DLTS with Bayesian Inference for Analyzing RPD Induced Defects in Bulk near SiO2/Si Interface

Abstract

Bulk defects are induced in the Si crystal near the SiO2/Si interface in a carrier-selective contact solar cell by indium tin oxide (ITO), a reactive plasma deposition (RPD) process. These electrical properties were evaluated by deep level transient spectroscopy (DLTS) with Bayesian inference. The bulk defects were distinguished from the interface by varying the pulse voltage Vp. The capacitance transient decay signals were converted to the DLTS spectra by Fourier transformation. The DLTS spectrum was deconvolved into signals originating from different types of defects using Bayesian optimization. Three types of electron traps, three types of hole traps in the bulk, and one interface defect were generated by RPD. The electron trap E1 and hole trap H1 had almost the same energy levels near the mid-gap of Si, and they had large capture cross-sections. Their concentrations were almost the same. Therefore, it is suggested that E1 and H1 signals were obtained from the same defect and that this defect may act as a recombination center because it efficiently captures both the electrons and holes.