Bidirectional Photoresponse in Perovskite‐ZnO Heterostructure for Fully Optical‐Controlled Artificial Synapse

Abstract

energy consumption and inherent speed limitations. Inspired by human neural system, communicating through synapses with the features of parallel distributed processing and integrated storage and computation, various neuromorphic devices including memristors and fieldeffect transistors have been employed to simulate synaptic functions to overcome the von Neumann bottleneck. [1][2][3] Synaptic plasticity such as long/short-term potentiation (LTP/STP), long/short-term depression (LTD/STD), and paired pulse facilitation/depression (PPF/PPD), have been mimicked by modulating device conductivity (i.e., synaptic weight). Among these synaptic devices, potentiation and depression are always triggered by electrical stimulation. As known the electronic technology suffers the problem of limited computing speed because of the bandwidth connection density trade-off. [4] Fortunately, optoelectronic synapses with the light stimuli exhibit potential for ultrafast computing with advantages of the low crosstalk, high bandwidth, and low power consumption, also can be applied in optical wireless communication. [5,6] Herein, the optoelectronic synapses are operated with direct light illumination and the output is interpreted from the change of electrical conductance. It is different from the photonic devices that are mentioned in the photonic integrated circuits, in which the light is illuminated through a waveguide and output is denoted by the transmittance/absorbance of light in the waveguide. [7,8] However, most of reported optoelectronic devices only show positive photoresponse corresponding to the potentiation behavior of synapses. The achievement of depression behavior still depends on electrical stimulation due to the unidirectional photoresponse of devices. Therefore, a device with bidirectional photoresponse to emulate both excitatory and inhibitory behaviors for a fully optical-controlled artificial synapse is highly desired.Recently, a few of fully optical-controlled artificial synapses have been reported, such as pyrenyl graphdiyne/graphene/ PbS heterostructure, [5] Bi 2 O 2 Se/graphene hybrid structure, [6] ZnO/PbS hybrid heterostructure, [9] oxygen-deficient/oxygenrich InGaZnO homojunction, and black phosphorus. [10][11][12][13] Both the excitatory and inhibitory synaptic behaviors of Compared with electrical synapses, optoelectronic synapses exhibit great potential for ultrafast computing and wireless communication in neuromorphic hardware with advantages of low crosstalk, high bandwidth, and low power consumption. However, due to the unidirectional photoresponse in most of optoelectronic synapses, the related researches still focus on the electrical stimulation of inhibitory behaviors. Herein, a synaptic device based on perovskite−ZnO heterostructure is prepared with the characteristic of bidirectional photoresponse and can realize both the potentiation and depression by ultraviolet (UV) and green light stimuli, respectively. Consecutive UV and green light stimuli for potentiation and depression reveal the reliability ...

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

“…The PPF index can be defined as the ratio (A 2 /A 1 ) of the first peak (A 1 ) to the second peak (A 2 ). 39 At the time interval of 2 s, the largest PPF index was 1.25. Nevertheless, it gradually decreased until it stabilized at a value of 1.00 with the increase in the time interval between the two pulses (Figure 3d).…”

“…After waiting for the pulse interval Δ t , the second stimulation was applied, and the amplitude A 2 was enhanced than A 1 . The PPF index can be defined as the ratio ( A 2 / A 1 ) of the first peak ( A 1 ) to the second peak ( A 2 ) . At the time interval of 2 s, the largest PPF index was 1.25.…”

Nanomaterial-Based Synaptic Optoelectronic Devices for In-Sensor Preprocessing of Image Data

“…Inspired by these responses, electrical synaptic devices exhibiting both excitatory and inhibitory behaviors have been used for neuromorphic computing, providing the advantages of ultrafast computational speed with low crosstalk, high bandwidth, and low power consumption . Recently, optically triggered artificial synapses that exhibit both excitatory and inhibitory behaviors have been developed by integrating semiconducting nanoparticles with different bandgaps. − The artificial synapse is composed of a CdS/a-IGZO synaptic phototransistor, in which the gate electrode is connected to a voltage divider comprising a CdSe photoresistor and an a-IGZO load resistor (Figure g) . Because the bandgap of CdS and a-IGZO is 2.36 and 3.69 eV, respectively, the CdS/a-IGZO synaptic phototransistor generated the photocurrent in response to the green light because of the charge transfer between the CdS and a-IGZO, which corresponds to excitatory behavior.…”

Electrospun Nanofiber‐Based Synaptic Transistor with Tunable Plasticity for Neuromorphic Computing

“…Moreover, synaptic behaviors experienced huge changes with perovskite doping in comparison with the one without doping. Considering the particularity of the perovskite to response to the light, [43,44] a presynaptic spike of 8 V amplitude (pulse width = 40 ms) could trigger an EPSC for the doped device under light irradiation (≈9.90 nA) of two orders of magnitude larger than the undoped device (≈302.53 pA) (see Figure 4b). The declined operating voltage for V ds (10 to 1 V) and the greatly improved synaptic current demonstrated the direct benefit of the surface doping technique and the photosensitive characteristics of the added perovskite materials.…”