Modeling microstructural effects of deformation resistance and thermal conductivity

Abstract

This is a study of the microstructural influences on thermo-mechanical behavior of selected metals and ceramics using computer simulation, with original contributions in both theoretical and applied aspects. There are three major thrusts. First, by constructing a many-body empirical potential for ZrCx and then carrying out MD simulations to calculate its lattice thermal conductivity, I obtain the first quantitative evidence ever that the vibrational contribution is only a small part of the total thermal conductivity of refractory carbides at realistic carbon vacancy concentrations. This is a long-standing problem which even the most recent review article on the subject give what I now believe is the wrong estimate. Second, ideal strengths are calculated for Ar,Cu,SiC crystals using both lattice and molecular dynamics methods. A set of homogeneous instability criteria are derived. Tension tests are performed on amorphous and nanocrystalline SiC at room temperature, based on which a grain size cutoff of ~20 nm is extrapolated for the Hall-Petch effect. Nano-indentation is performed on single-crystal and nanocrystalline Cu, and bursts of dislocation loops is observed at a local stress level consistent with recent experiments on Cu thin films. Third, an invariant loop summation similar to the J-integral is derived for the driving force on defect motion, but with the loop size now down to nanometers, and the summation now expressed in terms of interatomic forces instead of stress, a field concept which is hard to use in atomistic calculations and becomes ill-defined when defect separations approach the nanometer scale. It is shown first that the change in a system's total Helmholtz free energy due to a defect's move can be approximated by a local quantity involving only scores of atoms immediately surrounding the defect. Then, perturbation expansion is used to evaluate this local invariant for defect translation using only the current configuration. This driving force measure is then tested on a) self-interstitial diffusion near free surface in [alpha]-iron, b) crack-tip extension near a void in Si, c) screw dislocation translation in Si, with convincing results down to literally r = 1 nm, at a fraction of the cost of a full relaxation or free energy calculation for the whole system. This means that defect mobility can now be characterized by a universal and invariant standard, computable from a tiny atomistic calculation without relying on elasticity formulas or image summations. The standard is then used to determine the true Peierls-Nabarro stress in Si-like materials.