A ROBUST AUTOMATIC FOCUSING AND ASTIGMATISM CORRECTION METHOD FOR THE SCANNING ELECTRON MICROSCOPE

Abstract

In a previous project, a new approach to focusing and astigmatism correction based on the Fast Fourier Transforms (FFTs) of SEM images has been explored. From the FFTs, it is possible to obtain information on the severity of the defocus and astigmatism. This information is then processed by an algorithm to perform real-time focusing and astigmatism correction on the SEM. The algorithm has been tested on defocused and astigmatic images of different samples including those with highly directional features. Experiments show that the images obtained after running the algorithm are as good as what an experienced SEM operator can achieve. However, there are some problems with this technique. The main problem is the long correction time. Furthermore, the algorithm is only able to correct low to moderate defocus and astigmatism, besides being sensitive to noise. This thesis discusses the various developments to reduce the correction time and extend the technique to correct images with any degree of defocus, astigmatism and noise. Direct control of the SEM has been implemented to reduce the communication time with the SEM. An adaptive FFT threshold algorithm has been developed to determine the optimal value for thresholding the FFTs for optimal extraction of defocus and astigmatism information under noisy conditions. Two other algorithms, namely the coarse and fine focusing algorithms have also been developed to bring the SEM to the best focus point. The focusing and astigmatism correction algorithm has also been enhanced to make it more efficient and more robust to noise. All the techniques are then integrated to form a new enhanced focusing and astigmatism correction algorithm. Images with different degrees of defocus and astigmatism are used to test the new algorithm and the results show that the algorithm is able to perform the correction in a shorter time and at the same time, maintains the accuracy of the original algorithm even under noisy conditions.