From first-principles to material properties

Many modeling problems in materials science involve finite temperature simulations with a realistic representation of the interatomic interactions. These problems often necessitate the use of large simulation cells or long run times, which puts them outside the range of direct ab initio simulation. In ionic systems, it is possible to introduce physically motivated model potentials for the interactions, in which additional degrees of freedom provide a 'cartoon' of the response of the electronic structure of the ions to their changing coordination environments and allow the compact representation of many-body contributions to the interaction energy. These potentials may then be parameterized by fitting the predicted forces and multipoles to a large body of information generated from ab initio calculations. The resulting potentials are predictive, of ab initio accuracy and have a high degree of transferability between different systems. (c) 2006 Elsevier B.V. All rights reserved.