Modeling martensic transformations in crystalline solids: validity and redesign of geometric approaches

Abstract

Taking use of group theory, elegant mathematics allows transformations between unit cells and the estimate of intermediate structures. This, however, implies a (local) second order-type picture of what in reality is a first-order phase transition. Recent constant pressure molecular dynamics simulations combined with the transition path sampling approach opened a new level of detail for the mechanistic studies of pressure-induced phase transitions. For a selection of computationally favorable systems, we were able to observe local nucleation events, followed by subsequent phase growth and, provided large enough simulation cells, even the identification of phase growth from multiple centers leading to polycrystalline structures. On this basis, the validity of the long-established geometric approaches must be reconsidered. Phase interfaces are typically not smooth (which would justify a local second-order type modeling) but may be as sharp as a single layer of atoms. To account for this issue, a simple and generally applicable concept for the more appropriate design of the geometric modeling approaches to solid-solid transitions is outlined.