Excimer laser model based on a temporal convolutional neural network

Excimer lasers play a key role in deep ultraviolet lithography. To further study the voltage and energy features of the excimer laser and control it to work in a constant energy mode, a temporal convolutional neural network was designed to fabricate an excimer laser voltage-energy model. The proposed model uses the currently measured energy data to predict the subsequent output energy data. For the voltage-energy data that cannot be obtained, we simulated the initial energy data corresponding to the part of the voltage value based on the relationship between energy and voltage as the initial input of the model. The energy data of any voltage for the excimer laser at each moment were obtained. Finally, a continuous excimer laser voltage-energy model was established. The difference between the means of the measured and generated energy data is less than 0.5 mJ.