Chemisorption of Ultrathin Pd Layers on W(110) and

Transcription

1 Langmuir 1988,4, Chemisorption of Ultrathin Pd Layers on W(110) and W(100): Adsorption of H2 and COT,$ 1091 Paul J. Berlowitz and D. Wayne Goodman* Surface Science Division, Sandia National Laboratories, Albuquerque, New Mexico Received January 28, In Final Form: April 12, 1988 The structural and chemisorptive properties of thin films ( monolayers, ML) of Pd evaporated onto W( 110) and W(100) single-crystal surfaces have been examined by low-energy electron diffraction (LEED), Auger electron spectroscopy (AES), and thermal-programmed desorption (TPD). LEED results indicate that at coverages up to 1 ML the Pd films grow pseudomorphically with respect to the W(110) and W(100) substrates. AES data indicate layer-by-layer growth of Pd on both surfaces at 100 K. TPD of Pd from W(110) and W(lO0) yields two discrete features, which are related to Pd desorption from the monolayer and multilayer structures. For both surfaces heats of sublimation of an annealed Pd multilayer determined from the TPD data are consistent with three-dimensional island formation of Pd in excess of 1 ML. TPD results following the chemisorption of CO onto monolayer Pd films on both W(ll0) and W(100) indicate a significant weakening of the CO binding strength, corresponding to a decrease of approximately 200 K in the CO desorption temperature maximum, compared with CO on bulk Pd. TPD of Hz from similarly prepared Pd f ii shows evidence of adsorption on the overlayer Pd as well as adsorption at the Pd/W interface. Following a thermal treatment that leads to three-dimensional Pd islands, absorption of hydrogen into these islands is also observed. Introduction In recent years considerable attention has been focused on the properties of mixed-metal and bimetallic systems.' The modifications of the electronic and structural properties at a metal-metal interface are substantial and can lead to improvement in catalytic activity and selectivity, corrosion resistance, and other attributes of the mixed metal system over either of the individual components.2 A common method of studying these mixed-metal systems is through the use of vapor-deposited metal films.3 These systems, which are constructed by evaporating one metal onto a second metal single-crystal substrate, create a useful analogue to mixed-metal systems, which can then be characterized by employing a wide variety of surface analytical tools under ultrahigh vacuum. Combining surface analytical probes with "real world" conditions, for example, by use of a contiguous high-pressure catalytic reactor or an electrochemical treatment cell, provides a useful model of mixed-metal systems that is amenable to detailed st~dy.~ In this laboratory recent work into the structural, electronic, and catalytic properties of the C~/Ru(0001)>~~~ Ni/W(110), and Ni/W(100)7~8 systems has demonstrated that metal overlayer systems provide a valuable tool for studying the relative importance of electronic and structural modifications on catalytic activity. For Cu/Ru and Ni/W, the first metal monolayer adsorbed grows pseudomorphic with respect to the underlying substrate and thus is "strained" compared to the bulk metal. In this first layer the chemisorptive and geometric properties of the overlayer metal are modified, and these modifications have been directly linked to changes in the catalytic activity. For example, Peden and Goodman have shown that submonolayer amounta of Cu or Ru(0001) increase the specific activity of the Ru(0001) substrate by almost 1 order of magnitude for the cyclohexane dehydrogenation rea~tion.~ Greenlief et alas have shown that the poisoning of the t Presented at the symposium on "Bimetallic Surface Chemistry and Catalysis", 194th National Meeting of the American Chemical Society, New Orleans, LA, Sept 1-3, 1987; B. E. Koel and C. T. Campbell, Chairmen. t This work, performed at Sandia National Laboratories, was supported by the US. Dept. of Energy under Contract No. DE-, AC04-76DP /88/2404-lO91$01.50/0 methanation reaction through carbide buildup is altered because of the perturbation of hydrogen bonding on Ni/ W(110). These workers also found that the rate of ethane hydrogenolysis is strongly dependent on the geometry of the Ni surface by comparing bulk Ni to Ni/W(110) and Ni/ W( Palladium has been chosen for this study because of its catalytic importance and its unique hydrogen absorption properties. In addition, previous work on the Pd/Nb"12 and Pd/Ta13-16 systems has shown that Pd undergoes very large perturbations in electronic structure over the first several adsorbed monolayers and that these perturbations can dramatically affect the hydrogen absorption kinetics.12 In this study we add to the previous work by Bauer and co-workers on the structure of Pd layer on W(l10)"J8 and (1) Bauer, E. The Chemical Physics of Solid Surfaces and Heterogeneow Catalysis; King, D. A,, Woodruff, D. P., Eds.; Elsevier: New York, 1984; Vol. 3. (2) Sinfelt, J. H. Bimetallic Catalysts; Wiley: New York, (3) Peden, C. H. F.; Goodman, D. W. J. Catal. 1986,100, 520. (4) Goodman, D. W. Annu. Rev. Phys. Chem. 1986,37,425. (5) Houston, J. E.; Peden, C. H. F.; Blair, D. S.; Goodman, D. W. Surf. Sci. 1986, 167, 427. (6) Houston, J. E.; Peden, C. H. F.; Feibelman, P. J.; Hamann, D. R. Phys. Reu. Lett. 1986, 56, 375. (7) Berlowitz, P. J.; Goodman, D. W. Surf. Sci. 1987, 187, 463. (8) Greenlief, C. M.; Berlowitz, P. J.; Goodman, D. W.; White, J. M. J. Phys. Chem. 1984,91, (9) Sagurton, M.; Strongin, M.; Jona, F.; Colbert, J. Phys. Rev. B: Condens. Matter 1983,28, (10) El-Batanouny, M.; Hamann, D. R.; Chubb, S. R.; Davenport, J. W. Phys. Rev. B: Condens. Matter 1983,27, (11) El-Batanouny, M.; Strongin, M.; Williams, G. P. Phys. Reu. B: Condens. Matter 1983,27, (12) El-Batanouny, M.; Strongin, M.; Williams, G. P.; Colbert, J. Phys. Rev. Lett (13) Pick, M: A:; Davenport, J. W.; Strongin, M.; Dienes, G. J. Phys. Reu. Lett. 1979,43, 286. (14) Koel, B. E.; Berlowitz, P. J., in preparation. (15) Nieman, D. L.; Koel, B. E. in Physical and Chemical Properties of Thin Metal Ouerlayers and Alloy Surfaces; Zehner, D. M., Goodman, D. W., Eds.; MRS: Pittsburgh, 1987; P 143. (16) Ruckman, M. W.; Stringin, M.-Phys. Rev. B: Condens. Matter 1984,29, (17) Paraschkevov, D.; Schlenk, W.; Bajpai, R. P.; Bauer, E. Proc. 7th Intern. Vac. Congr. & 3rd Intern. Conf. Solid Surfaces, Vienna, 1977; p (18) Schlenk, W.; Bauer, E. Surf. Sci. 1980, 93, American Chemical Society

2 1092 Langmuir, Vol. 4, No. 5, 1988 W(100)19 and examine the chemisorptive and absorptive properties of Pd overlayers on these two W surfaces. Experimental Section A conventional UHV system, described previously,20 was used in this work. This apparatus was equipped with Auger electron spectroscopy (AES), low-energy electron diffraction (LEED), and temperature-programmed desorption (TPD) spectroscopy. Gas dosing was carried out by using a stainless steel needle with a local pressure enhancement of 25:l. The W(110) and W(100) crystals, sample mounting, and sample cleaning procedure have been described in detail previously.' Briefly, the samples were mounted on a manipulator capable of resistive heating to 1800 K, electron beam heating to 2500 K, and cooling to 100 K. Sample temperature was measured by a W/5% Re-W/26% Re thermocouple. Cleaning involved heating under lo-' Torr of oxygen at 1800 K followed by annealing in vacuo at 2500 K. Cleanliness was determined by AES, with all impurity concentrations determined to be el%. Pd metal was deposited by heating a 0.25-mm W filament wrapped with 0.1-mm, highpurity Pd wire. The Pd doser was located in line-of-sight to the mass spectrometer; thus evaporation rates could be precisely controlled by directly measuring the Pd flux. After evaporation of Pd onto the W substrates, no impurities were detected by AES analysis. Research purity H, (99.999%) and CO (99.99%) were supplied by Matheson and used without further purification. Gas exposures reported are corrected for the 251 enhancement of the doser and represent the approximate exposure to the front face of the sample. Gas and metal dosing were always carried out at K; TPD ramp rates were always 10 K/s. Results and Discussion Growth of Pd Films on W(110) and W(lO0). The growth of ultrathin Pd films on these substrates has been studied in detail by Bauer and co-w~rkers"-~~ and by Butz and Wagner.21 Since our results are in general agreement with these previously reported results, only a brief review of Pd film growth will be given with emphasis on the aspects of our work that are at variance with the published literature. Pd growth on W(ll0) at 100 K is layer-by-layer. The first layer grows with one-dimensional pseudomorphism and an atomic density slightly below that of W(110). Annealing this layer results in a (1x1) LEED pattern and a fully pseudomorphic Pd layer with an atomic density equal to that of W(110); the work function change from the clean W(110) surface following the deposition of 1 ML of Pd is approximately -600 mev.17j8 Pd layers with thicknesses of up to 4 ML in Bauer's ~ork'~-'~ and 5 ML in our work remained stable to annealing temperatures of K. Above this temperature all Pd layers subsequent to the first agglomerate into three-dimensional islands, leaving a large portion of the pseudomorphic first monolayer exposed. Desorption spectra of Pd from W(110) are shown in Figure la. These results are in general agreement with previous work.17j8 The multilayer peak at approximately 1320 K exhibits zero-order desorption kinetics indicative of bulk sublimation. The sublimation energy determined for this peak, 91 f 3 kcal/mol, is in excellent agreement with a bulk sublimation value of 93 kcal/mo1.2z The desorption energy for coverages of ML is 102 f 5 (19) Prigge, S.; Roux, H.; Bauer, E. Surf. Sci. 1981, 107, 101. (20) Goodman, D. W.; Yates, J. T., Jr.; Peden, C. H. F. Surf. Sei. 1985, 164, 417. (21) Butz, R.; Wagner, H. Proc. 7th Intern. Vac. Congr. & 3rd Intern. Conf. Solid Surfaces, Vienna, 1977; p (22) Handbook of Chemistry and Physics, 57th ed., Weast, R. C., Ed.; CRC: Boca Raton, FL, 1978; p D-62. 'Pd' I o L43 L30 Berlowitz and Goodman (a) IS00 I Figure 1. (a) TPD spectra of Pd from W(110) following deposition at 100 K. (b) TPD spectra of Pd from W(100) following deposition at 100 K. kcal/mol, within experimental error of the values reported by Schlenk and Bauer.18 The temperature separation between monolayer and multilayer peaks is substantially larger for Pd/W(110) than the temperature separation observed for the Ni/W(110),7~z3 Fe/W(110),z4 or Cu/W- (110) systemsz5 but is comparable to the monolayermultilayer peak temperature separation observed for Pd on Ta(iiO).14 Results for Pd on W(lO0) are in many respects similar to those for Pd on W(ll0). The Pd(330 ev)/w(169 ev) AES ratio versus the monolayer quantity of Pd deposited as determined by thermal desorption is consistent with layer-by-layer growth of Pd at a deposition temperature of 100 K. The AES data indicate that a Pd bilayer is stable to annealing near the Pd desorption temperature, while up to 20 Pd layers were stable in the as-deposited structure to a K anneal. TPD results indicate the atomic density of each layer of the first three Pd layers deposited at 100 K is approximately equal, implying a similarity among layer structures. LEED patterns of Pd coverages below 1 ML, and of coverages from 1-4 ML after a K anneal, show essentially the W(100) (1x1) pattern, although a slight increase in background intensity is evident. LEED patterns below 1 ML indicate that the Pd overlayer is predominantly pseudomorphic. Thus, for certain annealing conditions, several pseudomorphic Pd layers can be formed, resulting in a thin Pd(100) BCC film. The increase in background intensity is indicative of the considerable defect density in these strained Pd layers. Complex LEED structures were observed by Prigge et a1.18 for these Pd multilayers. (23) Kolaczkiewicz, J.; Bauer, E. Surf. Sci. 1984, 144, 495. (24) Berlowitz, P. J.; Goodman, D. W., in preparation. (25) Bauer, E.; Poppa, H.; Todd, G.; Bonczek, G. J. Appl. Phys. 1974, 46, 5184.

3 Pd Chemisorption on W(110) and W(100) Langmuir, Vol. 4, No. 5, I I I I Fi ire 2. TPD of CO v8 Pd coverage on W(110) for an exposure of approximately 2 langmuirs at 100 K. Thermal desorption measurements of Pd from W(100), shown in Figure lb, are in qualitative agreement with the results reported previously,18 although the temperatures for desorption of the Pd monolayer are K lower in this study. The multilayer peak exhibits zero-order desorption with a sublimation energy within experimental error of that determined for multilayer Pd/W(llO). A distinct feature at K, observed by Prigge et al.18 for Pd coverages below 1.5 ML, was not observed in this study. Adsorption of CO on Pd/W(110) and Pd/W(100). Thermal desorption spectra for saturation coverages of CO from Pd/W(110) are shown in Figure 2 as a function of Pd coverage and compared with thermal desorption spectra of CO from Pd(ll1) and W(110). At submonolayer Pd coverages two peaks are clearly evident at 375 and 285 K. The feature at 375 K corresponds to CO desorption from bare W(ll0) patches and thus decreases with increasing Pd coverage. The feature at 285 K with a shoulder at 250 K grows with increasing Pd coverage. These features correspond to desorption from strained-layer Pd. All Pd layers were annealed to 1100 K prior to CO adsorption. As discussed above, this anneal temperature results in the formation of three-dimensional islands from Pd coverages in excess of 1 ML. For Pd coverage of 1 ML, no CO desorption from W- (110) is evident, consistent with a uniform, ordered Pd overlayer. The peak desorption temperature maximum for CO from 1 ML of Pd is less than the corresponding value for bulk Pd by approximately 200 K. Increasing the Pd coverage above 1 ML, with the concurrent formation of three-dimensional Pd islands, introduces a shoulder into the CO desorption spectra at -375 K. This shoulder grows slowly with coverage and shifts toward higher temperatures. This peak corresponds to desorption from the three-dimensional Pd islands. That the temperature of this peak shifts slowly with coverage and the CO desorption feature covers a broad temperature range imply significant perturbation of the Pd atoms in the three-dimensional islands over several metal overlayer thicknesses. On the W(100) surface, the CO desorption experiments were performed after annealing to only 800 K. This results CO/W(110) I I 1 I Figure 3. TPD of CO vs Pd coverage on W(l00) for an exposure of approximately 2 langmuirs at 100 K. in metastable Pd layers, which exhibit the substrate (1x1) LEED pattern. In addition, this annealing procedure results in a relatively flat overlayer structure in that the amount of CO desorbing from the surface increases by only 10-20% as the Pd coverage incrased from 1 to 5 ML. These experiments could not be carried out on the Pd/ W(ll0) system since Pd layers in excess of 1 ML are not stable to temperatures required to desorb completely the co. CO desorption spectra from Pd/W(100) are shown in Figure 3 and contrasted with CO desorption from W(l00) and bulk Pd. The results for monolayer and submonolayer Pd depositions are quite similar to CO desorption from Pd on W(110). The desorption temperature maximum is shifted to a much lower temperature, 305 K, compared to bulk Pd. A broadening of the desorption peak shape for submonolayer coverages is observed because CO desorption from W (100) and from strained-layer Pd/ W (100) occurs at approximately the same temperature. For Pd coverages above 1 ML, perturbative effects on the CO desorption, which extend across several lattice spacings, are evident. From 1 to 4.8 ML, the peak maxima for CO desorption shifts from 305 to 370 K. Some broadening and a 10-20% increase in the quantity of CO are observed, likely a result of the nonuniform nature of the metastable metal overlayer surface. The results presented above are consistent with results reported previously for Pd overlayers on Nb ll and Ta. 16 In both cases large perturbations in the electronic structure were observed by photoemission. These results showed that the electronic structure of the overlayer Pd resembles that of a noble metal in that the Pd d-band was shifted to higher binding energies. The density of states at the Fermi level was significantly less than that for bulk Pd. The CO desorption results presented here, with TPD peak temperatures at or below 300 K, are intermediate between CO desorption from noble metals (e.g. for CO/Cu, T < 200

4 1094 Langmuir, Vol. 4, No. 5, 1988 K) and CO from group VI11 metals ( K). Furthermore, these results are similar to CO desorption from monolayer films of Pd/Ta and Pd/Nb,'"le both of which exhibit peak temperatures below 300 K. It should be noted that Neiman and Koel15 report altered chemisorption properties for Pd films on Nb of thicknesses up to five layers. The results presented here are in general agreement with the previous work by Prigge et alex on the CO/Pd/W(110) system. The study of Prigge et al. was performed at 300 K, thus CO chemisorption on the Pd monolayer was not observed. As in the present study, the amount of CO adsorbed directly on the W(110) substrate (the only peak measured in the study by Prigge et al. at submonolayer Pd coverages) decreased with increasing Pd coverage and reached zero at 1 ML of Pd. CO adsorption on the second monolayer of Pd was observed, consistent with CO desorption from Pd islands. The peak temperature for 1.8 ML of Pd in the Prigge et al. study, K, is similar to the temperature observed for the high-temperature shoulder here. Note that the appearance of a TPD peak associated with the three-dimensional Pd islands on W(ll0) is a clear indication that Pd does not form an alloy upon annealing. However, due to the limited surface area of the islands, large Pd coverages must be initially present in order for these islands to be detected by chemisorption. Similar chemisorptive behavior regarding three-dimensional island formation has been observed for the Ni/W ~ystem.~ On the W(100) surface, the stability of the Pd overlayers has allowed the evaluation of the strength of the perturbation of the W substrate on the overlayer electronic structure over several Pd layers. In contrast to the Ni/ W ~ystem,~ where chemisorption was perturbed only slightly in the second monolayer, the perturbative effect of W on Pd is much more significant and extends over several overlayer lattice spacings. Adsorption and Absorption of H2 on Pd/W(llO) and Pd/ W( 100). Adsorption of H2 was carried out at 110 K on the Pd/W(110) surface. As in the case for CO adsorption, the Pd layers were annealed to 1100 K prior to H2 adsorption, resulting in the formation of three-dimensional Pd islands for coverages greater than 1 ML. H2 TPD spectra are shown as a function of Pd coverage and compared with H2 desorption from W(110) and bulk Pd in Figure 4. Two main features in these spectra are clearly distinguishable: a broad peak at 510 K, which is similar to Hz desorption from W(llO), and a smaller feature at K, which appears only for Pd coverages > 1 ML. The TPD feature at 510 K has several interesting aspects. The area of this peak, including the low-temperature shoulder, decreases slightly with coverage. This feature has a peak maximum that is similar in temperature to the peak maximum of H2/W(110) but much higher in temperature than the maximum of chemisorbed H2/Pd.27 The low-temperature shoulder at K decreases in intensity for very large Pd coverages. At submonolayer coverages this 510 K feature could correspond to either adsorption on bare W(110), adsorption on the strained Pd surface, or absorption at the Pd/W interface. At Pd coverage greater than one monolayer, where there is no directly exposed W substrate, the H2 TPD spectrum resembles desorption from W(llO), rather than Pd, consistent with an interfacial hydrogen chemisorbed state. That hydrogen could penetrate a strained Pd layer to the (26) Prigge, D.; Schlenk, W.; Bauer, E. Surf. Sci. 1982, 123, L698. (27) Belm, R. J.; Penak, V.; Cattania, M. G.; Christmann, K.; Ertl, G. J. Chem. Phys. 1983, 78, H2 /Pd I I I I I I / I I \ Berlowitz and Goodman H2 /Pd/W(110) epd = 19.5 &- 5.0 A Figure 4. TPD of Hz vs Pd coverage on W(110). H2 exposure was approximately 4 langmuirs at 100 K. Pd/ W interface is not unexpected, considering the facility with which hydrogen penetrates bulk Pd Evidence presented below for Hz adsorption on Pd/W(100), as well as for adsorption on Fe/W(100)F4 support the presence of interfacial hydrogen. The shoulder in the TPD curve below the 510 K absorption state can be attributed to chemisorption on the Pd layer. The shoulder is most pronounced at 1 ML of Pd coverage, where there is little interference from desorption of Hz from the W substrate. The shoulder is less pronounced at very large coverages, where Pd islands begin to cover a larger proportion of the first Pd layer. The low-temperature peaks correspond well with the known desorption behavior of Pd.27932t33 The peak at approximately 200 K corresponds to desorption of hydrogen from a hydride phase,n*28 while the peak at 275 K is typical of desorption from chemisorbed hydrogen on Pd. Thus, these peaks are directly attributable to Pd threedimensional islands and exhibit properties similar to bulk Pd. Hydrogen desorption from Pd/W(100) is shown in Figure 5. For these experiments the Pd layer were annealed to 800 K to form metastable (1x1) layers. As observed for adsorption of H2 on Pd/W(110), two features are apparent. A broad feature with a peak maximum at 525 K grows with coverage and saturates at 1-ML coverage. Additional deposition of Pd does not affect the area or peak desorption temperature maximum. A second feature, ~~ (28) Gdowski, G. E.; Felter, T. E.; Stulen, R. H. Surf. Sci. 1987,181, L147. (29) Bucurr, R. V. Surf. Sci. 1977, 62, 519. (30) Peden, C. H. F.; Kay, B. D.; Goodman, D. W. Surf. Sci. 1986,175, 215. (31) Kay, B. D.; Peden, C. H. F.; Goodman, D. W. Phys. Reu. E Condens. Matter 1986, 34, 817. (32) Conrad, H.; Ertl, G.; Latta, E. E. Surf. Sci. 1974, 41, 435. (33) Cattania, M. G.; Penka, V.: Behm, R. J.; Christmann, K.; Ertl,G. Surf. Sci. 1983, 126, 382.

5 Pd Chemisorption on W(110) and W(lO0) Langmuir, Vol. 4, No. 5, Considering the behavior of bulk Pd, the ability of strained Pd layers to form an interfacial hydrogen state and to exhibit absorption of hydrogen into a hydride phase is not at all surprising. A thick layer of Pd on W(lO0) displays properties quite similar to the three-dimensional islands present for annealed Pd/W(110) in terms of the formation of chemisorbed and absorbed hydride species. Shifts in the Hz desorption peaks for these states from values for bulk Pd are relatively small, <50 K, and the long-range perturbative effects observed for CO are significantly less for H2. Interestingly, significant H2 adsorption on the Pd/W(110) monolayer is not observed. It is likely that the prominent low-temperature shoulder at K in these TPD may be due to hydrogen chemisorption; however, the area under these peaks is less than 50% of that corresponding to a monolayer. The fact that this peak is most prominent at 1 ML of Pd, and then decreases with increasing coverage as the first layer is covered with three-dimensional islands, supports the contention that this feature corresponds to chemisorbed hydrogen Figure 5. TPD of H2 vs Pd coverage on W(lO0). Hz exposure was approximately 4 langmuirs at 100 K. with a peak desorption temperature of 350 K at low Pd coverage, shifts to -300 K as the Pd coverage is increased to >1 ML. At large coverages a prominent low-temperature shoulder at 200 K is apparent. The high-temperature peak, which saturates at 1 ML and is directly proportional in area to the Pd coverage below 1 ML, can be assigned to interfacial hydrogen. In contrast to Pd/W(110), this feature does not overlap with desorption of hydrogen from W(100) and thus can be unambigously defined. The low-temperature feature undergoes a continuous transition with Pd coverage, converging to bulk Pd features at coverages of 1 ML or greater. Furthermore, this TPD feature sharpens and loses the high-temperature shoulder associated with Hz desorption from W(lO0). For coverages above 1 ML, the 200 K shoulder becomes more prominent and likely corresponds to absorption of hydrogen into the Pd layer to form the hydride phase.28 Note that for Pd coverages of 1 ML or greater, the total amount of hydrogen adsorbed and absorbed is at least double the amount adsorbed on clean W(lO0). This is a further indication of the presence of an absorbed interfacial hydrogen state. Summary At coverages of 1 ML and less, Pd forms a pseudomorphic layer on W(ll0) and W(l00). Annealing this layer to 1200 K does not perturb the AES, TPD, LEED, or chemisorption properties of the overlayer. On W(110) the first monolayer exhibits close to zero-order kinetics with an activation energy of 91 kcal/mol. Above 1 ML, Pd grows layer-by-layer when deposited at 100 K; however, annealing multiple layers at K results in the formation of three-dimensional islands. Multilayers of Pd on W(110) and W(100) desorb with apparent zero-order kinetics, with a desorption energy equal to the heat of sublimation of bulk Pd. Chemisorption of CO on either surface results in a decrease in binding strength compared to bulk Pd, as exhibited by a decrease in the TPD peak temperature of 200 K. Adsorption of H2 on multilayer Pd on W yields both chemisorbed hydrogen as well as interfacial hydrogen, which results from penetration of adsorbed hydrogen through the overlayer to the Pd/ W interface. Exposure of the Pd/W system to Hz in the presence of three-dimensional Pd islands leads to absorption of hydrogen into these islands. Acknowledgment. We acknowledge with pleasure the partial support of this work by the Department of Energy, Office of Basic Energy Sciences, Division of Chemical Sciences. We also acknowledge many helpful discussions with J. E. Houston. Registry No. Pd, ; W, ; Hz, ; CO,