Multi-resonator plasmonic metasurface biosensor with graphene enhancement for ultra-sensitive terahertz pregnancy detection using machine learning optimization

Abstract

This study presents a multi-resonator plasmonic metasurface biosensor operating in the terahertz range for detecting human chorionic gonadotropin (hCG), a primary pregnancy biomarker. The sensor consists of four resonators with different geometries and dimensions made from graphene, copper, aluminum, and gold. Its operation is based on surface plasmon resonance. Finite element simulations showed that transmittance varied from 98.428% to 30.736% as the graphene chemical potential changed from 0.1 to 0.45 eV. The optimized sensor achieved a sensitivity of 1000 GHz per refractive index unit (RIU) and a figure of merit of 13.333 RIU−1 . A Gradient Boosting Regressor model was used to predict sensor behavior. The model produced R 2 values between 0.90 and 1.00 for variations in incident angle, square ring geometry, and graphene chemical potential. Resonance frequency shifted from 0.32 to 0.30 THz with refractive index changes, following a linear relationship (R2 = 0.88947) that allows calibration for hCG detection.