Cavity photothermal oscillation spectroscopy generation in an optofluidic microbubble resonator for multi-component mixture solution detection

Optofluidic whispering gallery mode microbubble cavities offer several advantages, including high quality factors, small mode volumes, and real-time detection capabilities, making them well-suited for applications in high-precision sensing. The sensing mechanism of microcavity-based solution sensors typically relies on detecting mode shifts, broadening, or splitting induced by variations in the refractive index of the analyte. In this study, we introduce a novel, to the best of our knowledge, detection method that integrates cavity photothermal oscillation spectroscopy with a deep neural network. This hybrid technique enables high-precision detection of anhydrous ethanol concentrations in multi-component mixture solution while remaining insensitive to mode shifts. Experimental results demonstrate a predicted accuracy of 0.992 and a mean squared error of 5.29 × 10^-4 within the 0-1% concentration range.