Mr. Anurag Ghosh - Ph.D. Candidate
10/09/2026
אודיטוריום ע"ש דויד וואנג, בניין מידן, קומה 3
Recent advances in artificial intelligence (AI) have led to a growing demand for high-speed, low power computational systems. The traditional von Neumann architecture is limited by the physical separation between memory and processor units, leading to data transfer delays and high-power consumption. 2D material-based neuromorphic devices, addresses this challenge by integrating sensing, memory, and processing within a single device while continuing device miniaturization trend. This seminar presents different design strategies, guided by underlying device mechanisms, for realizing optically controlled neuromorphic devices combining ferroelectric α-In₂Se₃ and low-bandgap semiconductor PdSe₂.
Importantly, most of the reported 2D synaptic devices rely on hybrid operation; optical writing combined with electrical erasing, introducing hardware complexity and electrical crosstalk. In contrast, the presented PdSe₂/α-In₂Se₃ heterostructure (HS) FET exhibits bidirectional synaptic behavior without electrical gating. In particular, the device demonstrates excitatory responses under 642–980 nm and inhibitory responses under 406–520 nm illumination with reconfigurable SET/RESET process using only red and blue light illumination.
In addition, it is shown that the HS-FET exhibits limited infrared synaptic response due to the high dark current of the conductive PdSe₂ channel. To overcome this limitation, we introduce a vertical heterojunction FET architecture that extends the synaptic response to the 1310 nm optical communication band by suppressing dark noise. The device further achieves ultrafast (2.5 KHz), and ultralow-power (~10 pJ) synaptic operation under visible light, originating from a light-triggered quantum tunneling effect. Together, these design strategies establish a versatile platform for energy-efficient and broadband optoelectronic synapses with potential applications in optical communication, motion detection, and next-generation neuromorphic vision systems.