A Novel In-line Processing Method to Measure Thickness of a Thin Multi-layer Film

Abstract

A novel method to measure thickness of a thin multi-layer film (TMLF) is proposed by using the Daubechies wavelet (DBW) and the Maximum entropy method (Burg algorithm).
A signal is a reflective spectrum measured by a spectroscope when visible light is normally incident upon a TMLF. This spectrum is composed of interference and sometimes absorption. The DBW is very effective to separate them. It is impossible to achieve the same effect by an ordinary low pass filter.
After this pre-processing, thickness is calculated by the Burg algorithm. It is more accurate to estimate spectrum than the FFT algorithm. And also the tapered Burg algorithm that is to minimize error variance for the estimation of frequency is more effective for this objective.
Now a thickness gage has been developed that has a very high sensitive detector of a CCD area sensor that has plural rows of a CCD array with binning operation based on this novel method. It becomes possible to measure thickness of a TMLF in-line that means in production line in a non-destructive manner.