Amorphous systems such as metallic glasses, non-crystalline ceramics and foams exhibit interesting inelastic reponse to shear. Molecular dynamics investigations of an amorphous pairwise interacting two-component Lennard-Jones solid reveal analogous properties. We propose a constitutive model for the dynamic response to shear of an amorphous solid considered as a collection of microscopic two-state systems. This model accounts for the observed response of the aforementioned computer simulations. It is further conjectured that an understanding of this non-linear response sheds light on our recent simulations of brittle/ductile fracture behavior in computational models of amorphous solids as well as dynamic fracture experiments performed by others on polymeric systems in the laboratory.
[P1.02] First-Principles Effective Hamiltonians for Piezoelectric Behavior
Eric Cockayne, Karin M. Rabe (Yale University)
Over the last few years, effective Hamiltonians have been constructed from first-principles calculations for several perovskite ferroelectrics, resulting in simple models consisting of 3-component vectors on a simple cubic lattice. Classical Monte Carlo and molecular dynamics simulations have yielded predictions for ferroelectric transition temperatures, latent heats and temperature dependence of structural parameters in very good agreement with experiment. Simulations based on these models will now be extended to investigate piezoelectric behavior of these systems, including (i) temperature dependence of the piezoelectric activity; (ii) strain vs.~E-field behavior; and (iii) structural phase transitions induced by E-fields or applied stress. Preliminary results for single-crystal PbTiO_3 will be presented and compared with available experimental data. Mixed systems are currently considered the most promising for optimization of piezoelectric behavior necessary for a variety of applications. A procedure for symmetrizing and averaging the results of first-principles calculations on ordered supercells to predict the properties of systems without long range cation order will be discussed, and results presented for the T=0 piezoelectric tensor of Pb_0.75Ge_0.25Te. Finally, progress in extending the effective-Hamiltonian approach to study the finite-temperature behavior of mixed ferroelectrics will be summarized.
[P1.03] The tubule phase of a crystalline surface
Mark Bowick (Syracuse U.), Marco Falcioni (Suracuse U./UCLA), Gudmar Thorleifsson (Bielefeld, Germany)
Crystalline phantom surfaces with bending rigidity show a very interesting and rich phase diagram. The surface is crumpled in the high temperature (low bending rigidity) phase and flat in the cold (high bending rigidity) phase. Incorporating an anisotropy, which is an experimentally relevant feature, changes the phase diagram dramatically: between the high and low temperature phases there is a novel tubule phase, characterized by one extended direction and a crumpled direction. We report the status and the results of a large scale Monte Carlo simulation of an anisotropic system and we stress the importance of improved simulation methods. We also discuss the challenge of extending the simulation to self-avoiding surfaces. A preprint is available at the LANL archives.
[P1.04] Calculating Potentials of Mean Force with a Nonlocal Density Functional Theory
Laura Douglas Frink, Andrew Salinger (Sandia National Laboratories)
When macromolecules or colloidal particles interact at short range as is the case in crystallization, molecular recognition, and self-assembly, the solvent often cannot be considered a continuum. Rather, surface forces experiments have shown that the solvent is nonuniform and produces oscillatory forces between the surfaces or macromolecules of interest. In order to simulate any of the long-time phenomena listed above, it is necessary to obtain potentials of mean force (PMF) between the macromolecules given the solvent state point. Until now, PMF have been calculated considering electrostatics exclusively. In this paper we present the computational methodology for calculating PMF that include both electrostatic and solvation effects. This method employs a nonlocal density functional theory(DFT) in novel 2 and 3-dimensional implementations. We will discuss the efficiency of calculating PMF using DFT as compared to direct molecular simulation. In addition, we will discuss when it is appropriate to apply the Derjaguin approximation for calculating PMF between surfaces with complex geometry and nonuniform charge distribution.
[P1.05] Theoretical Calculation of Ground State Geometry and Electronic Structure for Semiconducting Crystals and Polymers
Bradford K. Dickerson, Garett W. Yoder, An-Ban Chen (Auburn University)
We have developed an ab initio LDA code using short-ranged (~3rd neighbor) orbitals, norm-conserved psuedopotential, and Harris approximation with a renormalized density derived from these compact orbitals. With the inclusion of the order-N density matrix approach and novel angular integration for exchange correlation matrix elements, this scheme is capable of modeling large systems. This method approaches a computation time of a tight-binding method without the need to fit the tight-binding parameters. Applied to over 20 semiconducting crystal we find bondlengths around 2% of experiment, elastic constants around 10% of experiment, and relative binding energies are predicted correctly. On the other hand, the electronic structure of organic polymers is more sensitive to geometry and cannot be modeled well with the LDA Hamilitonian alone. The well-known SSH model predicts the band gap of trans-polyacetelyene, but cannot predict the geometry of the system. Combining ideas from SSH and LDA methods we used a semi-empirical correction fixing the bondlengths and HOMO-LUMO separation in benzene and the band gap and dimerization in trans-polyacetylene as a small modification of the first neighbor \pi-\pi LDA matrix elements. With this correction, we were able to predict both structural properties and band gaps of cis-polyacetylene, poly(p-phenylene) and poly(p-phenylene vinylene) and are currently investigating polymers that contain Nitrogen and Sulfur.
[P1.06] Computational Insights into Potential Automotive Catalysts
K. C. Hass (Ford Motor Co.)
Although the design of new catalysts remains largely an Edisonian endeavor, first-principles quantum calculations are beginning to add greatly to our understanding of catalytic mechanisms and materials. This presentation will focus on recent progress made in modeling Cu-exchanged zeolite catalysts, which are among the most active candidates for controlling NOx emissions from ``lean-burn'' and diesel engines. The adsorption and subsequent reactions of various species at active, non-framework Cu sites in these materials are examined using finite cluster models and density functional theory. A complete, kinetically plausible cycle for the catalytic decomposition of NO by isolated, zeolite-bound Cu+ sites is identified. In contrast to previous suggestions, this cycle is initiated by the formation of an unusual Cu-ON intermediate, while the more stable Cu-NO species acts only as a spectator. The relative uniqueness of Cu-exchanged zeolites for catalyzing NO decomposition results from the ability of these materials to bind both atomic and molecular oxygen in a reversible manner.