Table of Contents

X-Ray Photoelectron Diffraction of MgO(111)-(1x1) M. Gajdardziska-Josifovska,(a,b) R. Plass,(a,b) and S. A. Chambers Supported by Office of Biological and Environmental Research, EMSL Operations, National Science Foundation. (a) EMSL User. (b) Department of Physics, University of Wisconsin at Milwaukee. The stability of polar oxide surfaces has long been a problematic question in surface science. A bulk-terminated polar surface has an infinite surface energy because alternating layers of oppositely charged ions produce a large dipole moment perpendicular to the surface. Theoretical studies have proposed three different mechanisms for the stabilization of the clean polar oxide surfaces with the rock salt structure: 1) microscopic faceting into neutral {100} planes; 2) octopolar 2 x 2 surface reconstruction model, which is essentially faceting into neutral planes at the nanoscale; 3) metallization of the 1 x 1 surface associated with surface relaxation. Adsorption of charged species on the surface opens a fourth mechanism, and an OH terminated 1 x 1 surface has also been predicted. In this project we are seeking to determine which, if any, of these mechanisms is correct. Very recent atomic force microscopy (AFM) and scanning electron microscope (SEM) studies have established that the model MgO(111) surface is not faceted to neutral {100} planes, thus eliminating mechanism 1 above (Plass and Gajdardziska-Josifovska 1998). High temperature anneals (1500-2000°C) have yielded three air-stable reconstructions, among which is the predicted 2 x 2 periodicity. However, transmission high energy electron diffraction (THEED) structure determination has eliminated the octopolar model structure (mechanism 2 above) and shown stabilization by oxygen trimers and single oxygen ions (Plass et al. 1998). At lower temperatures THEED, low energy electron diffraction (LEED), and reflection high energy electron diffraction (RHEED) reveal stable 1 x 1 patterns. Hence, we are employing a combination of scanned-angle x-ray photoelectron diffraction (XPD) and high-energy-resolution x-ray photoelectron spectroscopy to distinguish between mechanisms 3 and 4 above, and to determine the interlayer relaxations. The specimens were bulk single crystals that were mechanically polished. Some were etched in hot concentrated nitric acid, and furnace annealed in air. Once in ultrahigh vacuum, the specimens were further cleaned with an oxygen plasma at room temperature, followed by vacuum annealing at 800°C. Scanned-angle XPD at low take-off angles enhances the surface sensitivity of the technique. We have thus measured O 1s and Mg KLL azimuthal angular distributions at take-off angles ranging from 7° to 16°, as well as polar scans in high-symmetry azimuths. These data require extensive simulation with single- and multiple-scattering theories as a function of surface structure in order to extract interlayer relaxations, and this analysis is in progress. However, the surface termination is readily determined from qualitative considerations of polar scans, at least for this particular crystal type and surface orientation. The O 1s/Mg KLL intensity ratio is expected to increase (decrease) as grazing emission is approached if the surface is terminated with O (Mg), due simply to inelastic attenuation of outgoing photoelectrons. We have found that this ratio increases at low take-off angles in all low-symmetry azimuths, indicating that the surface is terminated with an oxygen layer. We show these intensity ratios in Figure 2.2a. However, this structure is likely to be highly polar and, therefore, unstable in the absence of surface relaxation leading to metallization of the top few atomic layers. High-energy-resolution O 1s spectra obtained at normal emission (q = 90°) and grazing emission (q = 10°) reveal that the terminal oxygen layer is actually a layer of OH. The presence of a monolayer of H associated with the terminal layer of oxygen removes the surface dipole and stabilizes the surface, independent of surface relaxation. XPD cannot detect the presence of H since H is such a weak electron scatterer. However, the chemical shifts of surface OH relative to lattice oxygen have been measured for a number of oxides, including MgO(001) (Liu et al. 1998). The spectra for MgO(111) are shown in Figure 2.2b. The weaker peak to higher binding energy is ascribed to the presence of a terminal layer of OH based on the nearly identical chemical shift measured for dissociative chemisorption of water on MgO(001) (Liu et al. 1998). In addition, this chemical shift is measurably larger than what we have determined for surface O bound to Cr cations on O-stabilized a-Cr2O3(0001). Therefore, it seems highly likely that this peak is indeed due to the presence of a terminal layer of OH. The angular dependence of the OH peak intensity relative to that of lattice oxygen shows that the amount of OH is equal to ~1 ML if the O 1s photoelectron escape length is assumed to be 30-35 Å, which is not unreasonable for a wide bandgap insulator such as MgO. Quantitative information on the interlayer relaxations is currently being sought from analysis of the XPD azimuthal scans. However, it appears that the fundamental issue of surface termination and surface dipole is answered by the presence of a terminal OH layer. Figure 2.2. O 1s/Mg KLL intensity ratio vs. take-off angle in three high-symmetry azimuths (a), and high-resolution O 1s spectra at take-off angles of 90° and 10° (b) for MgO(111). References Liu, P., T. Kendelewicz, G. E. Brown, Jr., and G. A. Parks, Surf. Sci. 412/413, 287 (1998). Plass, R., and M. Gajdardziska-Josifovska, Surf. Sci. 414, 26 (1998). Plass, R., K. Egan, C. Collazo-Davila, D. Grozea, E. Landree, L. D. Marks, and M. Gajdardziska-Josifovska, Phys. Rev. Lett. 81, 4891 (1998). William R. Wiley Environmental Molecular Sciences Laboratory Feedback: webmaster@emsl.pnl.gov Revised: June 12, 2001 Security & Privacy PNNL-13147