Environmentally Stable and Reconfigurable Ultralow-Power Two-Dimensional Tellurene Synaptic Transistor for Neuromorphic Edge Computing
Jeechan Yoon, Bolim You, Yuna Kim, Jina Bak, Mino Yang, Jihyang Park, Myung Gwan Hahm, Moonsang Lee
Abstract
While neuromorphic computing can define a new era for next-generation computing architecture, the introduction of an efficient synaptic transistor for neuromorphic edge computing still remains a challenge. Here, we envision an atomically thin 2D Te synaptic device capable of achieving a desirable neuromorphic edge computing design. The hydrothermally grown 2D Te nanosheet synaptic transistor apparently mimicked the biological synaptic nature, exhibiting 100 effective multilevel states, a low power consumption of ∼110 fJ, excellent linearity, and short-/long-term plasticity. Furthermore, the 2D Te synaptic device achieved reconfigurable MNIST recognition accuracy characteristics of 88.2%, even after harmful detergent environment infection. We believe that this work serves as a guide for developing futuristic neuromorphic edge computing.