Neural networks for the simulation of photoresist exposure process in integrated circuit fabrication

A neural network for the simulation of photolithographic resist exposure process is proposed for the first time. The model is constructed by means of a feed-forward neural network, using the back-propagation training method. The training can be done with only a few experimental measurements with satisfactory output results. This neural network is compared with the ABC parameter model, which is currently in use; it is found to produce exactly the same results, but it is much faster because of the parallel structure of neural networks. Finally, this neural network for the resist exposure simulation is integrated with a cellular automaton model for the resist etching simulation. This integrated neural network/cellular automaton model is successfully used for the simulated production of some standard photolithographic profiles.