Advanced terahertz gas sensor based on multi-material metasurface integration with machine learning optimization

This paper presents a gas sensor based on an integrated metasurface design incorporating borophene, germanene, phosphorene, and graphene. The sensor leverages the unique properties of these two-dimensional materials arranged in a precisely engineered geometric configuration of circular and square ring resonators. Comprehensive simulation and optimization studies demonstrate that the sensor achieves a maximum sensitivity of 700 GHz/RIU, a figure of merit of 9.459 RIU−1, and a detection accuracy of 13.514. Meanwhile, the extensive optimization procedure regarding the sensor’s performance was analyzed across varying parameters, including graphene chemical potential (0.1–0.9 eV), angle of incidence (0–80°), and resonator dimensions. In addition, a machine learning approach using locally weighted linear regression was implemented to optimize the sensor response, achieving prediction accuracies up to 100% R2. The proposed design offers a promising platform for real-time, highly sensitive toxic gas detection with potential applications in industrial safety and environmental monitoring.