow attached to the nearby AI adatoms. In any case, we envision a very direct interaction between C O and the Al adatoms. The small difference in Pauling’s electronegativities for A1 (1.61) and Ru (2.2) suggests the possibility of a weak electron charge transfer from A1 to Ru. This is, however, expected to be much smaller than for cases such as K on Ru(0001), where substantial charge transfer results in a marked indirect electronic influences on the Ru and consequent large changes in CO desorption beh a ~ i o r .The ~ ~ smaller charge transfer expected here requires that a more direct model for the AI-CO interaction be used to explain the very dramatic influences of A1 upon CO chemisorption. Very recently, Rao and co-workers have studied the adsorption of C O on Al-dosed Ni and Cu surfaces.25 They, too, found a very strong influence of A1 upon C O adsorption, evidenced both by a change in the valence energy levels and by a drastically reduced C-0 stretch frequency. They attributed this to an Alinduced precursor to dissociation and observed CO dissociation by 300 K. They did not comment on the thermal history of their A1 overlayers but mentioned that alloy formation had already occurred. They developed a molecular-orbital picture of the A1 effect by extended Hiickel calculations of Cu,Al-CO clusters. Their picture involved A1 bonding to the carbon end of the C O molecule via its 5 6 orbital, coupled with Cu-CO interactions via the la orbital. Similarly, Chen and co-workers, in the accompanying paper,l observe an EELS peak at 1370-1430 cm-I for CO on Ni/Al and assign this, for a number of reasons, to a Nix-C-0-AI, species which is a precursor to C O dissociation. In that case, the species dissociated completely between 450 and 700 K. N o C 0 recombinative desorption was seen by their maximum temperature (700 K). Given the similarities between Ni and Ru with respect to C O chemisorption, we propose a similar mechanism for the interaction of C O with A1 adatoms on Ru(0001). Thus, we envision the 0 end of adsorbed C O being attracted to the A1 adatom and forming a bond which eventually leads to C O dissociation. We thus tentatively assign the 900 K y-desorption state for CO to the recombinative desorption of C + 0. This occurs at a higher temperature than observed on clean Ru(0001), which is at -600 K,5323since now the oxygen and/or carbon adatoms are stabilized by interactions with Al.
+
Acknowledgment. We acknowledge useful discussion with Dr. Jan Hrbek and Professor John T. Yates, Jr. Both C.T.C. and D.W.G. acknowledge partial support of this work by the Department of Energy, Office of Basic Energy Sciences, Division of Chemical Sciences. Registry No. AI, 7429-90-5; Ru, 7440-18-8; CO, 630-08-0; A12Ru, 12043-25-3. (23) Lauderback, L. L.; Delgass, W. N. Surf.Sci. 1986, 172, 715. (24) de Paola, R. A.; Hrbek, J.; Hoffmann, F. M. J . Chem. Phys. 1985, 82, 2484. (25) Rao, C. N. R.; Rajumon, M. K.; Prabhakaran, K.; Hedge, M. S.; Kamath, P. V. Chem. Phys. Lett. 1986, 129, 130.
