Thickness Measurement of Quartz and Analysis of Interface Roughness Effects Based on Terahertz Time-Domain Spectroscopy

Abstract

Pulsed terahertz signal technology is increasingly being applied across various industries as an effective means of measuring material thickness. In terahertz time-domain spectroscopy systems, the presence of mixed noise affects the extraction of valid information and signal quality, making noise reduction essential for accurate thickness detection. Firstly, the system echoes are removed using double Gaussian inverse filtering, and then the system noise is removed using an adaptive variational mode decomposition method, which utilizes the minimum envelope entropy as the criterion, and through the sparrow search algorithm automatically optimizes the mode number and penalty factor to achieve adaptive and self-consistent noise removal. The terahertz signals of parallel-surface quartz specimens are reconstructed, and an automatic peak extraction algorithm is demonstrated. The influence of different surface roughness on reflective terahertz pulse signals and the correction parameters is quantitatively analyzed to achieve high-precision automated thickness detection.