Using Machine Learning potentials for barocaloric study of AgI

Barocalorics offers a sustainable path to more efficient cooling devices using solid-state materials. To date, experimental research has dominated the field — largely due to the complexity of simulating the phase transitions observed during the thermodynamic cycle, and the diverse family of materials with promising caloric properties. In this talk, I will introduce a protocol that uses Machine Learning potentials to obtain barocaloric effects from order-disorder phase transitions, paving the way for faster development and discovery of next-generation barocaloric materials. For this, I will present a comparison between experimental results and those obtained through simulations for the solid electrolyte AgI.