Optimizing the growth path for Cu(In,Ga)Se2 thin films co-evaporated at low temperatures on flexible polyimide foil

Abstract

Thin film solar cells with Cu(In,Ga)Se2 (CIGSe) absorbers prepared by co-evaporation reach efficiencies >20% even on flexible polymer substrates and when deposited at low growth temperatures. The use of flexible polyimide (PI) substrates has advantages in terms of roll-to-roll device fabrication. Also, PI-based devices are extremely light weight and have a shorter energy pay-back time compared to devices on glass substrates. Commercial PI foils, however, only tolerate temperatures below 500°C, which can lead to limitations during deposition by multi-stage co-evaporation with the result of reduced absorber quality and a pronounced Ga gradient. Furthermore PI has no intrinsic alkaline reservoir like soda-lime glass. In the historical development of the CIGSe device, major steps for an improved absorber quality were the introduction of a Cu rich growth phase during the deposition process, the incorporation of Ga to modify energy band gradients and the supply of Na by using sodalime glass substrates. In this work, we adapt these steps to our in-house low-temperature CIGSe technology on PI and show, how we could increase the solar cell efficiency from 12.6% up to 17.9%. Studying the growth dynamics during our multi-stage co-evaporation process at a substrate temperature of 450°C by in-situ real-time energy-dispersive X-ray diffraction and comparing them to the high-temperature grown CIGSe at 530°C underlined the importance of the Cu-rich growth stage for low temperature growth. Then, the Ga depth profile was adjusted by a modified deposition sequence in all stages yielding an increased minimal band-gap energy. The dominant CIGSe XRD pattern was influenced by the Se flux profile, which, hence, needed to be adjusted with respect to the cationic flux rates. Finally, the Na supply was changed from a NaF precursor to a NaF post deposition treatment. An improved performance of the device was observed when the CIGSe thin film was treated with KCN before CdS-CBD deposition. Retracing our learning curve confirms the close intercorrelation of the different device aspects to be optimized.