Deep learning-based prediction of band diagrams and mode dispersion in photonic crystal waveguides

Abstract

Photonic crystals are structures with a forbidden band gap in which the refractive index varies periodically in one, two, or three dimensions. This forbidden band gap in their structure is called the ‘photonic band gap’, which allows light to pass through the crystal only at specific wavelengths (frequencies). This manipulation of light enables many photonic designs. Those designs include wavelength filters, high-precision sensors, lasers, and solar cells. The group speed of light traveling at specific wavelengths within those band-gap structures can be reduced at certain rates depending on the change of photonic crystal design parameters (hole or dielectric rod radius, dielectric constant, background material, index difference), and this physical phenomenon is called the ‘slow light effect’. As the velocity of light decreases within a given medium, there is a direct correlation with the matter-field interaction occurring between the sensor and the targeted analyte to be measured. This phenomenon results in a marked increase in sensor sensitivity, making it a highly effective means of detection (Y. Zhao, Y. N. Zhang and Q. Wang “High Sensitivity Gas Sensing Method Based on Slow Light in Photonic Crystal Waveguide” Sensors and Actuators B: Chemical, vol. 173, 28-31, Oct. 2012).