A deep learning-based predictive simulator for the optimization of ultrashort pulse laser drilling

• DOI: 10.1038/s44172-022-00048-x
• Publication Date: 2023-01-07T09:06:50Z
• Authors: Kohei Shimahara; Shuntaro Tani; Haruyuki Sakurai; Yohei Kobayashi
• Keywords: Laser drilling; Microchannel

Abstract Ultrashort pulse laser drilling is a promising method for the fabrication of microchannels in dielectric materials. Due to complexity process, there strong demand numerical models (simulators) that can predict structures produced under specific processing conditions order rapidly find optimal parameters. However, validity conventional simulators dielectrics has been confined range strict interpolations data used during construction model, and thus, their usefulness limited. Here, we demonstrate simulator-based optimization ultrashort based on an iterative deep neural network which trained microchannel structure after small number irradiated pulses. Our approach predicts development hole shapes over wide variety allowed discovery 20% more energy efficient strategies than initial experimental data. More broadly, our address realistic problems considering parameters, thus enabling improved performance next-generation smart systems.