The Device Compact Model Based on Multi-gradient Neural Network and Its Application on MoS₂ Field Effect Transistors

Transistor compact model (TCM) is the key bridge between process technology and chip design. A TCM based on multi-gradient neural network (MNN) is developed to capture the nonlinear device electronic characteristics and their high order derivatives in high precision. The typical MNN model creation time of a single transistor is < 3 hours with accuracy > 95% (error < 5%). Further, this MNN model methodology is applied to the MoS 2 FET, and the created model is characterized and implemented for the simulations of logic circuits such as ring oscillator (RO) and various standard cells. Simulation results agree well with experimental data.