Predicting the $Q$ factor and modal volume of photonic crystal nanocavities via deep learning

A Deep Learning (DL) based forward modeling approach has been proposed to accurately characterize the relationship between design parameters and the optical properties of Photonic Crystal (PC) nanocavities. The proposed data-driven method using Deep Neural Networks (DNN) is set to replace conventional approaches manually performed in simulation software. The demonstrated DNN model makes predictions not only for the Q factor but also for the modal volume V for the first time, granting us precise control over both properties in the design process. Specifically, a three-channel convolutional neural network (CNN), which consists of two convolutional layers followed by two fully-connected layers, is trained on a large-scale dataset of 12,500 nanocavities. The experimental results show that the DNN has achieved a state-of-the-art performance in terms of prediction accuracy (up to 99.9999% for Q and 99.9890% for V ) and convergence speed (i.e., orders-of-magnitude speedup). The proposed approach overcomes shortcomings of existing methods and paves the way for DL-based on-demand and data-driven optimization of PC nanocavities applicable to the rapid prototyping of nanoscale lasers and integrated photonic devices of high Q and small V.