Optically Readable Electrochromic-Based Microfiber Synaptic Device for Photonic Neuromorphic Systems

Abstract

Optical synaptic devices possess great potential in both artificial intelligence and neuromorphic photonics. In this work, an optically readable electrochromic-based microfiber synaptic device was designed by the combination of a multimode fiber and an electrochromic device and using an external voltage to control the transmission of light in the fiber. The proposed synaptic device has the ability to imitate various basic functions of the biological synapses, such as synaptic plasticity, and paired-pulse facilitation (PPF), as well as the transition from short-term memory to long-term memory. Moreover, the proposed device decodes the output optical signal with the international Morse code to express the signal “HFUT” in two ways, and a 3 × 3 array composed of this device can simulate the perceptual learning process. The device can be easily prepared for a wide range of applications, and the incorporated microfibers can be replaced by planar optical waveguides, making it easy to be integrated into a more complex and versatile photonic neuromorphic system.

Feature-Enhanced Artificial Visual Perception Based on Superlinear Voltage-Reflectance Responses of Electrochromic Arrays

“…The reflectance values are 2.6, 4.7, 9.1, 53.0, 75.6, and 76.8% from −0.3 to −1.5 V, decrease to 76.4, 68.4%, 46.5, 29.0, 17.1, 9.1, 4.5, 2.3, and 1.9%, and then slightly increase to 2.0% from −1.2 to 1.5 V. The phenomenon also implies that the electrochromic device can adjust the reflectance by simply controlling the voltage. Therefore, the reflectance can serve as tunable inputs for neuromorphic computing. , To measure the switching time and cycle life of the PB/carbon electrochromic device, the reflectance changes at the wavelength of 794 nm are monitored under the square wave voltage (−1.5 V for 30 s and 1.5 V for 30 s) for 200 cycles. Switching time is defined as the time required for a 90% change in maximum reflectance modulation.…”

“…Therefore, the reflectance can serve as tunable inputs for neuromorphic computing. 41,51 To measure the switching time and cycle life of the PB/carbon electrochromic device, the reflectance changes at the wavelength of 794 nm are monitored under the square wave voltage (−1.5 V for 30 s and 1.5 V for 30 s) for 200 cycles. Switching time is defined as the time required for a 90% change in maximum reflectance modulation.…”