Neural network proximity effect corrections for electron beam lithography

The use of a neural network to compute corrections for images written by electron beams to eliminate the proximity effects caused by electron scattering is described. Optical local changes in the incident beam dose, found iteratively, require a prohibitively long time to compute for realistic pattern sizes. A neural network was trained to perform equivalent corrections, resulting in a significant decrease in computation time. Hardware implementations of the networks for this application, using both analog and digital electronic networks, were examined. Although the analog hardware was potentially faster, the digital network has the advantage of being much more flexible. Both networks had an acceptably small error of 0.5% compared to the results of the iterative computation. The neural network correctly generalized the solution of the problem to include patterns not contained in its training set. The experimental verification of the approach is described