Integrating computational and experimental thermodynamics of refractory materials at high temperature

We develop new computational and experimental methods to determine materials properties at high temperature, such as melting temperature, heat of fusion, heat capacity, and lattice constant. From density functional theory, we construct the small-size coexistence method and the SLUSCHI package to compute the properties accurately, as well as to fully automate the computation process. From experiment, we build experimental approaches, including ultra-high-temperature Drop-n-Catch (DnC) calorimetry and synchrotron X-ray diffraction on solid laser-heated aerodynamically levitated samples. Employing deep learning techniques, we build an ensemble graph-neural-networks model that predicts materials properties in milliseconds. The simultaneous development of computational and experimental approaches allows us to integrate these methods and the data generated by them.