Performance Analysis of DM JL Forksheet-NCFET Biosensor With ML-Driven LoD Analysis

This study evaluates the performance of a Junctionless (JL) vertically stacked two-stage forksheet (FS) device designed for biosensing applications. The device characteristics are thoroughly analyzed using 3D numerical simulations in Sentaurus TCAD. The proposed JL sensor demonstrates superior electrical performance and enhanced sensitivity at reduced metal gate work functions, outperforming comparable JL devices operating at higher gate work functions. Additionally, its simplified fabrication process, lower work function requirement, and reduced response time make it an efficient solution for biosensing applications. By combining the advantages of JL technology, the negative capacitance (NC) effect in ferroelectric (FE) materials, and the fin-type FS structure, we design a high-performance Dielectric Modulated (DM) JL FS-NCFET-based biosensor. This sensor is optimized for label-free detection of biomolecules with distinct dielectric constants, such as streptavidin (K’ = 2.1), biotin (K’ = 2.63), APTES (K’ = 3.57), and keratin (K’ = 8). A “T-shaped” metal gate with side cavities enables biomolecule detection, with differentiation achieved by analyzing critical electrical parameters like threshold voltage, OFF current, and their sensitivities. Furthermore, the effects of varying FE thickness, steric hindrance, and process-induced variations, such as random discrete dopants (RDD), on device reliability are systematically examined. Using a machine learning (ML) approach, we calculate the limit of detection (LoD) for the proposed sensor at different FE thicknesses. Our findings reveal that the LoD increases with increasing FE thickness for both N-FET and P-FET devices.