Influence of Surface Roughness on Material Classification for Reflective THz-TDS Measurements

THz spectroscopy is a useful tool for identification of materials. However, for its reliable use, the effects of surface roughness must be taken into account. Scattering, diffraction and interference can hinder reliable detection. In our experiments, we measure samples with defined roughness parameters at varying angles of incidence and reflection. To determine the effect on material classification, we train a machine learning algorithm with samples of low roughness and test it with samples of higher surface roughness.