HEC/PVDF-Coated Micro-Sphere Humidity Sensor Using CNN-Based Barcoding

Whispering gallery mode (WGM) sensors offer a promising alternative to conventional relative humidity (RH) sensors, due to their high quality (Q) factors and sensitivity to environmental changes. In this work, we demonstrate a WGM microsphere-based RH sensor featuring a 170 μm diameter resonator coupled with a tapered optical fiber. The resonator was coated with a hydroxyethyl cellulose/polyvinylidene fluoride (HEC/PVDF) composite to enhance humidity sensitivity, resulting in a wavelength shift sensitivity of 0.4 pm/%RH and an attenuation response of 0.125 dB/%RH. The sensor was further utilized to generate a dataset of barcoded spectral responses, which were used to train and test a convolutional neural network (CNN) classifier, achieving an accuracy of 98%.