Berlin 2018 – scientific programme

Parts | Days | Selection | Search | Updates | Downloads | Help

CPP: Fachverband Chemische Physik und Polymerphysik

CPP 49: Soft Matter Physics: Emerging Topics, New Instruments and Methods

CPP 49.6: Talk

Wednesday, March 14, 2018, 16:15–16:30, C 230

A generalized convex hull construction for computational materials discovery — •Andrea Anelli, Edgar Engel, and Michele Ceriotti — EPFL, Lausanne,Switzerland

Computational materials searches have the ultimate goal of discovering technologically relevant materials with desired physical properties. State of the art methods can generate thousands of locally-stable configurations. Selecting among these the few structures that can be experimentally synthesised is a needle-in-a-haystack kind of problem.

Conventionally, this screening is performed using a convex hull construction (CH), which can only identify structures stabilized by manipulation of a particular thermodynamic constraint (e.g. density or composition) chosen on the basis of experimental evidence or intuition.

The predictive power of such a formulation is weakened by its dependence on prior knowledge of the system, disregarding configurations that could appear as (meta-)stable under unconventional choices of constraints. We introduce a generalized CH based on an abstract representation of structural features, which are dependent solely on the cartesian coordinates of the systems and constitute a general, agnostic set of thermodynamic constraints. We further introduce a probabilistic scheme that is consistent with uncertainty quantification of the convex-hull construction, reducing the impact of modelling errors on the determination of synthesizable structures. We demonstrate the predictive power of our formulation by applying it to Hydrogen at TPa pressures, the binary high-pressure Oxygen-Hydrogen system and, increasing the structural complexity of the system, on Pentacene molecular crystals