A framework for quantifying uncertainty in DFT energy corrections

Abstract

In this work, we demonstrate a method to quantify uncertainty in corrections to density functional theory (DFT) energies based on empirical results. Such corrections are commonly used to improve the accuracy of computational enthalpies of formation, phase stability predictions, and other energy-derived properties, for example. We incorporate this method into a new DFT energy correction scheme comprising a mixture of oxidation-state and composition-dependent corrections and show that many chemical systems contain unstable polymorphs that may actually be predicted stable when uncertainty is taken into account. We then illustrate how these uncertainties can be used to estimate the probability that a compound is stable on a compositional phase diagram, thus enabling better-informed assessments of compound stability.

Ab Initio Quantum-Mechanical Predictions of Semiconducting Photocathode Materials

“…Toward Cs-rich compositions (left side of Figure 7 c) this trend is reverted, with the data from MP being slightly higher in energy for Cs Sb ( = 0.75). The reason for this behavior can be found in the empirical correction scheme applied on the data originating from MP [ 149 ]. Additionally, the Cs phase corresponding to the lowest energy is cubic, while in the OQMD, it is trigonal and, evidently, more stable.…”

Exploring the Cs-Te phase space via high-throughput density-functional theory calculations beyond the generalized-gradient approximation

“…This is a consequence of the correction scheme applied to the values stored in MP, which increases in magnitude at increasing relative Te content in the binary crystals. 63 In spite of these differences, the minimum of the convex hull in all three datasets is found for double amount of Cs atoms compared to Te ones (2:1 Cs:Te ratio). The absolute minimum (-1.26 eV/atom) is obtained with SCAN while with PBE we find -1.13 eV/atom in the MP dataset and -1.04 eV/atom in OQMD.…”

“…12,13 The discrepancy between MP and OQMD datasets, in spite of the analogous computational parameters adopted to generate them, originates from an empirical correction for Te-containing compounds applied to the formation energies in MP. 63 The convex hull shown in Fig. 2c) is obtained for each dataset (OQMD, MP, and our own SCAN results) by connecting the formation energies of the most stable phase for each composition including linear combinations of phases with different compositions.…”

Navigating the Ti-C-O and Al-C-O ternary systems through theory-driven discovery

“…II C), whereas total energies in the Materials Project are evaluated with the PBE functional [83]. Third, the Materials Project applies an energy correction to certain classes of materials, such as oxides and superoxides, with the goal of producing more accurate formation energies [84]; in this work, such corrections aren't applied and all compounds are treated at the same level of theory. Finally, the most stable crystal structure for a given compound can differ between the present work and the Materials Project.…”