Controlling Device Instabilities with i-ZnO Oxygen Content in Chalcogenide Solar Cells

Intrinsic zinc oxide (i-ZnO) is systematically used in combination with a buffer layer (generally cadmium sulfide) and a transparent conductive oxide (TCO) for the fabrication of the n-type side of thin film chalcogenide solar cells. However, despite this near ubiquitous empirical use, its role remains debatable. Moreover, a particularly intriguing factor regarding the performances is the role of oxygen content in the ZnO. Its concentration modifies mainly the oxide stoichiometry, conductivity and transparency. Oxygen is commonly used during the preparation of the transparent oxides, and allows improving the transmission to the detriment of the resistivity. While incorporating a little amount of oxygen during the TCO deposition is generally used for this reason , the role of oxygen in i-ZnO remains unclear but appears necessary to obtain a high resistivity layer, by preventing oxygen vacancies or zinc interstitials responsible of the n-type conductivity. Moreover, adding oxygen during the i-ZnO deposition could also affect the CdS buffer surface in a similar way as sputtered oxygenated CdS (CdS:O) commonly used in CdTe heterojunction solar cells, and could also reduce the film crystallinity, increase the optical bandgap, change the composition at the surface of the CdS and/or increase the defects density. As a result of these multiple interplays between oxygen content and film properties, the solar cells’ performance are significantly impacted.<br/>The work here presented brings insights on the role of the i-ZnO layer in the solar cell structure by investigating the effect of the oxygen incorporation during its preparation and its influence on the performance of the devices. By varying the oxygen ratio (from 0 to 15% of oxygen in the total gas mixture with Ar) during the i-ZnO deposition, several batches of solar cells (using both CZTS and CIGS absorbers) were prepared and characterized. Limited differences were observed in their optoelectronic properties, with all the oxygen ratios leading to comparable values under illumination. However, a clear difference was observed in the dark behaviour before illumination, with devices with high oxygen content requiring longer light soaking times to reduce instabilities. To highlight this instability, a series of J-V measurements in (1) dark, (2) red light, (3) white light and (4) again dark conditions was performed in CZTSe devices. These measurements were made after keeping the samples in dark for several days to avoid any possible effects from light-sensitive defects states. A first J-V measurement in dark (dark1) was recorded followed by a red illuminated J-V curve (using a 550 nm long pass filter with a 1 sun 1.5 AM spectrum white light), then a 1 sun 1.5 AM illuminated curve and finally another dark J-V curve (dark2). All the measurements are shown in the figure 1, and put in evidence a charging mechanism with white light illumination, that is required to stabilize the devices. It will be shown during the presentation that this relates to ionized acceptor defects at both the CdS and CdS/i-ZnO interfaces, which are neutralized when photogenerated holes are created with UV absorbed photons. The potential role of oxygen vacancies in this behaviour will be discussed. These results demonstrate the importance of controlling the defects density at interfaces to reduce device instabilities. More interestingly, this photo-induced mechanism, which shows reversibility after dark discharging, could be further exploited into specific light-based memory devices, which will be the developed in another presentation of MRS Spring 2023.