The development of emerging technologies, particularly artificial intelligence, has posed significant challenges to traditional computers based on the conventional “von Neumann” architecture, which suffer from high energy consumption and low computing efficiency, making them inadequate for increasingly demanding applications. Neuromorphic computing systems, inspired by the human brain, have emerged as a promising alternative for large-scale, high-density parallel computing owing to their integrated “operation-storage” capability. Brain-like artificial synapses, based on the principles of memristors, are designed to simulate the biological synaptic functions of neurons and provide essential support for neuromorphic computing. In recent years, 2D materials have emerged as promising candidates for brain-like artificial synapses. Among them, transition metal dichalcogenides (TMDs) have been widely used for simulating biological synaptic plasticity owing to their layer-dependent bandgap characteristics, high carrier mobility, and gate voltage-controlled optoelectronic properties.

In this paper, we present innovative experiments on brain-like artificial synapse devices based on three-terminal transistors using TMDs. Single-layer or few-layer WSe₂ microflakes of varying thicknesses were obtained through mechanical exfoliation, while oxygen plasma treatment was applied to the SiO₂/Si substrate surface to enable interfacial state tuning. The dry transfer technique was then applied to fabricate brain-like artificial synapse devices based on WSe₂ materials, followed by an annealing process at 300 ℃ under an argon gas atmosphere. Optical and scanning electron microscopy were employed to characterize the morphology and thickness of the WSe₂ material. To analyze the electrical properties of the devices, the transfer and I–V output curves were measured. The post-synaptic current (PSC) as a function of pulse number was evaluated under two conditions: 50 continuous cycles of a 50 V pulse and 50 cycles of a −10 V/−50 V pulse. Both measurements were conducted under the same test conditions, with a source-drain voltage of 3 V, a pulse duration of 30 ms, and a pulse period of 60 ms to assess the stability of the artificial synapse devices. Furthermore, the biological synaptic plasticity of the artificial synapse devices can be tuned by varying the duration of oxygen plasma treatment (10 s, 60 s, 120 s, and 180 s) on the surface of the SiO₂/Si substrate, thereby modifying the interfacial state density.

The results demonstrated that by controlling the gate voltage polarity, a transition from high-resistance states to low-resistance states can be achieved. The PSC at 0 V gate voltage after the application of a 50 V gate voltage pulse was greater than that observed after the application of a −10 V gate voltage pulse. Under a 50 V gate voltage, the device entered a low-resistance state, whereas under a −10 V gate voltage, the device remained in a high-resistance state, successfully enabling switching between high- and low-resistance states. During the long-term potentiation phase, the PSC of devices subjected to different plasma treatment durations continuously increases with the number of pulses. Moreover, as the number of pulses increases, the PSC exhibits a gradual saturation trend, which becomes more pronounced with longer treatment times. In contrast, during the long-term depression phase, the PSC decreases as the number of pulses increases, highlighting the effect of plasma treatment duration on biological synaptic function.

This paper presents an innovative experiment on brain-inspired artificial synapses based on transition metal dichalcogenide tungsten diselenide (WSe₂) transistors. The study encompasses device fabrication, characterization, and property measurement, integrating research with practical teaching. This approach not only fosters students’ interest but also helps cultivate their creativity and problem-solving abilities.