Defect-dominated resistive switching of 2D MoTe2 nanosheets for artificial synapses and decimal arithmetic

Abstract

In this work, the defect-dominated resistive switching (RS) behaviors and synaptic functions are studied in two-dimensional MoTe2 nanosheets. Different from previously reported phase-transition RS performance, the mechanically exfoliated MoTe2 nanosheets demonstrate analog RS behaviors. To further understand the RS mechanism, MoTe2 nanosheets with more Te vacancies are prepared by chemical vapor deposition method and exhibit digital RS behaviors, confirming the RS mechanism is dominated by Te vacancies. Furthermore, memristors with analog RS behavior are especially suitable for artificial synapses with diverse synaptic functions. The achieved 150 sets of long-term potentiation and long-term depression are employed to implement handwritten digit image recognition with a recognition accuracy as high as 93.8%. Moreover, decimal arithmetic of addition, subtraction, multiplication, and division can be well implemented by utilizing the good synaptic functions. All these studies suggest the neuromorphic computing potential of the defect-dominated memristors.