High-performance microcavity sensing combining MgCo2O4 modification with machine learning.

The whispering gallery mode (WGM) microbottle resonators, with their natural liquid delivery channel structure and high-quality factor (Q), provide a crucial technical platform for precise microfluidic measurement. To further enhance the detection sensitivity, this research constructs a highly sensitive ethanol and acetone sensor by modifying the inner wall of a hollow microbottle resonator (HMR) with MgCo2O4 nanomaterials. The results show that the sensitivity of the modified sensor is 0.0792 nm/% to ethanol solution, which is 2.1 times that of acetone. This indicates that the sensor has higher selectivity for ethanol. In addition, to further utilize the full-spectrum information, a multilayer perceptron (MLP) deep learning algorithm is introduced for intelligent analysis. The results indicate that the test accuracy for ethanol and acetone is 99.85% and 99.79%, respectively. With its high sensitivity and testing accuracy, this research provides support for the development of intelligent optical microcavity sensing.