RESEARCH ARTICLE Surface preparation for 10% efficient CZTSe solar cells Louis Grenet 1 | Fabrice Emieux 1 | Léo Choubrac 2 | José A. Márquez 2 | Eric De Vito 1 | Frédéric Roux 1 | Thomas Unold 2 1 Univ. Grenoble Alpes, CEA, Liten, Grenoble, 38000, France 2 Department Structure and Dynamics of Energy Materials, Helmholtz-Zentrum Berlin, Hahn-Meitner-Platz 1, Berlin, 14109, Germany Correspondence Louis Grenet, CEA, LITEN, 17, rue des Martyrs, F-38054, Cédex 09, Grenoble, France. Email: louis.grenet@cea.fr Funding information H2020 Programme, Grant/Award Number: H2020-NMBP-03-2016-720907 (STARCELL) Abstract Kesterite-based solar cells suffer from a large open-circuit voltage deficit, which largely arises from carrier recombination at the buffer interface. In this study, we compare two strategies to passivate the absorber surface in order to fabricate devices with power conversion efficiency higher than 10% and an open-circuit voltage deficit as low as 306 mV. These two strategies consist of annealing in air or performing a chemical etching of the absorbers before buffer deposition. They lead similarly to a significant reduction of the interface recombination but as well to a shortening of the minority carrier diffusion length from 1 μm to less than 500 nm. This latter effect limits the short-circuit current and fill factor of the devices but is largely compensated by the open-circuit voltage gain of more than 100 mV. For the absorber air annealing, which is the simplest solution to implement, absolute photoluminescence measurements reveal that the voltage gain is directly linked to a drop in the nonradiative losses in the absorber and to a small reduction of the band tailing. It is demonstrated that the removal of detrimental secondary phases at the surface of the absorber due to oxidation at elevated temperature and etching in the basic CdS solution is responsible for these improved opto-electronic properties. On the contrary, the apparent Cu-depletion observed after air annealing is totally recovered after the chemical bath and cannot be responsible for the improved performances. 1 | INTRODUCTION Cu 2 ZnSn(S,Se) 4 (CZTSSe) materials are promising candidates to replace Cu (In,Ga)(S,Se) 2 (CIGS) absorbers in thin film solar cell technology because of their similar opto-electronic properties but without using critical raw materials 1 However, the maximum certified power conversion efficiency (PCE) achieved with kesterite-based devices is only 12.6%, 2,3 which is far from the 23.35% demonstrated by CIGS devices. 4 The main limitation of kesterite devices lies in the large deficit in open-circuit voltage (V OC deficit expressed as V OC-SQ - V OC , where V OC-SQ is the maximum achievable V OC in the Shockley–Queisser (SQ) limit depending on the bandgap E G of the absorber 5 ). This V OC deficit arises either from the bulk material property or its interfaces with the front and back electrodes. 6 If recombination in the bulk absorber is said to be mainly responsible for the voltage limitation in optimized processes, a significant part of the V OC deficit is attributed to the absorber/buffer interface. 7 Thus, studying and optimizing this interface have been subject of intense research recently, and two main strategies have demonstrated the improvement of the devices photovoltaic (PV) properties 8 : (i) using a chemical etching of the absorber prior to CdS deposition and (ii) using a thermal posttreatment of the absorber. For the second case, several groups have noticed the positive impact of an annealing stage after the absorber synthesis, 9–14 but the methodology and the reasons for such an improvement are not clearly established: This stage is either performed in air, under vacuum, or under inert atmo- sphere and either on bare absorbers or with the CdS or CdS/TCO protecting layers 9–12 while various reasons are invoked to explain this behavior (Cu depletion or Zn enrichment at the surface, passivation with SnO x , …). 14,15 Received: 30 June 2020 Revised: 1 September 2020 Accepted: 29 September 2020 DOI: 10.1002/pip.3356 Prog Photovolt Res Appl. 2020;1–12. wileyonlinelibrary.com/journal/pip © 2020 John Wiley & Sons, Ltd. 1