The presence of small amounts of sodium has been shown to improve the electronic performance of Cu(In,Ga)Se-2 (CIGS) solar cells, but the origins of this effect have not yet been fully resolved. In this work, we have addressed the questions involving the role of sodium in CuInSe2 (CIS) using densityfunctional-theory-based calculations. We find no direct way how the creation of Na-related point defects in bulk CIS would enhance p-type conductivity. Instead, we demonstrate that Na reduces copper mass transport due to the capture of copper vacancies by Na-Cu defects. This finding provides an explanation for experimental measurements where the presence of Na has been observed to decrease copper diffusion. The suggested mechanism can also impede V-Cu-related cluster formation and lead to measurable effects on defect distribution within the material. The presence of small amounts of sodium has been shown to improve the electronic performance of Cu(In,Ga)Se 2 (CIGS) solar cells, but the origins of this effect have not yet been fully resolved. In this work, we have addressed the questions involving the role of sodium in CuInSe 2 (CIS) using density-functional-theorybased calculations. We find no direct way how the creation of Na-related point defects in bulk CIS would enhance p-type conductivity. Instead, we demonstrate that Na reduces copper mass transport due to the capture of copper vacancies by Na Cu defects. This finding provides an explanation for experimental measurements where the presence of Na has been observed to decrease copper diffusion. The suggested mechanism can also impede V Cu -related cluster formation and lead to measurable effects on defect distribution within the material.Effect of sodium incorporation into CuInSe 2 from first principles Effect of sodium incorporation into CuInSe 2 from first

We calculate the energetics of vacancies in CuInSe(2) using a hybrid functional (HSE06, HSE standing for Heyd, Scuseria and Ernzerhof), which gives a better description of the band gap compared to (semi)local exchange-correlation functionals. We show that, contrary to present beliefs, copper and indium vacancies induce no defect levels within the band gap and therefore cannot account for any experimentally observed levels. The selenium vacancy is responsible for only one level, namely, a deep acceptor level ε(0/2-). We find strong preference for V(Cu) and V(Se) over V(In) under practically all chemical conditions.

No abstract