Surface-structure dependent reaction of hydrogen-assisted reduction at O/Ni(110) surfaces studied by MIES and LEED

Surface-structure dependence of hydrogen-assisted reduction at oxidized Ni(110) surfaces was studied by metastable-induced electron spectroscopy (MIES) and low-energy electron diffraction (LEED). We prepared three different types of oxidized Ni(110) surfaces: (a) Low oxygen-coverage (3×1) surface; (b) moderate oxygen-coverage (2×1) surface; (c) high oxygen-coverage (3×1) surface. When exposed the surfaces (a) and (b) to hydrogen molecular gas at exposures 10∼100 L, their structures were changed to a hydrogen-adsorbed (1 × 1) structure. MIES clearly demonstrated a progress of hydrogen-assisted reduction by showing intensity decrease of the O2p-induced peak together with recovery in intensity of the Ni3d-induced peak and appearance of hydrogen-induced broad peak. Hydrogenated (1 × 1) surfaces, thus produced, were transformed to oxidized surfaces of the low-coverage (3 × 1) or (2 × 1) structure by being exposed to oxygen, depending on the oxygen exposure. These reduction/oxidation reactions took places reversibly. At the surface (c), however, such reactions were not observed in the range of hydrogen exposure up to 100 L at RT. This fact implies that the density of Ni-O-Ni atomic rows formed on the oxidized Ni(110) surfaces determines the progress of oxygen reduction reaction. [DOI: 10.1380/ejssnt.2006.170]


I. INTRODUCTION
Adsorption of atoms and molecules, and chemical reactions, such as oxidation or reduction, at Ni surfaces have been an important subject of study for a long time. This is recently prompted by the reason that knowledges about the surface preparation and the interface-formation process are necessary when one wants to apply magnetic nano-structures at these surfaces to some nano-devices. The Ni(110) surface, particularly, attracts considerable attention, because it provides many reconstruction phases depending on the adsorbed element and its coverage which may serve the study on the structure-dependency of those reactions.
The co-adsorption of oxygen and hydrogen and the oxidation/reduction reaction at Ni surfaces have been investigated with various techniques. Lescop, et al. showed that oxygen atoms are removed from a H/O(200 L)/Ni(111) surface in forms of water molecules, by MIES and UPS [1,2]. A TPD and XPS investigation by Vesselli, et al. demonstrated that oxygen is hardly removed by hydrogen at a H(10 L)/O(∼20 L)/Ni(110) surface at low temperatures (∼100 K) [3]. Experiments studying a relation between the progress of surface oxidation/reduction and the atomic and electronic structure of reconstructed substrates have not been reported yet.
In the present study, we treat adsorption of hydrogen on * This paper was presented at International Symposium on Surface Science and Nanotechnology (ISSS-4), Saitama, Japan, 14-17 November, 2005. † Corresponding author: tomonori@elcs.kyutech.ac.jp oxygen pre-adsorbed Ni(110) surfaces and intend to construct a picture of the hydrogen-assisted reduction process being consistent with observed changes in both the substrate structure and the surface electronic states. We applied metastable-induced electron spectroscopy (MIES) to the analysis of hydrogen adsorption induced changes in the local electronic states of the O/Ni(110) surface. We will show that a reversible reaction of oxidation/reduction proceeds at surfaces covered with low densities of the Ni-O-Ni row but does not at surfaces with its higher densities.

II. EXPERIMENTAL
The apparatus is composed of a He * source, a hemispherical energy analyzer (VG Scientific) and a low-energy electron diffraction (LEED) optics. The base pressure was around 1.0 × 10 −8 Pa. A beam of helium metastable atoms was produced by high-voltage discharge at a nozzleskimmer space. The main ingredient of the beam is He * -2 3 S (excitation energy=19.8 eV), which was confirmed by a Stern-Gerlach analysis. The sample is a single-crystal Ni(110) surface, cleaned by repeated cycles of Ar + ion sputtering and annealing. Its clean surface exhibited a clear (1 × 1) LEED pattern. The surface was exposed to oxygen and hydrogen molecular gas at room temperature.
At Ni(110) surfaces, whether clean or adsorbed with oxygen or hydrogen, He * atoms undergo resonance ionization followed by Auger neutralization owning to high work functions of those surfaces [4][5][6]. The electron excitation in this Auger neutralization process takes place nearly in the topmost layers, and the resultant emission spectrum reflects a self-convolution of local DOS of the e-Journal of Surface Science and Nanotechnology target surface. Each spectrum showed a supplementary high-energy cutoff due to a contribution of UV photons (hν=21.2 eV) being residual in the beam. We used this position for calibrating the electron energies of the MIES spectra.

III. RESULTS AND DISCUSSION
In our previous report [7], we showed variations in the atomic and electronic structures in the oxidation process of the Ni(110) surface in the range of oxygen exposure S ox below 5 L using MIES and LEED. The structures of this Ni(110) surface are classified into the following four different types depending on S ox : (a) At S ox = 0 L, nonreconstructed (1 × 1) surface; (b) at S ox ∼ 0.5 L, low oxygen-coverage (3 × 1) surface; (c) at 1 ≤ S ox ≤ 3 L, moderate oxygen-coverage (2 × 1) surface; (d) at 4 L≤ S ox : high oxygen-coverage (3 × 1) surface. According to a proposed model, Ni-O-Ni atomic rows stretching along the [001] direction on the surface plane have been created, and its density increases up to two thirds a monolayer at oxygen exposures around S ox ∼ 5 L.
At first we observed hydrogen-induced structural changes for each of the above mentioned types of the Ni(110) surface by LEED. Figure 1 shows a diagram of LEED patterns as a function of the hydrogen exposure for four different surfaces. When we exposed the Ni(110) clean surface to hydrogen molecular gas, a (1 × 1) pattern did not change even at exposures around 100 L. Although there are some references reporting a reconstruction to the (1 × 2) structure by hydrogen adsorption [8][9][10], we did not observe such a reconstruction in LEED. At the surface (b), the (3 × 1) structure was changed to the (1 × 1) through the streaky (1 × 1) with increasing the hydrogen exposure. Also the surface (c) was reconstructed to (1 × 1) through a streaky structure by hydrogen adsorption. However, the high oxygen-coverage (3 × 1) surface was not led to reconstruction even by heavier hydrogen

exposures.
After the hydrogenated (1 × 1) structure was formed by molecular hydrogen exposure of 100 L at the surfaces of type (b) or (c), the surface was exposed to oxygen again in order to check the reversibility of the whole reactions. The LEED showed that an oxygen exposure of 0.5 L leads to a (3×1) reconstruction and an exposure of 1 L to a (2× 1) structure. This observation suggests that the surface follows a reversible path in the hydrogenation of oxidized surfaces and the oxidation of hydrogenated surfaces.
Before the MIES experiment on the oxygen reduction, we performed hydrogenation of the clean Ni(110) surface. Figure 2 shows variations in the MIES spectrum taken from the Ni(110) surface with increasing the hydrogen exposure up to ∼100 L. The peak around 15.7 eV in Fig.  2, labeled P d , is due to emission from the Ni-3d induced states by the Auger neutralization process. A shoulder structure, labeled P h , appeared in the intermediate energy region around 11.5 eV. This structure would be due to emission by inter-atomic Auger transition involving a H-1s derived states, Ni-3d and the He * -1s state. An energy difference between P d and P h is approximately 4∼5 eV and corresponds to that between H-1s induced and Ni-3d states observed by photoemission for the H/Ni(110) surface [11]. The low energy cutoff of the spectrum in Fig. 2(a), denoted E c , which is positioned around 5.3 eV, slightly shifted upward by hydrogen exposure below 5 L, and increased abruptly at around 8∼10 L. This means non-linear increase in the surface potential barrier upon hydrogen adsorption. If adsorbates induce dipole moments quite locally, the potential barrier height should increase in proportion as the adsorption proceeds. This is, however, not the case here. At a clean surface, the density distribution of mobile electrons tends to be smoothed, which effectively reduces the work function. We suppose a redistribution of those electrons at the Ni(110) surface takes place at a certain coverage with hydrogen. Then the smoothing effect will be suppressed, and the work function will be increased. The total emission intensity in the MIES spectrum was decreased at the same H 2 -exposure range as the work-function was increased. This result supports the above consideration, because this decrease in the transition probability of Auger neutralization at the surface [7] is well explained by a disappearance of delocalized states from which an electron tunnels to the He-1s state. Figure 3 shows a series of MIES spectra taken from the O/Ni(110)-0.5 L surface (type (b)) in the progress of hydrogen adsorption. The spectrum (b), from a hydrogenfree O/Ni(110) surface, contains a strong peak, labeled P ox , at about 10.5 eV. Peak P d , on the other hand, is weakened by the oxidation. This has been attributed to a decrease in the probability of Auger neutralization transition connecting the Ni 3d-states by an influence of oxygen adsorption [7]. When this surface was exposed to hydrogen gas for exposures of ∼25 L, the intensity of peak P d was recovered rapidly (spectrum (d)). This suggests that the oxygen reduction proceeds and the surface restores 3d-electronic states due to bare Ni atoms. A similar effect has been found in the hydrogen-assisted reduction of the O/Ni(111) surface by Lescop, et al [1,2]. Figure 4 shows a series of MIES spectra which demonstrates variations in the electronic structure of the O/Ni(110)-1 L surface (type (c)) by hydrogen adsorption. The spectral variation in this case was similar to that for the O/Ni(110)-low-coverage (3 × 1) surface (type (b)). It is confirmed by these spectra that the reaction of oxygen reduction takes place also on such a surface with moderate coverages of oxygen. However, the reduction rate was found to be lower than the case of the type (b) surface. This is shown by the two facts: At first, the recovery of the intensity of P d at the (2 × 1) surface (type (c)) needs more amount of hydrogen exposure than for the type (b); Secondly, in the LEED observation, the appearance of the hydrogen-induced streaky (1× 1) structure at the type (c) surface needs again more amount of hydrogen exposure. These results may tempt us to lead a model in which the reduction rate depends on the Ni-O-Ni row density on the O/Ni(110) surface. As the shape of MIES spectra (e-f) in Fig. 3 and (e-f) in Fig. 4 substantially agrees with the spectra for the hydrogenated surfaces (Fig. 2), one can judge that oxygen reduction of these surfaces was completed by hydrogen exposures of 50∼100 L.
MIES spectra taken from hydrogen-adsorbed O/Ni(110)-4 L surfaces (type (d)) were shown in Fig. 5. Some influences of hydrogen exposure appeared: a slight decrease in the intensity of O-2p induced states, a gradual increase in the work function, and an appearance of a weak shoulder at energies corresponding to P h . However, the overall shape of the MIES spectrum was not changed even by heavier hydrogen exposures. As already mentioned, we confirmed by LEED that the (3 × 1) periodicity in the structure of the type (d) surface was not changed by the hydrogen exposure. Obviously hydrogen does not effectively work in reducing the O/Ni(110)-high-coverage (3 × 1) surface.
We discuss possible processes of the oxida- tion/reduction reaction at the Ni(110) surface being based on the two principal findings of the present experiment: (i) the reduction proceeds at the low-coverage (3 × 1) surface and the (2 × 1) surface, but does not at the high-coverage (3 × 1) surface; (ii) reduction of oxidized surfaces and oxidation of hydrogenated surfaces are the reversible reaction of each other. From the result (i), a mechanism of direct attack of hydrogen molecules to adsorbed-oxygen sites is ruled out. The finding (i) means that the dissociation of hydrogen molecules needs bare-Ni 3d electronic states. The high-coverage (3 × 1) surface contains Ni-O-Ni rows so densely that it may be hard for incident hydrogens to directly meet the Ni-3d orbitals. Since the Ni-O-Ni bonds are much stronger than the Ni-H-Ni bonds [3], one cannot explain the reversibility of the reaction (the result (ii)) only from the viewpoint of energetics. This problem can be tricked by assuming a fast step of production of intermediate compounds.
In this case H 2 O may be a proper candidate, and actually Vesselli, et al. detected such species in their TPD spectra.

IV. CONCLUSION
We applied MIES and LEED to the study of hydrogenation of the O/Ni(110) surface. The reaction of hydrogeninduced reduction took place at both the low-coverage (3 × 1) surface and the (2 × 1) surface, but did not at the high-coverage (3 × 1) surface. The reaction rate of oxygen reduction strongly depends on the surface structure, that is, on the density of Ni-O-Ni rows in the present case, which is consistent with the model of dissociation of hydrogen molecules at surface sites with the bare-Ni 3d electronic states. The reactions, reduction of oxidized surfaces and oxidation of hydrogenated surfaces, was shown to proceed reversibly. These results suggest an intermediate product, such as water molecules, which needs further experiments by other techniques.