Numerical Design and Machine Learning-Assisted Analysis of a Graphene–Au–Ag Hybrid Terahertz Metasurface for Isoquercitrin Detection
摘要
The precise quantification of isoquercitrin is essential for pharmaceutical quality control and herbal product standardization, motivating the development of rapid, label-free, and non-destructive intelligent sensor platforms beyond conventional analytical techniques. In this work, a novel Graphene–Au–Ag hybrid terahertz (THz) metasurface is proposed for ultra-sensitive isoquercitrin detection by exploiting the complementary plasmonic properties of silver and gold together with the electrically tunable conductivity of graphene. Unlike conventional graphene-based THz sensor, the proposed architecture systematically investigates three distinct material distribution configurations under varying graphene chemical potentials, incidence angles, and resonator geometries to optimize electromagnetic field confinement and sensing performance. Frequency-domain finite-element simulations performed in COMSOL Multiphysics demonstrate that the optimized sensor achieves a maximum sensitivity of 400 GHz/RIU, a peak figure of merit of 7.692, and a quality factor of 8.021, while exhibiting a strong linear relationship between resonance frequency and refractive index (R² = 0.947). Electric-field analysis confirms pronounced plasmonic localization at 0.375 THz, and an integrated Extreme Gradient Boosting (XGBoost) model attains prediction accuracies of up to 96% across varying operating conditions. These results demonstrate that the proposed hybrid metasurface provides a robust, scalable, and machine learning-assisted platform for high-performance THz biosensing with significant potential for pharmaceutical quality control, herbal product standardization and smart sensors technologies.