Inorganic p-Type Tellurium-Based Synaptic Transistors: Complementary Synaptic Pairs with n-Type Devices for Energy-Efficient Operation

Abstract

Neuromorphic computing is a rapidly emerging technology that can overcome the limitations of von Neumann-type architecture-based computing systems, offering the potential for implementing next-generation computing architectures. Here, we propose a p-type three-terminal synaptic device that successfully mimics the function of biological synapses. The proposed tellurium (Te) synaptic transistors incorporating SiO2 or Al2O3 gate dielectric layers modulate the synaptic weight-that is, the channel conductance-essential for realizing synaptic characteristics. Synaptic devices with optimal Al2O3 layers exhibit large hysteresis properties that efficiently induce conductance modulation, demonstrating low power consumption, good linearity, and short-/long-term plasticity. Furthermore, the proposed optimal Te synaptic transistor achieved a high recognition accuracy of 93.8%. These findings suggest that Te-based synaptic devices fabricated utilizing thin-film processes could enhance the efficiency of future neuromorphic computing systems.