High-Performance Terahertz Metasurface Biosensor Based on Graphene-Enhanced Gold-Silver Composite Architecture: Machine Learning-Assisted Design Optimization and Refractive Index Sensing

This work introduces a biosensor featuring a symmetrical design that combines graphene-enhanced plasmonic elements for high-sensitivity refractive index detection. COMSOL Multiphysics simulations demonstrate a maximum sensitivity of 565 GHz RIU −1 and a figure of merit of 16.607 RIU −1 , with absorption ranging from 75.33% to 79.11% across refractive indices from 1.3341 to 1.4238. The resonance frequencies are tunable from 0.345 to 0.327 THz, enabling versatile sensing applications. Parametric studies reveal that graphene chemical potential tuning (0.1–0.9 eV) is critical for enhancing absorption up to 75.565%, while the sensor maintains stable performance at incidence angles up to 80°. A strong linear correlation (R 2 = 0.86) between resonance frequency and refractive index supports precise sensing capabilities. Machine learning polynomial regression models achieve 89% accuracy in predicting sensor responses and successfully identify optimization trends for enhanced performance. Compared to conventional designs, this sensor offers superior sensitivity while maintaining practical fabrication feasibility through standard deposition methods, making it highly suitable for real-world biosensing applications.