Research On Parallel Fast Ellipsometric Analysis Method

Ellipsometry is a kind of non-contact and non-destructive method with high accuracy and sensitivity, and has been widely used in the fields of industry and scientific researches.This work is mainly aiming to enhance ellipsometric data acquisition speed and to improve the accuracy of the ellipsometry data through the measurement process. The content of this work mainly includes:1. A new type of ellipsometer with parallel measurement of polarization states of light was constructed and studied. By using an integrated analyzer composed of12sub-analyzers with different azimuth angles and a two-dimensional CCD array camera, the light intensity emerging from each sub-analyzer with the azimuth angles uniformly distributed in the range of about180°is measured simultaneously. Without any moving parts during measurement, the acquisition time for ellipsometric parameters is significantly reduced. The sample of bulk silicon wafer was measured with the newly developed ellipsometer, and the results are in good agreement with those taken by two conventional photometric ellipsometers for the same sample. Therefore, the feasibility of the new type of ellipsometer has been successfully demonstrated with the expectation and its potential application in the real time ellipsometric measurement process in the future.2. Both numerical simulation and experimental measurement were applied to study the systematic errors induced in the ellipsometric parameter measurement process to determine the optical constants by using the RPAE method. The systematical errors will be induced due to the digitalized Fourier transform process and random errors due to some uncertain factors in measurement. The subsequent data processing with two sets of equations may further make errors be amplified, and the mechanism and rules for the errors to affect the data quality in doing the data processing have been studied. The advantage and disadvantage of data processing methods using two different sets of equations are evaluated, and an approach to improve the accuracy of ellipsometry data in measurement is proposed.