Printability Prediction in Projection Two-Photon Lithography via Machine Learning Based Surrogate Modeling of Photopolymerization

Abstract

Abstract The two-photon lithography (TPL) direct laser writing process enables the fabrication of complex three-dimensional features with sub-micrometer resolution. While traditional TPL techniques have suffered from low throughput due to their serial writing scheme, the process has been recently parallelized through the development of the projection two photon lithography (P-TPL) process that allows printing of entire layers at once. In this work, we use machine learning-based surrogate modeling to predict the outcomes of P-TPL to > 97% of the accuracy of a physics-based reaction-diffusion finite element simulation. Training a classification neural network on data points generated from physics-based simulations allows for computationally efficient and accurate prediction of whether a set of printing conditions will result in precise and controllable polymerization and the desired printing, as opposed to no printing or runaway polymerization. We interrogate this surrogate model to investigate the parameter regimes that are promising for successful printing. We predict combinations of rate constants necessary to print for a given set of material properties, thereby generating the first printability map for the P-TPL process and allowing informed selection of photoresists and printing parameters in the future.