Automatic mark detection in electron beam nanolithography using digital image processing and correlation

Abstract

A method of mark detection for a vector scan eCF1-beam nanolithography system has been developed, in particular, taking into account that fairly low beam currents are used in high resolution writing. The digitally stored image of the scanned mark region is compared to a computer generated template using an advanced correlation technique. This method does not require a threshold determination. It exhibits high immunity to noise and poor contrast in the mark signal. The achieved accuracy is better than one beam step size (LSB) even for partially damaged marks. Due to specially chosen mark shapes, the probability of detecting wrong features is low. This paper describes the method and its hardware implementation along with experimental results for accuracy and repeatability of the mark position evaluation.