The occurrence of secondary phases in CuZnSn(S,Se) (CZTSSe) is highly expected due to the off-stoichiometry conditions required for the preparation of high efficiency devices. A strong attention has been given to minimization of the occurrence and selective removal of ternary Cu-Zn-(S,Se) and binary Zn(S,Se) and Cu-(S,Se). However, so far very little attention has been given to the role of other potential phases as Sn(S,Se). The selective removal of this phase is important because it can occur either by stoichiometric deviation or also from condensation from the annealing atmosphere, which usually contains Sn and the chalcogen. As processes are further optimized to achieve higher efficiency devices, it is important to increase the level of control on the presence of these potential detrimental low band gap phases.
In this work we present a simple chemical route for the selective removal of Sn(S,Se) by using (NH4)S solutions as an etchant in the concentration range of 0.5 M to 3.0 M. The analysis of the etch rate performed on the different secondary phases as well as on CZTSSe, both in powder forms, allow us to conclude that a 3.0 M solution of (NH)S is highly selective for Sn(S,Se) removal. Using SEM, EDX and XRD we show that there are two types of Sn(S,Se) aggregates on the CZTSSe surface: those related to stoichiometric deviations typically encrusted on the surface, and those related to the condensation from the annealing atmosphere during the cooling down process, typically observed as surface over-growths. Whereas a 0.5 M (NH)S solution does not remove the Sn(S,Se) aggregates, a 1.0 M solution seems to slightly affect the aggregates and the 3.0 M solution is very effective at removing both types of aggregates.
Using XRF we observe that both, the Cu/(Zn+Sn) and Zn/Sn ratios slightly increase, in agreement with XRD. The impact of the presence of these phases on the characteristics of the solar cells has been assessed by the analysis of devices produced both with unetched and (NH)S etched absorbers.
These data confirm that the etching with 0.5 M and 1.0 M solutions are only somewhat useful to remove Sn(S,Se) secondary phases, obtaining devices with almost the same optoelectronic parameters as those measured on the unetched samples. Conversely, devices obtained with layers etched in a 3.0 M solution exhibit a remarkable increase of the efficiency from typically 3.0% for the unetched sample, up to more than 5.0% for the etched ones. This is mainly related to an increase on the V and F.F. of the solar cells, while the J is almost unaffected. Using test samples with Sn(S,Se) phases intentionally grown onto the surface, we corroborate that this feature is not only related to the removal of the low band-gap Sn(S,Se) phases which explain the increase of the V, but also to the passivation of the surface. Finally, the impact of this chemical route in the further development of CZTSSe is presented.