Eric Mottay from h-nu chaired the Laser Materials Microprocessing track at ICALEO 2025 in Orlando, 13 to 16 October, with Jared Speltz of the University of Dayton.
He also gave two presentations.
“Machine Learning for Predictive Control of Silicon Thinning in Integrated Circuits”, in collaboration with ALPhANOV, used a gradient boosting regressor to predict ablation depth and surface roughness, and a generative adversarial network to predict the surface texture itself.
“Data-Efficient Hybrid Physics – Machine Learning Model for Ultrafast Laser Ablation” combined the two-temperature model with a neural network. Trained on silicon and transferred to stainless steel using ten datapoints selected by Bayesian optimization, it improved R² from -0.40 to 0.67.