Tunable graphene based terahertz sensor for biochemical sensing with ML-driven refractive index prediction

This paper presents the design and analysis of a tunable terahertz sensor at 2.65 THz for biochemical sensing. The Bandpass characteristics in the frequency response of the sensor are achieved by coupling a rectangular split ring resonator to the transmission line. Graphene layer is deployed in the design to facilitate the propagation of surface plasmons and to incorporate the tunability in the frequency response. The simulation results reveal that the designed sensor is capable of achieving the desired frequency response. By varying the applied voltage across the graphene the designed sensor can be tuned over a range of 80 GHz. The proposed design can be employed as a highly sensitive biosensor for detection of samples like ethanol, water, sodium chloride, glycerine and others. The designed sensor offers a maximal sensitivity of 200 GHz per refractive index unit change. Further, machine learning based regression models are considered for predicting the refractive index of the sample from the sensor response. The prediction is based on the frequency response of a sensor, which serves as a reliable indicator of changes in the analyte's refractive index. Results indicate that the Gaussian process regression model outperforms the other models with 0.0099 root mean square error.