BSIM4 modeling of drain voltage, gate length, and temperature-dependent I d – V g characteristics of bulk MOSFETs at cryogenic temperature using a conditional diffusion neural network

Abstract In this study, a neural-network-based approach was developed to model the drain-voltage, gate-length, and temperature-dependent drain-current–gate-voltage ( I d – V g ) characteristics of metal-oxide-semiconductor field-effect transistors (MOSFETs) at cryogenic temperatures. Using experimental data obtained at cryogenic temperatures, fifteen BSIM4 parameters were automatically tuned through this approach. As a result, a single set of BSIM4 parameters was extracted that reproduced the I d – V g characteristics of short-channel bulk MOSFETs with gate lengths of 90–250 nm over the temperature range of 4–100 K. These results contribute to continuity and convergence in Cryo-CMOS circuit simulations under self-heating effects, facilitating precise qubit control and the realization of large-scale fault-tolerant quantum computers.