What is the likelihood that a hypothetical material-the combination of a composition and crystal structure-can be formed? Underpinning the reliability of predictions for local or global crystal stability is the choice of thermodynamic potential. Here, we discuss recent advances in free energy descriptions for crystals including both harmonic and anharmonic phonon contributions to the vibrational entropy. We critically discuss some of the techniques and descriptors, including data-driven machine learning approaches, being developed to assess the stability and synthesisability of solids. Avenues are highlighted that deserve further attention including thermodynamic and kinetic factors that govern the accessibility of metastable structures away from equilibrium. Prediction of crystal stability and synthesisability is crucial for accelerated materials design. We discuss modern free energy methods for building more accurate models and data-driven approaches suitable for screening large chemical spaces.
Funding Agencies|K. T. acknowledges the Independent Research Fund Denmark for funding through the International Postdoctoral grant (0164-00015B). A. M. G. was supported by EPSRC Fellowship EP/T033231/1. J. K. acknowledges support from the Swedish Research Council (VR) prog [0164-00015B]; Independent Research Fund Denmark [EP/T033231/1]; EPSRC Fellowship [2021-00486]; Swedish Research Council (VR) program; EPSRC [EP/T033231/1] Funding Source: UKRI; Swedish Research Council [2021-00486] Funding Source: Swedish Research Council