Fig. 1. Overview of this review. Neurons and synapses reprinted with permission from [54]. Copyright 2013, American Institute of Physics. Synaptic plasticity reprinted with permission from [55]. Copyright 2022, Wiley-VCH. Neuromorphic computing reprinted with permission from [56], copyright 2022, Elsevier, and from [57], copyright 2022, Wiley-VCH. Neuromorphic visual systems reprinted with permission from [23], copyright 2020, American Chemical Society, and from [58], copyright 2020, Wiley-VCH. High-order learning behaviors reprinted with permission from [59]. Copyright 2020, Wiley-VCH. Memristors reprinted with permission from [23]. Copyright 2020, American Chemical Society. Transistors reprinted with permission from [60]. Copyright 2021, Wiley-VCH.

Fig. 2. (a) Structural schematic diagram of the biological synapse. (b) and (c) EPSC/IPSC of the perovskite-gated synaptic device triggered by an optical stimulus when the gate voltage is 5 V/−6 V. Reprinted with permission from [67]. Copyright 2022, John Wiley and Sons. (d) EPSC/IPSC triggered by a 980 nm/450 nm optical pulse for synaptic transistor based on the Pyr-GDY/Gr/PbS-QD heterojunction. Reprinted with permission from [68]. Copyright 2001, Elsevier. (e) and (f) EPSC as a function of pulse duration/power density for synaptic transistors based on the MAPbI3/SiNM heterojunction. Reprinted with permission from [23]. Copyright 2020, American Chemical Society. (g) Typical decay of the PSC as a function of time. Reprinted with permission from [69]. Copyright 2014, Nature Portfolio.

Fig. 3. (a) Schematic diagram of typical PPF/PPD behavior of synapses with two successive pulse stimuli. Reprinted with permission from [54]. Copyright 2013, American Institute of Physics. (b) and (c) IPSC/EPSC curves of heterojunction synaptic devices stimulated by two consecutive pulses. (d) and (e) The PPD/PPF index obtained as a function of stimulus pulses applied with different Δt, where the dashed line represents the fitted curve based on Eq. (2). (b)–(e) Reprinted with permission from [81]. Copyright 2019, Wiley-VCH. (f) PPF/PPD effect triggered by a pair of 532 nm/375 nm optical pulses. (g) PPF/PPD index as a function of Δt stimulated by 532 nm/375 nm optical pulses. (f) and (g) Reprinted with permission from [55]. Copyright 2022, Wiley-VCH. (h) PPF ratio as a function of pulse interval stimulated with different electrical pulse peaks. Reprinted with permission from [82]. Copyright 2019, Wiley-VCH.

Fig. 4. (a) Schematic diagram of a typical STP to LTP transition model. (b)–(e) Ids as a function of (b) presynaptic optical pulse wavelength, (c) presynaptic optical pulse density, (d) presynaptic optical pulse width, and (e) number of presynaptic optical pulses. (a)–(e) Reprinted with permission from [87]. Copyright 2022, Springer Nature. Four typical STDP learning rules are illustrated in (f) the antisymmetric Hebbian learning rule, (g) antisymmetric anti-Hebbian learning rule, (h) symmetric Hebbian learning rule, and (i) symmetric anti-Hebbian learning rule, simulated by a GST-based memristor. (f)–(i) Reprinted with permission from [88]. Copyright 2013, Nature Portfolio.

Fig. 5. (a) Schematic illustration of the MAPbI3-based optoelectronic synaptic memristor prefabricated on the SiO2 substrate. (b) Schematic diagram of the generation/annihilation process of VI˙/VI* under darkness (upper) and illumination (lower). (c) Dependence of the spontaneous decay of the MAPbI3-based memristor conductance value upon illumination (1.29  μW/cm2). (d) LTP/LTD of the MAPbI3-based optoelectronic synapse with applying electrical spikes (1 V, 10 ms) upon darkness/illumination (1.29  μW/cm2). (a)–(d) Reprinted with permission from [76]. Copyright 2018, American Chemical Society. (e) Schematic illustration of the Au/KI−MAPbI3/ITO optoelectronic synaptic memristor prefabricated on the glass substrate. (f) Dependence of the device conductance on the electrical stimulus (0.5 V, 2 ms, Vread=0.1  V) upon illumination (0,0.25,0.63  mW/cm2). (g) LTP/LTD behaviors of the Au/KI−MAPbI3/ITO-based synapse with applying consecutive positive/negative voltage spikes (1 V/−1 V, 2 ms, Vread=0.1  V) at various illumination intensities. (e)–(g) Reprinted with permission from [130]. Copyright 2021, Wiley-VCH. (h) Schematic illustration of the ITO/SnO2/CsPbCl3/TAPC/TAPC:MoO3/MoO3/Ag/MoO3 synaptic memristor with dual-mode operation. (i) EPSC triggered by two successive optical spikes (2.5  μW/cm2, 365 nm). (j) Dependence of the PSC on the pulse number upon various illumination intensities (from 12.5 to 50  μW/cm2). (k) Dependence of the SFDP index on the various illumination intensities ranging from 1.25 to 12.5  μW/cm2. (h)–(k) Reprinted with permission from [139]. Copyright 2021, Wiley-VCH.

Fig. 6. (a) Schematic illustrations of the BA2PbBr4-based SPT by inserting the IZTO layer. (b) Energy band diagram of the ITZO/BA2PbBr4 SPT under illumination. (a), (b) Reprinted with permission from [128]. Copyright 2021, Royal Society of Chemistry. (c) LTP and LTD of the ITZO/BA2PbBr4 SPT with applying 50 potentiation (100  μW/cm2, 1 s) light pulses and 50 depression (20 V, 2 s) gate pulses. Adapted with permission from [128]. Copyright 2021, Royal Society of Chemistry. (d) The IDS−VDS characteristic of the G-PQD SPT was evaluated under illumination and dark, respectively. Inset, schematic illustrations of the SPT based on G-PQD superstructure. (e) LTP and LTD of the G-PQD SPT with applying 20 consecutive potentiation (1.1  μW/cm2, 5 s) light spikes and consecutive depression (−0.5  V, 1 s) drain spikes. (f) LTP of the G-PQD SPT stimulated by 20 consecutive light spikes (1.1  μW/cm2, 5 s) with different VG. (d)–(f) Reprinted with permission from [138]. Copyright 2021, American Association for the Advancement of Science. (g) Schematic illustration of the PEA2SnI4/Y6 ambipolar SPT. (h) Wavelength-dependent ΔEPSC peaks. (i) Operation mechanism of synaptic plasticity in response to visible and NIR light pulse irritation. (g)–(i) Reprinted with permission from [60]. Copyright 2021, Wiley-VCH.

Fig. 7. (a) Excitation currents (EPSCi) by optical pulses are a linear function over the number of optical impulses (i, ranging from 1 to 16). (b) Schematic diagram of the principle for employing optical excitation in a synaptic device to achieve addition and subtraction operations. Diagram of the (c) addition operation “9+7,” (d) multiplication operation “5×3,” (e) subtraction operation “10−7,” and (f) division operation “15/10”. Reprinted with permission from [20]. Copyright 2020, Elsevier.

Fig. 8. (a) Schematic illustration of the multi-input light-stimulated CsPbBr3 QDs-based optoelectronic synaptic transistor. Diagram of the (b) “AND” and (c) “OR” logic functions tuned by multiple optical inputs. (a)–(c) Reprinted with permission from [158]. Copyright 2020, American Chemical Society. (d) Schematic diagram of a synapse for switching logic functions via electrical and optical signals. Input-output characteristics of the (e) “AND,” (f) “OR,” (g) “NAND,” and (h) “NOR” logic operations moderated synergistically by the optical and electrical inputs. (d)–(h) Reprinted with permission from [56]. Copyright 2022, Elsevier.

Fig. 9. (a) Schematic illustration of the changes in postsynaptic currents resulting from successive stimulation with optical and electrical signals. (b) Characteristic curves of optical pulse writing and electrical pulse erasing of the device. (c) Variation curves of handwritten digit recognition accuracy along with training epochs of different devices. (d) Schematic illustration of input number “8” and artificial neural network. (e) The initial state of the weight matrix is related to the input numbers. (f) The final state of the weight matrix is related to the input numbers. Reprinted with permission from [57]. Copyright 2022, Wiley-VCH.

Fig. 10. (a) Vpre and Vpost applied to perovskite-based memristors evoked both (b) the symmetric Hebbian learning rule and (c) the symmetric anti-Hebbian learning rule. (d) Vpre and Vpost applied to a memristor-based artificial retinal system evoked both (e) the symmetric Hebbian learning rule and (f) the symmetric anti-Hebbian learning rule. (a)–(f) Reprinted with permission from [161]. Copyright 2017, Nature Portfolio. (g) Schematic diagram of the concept of Pavlovian conditioned reflex. (h) Emulation of Pavlovian conditioned reflex by using CsPbBr3−QDs/MoS2 MVVH. (i) Emulation of Pavlovian conditioned reflex by setting photoelectric synergy training duration to 60 ms. (g)–(i) Reprinted with permission from [58]. Copyright 2020, Wiley-VCH. (j) Schematic diagram of the reward and punishment mechanisms that occur in creatures. (k) Emulation of punishment mechanism by synergistic control of optical spikes and positive voltage spikes. (l) Emulation of reward mechanism by synergistic control of optical spikes and negative voltage spikes. (j)–(l) Reprinted with permission from [131]. Copyright 2021, American Chemical Society.

Fig. 11. (a) Relationship between the maximum EPSC triggered by 30 light pulses and varying gate voltages. (b) EPSC is triggered by 30 light pulses at varying gate voltages. (c) Recognition of the letter “H” as the brain enters positive/neutral/negative mood states. (a)–(c) Reprinted with permission from [23]. Copyright 2020, American Chemical Society. (d) Diagram of the relationship between learning and memory under different emotional states. Reprinted with permission from [130]. Copyright 2021, Wiley-VCH.

Fig. 12. (a) Schematic diagram of the human visual system. (b) Schematic representation of habituated behavior when the nervous system is stimulated. (c) Schematic diagram of the structure of the DAVAN device. (d) Habituation behavior of the device when stimulated by 40 light pulses. (e) Optical photograph of the DAVAN device for the 3×3 array. (f) Schematic diagram of the DAVAN device array in photopic vision condition (left) and scotopic vision condition (right) when illuminated with different intensities of incident light. (g) and (h) Schematic diagram of the dynamic response process of the DAVAN device at the center of the array. Reprinted with permission from [59]. Copyright 2020, Wiley-VCH.

Table 1. Summary of MHP-Based Optoelectronic Synapses