Utilizing Forward Characteristics of Pocket Doped SiGe Tunnel FET for Designing LIF Neuron Model

In this paper, a single SiGe Tunnel FET is used to design a Leaky Integrate and Fire (LIF) neuron with significant improvement in area, energy and cost. SiGe Tunnel Field-Effect Transistor (FET) transfer characteristic with steep sub-threshold swing has been used to observe LIF neuronal characteristics. By employing calibrated simulation using Atlas 2D, we have verified that the TFET with LIF characteristics can effectively replicate neuron behavior without relying on external circuitry. The proposed LIF neuron, based on SiGe TFET, exhibits significantly reduced energy consumption, specifically 210 fJ per spike. This energy consumption is \(\approx \) $\approx$ 215 times lower compared to previously reported single-device neurons in existing literature. Additionally, we have achieved an impressive recognition precision of 91.3 % for Modified National Institute of Standards and Technology (MNIST) images.