A multilevel workflow for designing new photovoltaic
materials
based on high-throughput calculations is proposed, which consists
of a structure predictor coupled to a property calculator. With the
chemical composition as the only input, the workflow will automatically
predict structures with theoretically high spectroscopically limited
maximum efficiency (SLME). Based on this workflow, 4 thermodynamically
stable (one of which is new) and 31 metastable (22 of which are new)
non-toxic ABN2 (A = Mg, Ca, Sr, Zn; B = Sn, Ti, Zr) structures
with high SLMEs have been discovered. Among these, MgTiN2 (Fd-3m) and ZnSnN2 (Pna21, Pmc21, and P-4m2) structures are suggested to be promising
photovoltaic materials because of their good thermodynamic stability
(E
hull < 10 meV per atom), high SLME
(>20%), and small carrier effective mass.