Monitoring of porous silicon layers for epitaxial wafer production using inline reflectance spectroscopy

Fabrication of solar silicon wafers directly from the molten or the vapor phase has potential for significant cost savings as these methods skip the time-consuming and material-wasting diamond/slurry sawing. Among the kerfless techniques, so-called sintered porous silicon process enables the growth of low-cost but high-quality monocrystalline silicon wafers. During this process, a recyclable substrate wafer is first porosified electrochemically and then sintered, both to form a smooth growth surface for vapor-based epitaxy and to create cavities underneath for a later separation of the epitaxial wafer from the substrate. In this contribution, we evaluate the potential of inline reflectance spectrophotometry for industrial quality assurance of the as-etched porosified substrate. We characterize multilayer stacks of porous silicon with two different inline spectrophotometers and an offline reference counterpart and compare the appearance of the spectra. Subsequently, we utilize an optical model-based fitting approach to extract the thicknesses and the porosities of the layers. To maximize the speed of the evaluation, we rely on local optimization and accurate initial guesses based on pre-evaluation of the spectra, reaching a fitting rate of ~1 spectrum/s. To verify the results, we perform a comparison to scanning electron microscope (SEM) images and find that the thicknesses of the porous layers can be fitted with an accuracy of 5-11% in mean relative deviation between the reflectance-based analysis and SEM. As an advantageous, built-in feature, the inline setups enable profiling of the layer parameters along one spatial dimension as the wafers move under the spectrometer on the conveyor belt.