# Fileset

[Advanced Science - 2025 - Guo - A Van Der Waals Broadband Infrared Optical Synapse Enabling Orientation Detection.pdf](https://mdr.nims.go.jp/filesets/098acd3a-dc8a-4b64-98e8-efcf54a63f30/download)

## Creator

Dan Guo, Wenjing Li, Pingfan Gu, Weikang Dong, Xuyan Rui, [Kenji Watanabe](https://orcid.org/0000-0003-3701-8119), [Takashi Taniguchi](https://orcid.org/0000-0002-1467-3105), Yu Ye, Fawei Zheng, Jiadong Zhou, Shoujun Zheng

## Rights

[Creative Commons BY Attribution 4.0 International](https://creativecommons.org/licenses/by/4.0/)

## Other metadata

[A Van Der Waals Broadband Infrared Optical Synapse Enabling Orientation Detection](https://mdr.nims.go.jp/datasets/a52e98ed-9256-429d-b916-e71b052b70f2)

## Fulltext

A Van Der Waals Broadband Infrared Optical Synapse Enabling Orientation DetectionRESEARCH ARTICLEwww.advancedscience.comA Van Der Waals Broadband Infrared Optical SynapseEnabling Orientation DetectionDan Guo, Wenjing Li, Pingfan Gu, Weikang Dong, Xuyan Rui, Kenji Watanabe,Takashi Taniguchi, Yu Ye, Fawei Zheng, Jiadong Zhou,* and Shoujun Zheng*A vision system with efficient infrared-sensitive optical synapses is crucial forenabling infrared radiation detection and target recognition for somepredators hunting in the dark. Current 2D synaptic devices typically adopt thestrategy of charge trap and release to achieve intelligent sensing and visualrecognition. However, the response wavelength is limited in a narrow range(typically in visible) due to the intrinsic bandgap of these 2D materials. In thiswork, a broadband infrared optoelectronic synaptic device based on afew-layer graphene/CrOCl/few-layer graphene van der Waals (vdW)heterostructure is reported, featuring tunable spike timing-dependentplasticity and a broadband response range from the visible to the infrared(520–2000 nm). The broadband synaptic response in the tunneling device isattributed to the modulation of the tunneling barrier by strong interfacialcoupling and charge transfer-induced long-wavelength charge order at thevdW interface. Integrated with reservoir computing technique, the tunnelingdevice can efficiently detect images in different orientations, achieving arecognition accuracy exceeding 98%, and judge the possible escape directionsof a mouse. This work not only allows us to explore broadband opticalsynapses by controlling the vdW interfacial coupling but also offers apromising solution for developing advanced infrared detection systems.1. IntroductionBrain-inspired neuromorphic computing is increasingly beingexplored as an alternative for efficient information process-ing and transmission[1,2] to address the bottleneck issues ofD. Guo, W. Li, W. Dong, X. Rui, F. Zheng, J. Zhou, S. ZhengCentre for Quantum PhysicsKey Laboratory of Advanced Optoelectronic Quantum Architecture andMeasurement (MOE)School of PhysicsBeijing Institute of TechnologyBeijing 100081, ChinaE-mail: jdzhou@bit.edu.cn; szheng@bit.edu.cnThe ORCID identification number(s) for the author(s) of this articlecan be found under https://doi.org/10.1002/advs.202507530© 2025 The Author(s). Advanced Science published by Wiley-VCHGmbH. This is an open access article under the terms of the CreativeCommons Attribution License, which permits use, distribution andreproduction in any medium, provided the original work is properly cited.DOI: 10.1002/advs.202507530low efficiency and high energy consump-tion caused by the separation of memoryand computational units in von Neumannarchitectures.[3] Synapses are the funda-mental units of the brain’s neural system re-sponsible for detection, learning, and com-putation, playing a critical role in infor-mation transfer and processing.[4] Artificialsynapses are desired for the on-chip cir-cuit to mimic the plasticity of synapses fordetecting and learning environmental sig-nals, such as excitatory postsynaptic cur-rent (EPSC).[5,6] In recent years, electricaland optical artificial synaptic devices basedon 2D materials have been widely reportedfor neuromorphic computing.[7,8] However,these devices majorly rely on charge trapand release processes originating from de-fects or floating gates, leading to device in-stability, functional degradation, and com-plex modulation.[9] Therefore, it is essentialto develop 2D devices with advanced mech-anisms to simulate synaptic behaviors effec-tively.2D optical synaptic devices with visualrecognition are expected to be applied inmany practical fields, such as the Internet of Things smart sen-sors, biomedical electronics, and robotics.[10] To date, numerous2D optical synaptic devices have been reported to be capableof visual recognition based on various operational mechanisms,including persistent photoconductive effects,[11,12] photo-gatingP. Gu, Y. YeState Key Laboratory for Artificial Microstructure & Mesoscopic Physicsand Frontiers Science Center for Nano-Optoelectronics School of PhysicsPeking UniversityBeijing 100871, ChinaK. Watanabe, T. TaniguchiNational Institute for Materials Science1-1 Namiki, Tsukuba 303-0044, JapanJ. Zhou, S. ZhengFaculty of Marine Science and TechnologyBeijing Institute of TechnologyZhuhai, Guangdong 519088, ChinaAdv. Sci. 2025, 12, e07530 e07530 (1 of 10) © 2025 The Author(s). Advanced Science published by Wiley-VCH GmbHhttp://www.advancedscience.commailto:jdzhou@bit.edu.cnmailto:szheng@bit.edu.cnhttps://doi.org/10.1002/advs.202507530http://creativecommons.org/licenses/by/4.0/http://creativecommons.org/licenses/by/4.0/http://crossmark.crossref.org/dialog/?doi=10.1002%2Fadvs.202507530&domain=pdf&date_stamp=2025-09-30www.advancedsciencenews.com www.advancedscience.comeffects,[13] material defects and charge-trapping effects,[14] photo-generated carrier generation and recombination,[15] ferroelectric-induced separation of photogenerated carriers,[16] and tunnelingof photogenerated carriers.[17] However, most of these devices ex-hibit optical responses in the ultraviolet (UV) and visible (VIS)ranges due to the intrinsic band gap of 2Dmaterials. Research onthe optical synapse operating in the infrared (IR) range remainslimited.The key to designing IR optical synapse devices is to effectivelymodulate the lifetime of carriers generated by the IR photons.First, the 2D materials must exhibit a high photoresponse in theIR range. Although black phosphorus (BP) possesses a narrowband gap suitable for IR detection, its applications are severelylimited by its susceptibility to oxidation.[18] Graphene is anothercandidate for IR detection due to its gapless band structure;[19,20]however, the rapid recombination of hot charge carriers restrictsits use in optoelectronic synapses.[21,22] Second, van der Waals(vdW) heterostructures provide a promising platform for mod-ulating photon-generated carriers and enabling synaptic func-tions driven by interlayer charge transport and strong interfa-cial coupling effect. For example, as an indirect bandgap mate-rial, CrOCl typically has a large bandgap (≈2.77 eV),[23] whichis consistent with our absorbance spectrum result of 2.3 eV inFigure S1 (Supporting Information). The wide bandgap makesit a good insulator, which also means it is not suitable for di-rect detection of infrared light. However, vdW heterostructurescombined CrOCl with other 2D materials, such as graphene andtransition metal dichalcogenides (TMDs), can be used to modu-late their electronic properties via strong interlayer coupling andpotentially extend to infrared light detection. Specifically, vdWinteractions between CrOCl and other 2D materials have beenreported to reconfigure the polarity of 2D TMDs[24] and achievelarge on/off ratios in graphene field effect transistors.[25,26] Thestrong interlayer coupling and charge transfer between CrOCland 2D material allow us to modulate optoelectronic propertiesat the vdW interface. Therefore, vdWheterostructures can be em-ployed to develop air-stable IR optical synapse devices, which canmeet the demands of various practical applications, such as in-telligent imaging, autonomous driving, and space communica-tions.In this study, we fabricated a few-layer graphene/CrOCl/few-layer graphene (FLG/CrOCl/FLG) vdW heterostructure toachieve broadband IR optical synapse. Compared to previouslyreported synaptic devices, our device demonstrates efficientsimulation of optical synaptic behaviors with tunable spike-timing-dependent plasticity (STDP), spike-number-dependentplasticity (SNDP), spike-rate-dependent plasticity (SRDP), andpaired pulse facilitation (PPF) over a broad spectral range fromthe VIS to the IR (520–2000 nm). Theoretical calculations revealthat the strong interfacial coupling at the graphene/CrOCl inter-face induces the formation of long-wavelength charge order andreduces the tunneling barrier under illumination. Integratedwith reservoir computing (RC), the optical synapse is capable ofeffectively sensing IR signals in different orientations, achievingan accuracy exceeding 98%. Our IR synaptic device demon-strates that the vdW heterostructure is a promising platform fordesigning high-performance optical neuromorphic devices andexploring potential applications in IR intelligent imaging andretinomorphic computing.2. Results and Discussion2.1. Biomimetic IR Perception of SnakeAn IR-sensitive vision system is essential for IR radiation detec-tion and target imaging in some nocturnal animals such as pitvapor and vampire bat (shown in Figure 1a). These predators pos-sess a thermosensitive pit organ (consisting of a pit membraneand trigeminal nerves, shown in Figure 1b) that enables them todetect IR radiation and perceive temperature variations even indark environments.[27,28] The trigeminal nerve plays a crucial rolein detecting, transmitting, and processing IR signals through thefamous transient receptor potential cation channel A1 (TRPA1)channel.[29,30] Based on this, an IR optical synaptic device can the-oretically replicate the function of the pit organ by responding toand processing IR light pulses to capture IR targets. Integratedwith RC, the IR optical synapse can theoretically detect orienta-tions in response to synaptic stimulation, like tracking the direc-tions of prey movement (Figure 1c).To detect IR signals, we fabricated a FLG/CrOCl/FLG tun-neling device and measured the tunneling current under IRlight illumination, as illustrated in Figure 2a, in which few-layergraphene is commonly employed in IR detectors due to its gap-less band structure and high response to IR light. CrOCl, anemerging antiferromagnetic insulator that has been applied inheat dissipation devices[31] and memory devices[32] due to itslow-symmetry crystal structure, large magnetoelastic couplingeffects, and vacancy-tunable bandgap,[33] is used as both a tun-neling barrier and an interfacial coupling layer with graphene.The strong interfacial coupling between graphene and CrOCl hasbeen exploited to tune the quantum Hall state and open a gap ingraphene.[25]In our FLG/CrOCl/FLG tunneling device (optical imageshown in Figure 2b), a bias voltage is applied to the bottomgraphene layer, and the tunneling current is measured under var-ious light pulse stimulations. The device’s active region is en-capsulated by a ≈30 nm thick hexagonal boron nitride (h-BN)(highlighted by the red circle in Figure 2b). Figure 2c presentsthe transmission electron microscopy (TEM) image of the cross-section of the vdW heterostructure, where both graphene andCrOCl can be clearly distinguished. CrOCl is identified as 22 lay-ers thick, which is consistent with our atomic force microscopy(AFM) measurements (the inset in Figure 2b shows the CrOClthickness of 16 nm). The crystal structure of CrOCl is shown inthe scanning transmission electronmicroscopy high-angle annu-lar dark-field (STEM-HAADF) image in Figure 2d, revealing itshigh crystalline quality.We performed I–V test on the vdW tunneling device in the darkwithout applying gate voltage, which shows a ≈80 nA tunnelingcurrent at a ±1 V bias (Figure S2, Supporting Information). Forcomparison, we fabricated two Au/CrOCl/Au devices with CrOClthicknesses of 45 and 6.5 nm, respectively (Figure S3, Support-ing Information). Both devices exhibited insulating behavior ata ±1 V bias under different gate voltages, demonstrating thatCrOCl is a good insulator, consistent with reported literatures.[34]The difference in conductance between the two types of tunnel-ing devices indicates a strong interfacial coupling effect and effi-cient charge transfer at the vdW interface of graphene andCrOCl,which significantly modifies the tunneling barrier.Adv. Sci. 2025, 12, e07530 e07530 (2 of 10) © 2025 The Author(s). Advanced Science published by Wiley-VCH GmbH 21983844, 2025, 48, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/advs.202507530 by Kenji Watanabe - National Institute For , Wiley Online Library on [29/12/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.advancedscience.comwww.advancedsciencenews.com www.advancedscience.comFigure 1. Detection of a prey through an infrared optical synapse. a) The infrared radiation perception of a pit viper utilizing the pit organ located betweenthe mouth and eyes to detect the prey movement direction in the darkness. b) The detailed structures of a pit organ composed of the pit membrane (athin membrane-like tissue with the ability to sense infrared radiation) and the trigeminal nerve branch (transmits the infrared radiation signals detectedby the pit organ to the brain, enabling the snake to perceive the location and movement direction of warm-blooded prey). c) Recognition processing ofinfrared radiation pulse trains using reservoir computing for the image classification.2.2. IR Synaptic Plasticity of the Tunneling DeviceTo demonstrate the opto-synaptic plasticity of the vdW het-erostructure, we applied 1550 nm laser pulses with a 2 s pulsewidth to the tunneling device (shown in Figure 2e). Unless oth-erwise specified, all measurements in Figure 2 were conductedunder these conditions, with 1550 nm light illumination at abias of 1 V. In biological vision systems, SNDP is the essentialfunction for processing time-dependent signals by modulatingboth the sign and magnitude of synaptic strength, and SNDPwas achieved by applying successive pulse stimuli in our device(Figure 2e). As the number of light pulses increases, the synapticweight (An/A1) continuously increases, and the ratio of A60/A1approaches 16 (Figure S4, Supporting Information), suggestingthe potential for improved contrast in neuromorphic imagingand preprocessing applications. Figure 2f illustrates the SRDPperformance of the device, transitioning from 0.5 to 5 Hz un-der ten light pulses at varying frequencies, and the peak value ofEPSC increases with the frequency of the laser pulses. Moreover,the EPSC increase at a specific frequency is evaluated in terms ofgain, which is defined as the ratio of the maximum PSC inducedby the tenth light pulse (A10) to the maximum current inducedby the first light pulse (A1), and we observe that the gain (A10/A1)increases with the frequency of the pulses Figure S5, SupportingInformation).Unlike typical photodetectors, the photocurrent in our devicecontinuously increases with the light illumination power andgradually recovers slowly once the light stimulus is removed.The magnitude of the photocurrent increases approximately lin-early with light intensity (Figure S6, Supporting Information),and the device exhibits a longer rise time and falling time,which is attributed to the continuous modulation of the tunnel-ing barrier by photogenerated carriers produced within the topgraphene layer. Furthermore, the photocurrent increases linearlyAdv. Sci. 2025, 12, e07530 e07530 (3 of 10) © 2025 The Author(s). Advanced Science published by Wiley-VCH GmbH 21983844, 2025, 48, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/advs.202507530 by Kenji Watanabe - National Institute For , Wiley Online Library on [29/12/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.advancedscience.comwww.advancedsciencenews.com www.advancedscience.comFigure 2. The structure of the device and synaptic plasticity under 1550 nm laser optical stimulation. a) Schematic diagram of the few-layergraphene/CrOCl/few-layer graphene optoelectronic device, which is fully encapsulated in hBN. b) Optical image of the device. few-layer graphene andCrOCl are highlighted with black and yellow dashed lines. Inset shows the AFM morphology of the CrOCl, with a thickness of 16 nm. c) Cross-sectionalTEM image of the device. Scale bar is 5 nm. d) Enlarged cross-sectional STEM-HAADF image in c showing the lattice structure of CrOCl. Scale bar is 1nm. e) SNDP triggered by pulse numbers where a single optical pulse is 2 s and the power is 51 μWat Vg = 0 V and Vb = 1 V. f) SRDP triggered by differentfrequencies of optical pulse from 0.5 to 5 Hz s with the power of 51 μW at Vg = 0 V and Vb = 1 V. g) Pulse duration-dependent photoresponse withthe power of 51 μW at Vg = 0 V and Vb = 1 V. h) Memory characteristics induced by optical pulses for simulating human learning-forgetting-relearningbehavior. i) Dependence of the PPF ratio (defined as ΔA2/ΔA1 × 100%) on the pulse interval with the light power of 51 μW and pulse duration of 1 s atVg = 0 V and Vb = 1 V. Inset shows the PPF behavior stimulated by a pair of optical pulses.with pulse duration (Figure 2g; Figure S7, Supporting Informa-tion) and demonstrates clear STDP with a tunable EPSC. Fivecycles of on/off light pulses with 5 s illumination time and 10 sshut-off time were used to test the learning and forgetting pro-cesses (Figure 2h). The current increases from 65 to 92 μA dur-ing the first learning phase and then decreases to a value slightlyhigher than the initial state. The subsequent re-learning processfurther strengthens the memory function, and the ∆I differencebetween the first and fifth decayed currents represents the re-peated learning process, which follows the Ebbinghaus forgettingcurve[35,36] of the human brain about the learning and forgettingbehaviors.The paired pulse facilitation (PPF) index, defined as ΔA2/ΔA1× 100%, is commonly used to quantify short-term plasticity re-Adv. Sci. 2025, 12, e07530 e07530 (4 of 10) © 2025 The Author(s). Advanced Science published by Wiley-VCH GmbH 21983844, 2025, 48, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/advs.202507530 by Kenji Watanabe - National Institute For , Wiley Online Library on [29/12/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.advancedscience.comwww.advancedsciencenews.com www.advancedscience.comFigure 3. Broadband synaptic response of the graphene/CrOCl/graphene device. a) Optical synaptic behavior of the device with light excitation of fourwavelengths of 520, 1064, 1400, and 2000 nm at Vg = 0 V and Vb = 1 V. b) Summary of the PPF of previously reported optical synaptic devices[8,14,41,48–50]and our device. Our device exhibits significant synaptic plasticity in the infrared range.sponses between neurons, particularly the changes in synaptic re-sponse under consecutive stimuli. The PPF triggered by the 1550nm light illumination in our device is 184% (shown in Figure 2i),indicating that the response to the second pulse is 84% greaterthan that of the first. This reflects that our device exhibits an ef-ficient STDP, which is critical for information processing as wellas the short-time formation of learning and memory in the ner-vous system. These results suggest that our device serves as anideal IR optical synapse with tunable EPSC.2.3. Broadband Synaptic Response from VIS to IRArtificial synapses, in conjunction with visual recognition, havebecome an inevitable development trend for future technolo-gies, particularly in areas such as night vision imaging and au-tonomous driving. Most reported optical artificial synapses basedon 2D materials operate within a narrow VIS spectrum.[12,37,38]However, our device exhibits synaptic plasticity across a broadspectral range, extending from the VIS to the IR range. We testedour device under laser illumination at various wavelengths andobserved significant optical synaptic plasticity at wavelengths of520, 1064, 1400, and 2000 nm shown in Figure 3a. The depen-dences of the EPSC on laser pulse number, frequency, power,and duration are consistent with those observed at 1550 nm(Figures S8–S11, Supporting Information). Additionally, the PPFalso shows great tunability to pulse interval (Figure S12, Sup-porting Information). Figure 3b summarizes optical synaptic de-vices based on 2D materials (more reported 2D optical synapsesare summarized in Table S1, Supporting Information). The en-hanced synaptic characteristics of our device position it as apromising candidate for advancing nighttime imaging and recog-nition applications.To investigate CrOCl thickness-dependent synaptic responses,we fabricated tunneling devices with thicknesses of 2.5 and 40nm (Figure S14, Supporting Information) and tested them with520 and 1064 nm lasers. Similar to the 16 nm thickness, both the2.5 and 40 nm thick tunneling devices demonstrated strong op-tical synaptic plasticity. We compared the test results of deviceswith three different CrOCl thicknesses with 520 and 1064 nmlasers, using identical test conditions (Table S2, Supporting In-formation). The dark currents of the tunneling devices are 900 μAat 2.5 nm (thin), 80 nA at 16 nm (medium), and 6.5 nA at 40 nm(thick), respectively. Devices with all three thicknesses demon-strated significant optical synaptic plasticity and only slight vari-ations of ΔA10/ΔA1 and PPF.2.4. Mechanisms for the Broadband Synaptic ResponseThe broadband synaptic response in our device originatesfrom the strong interfacial coupling and charge transfer in thegraphene/CrOCl vdW heterostructure. Due to its intrinsic zeroband gap, graphene is widely utilized in IR detectors.[19,39] Wecompared the absorbance spectra of CrOCl and graphene/CrOClheterostructure and found that IR light ismajorly absorbed by thegraphene (shown in Figure 4a). Moreover, the low conductivity inthe Au/CrOCl/Au devices suggests that graphene plays a crucialrole in the IR optical synaptic response (see Figure S3, Support-ing Information). Photogenerated electrons in graphene tend totransfer to the adjacent CrOCl layer due to the strong interfacialcoupling. Electrons trapped within the CrOCl form a long-rangeAdv. Sci. 2025, 12, e07530 e07530 (5 of 10) © 2025 The Author(s). Advanced Science published by Wiley-VCH GmbH 21983844, 2025, 48, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/advs.202507530 by Kenji Watanabe - National Institute For , Wiley Online Library on [29/12/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.advancedscience.comwww.advancedsciencenews.com www.advancedscience.comFigure 4. Mechanism of the broadband synaptic response. a) Absorbance spectra of graphene/CrOCl (S1) and CrOCl (S2). Inset is the optical imageof the samples. b,c) Calculated band structures of the graphene/CrOCl vdW heterostructure before (b) and after (c) charge transfer from graphene toCrOCl. The conduction band of CrOCl (green dashed lines) is lowered after the interlayer charge transfer and electron doping. d) Schematic diagramof the graphene/CrOCl heterojunction used for KPFM testing. e) Surface potential images of the graphene/CrOCl heterostructure in the dark (top) andupon illumination with light of 520 nm (bottom). f) Surface potential differences derived from (e). g,h) The schematic band alignments of the devicebefore (g) and after (h) IR light illumination. The photo-generated electrons can transfer to the CrOCl layer, resulting in the decrease of the Schottkybarrier. i) Schematic of the graphene/CrOCl, showing the charges transfer between graphene and CrOCl.Coulomb superlattice[24,25,40] and do not contribute to the deviceconductance. We performed density functional theory (DFT) cal-culations to compare the band structures of graphene/CrOClvdW heterostructure before and after electron charge transfer.The calculation results show that electron doping from grapheneto CrOCl shifts down the conduction band of the adjacent CrOCllayer (see the green lines in Figure 4b,c; Figure S15, SupportingInformation), indicating that charge transfer at the vdW interfaceis the primary mechanism for modulating the EPSC behavior.To study the mechanism of IR synaptic photoresponse, weemployed Kelvin probe force microscopy (KPFM) to probe thesurface potentials of the graphene/CrOCl vdW heterostructure.A vdW heterostructure of graphene/CrOCl was used for KPFMtesting in Figure 4d (see optical image and surface morphologyin Figure S16, Supporting Information). Figure 4e shows theKPFM images under dark and 520 nm light illuminations, wherethe surface potentials of graphene, graphene/CrOCl, and CrOClcan be evaluated. Figure 4f is the plots following the dashed ar-row lines in Figure 4e, showing that the work function of CrOClis lowered under 520 nm illumination. The light-tunable sur-face potential indicates that light illumination can modulate thetunneling barrier of our device, which makes the electrons eas-ily transfer from graphene to the adjacent CrOCl layer. Notably,the surface potential of graphene located above CrOCl shows aslight decrease compared to the graphene above SiO2, hintingthe strong interlayer coupling at the vdW interface. For compar-ison, we fabricated a graphene/h-BN/graphene device to replacethe CrOCl with a thin h-BN flake (Figure S17, Supporting Infor-mation), and the device exhibits typical photo response under theillumination of 520 nm laser. The absence of the synaptic behav-Adv. Sci. 2025, 12, e07530 e07530 (6 of 10) © 2025 The Author(s). Advanced Science published by Wiley-VCH GmbH 21983844, 2025, 48, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/advs.202507530 by Kenji Watanabe - National Institute For , Wiley Online Library on [29/12/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.advancedscience.comwww.advancedsciencenews.com www.advancedscience.comior in the graphene/h-BN/graphene device signifies the key roleof the interlayer coupling at the CrOCl/graphene interface.We present a band model to elucidate the EPSC behavior inour FLG/CrOCl/FLG tunneling device before and after IR illumi-nation (shown in Figure 4g,h). Initially, IR light generates pho-togenerated electron-hole pairs in the few-layer graphene layer.Then, electrons from the top graphene layer are transferred tothe adjacent CrOCl layer due to strong interface coupling, whichlowers the tunneling barrier and increases the tunneling current.Upon the removal of light illumination, electrons in the long-range charge order gradually retreat from the long-range super-lattice and transfer back to the graphene layer in several min-utes (Figure 4i). The formation of the long-range superlattice byelectrons transferred from graphene continuously modulates thetunneling barrier and the EPSC, ultimately leading to the de-vice’s broadband light synaptic plasticity. Therefore, the broad-band synaptic plasticity is associated with the interfacial couplingbetween CrOCl and the graphene layer, and is less affected theCrOCl thickness, making it suitable for subsequent practical ap-plications.2.5. IR Orientation DetectionFast response and real-time recognition of IR images are crucialfor predators hunting in the dark. Due to the broadband synapticplasticity, our device can achieve feature recognition of IR im-ages, providing valuable insights for simulating various opticalsynaptic inputs.[41] Based on the linear change of the postsynapticcurrent with the duration of light exposure (Figure S7, Support-ing Information), we evaluate the feature recognition capabilityof the device under 1550 nm laser illumination using five imagesin different orientations with the RC system.[42] The actual im-ages were converted into binary digitized pixels, and each imageconsists of 25 pixels (shown in Figure 5a) and can be mappedinto a pulse train consisting of 5 pulses (Figure 5b). The “ON”and “OFF” states in the pulse train correspond to “1” and “0” ina basic binary image, respectively. Each pulse train applied to ourdevice can obtain an EPSC value, which encodes the informationof the orientation (Figure 5c). Based on this translation, all of theimages consisting of 25 pixels can be mapped into five bits forfurther processing (Figure S18, Supporting Information).The EPSC values weremeasured using I-t curves for all images(Figure S19, Supporting Information) to assess the detection oforientations. The device demonstrated long-time stability, capa-ble of withstanding tests for up to 3 months, allowing for ten rep-etitions per image in both simulation and classification tasks. Forthe orientation classification, the obtained EPSC values were pro-cessed through a virtual network consisting of 5 synapses and 25weights (the detailed process is shown in Figure S20, Support-ing Information). The loss approaches zero after 25 iterations,and the recognition accuracy of 98% is achieved after 60 train-ing epochs for the five directions (shown in Figure 5d), which iscomparable to our simulated recognition result with a variationof 0.2.Furthermore, we map eight distinct mouse images (each con-sisting of 23 × 23 pixels) to eight different feature directions(shown in Figure 5e) to mimic the brain processing of IR sig-nals recognition by a pit vapor. Gray-level extraction is performedon the eight images and pulse trains are generated based onrules outlined in Figure 5c. In conjunction with our RC sys-tem for orientation classification, we trained 23 × 8 weightsand calculated the recognition accuracy. The confusion matrixdemonstrates a high recognition accuracy for all eight directions(shown in Figure 5f), indicating the optical synapse based on ourFLG/CrOCl/FLG tunneling device is suitable for IR image real-time recognition, making it promising for future neuromorphicdevices and IR vision systems.3. ConclusionIn summary, we reported a few-layer graphene/CrOCl/few-layergraphene tunneling device that exhibits optical synaptic charac-teristics across a broadband range from 520 to 2000 nm. Thedevice converts infrared signals with high efficiency and simu-lates various synaptic behaviors, including EPSC, STDP, SNDP,SRDP, and PPF. Furthermore, integrated with RC, our devicecan effectively detect infrared signals and classify the IR im-ages with different orientations. Both experimental and simula-tion results achieve recognition rates exceeding 98% within 60epochs, providing valuable insights for the development of mul-tifunctional, intelligent perception systems. Consequently, thisvdW heterostructure holds significant promise for advancing op-toelectronic synaptic devices and enhancing the performance ofIR recognition systems.4. Experimental Sectionh-BN/ Graphene/CrOCl/ Graphene Heterojunction Fabrication: The h-BN/ graphene/CrOCl/ graphene heterojunction was fabricated using adry transfer technique. All materials were mechanically exfoliated using3 M scotch tape on flake blocks onto a silicon/silicon dioxide (285 nm)(Si/SiO2) substrate for optical microscopy examination (Olympus). First,a poly (Bisphenol A carbonate) (PC) film coated with polydimethylsilox-ane (PDMS) was used to pick up h-BN. Next, top graphene and CrOClwere sequentially picked up using the strong vdW force between h-BN and2D materials. The PC was prepared by dispersing 1 g of polycaprolactonein 10 mL chloroform (Aladdin Tech. Co.). Then, the picked-up h-BN, topgraphene, and CrOCl were aligned and released onto the bottom grapheneat 180 °C. Finally, the PC was removed by soaking in chloroform for 5 min.Use electron-beam lithography technology (Zeiss supra 55 and Raith EL-PHY Quantum) to pattern the electrons of the device, and deposit Cr/Au(5/50 nm) with thermal evaporation in a high vacuum of 10−5 Pa by a ther-mal evaporation coater (Chinese Academy of Sciences Shenyang ScientificInstruments Co.).Au/CrOCl/Au Device Fabrication: The mechanically exfoliated CrOClwas picked up from the substrate using a PC and released onto the pre-deposited bottom electrode Cr/Au (5/10 nm) at 180 °C. Then, the preparedsample was evaporated the top electrode Cr/Au (5/50 nm) in the regionwhere CrOCl overlaps with the bottom electrode.Characterization of Tunneling Devices: The thicknesses of materialswere performed using atomic force microscopy (Dimension Icon, Bruker).Absorption spectra were obtained by a UV–Vis-NIR Microscopic Spec-trophotometer (MSV-5700, Jasco). The TEM images were obtained by anFEI Themis Z with double aberration correctors, and the data was pro-cessed by the Velox software.Photoelectronic Measurement: Photoelectric measurements were per-formed in a probe (CRX-6.5K, Lakeshore) using a semiconductor analyzer(Keysight B1500A). 520, 1064, and 1550 nm laser (NBeT) were introduceddirectly into the probe shore cavity through an optical fiber and the light in-tensity was measured by a power meter (THORLABS GmbH). Lasers withAdv. Sci. 2025, 12, e07530 e07530 (7 of 10) © 2025 The Author(s). Advanced Science published by Wiley-VCH GmbH 21983844, 2025, 48, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/advs.202507530 by Kenji Watanabe - National Institute For , Wiley Online Library on [29/12/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.advancedscience.comwww.advancedsciencenews.com www.advancedscience.comFigure 5. Recognition of images in different orientations based on IR in-sensor RC system. a) Five images in different orientations used for detectionby our device. b) Pulse trains with a binarized number obtained from the orientation of each image in (a). The pulse width is 0.5 s and the pulse intervalis 1.5 s. c) I-t photo-response characteristics of four representative inputs of “00001,” “00011,” “01111,” and “11111”. d) Loss and recognition accuracyof our device for experimental and simulated results with a variation of 0.2. e) 8 mouse images consisting of 23 × 23 pixels mapped to eight differentfeature directions to mimic the detection process of a pit viper. f) False-color confusion matrix showing the classification results of our device comparedto the predicted output results.other wavelengths were introduced into the probe cavity directly throughoptical fibers via a continuous-wave laser.KPFM Measurement: The KPFM was based on the Bruker by a con-ducting AFM tip coated with Pt/Ir under ambient conditions. Before per-forming real measurements of the samples, the instrument was calibratedusing an Au electrode deposited on a SiO2/Si substrate. The KPFM mea-surements of the sample (graphene/CrOCl heterojunction) performed inthe dark and under 520 nm laser illumination used the same settings with-out lifting the tip or changing any parameters. Based on the levels of workfunctions in graphene and CrOCl, the band diagrams of this heterostruc-ture in the dark and under 520 nm laser illumination were estimated anddepicted.First-Principles Calculation: The density functional theory (DFT) calcu-lations were performed by using the Vienna ab initio simulation package(VASP).[43,44] The core electrons were described with the projector aug-mented wave method. The exchange correlation potential adopts the gen-eralized gradient approximation (GGA) in the Perdew–Burke–Ernzerhofform.[45] The GGA + U approach[46,47] was adopted with a value of U =2.7 eV for the Cr-3d electron, which had been tested in previous works.[40]The heterostructure comprises Bernal (ABA) stacked trilayer graphene andmonolayer CrOCl. The lattice constant for graphene was 2.46 Å, and thelattice constants of CrOCl in the a and b directions were 3.24 and 3.93 Å,respectively. A computational supercell of the heterostructure contains a 4× 2√3× 1 supercell of trilayer graphene and a 3× 2× 1 supercell of mono-layer CrOCl, resulting in a lattice mismatch of 1.2% and 0.66 for graphenein the a and b directions, respectively. Due to the different electronega-tivity, there exists electron transfer from graphene to the CrOCl substrate.Besides, the photoelectrons generated on graphene and transferred to thesubstrate strengthen the non-equilibrium charge distribution. To capturethis point, the electronic energy bands of trilayer graphene doped with 0.1Adv. Sci. 2025, 12, e07530 e07530 (8 of 10) © 2025 The Author(s). Advanced Science published by Wiley-VCH GmbH 21983844, 2025, 48, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/advs.202507530 by Kenji Watanabe - National Institute For , Wiley Online Library on [29/12/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.advancedscience.comwww.advancedsciencenews.com www.advancedscience.comhole doping and that of CrOCl under 0.1 electron doping were calculated.The energy cutoff for the plane wave basis was set to be 500 eV, and a4 × 6 × 1 k-point mesh was used to sample the first Brillouin zone. Theconvergence thresholds of energy and force were set to be 10−6 eV and10−2 eV Å−2, respectively. A vacuum of 15 Å was introduced to avoid theinteraction between periodic images.Supporting InformationSupporting Information is available from the Wiley Online Library or fromthe author.AcknowledgementsThis work was supported by the National Key Research and DevelopmentProgram of China (No. 2022YFA1203904), Natural Science Foundation ofChina grants (No. 62375018), the Beijing Natural Science Foundation ofChina (No. L233003), the funding Program of Beijing Institute of Technol-ogy (grant No. 3180021502353), and Beijing Institute of Technology Re-search Fund Program for Young Scholars. The authors acknowledged theAnalysis & Testing Center in Beijing Institute of Technology.Conflict of InterestThe authors declare no conflict of interest.Author ContributionsD.G conducted the experimental test and wrote the manuscript. W.L. andF.Z. conducted the theoretical calculations. P.G. and Y.Y. synthesized CrOClsamples. W.D. finished the TEM test. X.R. assisted in the test of all opticalsynapses. K.W. and T.T. provided hBN samples. S.Z. and J. Z. supervisedthe project. All authors participated in discussions and contributed to themanuscript.Data Availability StatementThe data that support the findings of this study are available from the cor-responding author upon reasonable request.Keywords2D materials, infrared optical synapse, interlayer coupling, orientation de-tection, van der Waals heterostructureReceived: April 25, 2025Revised: August 22, 2025Published online: September 30, 2025[1] C. Eckel, J. Lenz, A. Melianas, A. Salleo, R. T. Weitz, Nano Lett. 2022,22, 973.[2] B. J. Shastri, A. N. Tait, T. Ferreira de Lima, W. H. P. Pernice,H. Bhaskaran, C. D. Wright, P. R. Prucnal, P. A. Intelligence, N.Computing, Nat. Photon. 2021, 15, 102.[3] Y. Lee, H.-L. Park, Y. Kim, T.-W. Lee, Joule 2021, 5, 794.[4] R. Yang, H.-M. Huang, Q.-H. Hong, X.-B. Yin, Z.-H. Tan, T. Shi, Y.-X.Zhou, X.-S. Miao, X.-P. Wang, S.-B. Mi, C.-L. Jia, X. Guo, Adv. Funct.Mater. 2018, 28, 1704455.[5] J.-X. Shen, D.-S. Shang, Y.-S. Chai, S.-G. Wang, B.-G. Shen, Y. Sun,Adv. Mater. 2018, 30, 1706717.[6] N. Raeis-Hosseini, Y. Park, J.-S. Lee, Adv. Funct. Mater. 2018, 28,1800553.[7] X. Mou, J. Tang, Y. Lyu, Q. Zhang, S. Yang, F. Xu, W. Liu, M. Xu, Y.Zhou, W. Sun, Y. Zhong, B. Gao, P. Yu, H. Qian, H. Wu, Sci. Adv. 2021,7, abh0648.[8] Y. Liu, Y. Wu, H. Han, Y. Wang, R. Peng, K. Liu, D. Yi, C. W. Nan, J.Ma, Adv. Funct. Mater. 2023, 34, 2306945.[9] N. Ilyas, J. Wang, C. Li, D. Li, H. Fu, D. Gu, X. Jiang, F. Liu, Y. Jiang,W. Li, Adv. Funct. Mater. 2022, 32, 2110976.[10] H.-L. Park, H. Kim, D. Lim, H. Zhou, Y.-H. Kim, Y. Lee, S. Park, T.-W.Lee, Adv. Mater. 2020, 32, 1906899.[11] P.-Y. Huang, B.-Y. Jiang, H.-J. Chen, J.-Y. Xu, K. Wang, C.-Y. Zhu, X.-Y.Hu, D. Li, L. Zhen, F.-C. Zhou, J.-K. Qin, C.-Y. Xu,Nat. Commun. 2023,14, 6736.[12] K. Liu, T. Zhang, B. Dang, L. Bao, L. Xu, C. Cheng, Z. Yang, R. Huang,Y. Yang, Nat. Electron. 2022, 5, 761.[13] S. Wang, C. Chen, Z. Yu, Y. He, X. Chen, Q. Wan, Y. Shi,D. Zhang, H. Zhou, X. Wang, P. Zhou, Adv. Mater. 2018, 31,1806227.[14] J. Zha, Y. Xia, S. Shi, H. Huang, S. Li, C. Qian, H. Wang, P. Yang, Z.Zhang, Y. Meng, W. Wang, Z. Yang, H. Yu, J. C. Ho, Z. Wang, C. Tan,Adv. Mater. 2024, 36, 2308502.[15] T. Zhang, C. Fan, L. Hu, F. Zhuge, X. Pan, Z. Ye, ACS Nano 2024, 18,16236.[16] J. Zeng, G. Feng, G. Wu, J. Liu, Q. Zhao, H. Wang, S. Wu, X. Wang,Y. Chen, S. Han, B. Tian, C. Duan, T. Lin, J. Ge, H. Shen, X. Meng, J.Chu, J. Wang, Adv. Funct. Mater. 2024, 34, 2313010.[17] J.-Q. Yang, R. Wang, Y. Ren, J.-Y. Mao, Z.-P. Wang, Y. Zhou, S.-T. Han,Adv. Mater. 2020, 32, 2003610.[18] A. Ziletti, A. Carvalho, D. K. Campbell, D. F. Coker, A. H. Castro Neto,Phys. Rev. Lett. 2015, 114, 046801.[19] C. Liu, Y. Chang, T. B. Norris, Z. Zhong, Nat. Nanotechnol. 2014, 9,273.[20] W. Liu, J. Lv, L. Peng, H. Guo, C. Liu, Y. Liu, W. Li, L. Li, L. Liu, P. Wang,S. C. Bodepudi, K. Shehzad, G. Hu, K. Liu, Z. Sun, T. Hasan, Y. Xu, X.Wang, C. Gao, B. Yu, X. Duan, Nat. Electron. 2022, 5, 281.[21] M. W. Graham, S.-F. Shi, D. C. Ralph, J. Park, P. L. McEuen,Nat. Phys.2013, 9, 103.[22] M. Freitag, T. Low, F. N. Xia, P. Avouris, Nat. Photon. 2013, 7, 53.[23] Y. Tang, Y. Ping, X. Yang, J. Xing, J. Chen, X. Wang, J. Lu, H.Jing, K. Liu, J. Wu, X. Zhou, T. Zhai, H. Xu, Adv. Mater. 2025, 37,2405358.[24] Y. Guo, J. Li, X. Zhan, C. Wang, M. Li, B. Zhang, Z. Wang, Y. Liu,K. Yang, H. Wang, W. Li, P. Gu, Z. Luo, Y. Liu, P. Liu, B. Chen, K.Watanabe, T. Taniguchi, X.-Q. Chen, C. Qin, J. Chen, D. Sun, J. Zhang,R. Wang, J. Liu, Y. Ye, X. Li, Y. Hou, W. Zhou, H. Wang, et al., Nature2024, 630, 346.[25] K. Yang, X. Gao, Y. Wang, T. Zhang, Y. Gao, X. Lu, S. Zhang, J. Liu,P. Gu, Z. Luo, R. Zheng, S. Cao, H. Wang, X. Sun, K. Watanabe, T.Taniguchi, X. Li, J. Zhang, X. Dai, J.-H. Chen, Y. Ye, Z. Han,Nat. Com-mun. 2023, 14, 2136.[26] B. Ying, B. Xin, M. Li, S. Zhou, Q. Liu, Z. Zhu, S. Qin, W.-H. Wang, M.Zhu, ACS Appl. Mater. Interfaces 2024, 16, 43806.[27] F. M. Silva, R. A. Guerra-Fuentes, D. C. Blackburn, A. L. C. Prudente,Dev. Dyn. 2024, 253, 606.[28] R. C. Goris, J. Herpetol. 2011, 45, 2.[29] E. O. Gracheva, N. T. Ingolia, Y. M. Kelly, J. F. Cordero-Morales, G.Hollopeter, A. T. Chesler, E. E. Sánchez, J. C. Perez, J. S. Weissman,D. Julius, Nature 2010, 464, 1006.[30] E. O. Gracheva, J. F. Cordero-Morales, J. A. González-Carcacía, N. T.Ingolia, C. Manno, C. I. Aranguren, J. S. Weissman, D. Julius, Nature2011, 476, 88.Adv. Sci. 2025, 12, e07530 e07530 (9 of 10) © 2025 The Author(s). Advanced Science published by Wiley-VCH GmbH 21983844, 2025, 48, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/advs.202507530 by Kenji Watanabe - National Institute For , Wiley Online Library on [29/12/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.advancedscience.comwww.advancedsciencenews.com www.advancedscience.com[31] X. Zheng, Y. Wei, Z. Wei, W. Luo, X. Guo, X. Zhang, J. Liu, Y. Chen, G.Peng, W. Cai, S. Qin, H. Huang, C. Deng, X. Zhang, Nano Res. 2022,15, 9377.[32] P. Gu, C. Wang, D. Su, Z. Dong, Q. Wang, Z. Han, K. Watanabe, T.Taniguchi, W. Ji, Y. Sun, Y. Ye, Nat. Commun. 2023, 14, 3221.[33] T. Zhang, Y. Wang, H. Li, F. Zhong, J. Shi, M. Wu, Z. Sun, W. Shen, B.Wei, W. Hu, X. Liu, L. Huang, C. Hu, Z. Wang, C. Jiang, S. Yang, Q.-M.Zhang, Z. Qu, ACS Nano 2019, 13, 11353.[34] P. Gu, Y. Sun, C. Wang, Y. Peng, Y. Zhu, X. Cheng, K. Yuan, C. Lyu, X.Liu, Q. Tan, Q. Zhang, L. Gu, Z. Wang, H.Wang, Z. Han, K. Watanabe,T. Taniguchi, J. Yang, J. Zhang, W. Ji, P.-H. Tan, Y. Ye, Nano Lett. 2022,22, 1233.[35] S. Oeda, S. Hasegawa, Procedia Comput. Sci. 2023, 225, 1191.[36] W. Wang, Y. Wang, F. Yin, H. Niu, Y.-K. Shin, Y. Li, E.-S. Kim, N.-Y. Kim,Nano-Micro Lett. 2024, 16, 133.[37] Y. Lian, J. Han, M. Yang, S. Peng, C. Zhang, C. Han, X. Zhang, X. C.Liu, H. Zhou, Y. Wang, C. Lan, J. Gou, Y. Jiang, Y. Liao, H. Yu, J. Wang,Adv. Funct. Mater. 2022, 32, 2205709.[38] Y.-X. Hou, Y. Li, Z.-C. Zhang, J.-Q. Li, D.-H. Qi, X.-D. Chen, J.-J. Wang,B.-W. Yao, M.-X. Yu, T.-B. Lu, J. Zhang, ACS Nano 2021, 15, 1497.[39] J. Wei, Y. Li, L. Wang, W. Liao, B. Dong, C. Xu, C. Zhu, K.-W. Ang, C.-W.Qiu, C. Lee, Nat. Commun. 2020, 11, 6404.[40] Y. Wang, X. Gao, K. Yang, P. Gu, X. Lu, S. Zhang, Y. Gao, N. Ren, B.Dong, Y. Jiang, K. Watanabe, T. Taniguchi, J. Kang, W. Lou, J. Mao, J.Liu, Y. Ye, Z. Han, K. Chang, J. Zhang, Z. Zhang, Nat. Nanotechnol.2022, 17, 1272.[41] Z. Zhang, X. Zhao, X. Zhang, X. Hou, X. Ma, S. Tang, Y. Zhang, G. Xu,Q. Liu, S. Long, Nat. Commun. 2022, 13, 6590.[42] J. Lao, M. Yan, B. Tian, C. Jiang, C. Luo, Z. Xie, Q. Zhu, Z. Bao, N.Zhong, X. Tang, L. Sun, G. Wu, J. Wang, H. Peng, J. Chu, C. Duan,Adv. Sci. 2022, 9, 2106092.[43] G. Kresse, J. Furthmüller, Comput. Mater. Sci. 1996, 6, 15.[44] G. Kresse, J. Furthmüller, Phys. Rev. B 1996, 54, 11169.[45] J. P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 1996, 77, 3865.[46] V. I. Anisimov, I. V. Solovyev, M. A. Korotin, M. T. Czyżyk, G. A.Sawatzky, Phys. Rev. B 1993, 48, 16929.[47] V. I. Anisimov, J. Zaanen, O. K. Andersen, Phys. Rev. B 1991, 44, 943.[48] T. Ahmed, S. Kuriakose, E. L. H. Mayes, R. Ramanathan, V.Bansal, M. Bhaskaran, S. Sriram, S. Walia, Small 2019, 15,1900966.[49] C.-M. Yang, T.-C. Chen, D. Verma, L.-J. Li, B. Liu, W.-H. Chang, C.-S.Lai, Adv. Funct. Mater. 2020, 30, 2001598.[50] F. Guo, M. Song, M.-C. Wong, R. Ding, W. Io, S.-Y. Pang, W. Jie, J. Hao,Adv. Funct. Mater. 2022, 32, 2108014.Adv. Sci. 2025, 12, e07530 e07530 (10 of 10) © 2025 The Author(s). Advanced Science published by Wiley-VCH GmbH 21983844, 2025, 48, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/advs.202507530 by Kenji Watanabe - National Institute For , Wiley Online Library on [29/12/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.advancedscience.com A Van Der Waals Broadband Infrared Optical Synapse Enabling Orientation Detection 1. Introduction 2. Results and Discussion 2.1. Biomimetic IR Perception of Snake 2.2. IR Synaptic Plasticity of the Tunneling Device 2.3. Broadband Synaptic Response from VIS to IR 2.4. Mechanisms for the Broadband Synaptic Response 2.5. IR Orientation Detection 3. Conclusion 4. Experimental Section Supporting Information Acknowledgements Conflict of Interest Author Contributions Data Availability Statement Keywords