The kesterite semiconductor Cu2ZnSnS(e)4 is seen as a suitable absorber layer to replace Cu(In,Ga)Se2 in thin film solar cells, if thin film photovoltaics are to be deployed on the terawatt scale. Currently the best devices, and hence the best kesterite absorber layers are grown away from stoichiometry and are zinc rich and copper poor, presumably leading to the formation of ZnS(e). However, it has been shown that secondary phases present in an absorber layer reduce device performance. If growth in Zn rich conditions seems to be mandatory, then any secondary phases formed should be grown on the surface of the absorber layer so that they may be easily removed by etching. Therefore, it is important to know how and why secondary phases form, and if possible, how to segregate them to the surface of the absorber layer.
Here we show that ZnSe is formed at the initial stages of absorber formation from annealing metal stacks in selenium vapor. Further we demonstrate that the way the precursor metals are distributed on the substrate leads to different absorber layer performances in full devices. The importance of selenium vapor pressure is highlighted in respect to the order of selenisation of the metals, Zn before Cu. Additionally, the importance of selenium and tin selenide vapor pressure during annealing is reviewed with regard to avoiding a decomposition of the Cu2ZnSnSe4 to ZnSe and Cu2Se phases. Regardless of the atmosphere above the absorber, the reaction of the absorber with molybdenum appears unavoidable without the use of a passivation strategy. Counter-intuitively, it is demonstrated that for our absorber layers grown under Zn-rich conditions, removal of the ZnSe is harmful for device performance.