The highest melting point material: Searched by Bayesian global optimization with deep potential molecular dynamics

The interest in refractory materials is increasing rapidly recent decades due to the development of hypersonic vehicles. However, substance that has highest melting point (Tm) keeps a secret, since precise measurements extreme conditions are overwhelmingly difficult. In present work, an accurate deep potential (DP) model Hf–Ta–C–N system was first trained, and then applied search for material by molecular dynamics (MD) simulation Bayesian global optimization (BGO). predicted points agree well with experiments confirm carbon site vacancies can enhance rock-salt-structure carbides. solid solution N verified as another new more effective enhancing approach HfC, while conventional routing Ta (e.g., HfTa4C5) not suggested result maximum point. (~4236 K) achieved composition HfC0.638N0.271, which ~80 K higher than value Hf–C binary system. Dominating mechanism addition believed be unstable C–N N–N bonds liquid phase, reduces phase entropy renders less stable. improved gas generation during oxidation provide modify thermal protection