TY - JOUR AU - Chen, Jiun-Shen AU - Wub, Tzong-Daw PY - 2017/04/22 Y2 - 2024/03/29 TI - Applying Artificial Neural Networks to Monitor Deposition Rate of Roll-to-roll Sputtering System in Real Time JF - Asian Journal of Applied Sciences JA - AJAS VL - 5 IS - 2 SE - Articles DO - 10.24203/ajas.v5i2.4442 UR - https://ajouronline.com/index.php/AJAS/article/view/4442 SP - AB - <p>A back propagation neural network (BPNN) is applied to determine the deposition rate of a roll-to-roll magnetron sputtering system in real time. Because the transmittance spectra of thin films are highly related to their thicknesses, the spectrum is a function of the thickness. Thus, determining deposited thickness through the functions is possible. However, these functions are not simple one-to-one functions; solving the inverse function to find thicknesses from spectra is difficult. Therefore, BPNNs are introduced to build approximate functions of spectra and output thicknesses. They are trained with various spectra which correspond to different film thicknesses, and will have abilities to estimate thicknesses of thin films. In this study, the estimation error of BPNNs was less than 0.6%. The results of low error and real-time response make BPNNs a promising method for monitoring a deposition process.</p> ER -