High-precision liquid crystal cell gap estimation via machine learning

Abstract

Liquid crystal (LC) cell gap measurement is crucial for determining key electro-optical properties, but traditional methods are often time-consuming and complex. In this study, we present the application of a multilayer perceptron (MLP) model for predicting empty LC cell gaps using spectral data. The proposed MLP model achieves high predictive accuracy, with correlation coefficients exceeding 0.99 and a mean squared error (MSE) of approximately 0.6 across various conditions. Different training algorithms, activation functions, and weight decay parameters were systematically tested to optimize performance. Experimental results showed that the MLP model achieved low MSE values while maintaining robustness under varying random seeds and train-test split ratios, confirming its strong generalization capability. This approach simplifies the cell gap measurement process and highlights the potential of machine learning for advancing LC device characterization, offering a reliable and efficient alternative to conventional methods.