Ultra-High Sensitivity Terahertz Detection Using a 2D-Material-Based Metasurface: Design, Tuning, and Machine Learning Validation

The quantification of dopamine, a critical catecholamine neurotransmitter, remains a significant challenge in neurological research and clinical diagnostics due to its low physiological concentrations and interference from structurally similar compounds. This study presents a simple metasurface sensor employing graphene-enhanced surface plasmon resonance for ultra-sensitive dopamine detection. Finite element method simulations using COMSOL Multiphysics 6.2 demonstrate exceptional performance with a maximum sensitivity of 500 GHzRIU⁻¹ at 0.805 THz, achieving a figure of merit of 2.110 and quality factor ranging from 3.376 to 3.435. The sensor exhibits tunable response through graphene chemical potential modulation (0.1–0.9 eV), with transmittance varying from 81.6% to 16.4%. Angular stability analysis reveals consistent performance across incidence angles from 0° to 80°. Machine learning integration using XGBoost regression achieves 92–100% prediction accuracy, enabling real-time performance optimization. The proposed sensor surpasses existing designs in sensitivity while maintaining broad refractive index detection range,positioning it as a promising platform for advanced neurochemical sensing applications in Parkinson's disease, schizophrenia, and substance abuse disorder diagnostics.