Machine Learning-Driven Optimization of Photonic Crystal Structures for Superior Optical NOR Gate Performance

In this study, we employ a two-dimensional photonic crystal structure to design a NOR logic gate, utilizing dielectric rods in air. The compact size and simplicity of the design make this optical gate particularly suitable for integration into photonic integrated circuits. To optimize the optical NOR gate design and achieve superior results, we leverage machine learning techniques, specifically XGBoost and RandomForestRegressor. By fine-tuning the radii of defect rods within the photonic crystal lattice, we maximize output power and ensure optimal gate functionality across various input scenarios. Through extensive simulations and comparative analyses, we showcase the effectiveness of our approach in accurately predicting optimal rod radii and enhancing NOR gate performance. Notably, our design utilizes only two defect rods, highlighting the structure’s efficiency. Moreover, alongside the simplicity of our proposed design, it boasts high output power for logical 1 and low power for logical 0. This feature contributes to minimized errors in logical output detection, further underscoring the practicality and efficacy of our approach.