RESUMO
Modeling of phase diagrams and, in particular, the anomalous re-entrant melting curves of alkali metals is an open challenge for interatomic potentials. Machine learning-based interatomic potentials have shown promise in overcoming this challenge, unlike earlier embedded atom-based approaches. We introduce a relatively simple and inexpensive approach to develop, train, and validate a neural network-based, wide-ranging interatomic potential transferable across both temperature and pressure. This approach is based on training the potential at high pressures only in the liquid phase and on validating its transferability on the relatively easy-to-calculate cold compression curve. Our approach is demonstrated on the phase diagram of Rb for which we reproduce the cold compression curve over the Rb-I (BCC), Rb-II (FCC), and Rb-V (tI4) phases, followed by the high-pressure melting curve including the re-entry after the maximum and then the minimum at the triple liquid-FCC-BCC point. Furthermore, our potential is able to partially capture even the very recently reported liquid-liquid transition in Rb, indicating the utility of machine learning-based potentials.
