Spectroscopic ellipsometry for analysis of polycrystalline thin-film photovoltaic devices and prediction of external quantum efficiency

Abstract

Complete polycrystalline thin-film photovoltaic (PV) devices employing CuIn1−xGaxSe2/CdS and CdS/CdTe heterojunctions have been studied by ex situ spectroscopic ellipsometry (SE). In this study, layer thicknesses have been extracted along with photon energy independent parameters such as compositions that describe the dielectric function spectra ε(E) of the individual layers. For accurate ex situ SE analysis of these PV devices, a database of ε(E) spectra is required for all thin film component materials used in each of the two absorber technologies. When possible, database measurements are performed by applying SE in situ immediately after deposition of the thin film materials and after cooling to room temperature in order to avoid oxidation and surface contamination. Determination of ε(E) from the resulting in situ SE data requires structural information that can be obtained from analysis of SE data acquired in real time during the deposition process. From the results of ex situ analysis of the complete CuIn1−xGaxSe2 (CIGS) and CdTe PV devices, the deduced layer thicknesses in combination with the parameters describing ε(E) can be employed in further studies that simulate the external quantum efficiency (EQE) spectra of the devices. These simulations have been performed here by assuming that all electron-hole pairs generated within the active layers, i.e. layers incorporating a dominant absorber component (either CIGS or CdTe), are separated and collected. The active layers may include not only the bulk absorber but also window and back contact interface layers, and individual current contributions from these layers have been determined in the simulations. In addition, the ex situ SE analysis results enable calculation of the absorbance spectra for the inactive layers and the overall reflectance spectra, which lead to quantification of all optical losses in terms of a current density deficit. Mapping SE can be performed given the high speed of multichannel ellipsometers employing array detection, and the resulting EQE simulation capability has wide applications in predicting large area PV module output. The ultimate goal is an on-line capability that enables prediction of PV sub-cell current output as early as possible in the production process.