Mark P. Sears

Oxygen-induced restructuring of the TiO2(110) surface: a comprehensive study

We report a comprehensive experimental and theoretical study of the eVect of oxidizing a TiO 2 (110) surface at moderate temperatures. The surfaces are investigated with scanning tunneling microscopy (STM ), low-energy He+ ion scattering (LEIS ) and static secondary ion mass spectroscopy (SSIMS ). Flat (1◊1)-terminated TiO 2 (110) surfaces are obtained by sputtering and annealing in UHV at 880 K. These surfaces are exposed to oxygen gas at elevated temperatures in the range 470‐830 K. Formation of irregular networks of pseudo-hexagonal rosettes (6.5 A ˚ ◊ 6A ˚ ) and small (11:0] oriented (1◊1) islands along with {001}-oriented strands is induced at temperatures from 470 to 660 K. After annealing above 830 K, only regular (1◊1) terraces and white strands are observed. The composition of these oxygen-induced phases is quantified using 18O 2 gas in combination with LEIS and SSIMS measurements. The dependence of the restructuring process on annealing time, annealing temperature, and sample history is systematically investigated. Exposure to H 2 18O and air in the same temperature regime fails to induce the restructuring. UHV annealing of restructured, oxygen-enriched TiO 2 (110) surface smooths the surfaces and converts the rosette networks into strands and finally into the regular (1◊1) terraces. This is reported in an accompanying paper [M. Li, W. Hebenstreit, U. Diebold, Phys. Rev. B (1999), submitted ]. The rosette model is supported by first-principles density functional calculations which show a stable structure results, accompanied by significant relaxations from bulk-truncated positions. A mechanism for the dynamic processes of the formation of rosettes and (1◊1) islands is presented and the importance of these results for the surface chemistry of TiO 2 (110) surfaces is discussed.

All-atom ab initio energy minimization of the kaolinite crystal structure

Abstract Calculations that minimize the energy and optimize the geometry of all atomic coordinates for two proposed kaolinite crystal structures were performed using a first-principles, quantum chemical code based on local density functional theory. All calculations were performed using published unit-cell parameters. Inner- and interlayer H atom positions agree well with those determined by Bish (1993) from neutron diffraction data and confirm a unit cell with C1 symmetry.

Ab initio calculations of Ru, Pd, and Ag cluster structure with 55, 135, and 140 atoms

A massively parallel ab initio computer code, which uses Gaussian bases, pseudopotentials, and the local density approximation, permits the study of transition-metal systems with literally hundreds of atoms. We present total energies and relaxed geometries for Ru, Pd, and Ag clusters with N=55, 135, and 140 atoms. The N=55 and 135 clusters were chosen because of simultaneous cubo-octahedral (fcc) and icosahedral (icos) subshell closings, and we find icos geometries are preferred. Remarkably large compressions of the central atoms are observed for the icos structures (up to 6% compared with bulk interatomic spacings), while small core compressions (∼1%) are found for the fcc geometry. In contrast, large surface compressive relaxations are found for the fcc clusters (∼2%–3% in average nearest neighbor spacing), while the icos surface displays small compressions (∼1%). Energy differences between icos and fcc are smallest for Pd, and for all systems the single-particle densities of states closely resembles bu...

Ab initio calculations of adsorbate hydrogen-bond strength: ammonia on Pt(111)

Seven-layer slab results for 14 monolayer of NH3 on Pt(111) (so called α-NH3) are compared with NH3 on a 91-atom Pt cluster; we find that the latter closely mimics the extended surface. The calculations predict atop site occupancy for α-NH3 with N-down. The H-bond between α-NH3 and an additional N-down molecule (β-NH3) approaching from the gas-phase is then compared with that of two molecules in the gas phase; we discover the H-bond on the surface is almost three times stronger and the bond length appreciably shorter. Geometry relaxation then results in a 65 ± 5 degree tilt of the β-NH3 axis. Finally, slab calculations with 14ML each of α- and β-NH3 support this geometry over symmetrically coordinated β-NH3 and predict an adsorption energy in good agreement with experiment.

A new efficient method for density functional theory calculations of inhomogeneous fluids

The accurate computation of the effects of solvation on chemical systems can be done using density functional theories (DFT) for inhomogeneous multicomponent fluids. The DFT models of interest are non-local theories which accurately treat hard-sphere fluid mixtures; attractive inter-particle potentials (Lennard-Jones) are added as perturbations. In this paper, we develop and demonstrate a new efficient method for an accurate non-local DFT. The method described here differs from previous work in the use ot fast fourier transform (FFT) methods to carry out the convolutions. As with our previous real space work (J. Comput. Phys. 159(2) (2000) 407, 425), we demonstrate that the Fourier space approach can be solved with a Newton-GMRES approach; however, we now employ a very efficient matrix-free algorithm. A simple but effective preconditioner is presented. The method is demonstrated with calculations performed for one-, two-, and three-dimensional systems, including problems with single and multicomponent fluids. Timing comparisons with previous implementations are given.

A radar simulation program for a 1024-processor hypercube

We have developed a fast parallel version of an existing synthetic aperture radar (SAR) simulation program, SRIM. On a 1024-processor NCUBE hypercube it runs an order of magnitude faster than on a CRAY X-MP or CRAY Y-MP processor. This speed advantage is coupled with an order of magnitude advantage in machine acquisition cost. SRIM is a somewhat large (30,000 lines of Fortran 77) program designed for uniprocessors; its restructuring for a hypercube provides new lessons in the task of altering older serial programs to run well on modern parallel architectures. We describe the techniques used for parallelization, and the performance obtained. Several novel parallel approaches to problems of task distribution, data distribution, and direct output were required. These techniques increase performance and appear to have general applicability for massive parallelism. We describe the hierarchy necessary to dynamically manage (i.e., load balance) a large ensemble. The ensemble is used in a heterogeneous manner, with different programs on different parts of the hypercube. The heterogeneous approach takes advantage of the independent instruction streams possible on MIMD machines.

Application of a High Performance Parallel Eigensolver to Electronic Structure Calculations

In this paper we report the development of a very high performance parallel eigensolver based on the portable ScaLAPACK library, and its application to electronic structure calculations in the MP-Quest code. This work was done on ASCI-Red, a supercomputer based on over 4600 dual-processor Pentium Pro1 nodes at Sandia National Laboratories. We report sustained performance in the code of 605GFlops and peak performance in the eigensolver of 684GFlops. This is comparable to performance obtained from MP-Linpack on a similar sized problem. For a smaller problem we have sustained performance of 420GFlops in the application and peak performance in the eigensolver of 563GFlops. Impact of this work on the specific application is important, but the development of significant improvements to a portable eigensolver and other libraries will also benefit a number of applications.

DIET in the bulk: Evidence for hot electron cleavage of Si-H bonds in SiO2 films

Abstract The observed increase in leakage current through SiO2 films after hot electron exposure is ascribed to dissociation induced by electronic transitions (“DIET”) of bulk SiH bonds, producing mobile hydrogen. We use ab initio supercell bandstructure calculations at the local density functional level to locate features produced by hydrogen-containing defects in α-SiO2. The edge of the SiH σ∗ resonance is found to be about 2.7 eV above the conduction band rise, in good agreement with the observed threshold for hot electron induced damage in amorphous SiO2 films grown on Si substrates. The OH σ∗ resonance is almost 4 eV higher. Removing H from OH in the supercell does not affect the gap region (O− forms); however, removing H from SiH produces a mid-gap state, suggesting leakage current by hopping conductivity between Si dangling bonds. A Morse potential model is used to explore the dynamics of bond scission by short-lived ( σ∗ capture. Supercell calculations on interstitial atomic hydrogen indicate the energy cost to break an embedded SiH bond is about 0.6 eV less than in the gas phase. The DIET yield is substantially increased by reducing both ground and electron-attached state binding by this amount. While uncertainty over the displaced equilibrium in the electron-attached excited state remains, the computed DIET cross-section for reasonable parameters is ≈10−18 cm2, and is in agreement with the semi-empirically derived value for trap creation. Comparisons are made to surface DIET processes involving SiH bonds.

Linear Algebra for Dense Matrices on a Hypercube

A set of routines has been written for dense matrix operations optimized for the NCUBE/6400 parallel processor. This work was motivated by a Sandia effort to parallelize certain electronic structure calculations [l]. Routines are included for matrix transpose, multiply, Cholesky decomposition, triangular inversion, and Householder tridiagonalization. The library is written in C and is callable from Fortran. Matrices up to order 1600 can be handled on 128 processors. For each operation, the algorithm used is presented along with typical timings and es timates of performance. Performance for order 1600 on 128 processors varies from 42 MFLOPs (Householder tridiagonalization, triangular inverse) up to 126 MFLOPs (matrix multiply). We also present performance results for communications and basic linear algebra operations (saxpy and dot products).

Predicting Function of Biological Macromolecules: A Summary of LDRD Activities: Project 10746

This LDRD project has involved the development and application of Sandias massively parallel materials modeling software to several significant biophysical systems. They have been successful in applying the molecular dynamics code LAMMPS to modeling DNA, unstructured proteins, and lipid membranes. They have developed and applied a coupled transport-molecular theory code (Tramonto) to study ion channel proteins with gramicidin A as a prototype. they have used the Towhee configurational bias Monte-Carlo code to perform rigorous tests of biological force fields. they have also applied the MP-Sala reacting-diffusion code to model cellular systems. Electroporation of cell membranes has also been studied, and detailed quantum mechanical studies of ion solvation have been performed. In addition, new molecular theory algorithms have been developed (in FasTram) that may ultimately make protein solvation calculations feasible on workstations. Finally, they have begun implementation of a combined molecular theory and configurational bias Monte-Carlo code. They note that this LDRD has provided a basis for several new internal (e.g. several new LDRD) and external (e.g. 4 NIH proposals and a DOE/Genomes to Life) proposals.