Charging and discharging of defect states in CIGS/ZnO junctions A.E. Delahoy a, * , A. Ruppert a , M. Contreras b a Energy Photovoltaics, Inc., P.O. Box 7456, Princeton, NJ 08543, USA b National Renewable Energy Laboratory, Golden, CO 80401, USA Abstract The formation of photovoltaic junctions in thin ®lm CIGS has been explored through the deposition of ZnO directly onto the CIGS by the reaction of Zn and atomic oxygen. Most of these devices exhibit strong light soaking effects in which V oc (at 1 sun) can increase from 250 mV to over 550 mV. A 12.3% ef®cient ZnO/CIGS solar cell was achieved. The soaking effect, which is the main topic of this paper, was found to be bias or ®eld driven. Detailed investigations were conducted using a `drive and interrogate' procedure, in which the device is driven between two steady states corresponding to two different voltage biases, with brief interrogation of V oc at various times. Data spanning the temperature range 31±1258C were acquired. It was found that the V oc versus log(t) relaxation curves can be described by a stretched exponential, and scale with t´exp(2E/kT) as a parameter, yielding E  0:51 eV. Junction capacitance data are also reported. The underlying process, whether electronic or ionic, possesses a wide range of time constants. q 2000 Elsevier Science S.A. All rights reserved. Keywords: CIGS, Cu(InGa)Se x ; Zinc oxide; Junctions; Field; Defects; Metastability 1. Introduction Junction formation in CIGS is typically accomplished by deposition of CdS, either by vacuum evaporation or chemi- cal bath deposition, followed by overcoating with doped ZnO to reduce sheet resistance effects. The functions of CBD CdS are to increase band bending in the CIGS (either by introducing donors at the CIGS surface, or in the near surface region), to serve as an insulating buffer layer, and to minimize sputter damage. Omission of the CdS usually results in solar cells of poor ef®ciency, or whose ef®ciencies are at best reduced by a factor of 0.8. This latter factor has been reproducibly achieved at NREL through appropriate choice of buffer layer (i-ZnO) sputtering conditions. In this work we explore the performance of ZnO/CIGS junctions formed by a novel method, viz. the deposition of ZnO by the simultaneous supply of ¯uxes of Zn and atomic oxygen (AO). We term this the ROMEAO process (reaction of metal and atomic oxygen). The Zn source consists of a small Knudsen cell containing both Zn and In. The AO source consists of a quartz tube ®tted with external electro- des into which O 2 is introduced and excited by RF, the discharge products exiting through a small hole [1]. The discharge was operated at about 50 Pa and low RF power, and the Zn and AO ¯uxes were arranged to intersect about 4 cm from the substrate. Because of the remote nature of the AO source, the ROMEAO process allows ZnO deposition onto CIGS without sputter damage or ion bombardment, and it was therefore investigated as a possible means of elim- inating CdS. The CIGS, prepared for these studies at NREL, was generally used without any cleaning, and the substrate temperature was about 145±1508C. The ZnO ®lms were found to be always moderately conductive. 2. Results The resulting ZnO/CIGS cells exhibit a variety of light soaking effects, the most noticeable being an increase in V oc . An increase in ®ll factor is also usually observed. The changes are rapid at ®rst, but can continue at a reduced rate for long times. The effects relax in the dark. It was found that the improvement can also be induced by forward voltage bias, regardless of whether light is present. Exam- ples of the I±V curves obtained before and after soaking are shown in Fig. 1 below for cells C1067-60 and C1080-73b. Initial V oc 's can range from 100 to over 400 mV. From the point of view of ®nding a viable non-CdS process it is encouraging that a V oc of 565 mV has been observed (see Fig. 1, C1080-73b). This cell exhibited an ef®ciency of 12.3% after soaking. With ROMEAO ZnO on NREL supplied CIGS/i-ZnO, a V oc of 605 mV has been achieved. To a ®rst approximation, there is usually no observable change in J sc upon soaking. In some cases however, after anti-soaking by application of reverse bias (see below), the Thin Solid Films 361±362 (2000) 140±144 0040-6090/00/$ - see front matter q 2000 Elsevier Science S.A. All rights reserved. PII: S0040-6090(99)00787-7 www.elsevier.com/locate/tsf * Corresponding author. Tel.: 11-609-587-3000; fax: 11-609-587- 5355.