Multi-Geometry Parameters Optimization of Large-Area Roll-to-Roll Nanoimprint Module Using Grey Relational Analysis and Artificial Neural Network

Abstract

Micro- and nanofabrication on polymer substrate is integral to the development of flexible electronic devices, including touch screens, transparent conductive electrodes, organic photovoltaics, batteries, and wearable devices. The demand for flexible and wearable devices has spurred interest in large-area, high-throughput production methods. Roll-to-roll (R2R) nanoimprint lithography (NIL) is a promising technique for producing nano-scale patterns rapidly and continuously. However, bending in a large-scale R2R system can result in non-uniform force distribution during the imprinting process, which reduces pattern quality. This study investigates the effects of R2R imprinting module geometry parameters on force distribution via simulation, using grey relational analysis to identify optimal parameter levels and ANOVA to determine the percentage of each parameter contribution. The study also investigates the length and force ratio on a backup roller used for bending compensation. The simulation results and the artificial neural network (ANN) model enable the prediction of nip pressure and force distribution non-uniformity along the roller, allowing the selection of the optimal roller geometry and force ratio for minimal non-uniformity on a specific R2R system. An experiment was conducted to validate the simulation results and ANN model.