An increasing number of studies have been reported in the modification of the chemical and physical properties of metal surfaces by depositing other metals. The modification is often important for technological applications in metal catalysts, metallurgy and microelectronics. Chemisorption of molecules is sometimes quite sensitive to the electronic structure change of substrates. Studies of molecular adsorption on metal overlayers would thus be suggestive for the understanding of modified properties of thin metal films.

Recently, we have investigated adsorption of sulfur dioxide SO2 on Ni [1] and Pd [2] single-crystal surfaces by means of XAFS (X-ray-absorption fine structure) spectroscopy. It was revealed that SO₂ is lying flat on the Ni surfaces, while the molecular plane of SO₂ is perpendicular to the Pd surfaces. Such a drastic difference between Ni and Pd is rather surprising because both metal elements belong to Group 10 in the periodic table and the electronic structures are similar to each other. It is thus quite interesting to investigate adsorption of SO₂ on thin Pd/Ni bimetallic surfaces for the understanding of the chemical and electronic properties of thin Pd films compared to the bulk one. The Ni substrate employed in the present study is more active than the bulk Pd concerning adsorption of SO₂; molecular adsorption of SO₂ is stable at room temperature on Pd, while SO₂ dissociates on Ni below room temperature. One can expect that the adsorbate-substrate interaction would be strengthened in the present thin Pd films on a more active Ni substrate.

Figure 1 shows the polar-angle dependent S K-edge NEXAFS spectra of submonolayer SO2 adsorbed on the 1.2 ML Pd film grown on the Ni(111) surface, together with those on Ni(111) and Pd(111), and of randomly oriented multilayer SO2. It should be pointed out from our previous STM observations [3] that at the 1.2 ML Pd coverage the surface is completely covered with Pd atoms and no Ni atoms remain on the topmost layer. Two intense features are observed in Fig. 1; the lower-energy peak (2473 eV) is attributed to the S1s-to-pi* resonance, while the higher-energy one to the S1s-to-sigma* (S-O) resonance. One can recognize a clear contrast in polarization dependence of the pi* and sigma* peaks between the spectra on Ni(111) and Pd(111). Such angular dependence clearly shows that SO2 is lying flat on Ni(111), while the molecular plane of SO2 is perpendicular to the Pd(111) surface. For the spectra of SO2 on 1.2 ML Pd/Ni(111), the pi* peak was enhanced at grazing incidence. This implies that the transition moment of the pi* orbital was normal to the surface, leading to the flat-lying orientation of the molecule on the thin Pd film surface. The deconvolution analysis of the pi* peaks gives the average angle omega between the molecular plane and the surface. The analysis yields omega=16±10 deg., which is close to the case of Ni(111) (0 deg.) and is much different from that for Pd(111) (90 deg.). We can thus conclude that although SO2 interact directly with the surface Pd atoms, the adsorbate structure is not similar to that on Pd(111) but is close to that on Ni(111).

Fig. 1: S K-edge NEXAFS spectra of submonolayer SO2 on 1.2 ML Pd grown on Ni(111), together with those of multilayer SO2 and submonolayer SO2 on Ni(111) and on Pd(111). Solid lines are the spectra taken at normal incidence (theta=90 deg.), dashed lines at theta=55 deg. and dotted lines at grazing incidence [theta=1 deg. for Pd(111), 15 deg. for Ni(111) and Pd/Ni(111)]. Note here that in the case of the previous Pd(111) work the grazing incidence spectrum was taken at the total-reflection condition of theta=1 deg. in order to avoid intense elastic scattering X-rays.

The intensity of the pi* resonance is regarded as a measure of the strength of the interaction between the metal and the SO₂ pi* orbital. One can directly compare those taken at 55-deg. incidence with each other. The pi* resonances of the submonolayer spectra are found to be weakened compared to the one of randomly oriented multilayer SO₂. Such intensity reduction can be ascribed to the charge transfer from the substrate to the SO₂ pi* orbital. When we assume that the p* intensity is simply proportional to the vacant density of the pi* level, the amount of the charge transfer can be estimated by comparing the intensity with that of multilayer, whose pi* state is completely vacant. The result for Pd/Ni(111) was 0.85 electron, which is smaller than that of SO₂/Ni(111) (1.2 electron) and is larger than that on Pd(111) (0.6 electron).

Figure 2 shows the resultant k²chi(k) functions. The EXAFS functions were Fourier transformed into r space. They are shown in Fig. 3. The peak at 1.0 Ang. can be attributed to the S-O bonds. Two peaks distinctly observed at 1.7 and 2.2 Ang. are ascribed to the S-Pd contribution. The S-O distance was determined to be 1.47±0.03 Ang., irrespective of the x-ray incidence. The elongation of the S-O bond compared to that of gaseous one (1.43 Ang.) can be ascribed to the charge-transfer from the substrate to the antibonding SO₂ pi* orbital. The effective coordination numbers yield the angle of the SO₂ molecular plane of a=82±10 deg. from surface normal. When we assume that one of the S-O bonds lies on the surface and another is standing up, keeping the OSO bond angle of 119.2 deg., the EXAFS result gives the polar angle of 75 deg. for the other S-O bond, leading to omega=18 deg., which is close to the NEXAFS result. As shown in Fig. 3, the S-Pd contribution was enhanced at grazing incidence, implying that the S-Pd bonds stand up on the surface. The S-Pd bond distance was determined to be 2.29±0.04 Ang.

Fig. 2: S K-edge EXAFS oscillation functions k2chi(k) of SO2 on 1.2 ML Pd/Ni(111) taken at theta=90, 55 and 15 deg.	Fig. 3: Fourier transforms of the EXAFS oscillation functions k2chi(k) in Fig. 2.

Figure 4 shows the S K-edge NEXAFS spectra of SO2/Pd/Ni(111) by varying Pd film thickness. The spectra correspond to 1.2, 3.0 and 6.0 ML Pd. Comparing the pi* features in the 6.0 ML spectrum with the 1.2 ML ones, we can clearly find that polarization dependence is drastically changed. On the 1.2 ML Pd film, the pi* peak was enhanced at grazing incidence, this corresponding to the flat-lying orientation of the molecule as discussed above. On the other hand, in the 6 ML Pd film, the pi* peak is enhanced at normal incidence. Such polarization dependence of the NEXAFS spectra corresponds to the perpendicular orientation of the SO2 molecular plane with respect to the surface. As shown in Fig. 1, these features closely resemble those on Pd(111). One can thus recognize that the adsorption structure of SO2 on 6 ML Pd /Ni(111) is essentially the same as that on Pd(111). Fig. 4: Pd thickness dependence (1.2, 3.0 and 6.0 ML) of S K-edge NEXAFS spectra from SO2/Pd/Ni(111). Solid lines are the spectra taken at normal incidence (theta=90 deg.), dashed lines at theta=55 deg. and dotted lines at theta=15 deg.

Figure 5 shows a schematic surface structure of SO₂/1.2 ML Pd/Ni(111), together with those on Ni(111) and Pd(111) for comparison. Table 1 summarized numerical results obtained by the present and previous analyses of NEXAFS and SEXAFS. On the 1.2 ML Pd film, the SO₂ molecule is nearly lying flat on the surface, this being quite similar to that on Ni(111), although the molecule interacts directly with the surface Pd atoms. Such a surface structure clearly excludes the effect of local geometry of the substrate and suggests that the electronic structure of the 1.2 ML Pd film on Ni(111) plays a key role.