Affiliation:
1. FG Nanotechnologie Institut für Mikro- und Nanelektronik and Institut für Mikro- und Nanotechnologien MacroNano and Institut für Werkstofftechnik TU Ilmenau 100565 Postfach Germany
Abstract
Sufficient energy consumption for conventional information processing makes it necessary to look for new computational methods. One of the possible solutions to this problem is neuromorphic computations using memristive devices. Memristors based on molybdenum disulfide () are a promising way to provide a sizeable amount of hysteresis at low energy costs. Herein, different configurations of memristors as well as the mechanisms involved in hysteresis formation are shown. Bottom gated configuration is beneficial in terms of hysteresis area and energy efficiency. The impact of device channel dimensions on the hysteresis area and energy consumption is discussed. Different operation conditions with triangular, rectangular, sinusoidal, and sawtooth drain‐to‐source pulses are simulated, and rectangular pulses demonstrate the highest energy efficiency. The study shows the potential to realize low‐power neuromorphic systems using memristive devices.
Subject
Materials Chemistry,Electrical and Electronic Engineering,Surfaces, Coatings and Films,Surfaces and Interfaces,Condensed Matter Physics,Electronic, Optical and Magnetic Materials