An efficient temperature dependent compact model for nanosheet FET for neuromorphic computing circuit

In this work, a temperature-dependent compact model is proposed for the three-sheet (3S) Nanosheet (NS) FET. This model is developed because a computationally efficient model is needed for large-scale circuit design. The model is based on the virtual source (VS) principle, which is chosen because for its simple mathematical formulation and minimal parameter requirements. This allows the model to accurately capture the performance characteristics of the 3S NSFET. The model is validated using TCAD results, which are well-calibrated with experimental data. It is then implemented in Verilog-A code for neuromorphic circuit simulations. Herein, we analyses the important parameters such as power, energy, and spiking frequency in NSFET-based leaky integrate-and-fire (LIF) neurons, with temperature variations. The results show that as the temperature increased from 25 °C to 125 °C, the spiking frequency increased by 36.64 %, due to higher current in the subthreshold operation of the device.