Accurate Ab Initio Calculations on Various PV-Based Materials: Which Functional to Be Used?
Adrien Stoliaroff, Camille Latouche
Abstract
Herein we report a thorough ab initio investigation targeting the optical and electronic properties of 14 binary, ternary, and quaternary compounds (CuInSe2, CuGaSe2, ZnO, ZnS, CuAlSe2, CuInS2, AgInSe2, CuGaS2, AgAlSe2, AgGaSe2, CdIn2S4, Cu2SnZnSe4 (CZTSe), CdS, and α-SiO2 (quartz)) and 1 elemental solid: C (diamond). To perform the present benchmark, different classes of functionals have been chosen in order to sample modern computational methods (e.g., GGA, global hybrids PBE0 and B3LYP, screened hybrid functional HSE, and many-body methods). Our results demonstrate that the PBE-GGA functional reproduces the structure well but is unable to predict optoelectronic properties. However, B3LYP is the worst performing functional for reproducing the structure but, astonishingly, provides accurate band gaps. PBE0 and HSE usually overestimate and underestimate band gaps, respectively. Single-shot many-body calculations (G0W0) were tested on top of the PBE structure and density for several compounds.