Locally weighted linear regression–optimized graphene–metal metasurface sensor for high-sensitivity organic compound detection in terahertz regime
Jacob Wekalao, Ahmed Mehaney, Bashir Salah, Mostafa R. Abukhadra, Hussein A. Elsayed
Abstract
Abstract This study presents the design and optimization of a graphene–gold–silver metasurface sensor for the highly sensitive detection of organic compounds in the terahertz (THz) regime. By leveraging the plasmonic properties of gold and silver combined with the tunable electronic characteristics of graphene, the sensor demonstrates enhanced molecular interaction and improved detection performance. COMSOL Multiphysics simulations show a high refractive index sensitivity up to the value of 810 GHzRIU −1 and excellent resonance characteristics with a quality factor (Q) of 11.197. The sensor achieves a detection limit (DL) as low as 0.075 RIU and a dynamic range (DR) from 2.765 to 2.555, indicating exceptional sensitivity and broad applicability. Additionally, the integration of machine learning optimization, particularly Locally Weighted Linear Regression (LWLR), improves prediction accuracy, achieving a value of R 2 near to 1. The results highlight the sensor’s real-time, label-free detection capabilities, with a signal-to-noise ratio (SNR) up to the value of 0.377. These findings demonstrate the sensor’s potential for applications in environmental monitoring, biomedical diagnostics, and industrial quality control, offering a significant advancement in THz-based sensing technologies.