CO-Assisted Subsurface Hydrogen Trapping in Pd(111) Films - The

Letter pubs.acs.org/JPCL

CO-Assisted Subsurface Hydrogen Trapping in Pd(111) Films J. I. Cerdá,*,† B. Santos,‡,∥ T. Herranz,‡,⊥ J. M. Puerta,† J. de la Figuera,‡ and K. F. McCarty§ †

Instituto de Ciencia de Materiales de Madrid, CSIC, Madrid 28049, Spain Instituto de Química-Física Rocasolano, CSIC, Madrid 28006, Spain § Sandia National Laboratories, Livermore, California 94550, United States ‡

S Supporting Information *

ABSTRACT: We use low-energy electron microscopy to image CO displacing adsorbed H from the Pd(111) surface. Quantitative electron diffraction reveals that upon codosing atomic H and CO the latter assists the absorption of hydrogen, which ends up trapped between the first and second palladium layers, and blocks its desorption. Density functional calculations reproduce this effect, which is found to originate from the antibonding character of the interaction between the adsorbed hydrogen and the CO π states.

SECTION: Surfaces, Interfaces, Catalysis

A

CO-induced absorption of H into ultrathin palladium films. Spectral information contained in electron reflectivity is used to image CO displacing adsorbed H. To overcome the limits of dosing H2 under ultrahigh vacuum conditions,19,22 we use atomic H to increase the absorbed H concentration. Quantitative electron diffraction and DFT-based theory reveal that coadsorption of CO and atomic H at 170 K efficiently traps the later at octahedral sites between the first two Pd layers. The experiments were carried out in a Elmitec LEEM III with a base pressure below 10−10 Torr. LEEM is a nonscanning microscopy technique where a beam of electrons reflected from a surface is magnified by lenses, giving a lateral resolution of ∼10 nm, all while the sample is heated or while dosing gases at chamber pressures below ∼10−5 Torr.23,24 Pd(111) films were grown on a Ru(0001) crystal by molecular beam epitaxy during LEEM observation.25 Films between 2 and 20 atomic layers thick were studied to explore finite-size effects in H absorption. H2 and CO were separately dosed from gas reservoirs using leak valves. Atomic hydrogen and CO were codosed from a cracker consisting of a hot tungsten filament in a water-cooled tube through which H2 was leaked. (The cracker also generated CO, giving a flux estimated to be ∼10−3 times the H flux.) Gas dosing was performed at 170 K, the lower limit of the sample manipulator, and exposures are given in Langmuirs (1 L = 10−6 Torr × s, with the pressure measured by an ionization gauge). Figure 1a shows a LEEM image of a typical Pd film, which consists of atomically flat terraces separated by atomic steps. Information about the concentration and type of adsorbate

classic system of gas−metal interactions is hydrogen and palladium.1 This system lies at the heart of applications ranging from purifying hydrogen to catalysis. The interplay of H and carbon monoxide on Pd is also of great interest because of their role in gas sensors,2 hydrogen purification,3 and heterogeneous catalysis. In the latter, Pd is used in selective hydrogenation reactions,4,5 CO oxidation, methanol synthesis,6 and steam reforming of alcohols to obtain hydrogen and CO.7,8 H and CO coadsorption on Pd surfaces has been studied with a wealth of surface-sensitive techniques, including thermal desorption,9 electron diffraction,10 vibrational spectroscopies,11 scanning tunneling microscopy,12 and theoretical simulations based on density functional theory (DFT).13 (See refs 11. and 13 for a literature review.) The results reveal a complex phase diagram strongly dependent on pressure and deposition temperature as well as on the order in which the species are adsorbed.11 As a general rule, CO and H on Pd surfaces repel each other,12 so that coadsorption at low temperatures gives pure domains of H and CO that compress as their respective coverages increase. In sequential dosing, the first-dosed species blocks adsorption of the second species.11 However, when a H precovered surface is raised to ∼125 K, CO adsorption leads to the complete removal of the adsorbed H.11 Part of the H desorbs after recombination and another portion dissolves into the Pd interior,10,14,15 leaving pure CO superstructures on the surface. Absorbed H in Pd is known to affect the rate and selectivity of catalytic reactions,4,16 and theory suggests that it is most strongly bound between the topmost two Pd layers,17 a proposal that has some experimental support.18,19 How much H is displaced into Pd by CO, its interior location, and whether its concentration can be controlled are open questions. Here we use low-energy electron microscopy (LEEM)20,21 to study © 2011 American Chemical Society

Received: November 2, 2011 Accepted: December 12, 2011 Published: December 12, 2011 87

dx.doi.org/10.1021/jz201455s | J. Phys. Chem.Lett. 2012, 3, 87−91

The Journal of Physical Chemistry Letters

Letter

of low-energy electron diffraction (LEED) from selected areas, the absorbed H can be sensitively detected by measuring the interlayer spacing of the outermost Pd layers.25 We do this by measuring the intensity of the LEED spots versus electron energy (voltage) and analyzing this LEED-I(V) data using multiplescattering simulations.28 This approach measures the expansion of the top Pd interlayer spacings with high precision, well below 0.1 Å, providing a sensitive probe for absorbed H. For instance, the formation of an ordered β-PdH phase should increase the Pd interlayer spacing by 3%.18 Even with or without a longranged ordered adsorbate phase,18 a reliable LEED-I(V) analysis may be still performed by considering only the integer (Pd-related) diffraction spots. Such approach is further justified given the small scattering strengths of H and CO. (In the SI, we provide all details of the analysis.) The first three columns of Table 1 give the spacings between the top two Pd layers (d12) Table 1. Interlayer Spacings Deduced from the LEED-I(V) Analysis for Pd(111): Clean, Saturated with Adsorbed H by Dosing with Molecular Hydrogen, Saturated with Adsorbed CO by Dosing with Pure CO, after CO Displaces Adsorbed H (Figure 1c−h) and after Co-Dosing Atomic H and COa

Figure 1. (a) LEEM image of a Pd(111) film on Ru(0001). Majority of film is six Pd layers (medium contrast) with smaller amounts of seven (dark) and five layers (bright). Smooth bands result from bunches of underlying Ru steps. Field of view is 9.3 μm. (b) Low-energy electron reflectivity curves for clean (bottom), H-covered (middle, red, 0.7 kLH), and CO-covered (top, green, 63 LCO) Pd films, respectively. (c−h) Sequence of LEEM images acquired while exposing a H-covered surface to CO. Each image combines frames from energies that gives strong intensities from adsorbed H (10 eV, red) and CO (16 eV, green), respectively. CO doses in panels c−h are 6, 27, 47, 55, 60, and 63 LCO.

Pd clean

Pd+H

Pd+CO

2.30 (1.1) 2.28 (0.2)

2.29 (0.7) 2.29 (0.7)

2.30 (1.1) 2.30 (1.1)

Pd+CO

Pd+CO+H

(H displaced) d12 d23

2.31 (1.5) 2.28 (0.2)

2.37 (4.2) 2.28 (0.2)

a

Values in parentheses give the percentage increase with respect to the average interlayer spacing of the inner Pd layers (2.275 Å). Only results for the 6 ML thick Pd films are shown. All distances are in angstroms. Error bars for d12 (d23) are always smaller than ±0.03 Å (±0.05 Å). (See the SI for further details.) H-covered = 4 kLH2, COcovered = 5 LCO, CO + H = 5 kLH + ∼5 LCO.

comes from measuring how the electron reflectivity varies with the energy of the incident electrons. Spatially resolved measurements are obtained by measuring the reflectivity from a series of LEEM images obtained as a function of energy. The reflectivity spectra provide fingerprints for adsorbates such as CO23,24 or H.26 Bare Pd(111) has a broad reflectivity peak around 22 eV (bottom curve in 1b). Exposing to either atomic or molecular hydrogen induces a broad new peak around 8− 11 eV (middle curve). In contrast, adsorbed CO shifts the broad peak at 22 to ∼16 eV (upper curve). (The narrower peaks below ∼15 eV result from interference effects in scattering from the thin film.27) Heating the film desorbs the adsorbates giving back the reflectivity of the bare film. Next, we show how the displacement of adsorbed H by CO can be imaged in real space. In Figure 1c−h, we show composite images of two LEEM frames, one sensitive to adsorbed H (10 eV) and the other to CO (16 eV). The frames were colored red and green, respectively, with brightness proportional to LEEM intensity and merged to form a color image. [See the Supporting Information (SI) for further details of the procedure.] Figure 1c shows the H-covered surface (red) after saturating the film of Figure 1a with 22 LH2. Figure 1d−h shows the evolution during the subsequent CO dosing. Starting from step bunches, the CO domains (green) nucleate and grow at the expense of the H areas (red), which finally disappear. This result demonstrates the power of LEEM-based electron reflectivity to image the time-dependent spatial distribution of coadsorbates. The reflectivity technique is not sensitive to any H inside the film. Therefore, another approach is needed to measure the amount and location of the H displaced from the surface to the interior by the CO. Because LEEM also allows the acquisition

and the second and third layers (d23) for clean, H- and COcovered surfaces. For the latter two, we find a very slight expansion between the topmost two Pd layers (Δd12 ≈ 1%), which is incompatible with a significant population of hydrogen within the first three surface layers.18 (Some dissolution into the Pd bulk cannot be excluded, however.) The same conclusion is reached for dosing atomic or molecular H. So the question arises about the fate of H displaced by CO adsorption (as in Figure 1c−h), some of which is injected into the Pd interior instead of desorbing.10,14 The LEED-I(V) analysis for the final CO-covered film gives again a very small expansion of the topmost Pd−Pd layers (fourth column of Table 1), showing that the amount of H displaced by CO from a H-saturated surface is insufficient to give an appreciable H concentration at the first subsurface sites. We next show that codosing of atomic H and CO, however, does lead to a significant H concentration between the first and second Pd layers. We first establish that codosing atomic H in a small background of CO leads to adsorbed H that is subsequently displaced. Figure 2 shows how the electron reflectivity changes during the codosing. First, the surface becomes covered with H (10 eV peak). Then, the CO adsorption begins (16 eV peak) and H is removed from the surface. On the basis of the distinctive LEED patterns,11 we estimate the final surface to be covered by 0.6 to 0.7 of a CO monolayer (ML). Unlike the case when CO displaced H from a H-saturated surface (Figure 1c−h and fourth column of Table 1), however, codosing using atomic 88



Letter

Figure 2. Low-energy reflectivity data from a six-layer Pd film codosed with atomic H and CO. Black: Clean film at 170 K. Red: After 500 LH. Yellow: after 2, 2.5, 3.5, and 4.5 kLH. Green: After 5 kLH. The data sets are offset for clarity. Figure 3. EPs giving the evolution of the total energy versus the normal distance of the H atom in the film, zH, for the six different unit cells considered. Dark dashed lines correspond to the (1 × 1)+H cell; dark solid to the √3+H; solid blue, green, and red to the √3+nCO +H, with n = 1, 2 and 3, respectively; dashed blue to the (c2)2CO+H. The energy origin for each cell is set to that obtained at the subsubsurface octahedral site (see SI), whereas the zH origin is located at the topmost Pd plane. Dashed vertical gray lines give the position of the Pd layers. Top views of each cell surround the EPs with frames following the same color code as lines in the plot. The H atoms (small blue circles) are always placed at fcc sites, whereas red circles correspond to the CO molecules.

H causes a large expansion for d12 of ∼4% whereas d23 remains essentially unchanged from the clean surface (fifth column of Table 1). Similar expansions were found for 3−6 layer films (see the SI) as well as for a 20 layer film (3%). Because the later film already adopts the bulk Pd lattice parameter, codosing a bulk Pd substrate should give the same expansion. In summary, we observe a significant subsurface H population only when dosing atomic H onto CO-covered surfaces. To understand this surprising H trapping mechanism and gain further insight into the role played by the CO, we calculated for different CO coverages the energy profiles (EPs) associated with displacing a H atom through the film along the surface normal. All calculations were performed with the DFT-based SIESTA29 code and under the local density approximation (LDA30) because it lead to interatomic distances closer to the experimental ones than those obtained under the generalized gradient approximation (GGA31). Nevertheless, we have checked for several cases that both functionals provide the same qualitative picture. The Pd films are modeled via a slab comprising four Ru layers on which four Pd layers are stacked following an fcc sequence.25 We have calculated the EPs by varying the normal coordinate of the H atom, zH, in 0.05 Å steps from 2 Å above the Pd surface plane down to a subsubsurface site. Here we define the subsurface sites to be between the first and second Pd layers and the subsubsurface sites to be between the second and third layers. For a given unit cell (see below) and at each fixed zH value we performed a total energy optimization relaxing all atoms in the slab except the two lowest Ru planes, which were fixed to bulk positions. In Figure 3, we present the six unit cells considered and their associated EPs, assuming that the adsorbed H is at an fcc site. (In the SI, we provide the EPs for H at hcp and top sites.) Let us first concentrate on the CO-free surfaces, for which we considered two unit cells: a (1 × 1)-H (dashed dark line) and a (√3 × √3)R30°-H (√3, solid dark), corresponding to hydrogen coverages of ΘH = 1.0 and 0.33 ML, respectively. The EPs are very similar for both cases, presenting a deep minimum for adsorbed H (Ha) and a large energy barrier (>0.5 eV) for the transition toward the subsurface octahedral site (Hs), which is ∼0.4 eV less stable. The subsubsurface tetrahedral site (Hss) is even less favorable, also presenting large barriers of 0.4 to 0.5 eV from the Hs site. Our EPs are in good accordance with recent related theoretical work,17,32 despite different H coverages and exchange-correlation functionals. Therefore, the main conclusion is that in the absence of CO, H penetration into the

film is highly unfavorable, regardless of the H coverage. This fact rules out H incorporating into the Pd within the H domains (red areas in Figure 1c−h). Rather, H at the boundaries between CO and H domains or where H is mixed with the CO domains should go subsurface during displacement by CO. We also note that the first two interlayer spacings calculated for the Ha geometry (d12 = 2.29 and d23 = 2.28 Å) are hardly expanded from the clean surface, in close agreement with the experimental values for the H-covered films (Table 1). We next addressed the effect of coadsorbing CO and H onto the film by inserting 1, 2, and 3 CO molecules in the √3 cell (blue, green, and red lines in Figure 3, respectively) so that for a fixed coverage ΘH = 0.33 ML, we have CO coverages of ΘCO = 0.33, 0.50, and 1 ML, respectively. Because of the limited cell size, we placed the CO at hcp sites in all cases, thus disregarding the bridge and top adsorption that is known to occur at high coverages.13 Nevertheless, the conclusions to be drawn below should still hold for more realistic geometries. For the sake of completeness, we also calculated the EP for a c(2 × 4) cell (c2) with one CO molecule placed at an hcp site and a second at an fcc site (ΘCO = 0.5 ML and ΘH = 0.25 ML, dashed blue lines in Figure 3). The repulsive character of the Pdmediated CO−H interactions is immediately apparent in the EPs; as the CO concentration increases, the Ha minimum shifts upward in energy so that for ΘCO = 0.33 ML we already find an inversion in the hierarchy among the different minima with the Hs and Hss sites, both almost equally stable, becoming ∼0.1 eV more favorable. For larger CO coverages, the H adsorbed at the fcc sites is destabilized, and its associated minimum is no longer present. Note that for the c2 cell (ΘCO = 0.5 ML) the Ha minimum still remains, with a substantial energy barrier of ∼0.4 eV for diffusion toward the Hs site. This is mainly because in this particular geometry the H only has one CO at its firstnearest-neighbor three-fold sites, whereas our results indicate 89



Letter

that at least two COs at such close sites are required to destabilize the Ha. (See the cases of ΘCO = 0.66 and 1.0 ML and c2 for H adsorbed at the hcp site in the SI.) As regards the interlayer spacings, DFT indicates that all COcovered cells, regardless of if the Ha is present, yield a slight expansion of the first interlayer spacings (d12 = 2.32 and d23 = 2.28 Å). On the contrary, when subsurface Hs sites are occupied, d12 shows large expansions ranging between 2.39 and 2.44 Å, whereas d23 = 2.29 to 2.31 Å shows much smaller expansions. When compared against the experimental distances in Table 1, we confirm that the large expansion for d12 can only be due to a substantial population of subsurface H. Subsubsurface sites, on the other hand, may be disregarded because they lead to far too large second interlayer spacings, d23 =2.33 to 2.34 Å, not compatible with the experimental values given in Table 1. With the aid of the EPs, we can reach a satisfactory explanation for most of the experimental findings. The absence of the Ha minimum at high CO coverages reproduces the well-reported effect whereby a CO-covered surface blocks H adsorption.10 At low coverages, the two species segregate into different domains because of the CO-H repulsion. As the coverages increase, the two phases compress until at the domain boundaries close proximity between the H and the CO occurs. This destabilizes the H, which may either desorb via recombination or occupy a subsurface site. Further diffusion into the bulk, however, is hindered because the EPs show that the energy barriers from Hs to Hss are always >0.2 eV, increasing with decreasing CO coverage. The fundamental question that remains open is why subsurface H reaches significant concentrations only when atomic H impinges on a CO-covered surface. A possible reason is that at coverages in the range of 0.5 to 0.6 MLs, close to the limit where the Ha minimum disappears, the energy barrier is sufficiently low for incoming H atoms to overcome. In fact, atomic H is known to penetrate through a complete layer of adsorbed H on metals such as Ni.33 In contrast, dissociated protons resulting from impinging H2 molecules cannot surpass the barrier because their momentum along the surface normal should be very small. An electronic structure analysis offers a simple explanation of why CO destabilizes adsorbed H. Mulliken population analysis shows that a CO donates around 0.4e to the metal. Ha also donates charge to Pd but a smaller amount (∼0.1e). The charge donation by the two types of adsorbates is evident in the charge density contour plot of Figure 4a calculated for the √3+CO+H case, where the blue iso-surfaces indicate charge-depletion regions. When both species come closer to each other, as for the (c2)2CO+H case (Figure 4b), the overlap between the two depletion regions results in a repulsive potential, which shifts the H adsorption minima to smaller binding energies (Figure 3). The density of states projected (PDOS) onto the surface atoms for the latter case is presented in Figure 4c. In the 12−10 eV region, peaks A and B in the Figure, the CO-σ2s/2pz molecular orbitals, interact directly with the Pd. In the 8−6 eV range we find a strong hybridization between the H, the CO-π2pxy states and the Pd (peaks C and D). If we now focus on the crystal orbital overlap populations (COOPs) displayed in Figure 4d, then we find that the interaction between the adsorbed H closest to the CO and the CO-π2pxy states (red lines) has a strong antibonding character; that is, peak D is negative. This destabilizes the adsorbed H. Notice, however, that when the H is at the octahedral subsurface site (blue lines), it shows hardly any bonding to the CO, whereas bonding to the Pd atoms is

Figure 4. (a,b) Top view of an iso-surface of the difference in charge density, δρ(r), of the complete system minus the sum of the individual atoms for H adsorbed at the fcc site within (a) the √3+CO and (b) the (c2)2CO cell. The positions of the H and CO adsorbates are indicated by white and dark circles, respectively, at the top of each image. (c) DOS projected on the adsorbed H at the fcc site (blue), the COs (red), and the first Pd layer (dark) for the (c2)2CO cell. (d) COOPs between the H atom and the two inequivalent COs (bottom plots) and the first two Pd layers (top plots) for the same cell as in (c). Red lines correspond to the H adsorbed at an fcc site and blue lines to the H at the octahedral subsurface site. Blue lines have been shifted vertically for the sake of clarity.

shifted to larger binding energies (peak C). The amphoteric character of H becomes apparent as it accepts around 0.1e from the surrounding metal atoms at the subsurface site. In summary, we have observed CO displacing H adsorbed on Pd(111), demonstrating how the spatial distribution of coadsorbates can be imaged using electron reflectivity. Co-dosing Pd(111) with CO and atomic H gives a CO-covered surface and substantial concentrations of H only between the top two Pd layers. DFT calculations show that the destabilization of the adsorbed H and the blocking of subsurface H desorption results from the antibonding character of the interaction between the adsorbed H and the CO π states.

■

ASSOCIATED CONTENT

S Supporting Information *

Description of the LEED procedure, nonmerged LEEM frames of Figure 1 and further details of first-principles calculations. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION Corresponding Author *E-mail: [email protected]. Present Addresses ∥ Elettra Sincrotrone S.C.p.A, Trieste, Italy. ⊥ Instituto de Catálisis y Petroleoquímica, CSIC, Madrid, Spain.

■

ACKNOWLEDGMENTS This research was supported by the Spanish Ministry of Innovation and Science under project nos. MAT2009-14578C03-01 and MAT2010-18432. Sandia National Laboratories is a multiprogram laboratory managed and operated by Sandia Corporation, a wholly owned subsidiary of Lockheed Martin Corporation, for the USDOE’s National Nuclear Security 90



Letter

Administration under contract DE-AC04-94AL85000. The work at Sandia was partially supported by the U.S. Department of Energy, Office of Basic Energy Sciences, Division of Materials Sciences and Engineering.

■

REFERENCES

(1) Christmann, K. Surf. Sci. Rep. 1988, 9, 1−163. (2) Favier, F.; Walter, E. C.; Zach, M. P.; Benter, T.; Penner, R. M. Science 2001, 293, 2227−2231. (3) Adhikari, S.; Fernando, S. Ind. Eng. Chem. Res. 2006, 45, 875− 881. (4) Teschner, D.; Borsodi, J.; Wootsch, A.; Revay, Z.; Hävecker, M.; Knop-Gericke, A.; Jackson, S.; Schlögl, R. Science 2008, 320, 86−89. (5) Doyle, A.; Shaikhutdinov, S.; Jackson, S.; Freund, H.-J. Angew. Chem., Int. Ed. 2003, 42, 5240−5243. (6) Poutsma, M.; Elek, L. F.; Ibarbia, P. A.; Risch, A. P.; Rabo, J. A. J. Catal. 1978, 52, 157−168. (7) Casanovas, A.; Llorca, J.; Homs, N.; Fierro, J. L. G.; de la Piscina, P. R. J. Mol. Catal. A 2006, 250, 44−49. (8) Nowitzki, T.; Borchert, H.; Jürgens, B.; Risse, T.; Zielasek, V.; Bäumer, M. Chem. Phys. Chem 2008, 9, 729−739. (9) Conrad, H.; Ertl, G.; Latta, E. E. J. Catal. 1974, 35, 363−368. (10) Kok, G.; Noordermeer, A.; Nieuwenhuys, B. Surf. Sci. 1983, 135, 65−80. (11) Morkel, M.; Rupprechter, G.; Freund, H.-J. J. Chem. Phys. 2003, 119, 10853−10866. (12) Rose, M. K.; Mitsui, T.; Dunphy, J.; Borg, A.; Ogletree, D. F.; Salmeron, M.; Sautet, P. Surf. Sci. 2002, 512, 48−60. (13) Rupprechter, G.; Morkel, M.; Freund, H.-J.; Hirschl, R. Surf. Sci. 2004, 554, 43−59. (14) Ratajczykowa, I. Surf. Sci. 1986, 172, 691−714. (15) Eriksson, M.; Ekedahl, L. G. App. Surf. Sci. 1998, 133, 89−97. (16) Dohnálek, Z.; Kim, J.; Kay, B. D. J. Phys. Chem. C 2008, 112, 15796−15801. (17) Hong, S.; Rahman, T. Phys. Rev. B 2007, 75, 155405−10. (18) Felter, T. E.; Sowa, E. C.; Hove, M. A. V. Phys. Rev. B 1989, 40, 891−899. (19) Okuyama, H.; Nakagawa, T.; Siga, W.; Takagi, N.; Nishijima, M.; Aruga, T. J. Phys. Chem. B 1999, 103, 7876−7881. (20) Bauer, E. Rep. Prog. Phys. 1994, 57, 895−938. (21) Altman, M. S. J. Phys.: Condens. Matter 2010, 22, 084017−14. (22) Ceyer, S. Acc. Chem. Res. 2001, 34, 737−744. (23) Schmid, A. K.; Swiech, W.; Rastomjee, C. S.; Rausenberger, B.; Engel, W.; Zeitler, E.; Bradshaw, A. M. Surf. Sci. 1995, 331−333, 225− 230. (24) Yim, C.; Man, K.; Xiao, X.; Altman, M. Phys. Rev. B 2008, 78, 155439−9. (25) Santos, B.; Puerta, J. M.; Cerda, J. I.; Herranz, T.; McCarty, K. F.; de la Figuera, J. New J. Phys. 2010, 12, 023023−21. (26) Lindroos, M.; Pfnür, H.; Feulner, P.; Menzel, D. Surf. Sci. 1987, 180, 237−251. (27) Jonker, B.; Bartelt, N.; Park, R. Surf. Sci. 1983, 127, 183−199. (28) de la Figuera, J.; Puerta, J.; Cerda, J.; Gabaly, F. E.; McCarty, K. Surf. Sci. 2006, 600, L105−L109. (29) Soler, J. M.; Artacho, E.; Gale, J. D.; Garcia, A.; Junquera, J.; Ordejón, P.; Sanchez-Portal, D. J. Phys.: Condens. Matter 2002, 14, 2745−2779. (30) Perdew, J. P.; Zunger, A. Phys. Rev. B 1981, 23, 5048−5079. (31) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77, 3865−3868. (32) Ozawa, N.; Roman, T. A.; Nakanishi, H.; Kasai, H.; Arboleda, N. B.; Diño, W. A. J. Appl. Phys. 2007, 101, 123530−6. (33) Kammler, T.; Wehner, S.; Küppers, J. Surf. Sci. 1995, 339, 125− 134.

91


CO-Assisted Subsurface Hydrogen Trapping in Pd(111) Films - The

Recommend Documents