Optimization of a Hybrid Graphene–Copper Terahertz Gas Sensor Using Machine Learning

This work presents a hybrid copper–graphene terahertz (THz) sensor for multigas detection with tunable performance. The design achieves frequency reconfiguration from 5.235 THz for$\mu _{c} = 0$eV to 5.265 THz for$\mu _{c} = 0.5$eV, confirming the strong plasmonic control of graphene. Sensitivity analysis shows values of 405.4 GHz/RIU for CH4, 816.3 GHz/RIU for CO2, 847.5 GHz/RIU for H2O, and 606.1 GHz/RIU for NH3. To further enhance prediction accuracy, an eXtreme Gradient Boosting (XGBoost) regression model was employed, achieving$R^{2} = 0.998$. After optimization, the sensitivities were improved to 603.0, 960.7, 1003.2, and 604.5 GHz/RIU, respectively. The proposed approach highlights the dominant role of graphene chemical potential in resonance tuning and sensitivity enhancement, establishing a compact and selective platform for advanced THz gas sensing.