Micron-scale characterization of laser processed silicon via low temperature micro-photoluminescence spectroscopy

Abstract

Laser processing is now regarded as a promising tool to reduce the cost and complexity of fabricating the formation of localized contacts between heavily doped silicon and metal, features which have become an important element in high efficiency silicon solar cells, such as a passivated emitter and rear cell (PERC) and an interdigitated back contact cell (IBC). However, characterization of localized features with conventional PV characterization tools is challenging, mainly due to the limitations of spatial resolution. This thesis develops and applies novel characterization methods to these localized features using low temperature micro-photoluminescence spectroscopy (μ-PLS). This technique demonstrates that localized features, even single laser pulse processed regions typically tens of micrometres in scale, can be investigated directly without the need for specific sample structures and their electronic properties can be mapped spatially in the sub-micrometre regime. Utilizing the sub-micron precision of these measurements, the laser-induced crystallographic damages were investigated at various positions within the laser-processed region, particularly at specific points such as the boundary/edge of processed and unprocessed regions. It was found that the edge, or pulse overlapped regions, were significantly more defective than the centre region. The impact of laser parameters, such as laser pulse fluence and number of repeat pulses, on laser-induced damage was also analysed. Significantly different levels of defect-related PL signals were observed after laser processing of the two different substrate surface conditions. This suggests that wafer surface preparation can be an important factor impacting on the quality of laser-processed silicon. The doping profiles of thermally boron-diffused silicon samples, which have Gaussian function type doping profiles, can be estimated from the measured PL spectra alone. The wavelength of the doping-related PL peak (doping peak) has a reliable and simple linear relationship with the surface dopant density on a semi-log plot. The PL intensity of the doping peak also shows a linear relationship with the doping depth metric (depth factor), but only after considering the reduction of PL intensity due to enhanced incomplete dopant ionization at low temperature. Doping profiles can be easily reconstructed based on these two linear relationships and their vi accuracy was verified by comparisons with existing doping profiles (via ECV profiling). Mapping of the surface dopant density and the depth factor of micron-scale locally diffused features was undertaken using 2-D mapping with μ-PLS measurements at 2 μm spatial resolution. This method was also applied to 532 nm laser-doped silicon to show its effectiveness on locally laser-doped features. The doping profiles of laser-doped silicon were also successfully estimated from PL spectra measurements alone, along with 2-D maps of the surface dopant density and the depth factor of the laser-doped silicon. In addition, the impact of temporal pulse parameters, such as pulse duration and temporal pulse shapes, on the doping profiles and recombination properties of laser-doped silicon were investigated. By correlating defect-related PL band counts with the quantified recombination parameters determined by the luminescence-coupled numerical device simulations, it was shown that μ-PLS measurements are able to perform quantitative measurements of recombination properties. The last chapter of this thesis demonstrates an application of an advanced laser doping process using a stack of intrinsic amorphous silicon (Si:H(i)) and boron-doped amorphous silicon (a-Si:B). The results showed that this stack is able to provide excellent surface passivation as well as a sufficient amount of dopant source for laser doping. The method presented in this thesis is a very effective, simple and rapid characterization for analysing localized features, in particular spatially inhomogeneous laser-processed features on the micron-scale. This method enables the observation of the variation in properties within localized features which is not possible using conventional methods. It allows for a more in-depth study of laser processing and promotes further development of laser technologies for high efficiency cell fabrication.