One of the ultimate goals of computational materials science is to accurately predict material structure evolution under a set of user defined external conditions. This goal has become increasingly realistic with the emergence of faster computers and the application of multiscale algorithms (e.g., ab initio calculations, development of empirical interatomic potentials, molecular dynamics simulations based upon empirical potentials, Monte Carlo simulations, and continuum calculations). Our group focuses on a few of the multiscale algorithms. In particular, we develop empirical interatomic potentials and apply them in molecular dynamics to simulate the growth of thin film materials from vapor phases.

In a molecular dynamics simulation of vapor deposition, the transition of atoms from vapor to a solid surface and the subsequent surface reconstruction is simulated by tracking the positions of atoms using Newtons equations of motion. This correctly identifies the detailed atomic structures of a deposited film, and reveals many of the mechanisms active throughout the deposition process. However, the accuracy of the simulation depends upon the fidelity of the interatomic potentials used to calculate the interatomic forces. We have extended the potentials to a wide range of materials with different types of bondings including metallic, ionic, and covalent bondings.

Metallic Systems
The embedded atom method (EAM) potential initially developed by Daw and Baskes accounts for the local environment dependence of the potential, and therefore has been successfully used to for many metal systems. By normalizing the EAM elemental potentials, we have successfully developed an EAM potential database for metal alloys that can be any combinations of 18 metals (Cu, Ag, Au, Ni, Pd, Pt, Al, Pb, Fe, Mo, Ta, W, Mg, Co, Ti,Zr, Cr, and V). This EAM potential database has been used to simulate the growth of giant magnetoresistance metallic multilayers. These simulations indicated that increasing adatom energy flattens the interfaces of the multilayers due to impact induced atom displacement, but it also causes mixing between adjacent layers due to impact induced atom exchange. The results are supported by a direct examination of atomic scale structure using three dimensional atom probe analysis. They account for the observed properties of these metallic multilayers. Simulations also indicated that the addition of silver during deposition of copper can flatten the copper surface because silver segregates to the surface and mediates the surface morphology. These predictions were verified by Auger and atomic force microscopy analyses. The insights revealed by the simulations resulted in the invention of a modulated energy deposition scheme, where a low adatom energy is used to deposit the first a few atomic layers of a new material layer to minimize the exchange and mixing between dissimilar species, and a high adatom energy is then used to grow the remainder of that material layer to flatten the surface and interface (when another layer is deposited).

Ionic Systems
To study the atomic assembly of mixtures of metals and metal oxides, interatomic potentials must switch from ionic interactions in an oxide to metallic interactions in a metal. This requires an environment dependent charge on the atoms (ions). Our recently developed modified charge transfer ionic potential (CTIP) + EAM potential begins to enable such charge transfer behavior to be captured in oxygen metal alloy systems involving any number of metal elements. This provides a robust method to study the metal/metal oxide interfaces. This CTIP + EAM potential has been parameterized for O-Al-Zr and O-Al-Ni-Co-Fe systems. The O-Al-Ni-Co-Fe potential has been used to simulate the growth of magnetic tunnel junctions, consisting of a pair of ferromagnetic layers separated by a thin dielectric layer. The results indicated that there exists a threshold dielectric layer thickness around 10 Å below which the dielectric layer is always discontinuous. This discovery is verified by the high resolution transmission electron microscopy and three atom probe experiments, and can well account for the observed multilayer properties.

Covalent Systems
We have explored a variety of angular dependent covalent potentials for use in molecular dynamics simulations. We discovered that the bond-order potential (BOP) derived by Pettifor et al. from a tight binding approximation to multi-body quantum mechanics significantly improves over other more empirical potential in terms of correct description of energies and geometries of a variety of phases. Collaborating with Pettifor, we have now parameterized a new form of the GaAs BOP with both σ and π covalent bonding included. Simulation of GaAs growth using this potential revealed many new insights governing the crystalline quality of the GaAs films as a function of vapor flux ratio and substrate temperature. The results are in good agreement with experiments .