THz computed tomography for non-destructive testing

Abstract

This work describes a complete THz computed tomography imaging system. Starting from a physical imaging model based on geometrical optics, an image reconstruction algorithm is derived. This algorithm uses the raw time-domain measurement signals acquired from THz time-domain spectroscopy measurements of the sample in order to calculate the sample cross-section. In particular, the phase of the transmitted THz pulse is evaluated for image reconstruction. The results show a significant improvement compared to the reconstruction based solely on the absorption. If preliminary knowledge about the sample geometry is available, refraction effects occurring on the sample interfaces can be considered in order to obtain a more accurate measurement result. This is especially beneficial in industrial non-destructive testing, where a sample is typically tested for deviations from a predefined specification or geometrical shape. Typical accuracies around 150 µm are demonstrated using 3d printed benchmark samples.