We studied the hydrogen-induced reconstruction of the W(110) surface using the pseudopotential plane wave approach. The calculations for a full monolayer of hydrogen coverage showed that the quasithreefold hollow site (distorted bridge) has the lowest energy, and that for this geometry a surface reconstruction, consisting of a small uniform shift of the W top layer in the [1(1) over bar0$] direction, is energetically favorable. We also studied the surface states for clean and H-covered W(110) and investigated the effect of the reconstruction on electronic structure.

We studied total energies of various ordered structures of PdHx (in which hydrogen occupies the octahedral sites within the fee Pd lattice) using the pseudopotential method and a plane-wave basis within the local-density-functional approximation. The structures considered include the (420)-plane ordering of hydrogen atoms at different concentrations. For x greater than or equal to 1/2 we found that the NiMo- and Ni4Mo (D1(a))-type structures at x=1/2 and x=4/5, respectively, were energetically favored phases, in agreement with the superlattice reflections found in previous neutron-scattering measurements. For the intermediate concentrations, linear variation of the formation energy as a function of x in several (420)-ordered structures explained the observed short-range order. In contrast to an earlier proposal, we did not find the Fermi surface imaging effect responsible in this case. The overall energy variation in different phases indicates the importance of going beyond pairwise interactions between interstitial hydrogen atoms in this system.

We used the embedded-atom method potential to study the structures, adsorption energies, binding energies, migration paths, and energy barriers of the Ir adatom and small clusters on fcc Ir (100), (110), and (111) surfaces. We found that the barrier for single-adatom diffusion is lowest on the (111) surface, higher on the (110) surface, and highest on the (100) surface. The exchange mechanisms of adatom diffusion on (100) and (110) surfaces are energetically favored. On all three Ir surfaces, Ir-2 dimers with nearest-neighbor spacing are the most stable. On the (110) surface, the Ir-2 dimer diffuses collectively along the (110) channel, while motion perpendicular to the channel walls is achieved by successive one-atom and correlated jumps. On (111) surface, the Ir-2 dimer diffuses in a zigzag motion on hcp and fee sites without breaking into two single atoms. On the (100) surface, diffusion of the Ir-2 dimer is achieved by successive one-atom exchange with the substrate atom accompanying by a 90 degrees rotation of the Ir-2 dimer. This mechanism has a surprisingly low activation energy of 0.65 eV, which is 0.14 eV lower than the energy for single adatom exchange on the (100) surface. Trimers were found to have a one-dimensional (1D) structure on (100) and (110) surfaces, and a 2D structure on the (111) surface. The observed abrupt drop of the diffusion barrier of tetramer, I-gamma 4 on the Ir (111) surface was confirmed theoretically.

We report the first atomically resolved images of ordered Au trimers on Si(111)root 3X root 3R30 degrees-Au using wave-front reconstruction of scanned-energy glancing-angle Kikuchi electron spectra. Each Au image has a resolution (full width at half magnitude) of less than 1 Angstrom. The images indicate that Au trimers art ordered and nonrotated within the surface plane and with respect to the second-layer Si plane providing direct evidence of the conjugate honeycomb-chained-trimer model for the Au-root 3 system.

n/a

We report the first implementation of orthonormal wavelet bases in self-consistent electronic structure calculations within the local-density approximation. These local bases of different scales efficiently describe localized orbitals of interest. As an example, we studied two molecules, H-2 and O-2, using pseudopotentials and supercells. Considerably fewer bases are needed compared with conventional plane-wave approaches, yet calculated binding properties are similar. Our implementation employs fast wavelet and Fourier transforms, avoiding evaluating any three-dimensional integral numerically.

n/a