A Perspective of Materials Modeling

Abstract

The impossible combinations of materials properties required for essential industrial applications have made the present paradigm of empirically based experimental synthesis and characterization increasingly untenable. Since all properties of all materials are in principle describable by quantum mechanics (QM), one could in principle replace current empirical methods used to model materials properties by first principles or de novo computational design of materials and devices. This would revolutionalize materials technologies, with rapid computational design, followed by synthesis and experimental characterization only for materials and designs predicted to be optimum. From good candidate materials and processes, one could iterate between theory and experiment to optimize materials. The problem is that direct de novo applications of QM are practical for systems with ∼10^2 atoms whereas the materials designer deals with systems of ∼10^(22) atoms. The solution to this problem is to factor the problems into several overlapping scales each of which can achieve a scale factor of ∼10^4. By adjusting the parameters of each scale to match the results of the finer scale, it is becoming possible to achieve de novo simulations on practical devices with just ∼ 5 levels. This would allow accurate predictions of the properties for novel materials never previously synthesized and would allow the intrinsic bounds on properties to be established so that one does not waste time on impossible challenges.