Multi‐Functional Platform for In‐Memory Computing And Sensing Based on 2D Ferroelectric Semiconductor α‐In₂Se₃

Abstract

Abstract2D layered semiconductors with excellent light−matter interaction and atomic‐scale thickness have been envisioned as promising candidates for more than Moore and beyond Moore technologies. Here, for the first time, a multi‐functional platform is reported that is fabricated entirely from wrinkle‐free 2D ferroelectric semiconductor α‐In2Se3 integrated with a photodetector, reconfigurable logic switching, and visual perception processing functions. The intensity‐ and wavelength‐dependent resistance is used to demodulate broadband optical information into electrical signals and perform reconfigurable logic switching. Moreover, the platform offers dynamically modulated photosensitive visual sensing in different working modes at the pixel level. Using photo‐assisted piezoresponse force microscopy, the optical‐engineered ferroelectric polarization switch behavior is explored. The platform has excellent sensitivity to both optical and electrical stimuli and can respond to lights in the visible to short‐wavelength infrared region in volatile/non‐volatile manner under gate voltage modulation, with a response of 98 mA W−1 (to 1800 nm light), a current on/off ratio of over 106, and a high field‐effect mobility of 137.55 cm2 V−1 s−1. With its simple structure, unique photoelectric interaction, and controllable operating mechanism, the platform has the potential to simplify the complexity of neuromorphic computing circuitry systems, paving the way for high‐performance hybrid technologies suitable for artificial intelligence applications.

Boosting flexible electronics with integration of two‐dimensional materials

“…Changes in the magnetic field of the sensor, induced by the piezomagnetic effect, impact the magnetic resistance and alter the conductivity. Chen et al 237 used the ferromagnetism of In 2 Se 3 to realize a photodetector for visual perception processing. This involved using a magnetic field to manipulate the resistor, resulting in a responsivity of 98 mA/W and a carrier mobility of 137.6 cm 2 V À1 s À1 .…”

Multisensory Ferroelectric Semiconductor Synapse for Neuromorphic Computing

“…The electrical signals are modulations of the nonlinear transformation of optical signals, such as the relaxation time and the intensity of the optical response. [22,25,26] That is, they achieve nonlinear transformation of light signals through the characteristics of the device, with the role of electric pulses being to regulate the intensity of this transformation, while the outputs still represent the sensory signals themselves. [22,25,26] To facilitate signal computing, linear modulation of network weights is imperative, but current methods are limited to either linear modulation of conductance or photoresponsivity.…”

“…[22,25,26] That is, they achieve nonlinear transformation of light signals through the characteristics of the device, with the role of electric pulses being to regulate the intensity of this transformation, while the outputs still represent the sensory signals themselves. [22,25,26] To facilitate signal computing, linear modulation of network weights is imperative, but current methods are limited to either linear modulation of conductance or photoresponsivity. Furthermore, the linear modulation of photoresponsivity is challenged by issues of volatility.…”

Flash In2Se3 for neuromorphic computing

“…After 100 cycles of electrical stimuli, the transfer curve of 100th sweep is almost identical with that of 1st sweep, meaning that the devices are very stable and robust under repetitive electrical stimuli. In addition, a large clockwise hysteresis is observed when the VG is double swept between -40 and 40 V at a fixed VD of 0.1 (black), 1.0 (red), and 2.0 V (blue) in Fig.2e, indicating that the device can exhibit a memory operation under VG sweep 8,13,15,39. In addition, the hysteresis memory window can be tuned by using different gate sweep ranges (Fig.2h).…”