Efficient DFT prediction of chemical and structural stability using van der Waals correction: application for A3B2Ga3O12 garnets (A = Lu, Y and B = Al, Sc)

Abstract

Several seamless van der Waals (vdW) correction methods available for a wide range of systems could be expected to enhance stability predictions by accounting for the vdW effect. The stability of material can be evaluated using chemical potential phase diagram (CPD) which reveals the elemental chemical potential conditions for a successful synthesis. In this work, viability of various vdW correction approaches in improving the accuracy of stability prediction for A3B2Ga3O12 garnets (A = Lu, Y and B = Al, Sc) has been studied. From the results, we have found that vdW-df-cx, Grimme-D3, vdW-df-c09, and vdW-df2-c09 significantly improve ΔH prediction with MAPE of >5.0% lower than PBE, which exhibit their potential for stability prediction based on the CPD analysis. For CPD construction whose reliability is based on ΔH prediction, vdW-df-cx which can minimize the MAPE in ΔH, relative to experimental data, is selected as the best method among all studied vdW approaches. A more accurate description of total energy of O2 molecule and the competing compounds with layered structure can be also acquired by incorporating vdW interaction. However, the MAPE in lattice constant reveals that there is no significant improvement of lattice constant prediction for the studied garnets and their competing compounds. The vdW method which gives the MAPE in lattice constant slightly lower than that of PBE is vdW-df2-b86r. Although we found that the vdW corrections can improve material stability prediction, there is still room for the development of a novel DFT-based vdW method capable of accurately predicting both the lattice constant and ΔH of solids, including complex materials like garnets.