Multilayer WS 2 /Borophene/Au/MXene/Cu Metasurface Terahertz Biosensor With Machine Learning‐Augmented Prediction for Label‐Free Brain Tumor Detection

Vollständiger Abstract

Worum geht es in dieser Arbeit?

High‐performance terahertz metasurface biosensor is numerically developed for label‐free brain tumor detection. The sensor employs six multilayer elliptical resonators integrating WS 2 , borophene, Au, MXene, and Cu on a graphene‐coated SiO 2 substrate. Three material configurations are compared, with the hybrid WS 2 /borophene/MXene/Cu structure (Case 3) enabling plasmon–exciton (plexcitonic) coupling and delivering the best performance. Full‐wave FEM simulations in COMSOL incorporate dispersive material models, graphene conductivity based on the Kubo formalism, and Floquet–Bloch periodic boundaries. Parametric optimization covers graphene chemical potential, incidence angle, resonator size, and elliptical radii. The optimized Case 3 reaches 99.756% absorption at μ c = 0.3 eV, 0.995 THz/RIU sensitivity, 36.538 RIU −1 FOM, Q = 67.833, a 0.009 RIU detection limit, and R 2 = 99.361% linearity across refractive indices from 1.3333 to 1.4833 RIU. Near‐field analysis shows field enhancement approaching ×59. A 1D convolutional neural network surrogate accurately predicts THz absorption spectra with R 2 > 0.9999 and MAPE < 0.052%, substantially reducing computational cost. Overall, the proposed platform combines high sensitivity, strong electromagnetic confinement, and machine‐learning‐assisted prediction for efficient label‐free brain tumor diagnostics. Across eleven biological tissue conditions, the sensing range spans normal cerebrospinal fluid to high‐grade glioblastoma, supporting reliable discrimination of healthy and malignant brain tissue samples.