Design of zinc oxide enhanced surface plasmon resonance sensor optimized by Gaussian process regression

This paper reports a prism/Au/ZnO surface plasmon resonance (SPR) sensing structure utilizing the theory of SPR and the stable physicochemical properties of zinc oxide (ZnO). The structural model of the sensor is established using finite element analysis (FEA) and simulations are conducted to verify and compare performance in different refractive index environments. Furthermore, the sensitivity and quality factor are optimized by adjusting both the resonance wavelength of the incident light and the thickness of the metal film. Performance of the sensor is simulated using the transmission matrix method, and a dual-input single-output dataset is constructed based on the design structure. In particular, the reduced figure of merit (RFOM) is introduced as a key index for a comprehensive evaluation of sensor performance. The optimal combination of incident light wavelength (955 nm) and gold film thickness (46 nm) is predicted using a Gaussian process regression (GPR) model. Experimental findings demonstrate that the GPR model achieves good precision and reliability in predicting sensor performance, with an error of only 4.7707 RIU−1 when comparing the predicted RFOM values with those from experiments, thereby confirming that the model can enhance the SPR sensor’s performance during optimization. In summary, the prism/Au/ZnO SPR sensor designed in this article effectively enhances sensor performance by leveraging the prediction and optimization capabilities of the GPR model. This advancement provides strong support for the utilization of SPR-based sensing platforms for environmental surveillance, clinical diagnosis, industrial process regulation, and various additional domains.