Quantifying components of entropy in the latent heat of melting

The latent heat of melting in materials, L= T_mΔS_fus, is equal to the melting temperature, T_m, times the entropy of fusion, ΔS_fus. While many studies focus on predicting the melting temperature of materials, the empirical Richard’s rule of approximately 1.1 k_B/atom is often still used to estimate the entropy of fusion despite calorimetric measurements of ΔS_fus spanning nearly an order of magnitude. Here, we combine inelastic neutron scattering and molecular dynamics informed by machine learning interatomic potentials to quantify the vibrational, configurational, and electronic contributions to the entropy of fusion in Ge, Si, Bi, Sn, Pb, and Li. Our results reveal a correlation between the total and vibrational entropies across the melt and, using the framework of potential energy landscape theory, imply that the change in the number of basins and the change in their curvature upon melting are inversely related.