Breaking Badly: DFT-D2 Gives Sizeable Errors for Tensile Strengths in Palladium-Hydride Solids

Dispersion interactions play a crucial role in noncovalently bound molecular systems, and recent studies have shown that dispersion effects are also critical for accurately describing covalently bound solids. While most studies on bulk solids have solely focused on equilibrium properties (lattice constants, bulk moduli, and cohesive energies), there has been little work on assessing the importance of dispersion effects for solid-state properties far from equilibrium. In this work, we present a detailed analysis of both equilibrium and highly nonequilibrium properties (tensile strengths leading to fracture) of various palladium-hydride systems using representative DFT methods within the LDA, GGA, DFT-D2, DFT-D3, and nonlocal vdw-DFT families. Among the various DFT methods, we surprisingly find that the empirically constructed DFT-D2 functional gives extremely anomalous and qualitatively incorrect results for tensile strengths in palladium-hydride bulk solids. We present a detailed analysis of these effects and discuss the ramifications of using these methods for predicting solid-state properties far from equilibrium. Most importantly, we suggest caution in using DFT-D2 (or other coarse-grained parametrizations obtained from DFT-D2) for computing material properties under large stress/strain loads or for evaluating solid-state properties under extreme structural conditions.