Computational materials science from first principles

Abstract

We review recent developments in atomistic computer simulations of matter incorporating both quantum and classical statistical mechanical elements. These methods treat the electronic and geometric structure of solids, liquids and molecules on an equal footing and require no a priori knowledge (i.e. they are based on “first principles”, the basic laws of quantum and classical physics, not on experimental information). They borrow elements from quantum chemistry and solid state electron theories on one side and classical mechanics on the other. Such unified approach leads to more reliable computer-based predictions of materials properties over traditional simulation methods but requires substantial supercomputing resources. We present two typical applications in chemistry (nanoscale clusters of the element phosphorus) and solid state physics (the element hydrogen at extremely high compression).