# Fileset

[d5tc01712b (1).pdf](https://mdr.nims.go.jp/filesets/a5f1954f-3fe1-4b60-a739-bc87fc4b2b61/download)

## Creator

[Ryoma Hayakawa](https://orcid.org/0000-0002-1442-8230), Yuho Yamamoto, Kosuke Yoshikawa, Yoichi Yamada, [Yutaka Wakayama](https://orcid.org/0000-0002-0801-8884)

## Rights



## Other metadata

[Reconfigurable artificial synapses with an organic antiambipolar transistor for brain-inspired computing](https://mdr.nims.go.jp/datasets/5c636212-5cdf-4075-9bd7-dc3eca9f03da)

## Fulltext

Reconfigurable artificial synapses with an organic antiambipolar transistor for brain-inspired computing14234 |  J. Mater. Chem. C, 2025, 13, 14234–14241 This journal is © The Royal Society of Chemistry 2025Cite this: J. Mater. Chem. C, 2025,13, 14234Reconfigurable artificial synapses with an organicantiambipolar transistor for brain-inspiredcomputing†Ryoma Hayakawa, *a Yuho Yamamoto,ab Kosuke Yoshikawa,ab Yoichi Yamada band Yutaka Wakayama *aNeuromorphic computing, a nonvon Neumann architecture, holds promise for low-power, high-efficiency data processing. Herein, we demonstrated reconfigurable artificial synapses using a floating-gate-type organic antiambipolar transistor (FG-OAAT) to mimic biological synapses. The FG-OAATexhibited a L-shaped transfer curve with negative differential transconductance. A two-dimensionalcontinuous Au film was used as the floating gate to induce a large peak voltage shift in the L-shapedtransfer curve by controlling hole- and electron-trapping processes in the floating gate. This featureenabled reconfigurable synaptic operations. Long-term potentiation/depression, excitatory/inhibitory,and paired-pulse facilitation/depression functions were electrically reconfigured by tuning the chargeconditions in the floating gate. These versatile synaptic operations were induced by a consistent pre-synaptic signal, with fixed polarity, applied voltage, and pulse width. These behaviors closely resembledthose of biological synapses, highlighting the potential for a brain-like computing architecture thatsurpasses current von Neumann systems.1. IntroductionThe development of high-performance artificial intelligence(AI) is crucial for the upcoming Internet of Everything societyowing to the rapid increase in data volumes from versatileinformation and communication systems including automo-bile systems, healthcare sensors, and industrial robots.1–6However, current AI systems based on the von Neumann archi-tecture exhibit high power consumption.7,8 For instance, thepower consumption of supercomputer ‘‘K’’ reaches 10 MW.9This issue arises from the increased access frequency betweenlogic and memory units, known as the von Neumann bottle-neck. Therefore, the evolution of energy-efficient AI systems isan essential challenge in the big data era.Neuromorphic computing systems, a type of non von Neu-mann architecture, offer a solution for energy-efficient AI.10–12These systems mimic the brain, which operates with ultralowpower consumption (20 W) through parallel data processing.13Neuromorphic devices integrate logic and memory units, withnonvolatile memories such as magnetic random-access mem-ory and ferroelectric memory widely used.14–17 This deviceconfiguration enables extremely low-power consumption andhigh-speed parallel data processing. In addition, the adoptionof pulse-based analog operations, such as spiking neural net-works, is expected to drive the evolution of highly energy-efficient AI systems.18–20Organic transistors with nonvolatile memories are widelyemployed in neuromorphic applications because of their attrac-tive features, including light weight, flexibility, and biocom-patibility.21–24 The recognition of complex images and voicepatterns has already been demonstrated.25,26 However, conven-tional organic transistors usually support only synaptic opera-tions for an input signal because they exhibit unipolar carriertransport.21–24 Meanwhile, biological synapses exhibit diverseresponses to identical input signals, with synaptic operationsreconfigured by neuromodulatory commands.27–30 Accordingly,the development of reconfigurable artificial synapses is a keychallenge in the construction of brain-inspired AI systems.To address this, we demonstrate reconfigurable neuromorphicoperations based on a floating-gate-type organic antiambipolartransistor (FG-OAAT).An OAAT is a heterojunction transistor with at least one p–njunction in the transistor channel, which induces negativedifferential transconductance and produces a L-shaped trans-fer curve.31–37 The drain current increases and then decreasesa Research Center for Materials Nanoarchitectonics (MANA), National Institute forMaterials Science (NIMS), 1-1 Namiki, Tsukuba 305-0044, Japan.E-mail: WAKAYAMA.Yutaka@nims.go.jp; Tel: +81-29-860-4403b Faculty of Pure and Applied Sciences, University of Tsukuba 1-1-1 Tennodai,Tsukuba 305-8573, Japan† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d5tc01712bReceived 28th April 2025,Accepted 10th June 2025DOI: 10.1039/d5tc01712brsc.li/materials-cJournal ofMaterials Chemistry CPAPEROpen Access Article. Published on 12 June 2025. Downloaded on 10/20/2025 10:53:01 AM.  This article is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported Licence.View Article OnlineView Journal  | View Issuehttps://orcid.org/0000-0002-1442-8230https://orcid.org/0000-0001-8187-3409https://orcid.org/0000-0002-0801-8884http://crossmark.crossref.org/dialog/?doi=10.1039/d5tc01712b&domain=pdf&date_stamp=2025-06-17https://doi.org/10.1039/d5tc01712bhttps://doi.org/10.1039/d5tc01712bhttps://rsc.li/materials-chttp://creativecommons.org/licenses/by-nc/3.0/http://creativecommons.org/licenses/by-nc/3.0/https://doi.org/10.1039/d5tc01712bhttps://pubs.rsc.org/en/journals/journal/TChttps://pubs.rsc.org/en/journals/journal/TC?issueid=TC013028This journal is © The Royal Society of Chemistry 2025 J. Mater. Chem. C, 2025, 13, 14234–14241 |  14235with increasing gate voltage. This unique carrier transport inOAATs has enabled the creation of logic circuits, includingternary and quaternary inverters, ternary logic-in-memory, andreconfigurable two-input logic circuits.38–43Herein, we applied OAATs to a reconfigurable artificialsynapse, using 2,7-dioctyl[1]benzothieno[3,2-b][1]benzothiophene(C8-BTBT) and PhC2H4-benzo[de]isoquinolino[1,8-gh]quino-lone diimide (PhC2-BQQDI) films as the p-type and n-typetransistor channels, respectively. Notably, a Au film was intro-duced as the floating gate (FG) to induce a large peak voltageshift in the L-shaped transfer curve by controlling the hole- andelectron-trapping processes in the FG. This feature enabledunique synaptic operations, with long-term potentiation (LTP)/depression (LTD), excitatory/inhibitory, and paired-pulse facil-itation (PPF)/depression (PPD) electrically reconfigured byadjusting the charge conditions of the Au FG. These findingssuggest the potential to realize a new computing architecturebeyond the current von Neumann computing.2. Experimental methods2.1. Formation of FG-OAATsFG-OAATs with C8-BTBT and PhC2-BQQDI were fabricated ona highly doped p-type Si(100) substrate (o 0.01 O cm) with a200 nm-thick SiO2 layer. First, a Au (30 nm)/Cr (5 nm) elec-trode was deposited as a back gate (BG) on the SiO2/Sisubstrate via thermal vacuum deposition. Subsequently, aHfO2 film (35 nm) was prepared as the gate insulating layerusing atomic layer deposition (ALD) (SUGA Co. Ltd., SAL1000), with the substrate temperature set at 175 1C. Here,tetrakis(dimethylamino)hafnium and water were used as thehafnium and oxygen sources, respectively. Afterward, an Au(10 nm)/Cr (5 nm) film was thermally grown on the HfO2 layeras the FG, with a deposition rate of 5 Å s�1 to ensure a smoothsurface morphology of the Au thin film. Then, a 10 nm-thickHfO2 tunneling layer was formed via ALD, following the sameprocess as the gate insulating layer. To passivate carrier trapsites on the HfO2 surface, a 10 nm-thick polystyrene (PS)(Sigma-Aldrich, 182427, average molecular weight: 280 000)layer was spin-coated on the HfO2 surface. Subsequently,C8-BTBT (13 � 3 nm) and PhC2-BQQDI (8 � 2 nm) films weregrown as p- and n-type organic channels, respectively, via thermalvacuum deposition at a background pressure of 10�7 Pa. Finally,30 nm-thick Au films were deposited for the source (S) anddrain (D) electrodes via thermal vacuum deposition, withtypical width and length dimensions of 400 and 200 mm,respectively.2.2 Transistor measurementTransistor measurements of the FG-OAATs and neuromorphicoperations were performed using a source measure unit(Keysight Technology, B2912B) under atmospheric conditions.All the measurements were conducted with a four-probe systemat room temperature.3. Results and discussionFig. 1a and b show the device structure and optical microscopeimage of the FG-OAAT. The transistor features a BG-typestructure, with 14 nm-thick C8-BTBT and 8 nm-thick PhC2-BQQDI films as the p- and n-type transistor channels, respec-tively. The detailed optimization processes of FG-OAATs aredescribed in ESI† (Fig. S1–S5). Fig. 1c and d exhibit a typicaldrain current (ID)–BG voltage (VBG) curve and the differentialtransconductance (dID/dVBG) curve for the n-type operation ofthe FG-OAAT, with the D voltage (VD) fixed at 3.0 V. Thetransistor exhibited a L-shaped transfer curve (Fig. 1c), withno hysteresis appearing in the forward and reverse VBG sweeps(black and red solid lines, respectively, in Fig. 1c), indicating nocarrier trapping in the Au FG within the applied voltage range.ID started to increase at VBG (Von) = 0.2 V, reaching 126 nA at VBG(Vpeak) = 1.8 V. Thereafter, ID was completely suppressed at VBG(Voff) = 2.6 V in the forward VBG sweep from 0 to 3.0 V (blacksolid line in Fig. 1c). A distinct negative differential transcon-ductance was observed between 1.9 and 2.6 V in the forwardVBG sweep of up to 3.0 V (Fig. 1d). Additionally, it is noteworthythat the low voltage operation of below 3.0 V was anotheressential achievement, which is induced by the employmentof HfO2 layer as the gate insulating layer. Usually, operationvoltages of more than 5.0 V are required for OAATs.31–38In our previous study, the L-shaped transfer curves inOAATs were interpreted analogously to the shoot-through cur-rent in complementary metal-oxide-semiconductor inverters.Namely, ID of FG-OAATs is the overlapped current of theconstituent n-type and p-type transistors described using thefollowing equations:44–47ID;n ¼W2LnmnCi VBG � Vth;n� �2VBG � Vpeak� �(1)ID;p ¼W2LpmpCi VD � VBG þ Vth;p� �2VBG 4Vpeak� �(2)where Ln and Lp are the channel lengths of the PhC2-BQQDIand C8-BTBT channels, respectively, W is the common channelwidth, and Ci is the capacitance of the gate insulator per unitarea. Additionally, mn and Vth,n are the electron mobility andthreshold voltage of the PhC2-BQQDI transistor, respectively.Similarly, mp and Vth,p represent the hole mobility and thethreshold voltage of the C8-BTBT transistor, respectively.Accordingly, Von and Voff in Fig. 1c correspond to the thresholdvoltages of the PhC2-BQQDI and C8-BTBT transistors, respec-tively, while Vpeak is the value of VBG at the intersection of thetransfer curves of both transistors.Based on the above argument, the carrier transport processin FG-OAATs can be explained using the illustrations shown inFig. 1e and f. In VBG o Von region (Fig. 1e), no electron currentflows because VBG is below the Vth,n of the PhC2-BQQDIchannel. However, holes accumulate in the C8-BTBT channelowing to the applied effective gate voltage (VD–VBG). However,the hole current is suppressed by the potential barrier at thep–n junction, resulting in no ID in the VBG range. In VBG 4 VonPaper Journal of Materials Chemistry COpen Access Article. Published on 12 June 2025. Downloaded on 10/20/2025 10:53:01 AM.  This article is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported Licence.View Article Onlinehttp://creativecommons.org/licenses/by-nc/3.0/http://creativecommons.org/licenses/by-nc/3.0/https://doi.org/10.1039/d5tc01712b14236 |  J. Mater. Chem. C, 2025, 13, 14234–14241 This journal is © The Royal Society of Chemistry 2025(Fig. 1f), electrons are introduced from the S electrode and flowtoward the D electrode. Simultaneously, the accumulated holesin the C8-BTBT channel begin to flow toward the S electrode,generating ID in FG-OAATs. A further increase in VBG (VBG 4 VoffFig. 1g) hinders ID because the C8-BTBT channel enters theoff state.Next, the L-shaped transfer curve in the FG-OAAT can becontrolled using the Au FG. Fig. 2a–c illustrate the processes oferasing, electron trapping, and hole trapping. The corres-ponding ID–VBG curves are depicted in Fig. 2d, where the solidand dotted lines at each state represent the forward and reverseVBG sweeps, respectively. First, the Au FG was grounded to erasethe carriers (electrons or holes) in the Au FG, which is definedas the erasing process (Fig. 2a). The resulting ID–VBG curve isexhibited by the black solid and dotted lines in Fig. 2d. Vpeakwas estimated at 2.1 V. Then, VBG = 5.0 V was applied for 10 s totrap electrons in the Au FG (Fig. 2b), where the S and Delectrodes were grounded. This operation shifted the ID–VBGcurve higher VG, as shown by the blue solid and dotted lines inFig. 2d. Vpeak shifted from 2.1 to 3.1 V. Importantly, no hyster-esis was observed in the shifted ID–VBG curve, revealing that thetrapped electrons in the Au FG were retained during the VBGsweeps. After the erasing process (Fig. 2a), the opposite BGvoltage, VBG = �5.0 V, was applied for 10 s to trap holes in theAu FG (Fig. 2c). As a result, Vpeak shifted to 1.0 V (red solid anddotted line in Fig. 2d). No hysteresis appeared in the ID–VBGcurve, similar to the electron-trapped sate. Consequently, thetotal variation in Vpeak reached 2.1 V by controlling the chargeconditions (holes or electrons) in the Au FG. This value is muchlarger than that observed in our previous study with a zinc-phthalocyanine-core star-shaped polymer (ZnPc-PS4) as theFG.43,48 This improvement in Vpeak shift in this study benefitsfrom using the Au FG.Fig. 2e shows the retention property of electron- and hole-trapped states, where the potential of the Au FG was moni-tored. The trapped electrons and holes were retained forat least 1800 s. The switching behavior between the erasedand electron-trapped states of the FG-OAATs is depictedFig. 1 (a) Device structure and (b) optical microscope image of the FG-OAAT, utilizing PhC2-BQQDI and C8-BTBT films as transistor channels, with anAu film serving as the floating gate. (c) ID–VBG and (d) differential transconductance curve of the FG-OAAT measured at VD = 3.0 V. (e) Illustrations ofcarrier transport in the FG-OAAT across VBG ranges: (a) VBG o Von, (b) Von o VBG o Voff, and (c) Voff o VBG.Journal of Materials Chemistry C PaperOpen Access Article. Published on 12 June 2025. Downloaded on 10/20/2025 10:53:01 AM.  This article is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported Licence.View Article Onlinehttp://creativecommons.org/licenses/by-nc/3.0/http://creativecommons.org/licenses/by-nc/3.0/https://doi.org/10.1039/d5tc01712bThis journal is © The Royal Society of Chemistry 2025 J. Mater. Chem. C, 2025, 13, 14234–14241 |  14237in Fig. 2f and g. After 10 switching cycles, the Vpeak positionsafter the electron-trapping and releasing processes werealmost identical.Based on these nonvolatile memory properties, we appliedthe FG-OAAT to a neuromorphic device. For synaptic operationswith the FG-OAAT, the BG and D electrodes function asthe presynaptic input and postsynaptic output terminals,respectively. ID was monitored as the postsynaptic current(PSC). Fig. 3a shows ID–VBG for the erased (black solid line),electron-trapped (blue solid line), and hole-trapped (red solidline) states. As shown in Fig. 3b–d, a variety of synapticoperations were reconfigured by adjusting the initial chargecondition in the Au FG. First, Fig. 3b illustrates the transitionfrom LTP to LTD with the FG-OAAT. Prior to this measurement,the initial ID–VBG curve was set to the electron-trapped state(blue solid line in Fig. 3a) by applying a VBG of 5.0 V for 10 s.PSC (ID,read) was monitored at a VG (VG,read) of 1.8 V and VD(VD,read) of 3.0 V. When negative VBG pulses (VBG = �5.0 V, plusewidth (Pwidth) = 100 ms) were applied, ID,read started increasingand was maintained even after the VBG pulses were turned off,indicating the LTP operation. Thereafter, ID,read decreased withidentical positive VBG pulses (VBG = 5.0 V, Pwidth = 100 ms),corresponding to the LTD operation. Next, the opposite transi-tion from LTD to LTP was observed as shown in Fig. 3c evenwith identical sequence of presynaptic pulses (VBG = �5.0 V,Pwidth = 100 ms and VBG = 5.0 V, Pwidth = 100 ms) applied to thetransistor. The difference of Fig. 3b and c is the initial chargeconditions in the Au FG.In Fig. 3c, the initial ID–VBG curve was set to the erased state(black solid line in Fig. 3a). As a result, the transition from LTDto LTP was induced by applying the same sequence of VBGpulses with the negative-to-positive polarity change. Finally,Fig. 2 (a) Erasing, (b) electron-trapping, and (c) hole-trapping processes of the FG-OAATs. (d) ID–VBG curves for the erased (black line), electron-trapped(blue line), and hole-trapped (red line) states, with solid and dotted lines representing forward and reverse VBG sweeps, respectively. (e) Retentionproperty of the FG-OAAT. (f) Switching behavior of the FG-OAAT. (g) Variation in Vpeak values in electron-trapped and erased states as a function ofswitching cycles.Paper Journal of Materials Chemistry COpen Access Article. Published on 12 June 2025. Downloaded on 10/20/2025 10:53:01 AM.  This article is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported Licence.View Article Onlinehttp://creativecommons.org/licenses/by-nc/3.0/http://creativecommons.org/licenses/by-nc/3.0/https://doi.org/10.1039/d5tc01712b14238 |  J. Mater. Chem. C, 2025, 13, 14234–14241 This journal is © The Royal Society of Chemistry 2025Fig. 3d shows that the transition from LTP to LTD can beinduced by applying only negative VBG pulses (VBG = �5.0,Pwidth = 100 ms) continuously. (VBG = �5 V, Pwidth = 100 ms)(Fig. 3d). Here, the initial ID–VBG curve was set to the electron-trapped state (blue solid line in Fig. 3a).The abovementioned results clearly reveal that reconfigur-able synaptic operations, transitioning from LTP to LTD andvice versa, can be achieved by controlling the initial chargeconditions of the Au FG. Such synaptic plasticity has typicallynot been demonstrated in conventional transistors, whichexhibit unipolar carrier transport.21–26 By contrast, biologicalsynapses are known to display different responses to synapticplasticity under neuromodulatory control, even when the samepolarity, applied voltage, and pulse width are applied.27–30 Thissuggests that our proposed transistor has the potential to trulymimic biological synapses. Notably, the continuous VBG pulseapplications in Fig. 3d enabled a smooth transition from LTP toLTD, while a sharp drop in ID,read was observed in Fig. 3b whenthe VBG pulse polarity was reversed. The nonlinearity coeffi-cients and asymmetric factors of LTP/LTD behaviors were alsoimproved by the continuous VBG pulse applications (Fig. S6 inESI†). This improvement offers an additional advantage ofusing our proposed transistor.In a similar manner to the LTP/LTD behaviors, excitatory/inhibitory and PPF/PPD operations were electrically reconfi-gured, as shown in Fig. 4. Fig. 4a shows the ID–VBG curvesbefore (black dotted line) and after (black solid line) theapplication of a negative VBG pulse (VBG = �5 V for 100 ms),where no carriers were accumulated in the Au FG in the initialstate (black dotted line). The corresponding variation in ID,readis depicted in Fig. 4b, where ID,read was monitored at VG,read =1.2 V and VD,read = 3.0 V. A sharp increase in ID,read from 25 to40 nA was observed by the application of a negative VBG pulse(VBG = �5 V for 100 ms). Importantly, ID,read was maintained at30 nA after the VBG pulse was turned off, due to the thresholdvoltage shift induced by the hole-trapping process in the Au FG.This variation in ID,read corresponds to the excitatory synapticoperation. The synaptic plasticity was changed by the presy-naptic input pulse. Moreover, the PPF behavior distinctlyappeared with the application of double-negative VBG pulses(VBG = �5 V for 100 ms) (Fig. 4c). The FFP ratio (ID,read,change)was calculated using the following equation:ID;read;change ¼I2;read � I1;read� �I1;read� 100ð%Þ (3)where I1,read and I2,read represent the ID,read values after the firstand second VBG pulses, respectively. The change in ID,read wasenhanced by shortening the pulse intervals (Pinterval). The FFPratio increased from 8.0% at Pinterval = 3.5 s to 12.0% at Pinterval =0.5 s. Moreover, the FFP curve was well fitted by doubleexponential equations (Fig. S7a in ESI†), exhibiting a similarresponse of biological synapses.Strikingly, the opposite synaptic behaviors, namely inhibi-tory and PPD operations, are shown in Fig. 4d–f, even thoughthe same VBG pulses as in Fig. 4a–c were applied to thetransistor. First, the ID–VBG curve was set to the red dotted linein Fig. 4d, where holes were trapped in the Au FG. Then, theapplication of a negative VBG pulse (VBG = �5 V for 100 ms)shifted the transfer curve from the red dotted line to the solidline (Fig. 4d), leading to a reduction in ID,read, as shown inFig. 4e. Similarly, PPD was obtained by applying the samedouble-negative VBG pulses as in Fig. 4c. The PPD ratio wasFig. 3 (a) ID–VBG curves at the erased (black line), electron-trapped (blue line), and hole-trapped (red line) states. (b) LTP and LTD operations as a function ofpresynaptic pulse numbers, with the electron-trapped state defined as the initial state. (c) LTD and LTP as a function of presynaptic pulse numbers, following anerasing operation. I.e., no carriers were trapped in the Au FG. (c) Transition from LTP to LTD induced by continuous presynaptic pulses.Journal of Materials Chemistry C PaperOpen Access Article. Published on 12 June 2025. Downloaded on 10/20/2025 10:53:01 AM.  This article is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported Licence.View Article Onlinehttp://creativecommons.org/licenses/by-nc/3.0/http://creativecommons.org/licenses/by-nc/3.0/https://doi.org/10.1039/d5tc01712bThis journal is © The Royal Society of Chemistry 2025 J. Mater. Chem. C, 2025, 13, 14234–14241 |  14239varied from �2.5% at Pinterval = 3.5 s to �28.0% at Pinterval = 0.5 swith a reduction in pulse intervals.As shown, we realized reconfigurable synaptic operations.Namely, LTP/LTD, inhibitory/excitatory, and PPF/PPD opera-tions were electrically reconfigured by adjusting the chargingconditions of the Au FG. Such reconfigurable synaptic opera-tions have not been attained in other neuromorphic devices.Conversely, our transistor enables versatile operations with-out changing the presynaptic signals, similar to biologicalsynapses. Thus, our proposed transistors have the potentialto enable brain-like computing architectures, surpassing thelimitations of the current von Neumann model.4. ConclusionsWe achieved reconfigurable artificial synapses with FG-OAAT,using the Au film as the FG. The transistor exhibited a typicalL-shaped transfer curve. Notably, the Au FG induced a largeVpeak shift in the transfer curve by controlling hole- andelectron-trapping processes. This feature was applied to recon-figurable synaptic operations, where excitatory/inhibitory, LTP/LTD, and PPF/PPD behaviors were electrically reconfigured byadjusting the charge conditions of the Au FG. Importantly,these versatile synaptic operations were induced by identicalpresynaptic input signals, similar to those of the human brains.These findings are expected to pave the way for highly brain-like AI systems, advancing the evolution of computing systems.Author contributionsThis manuscript was written through the contributions of allauthors.Conflicts of interestThe authors declare no competing financial interest.Fig. 4 (a) ID–VBG curves before (black dotted line) and after (black solid line) a presynaptic pulse application (VBG = �5 V, Pwidth = 100 ms), with nocarriers trapping in the Au FG at the initial state. (b) Excitatory and (c) PPF operations of the FG-OAAT, with VD,read and VG,read set to 3.0 and 1.2 V,respectively; no carriers were accumulated in the Au FG at the initial state in both cases. (d) ID–VBG curves before (red dotted line) and after (red solid line)a presynaptic pulse application (VBG =�5 V, Pwidth = 100 ms), with holes trapped in the Au FG at the initial state. (e) Inhibitory and (f) PPD operations of theFG-OAAT, where holes were accumulated in the Au FG at the initial state.Paper Journal of Materials Chemistry COpen Access Article. Published on 12 June 2025. Downloaded on 10/20/2025 10:53:01 AM.  This article is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported Licence.View Article Onlinehttp://creativecommons.org/licenses/by-nc/3.0/http://creativecommons.org/licenses/by-nc/3.0/https://doi.org/10.1039/d5tc01712b14240 |  J. Mater. Chem. C, 2025, 13, 14234–14241 This journal is © The Royal Society of Chemistry 2025Data availabilityThe data that support the findings of this study are availablefrom the corresponding authors upon reasonable request.AcknowledgementsThis research was supported by Research Center for MaterialsNanoarchitectonics (MANA) of National Institute for MaterialsScience (NIMS), Tsukuba, Japan, and JSPS Kakenhi grantnumbers 19H00866, 22K18268, 23H00269 and 24K01564, theCanon Foundation, Innovative Science & Technology Initiativefor Security, and Advanced Research Infrastructure for Materi-als and Nanotechnology in Japan (ARIM) of the Ministry ofEducation, Culture, Sports, Science and Technology (MEXT)grant numbers JPMXP1223NM5170 and JPMXP1224NM5158.Notes and references1 S. Feng, H. Sun, X. Yan, H. Zhu, Z. Zou, S. Shen andH. X. Liu, Nature, 2023, 615, 620–627.2 S. Feng, X. Yan, H. Sun, Y. Feng and H. X. Liu, Nat.Commun., 2021, 12, 748.3 M. Yip, S. Salcudean, K. Goldberg, K. Althoefer, A.Menciassi, J. D. Opfermann, A. Krieger, K. Swaminathan,C. J. Walsh, H. H. Huang and I.-C. Lee, Science, 2023, 381,141–146.4 M. Chen, D. Cui, H. Haick and N. Tang, Adv. Sens. Res.,2024, 3, 2300009.5 C. Xu, S. A. Solomon and W. Gao, Nat. Mach, Intell., 2023, 5,1344–1355.6 S. Ambrogio, P. Narayanan, A. Okazaki, A. Fasoli, C. Mackin,K. Hosokawa, A. Nomura, T. Yasuda, A. Chen, A. Friz,M. Ishii, J. Luquin, Y. Kohda, N. Saulnier, K. Brew,S. Choi, I. Ok, T. Philip, V. Chan, C. Silvestre, I. Ahsan,V. Narayanan, H. Tsai and G. W. Burr, Nature, 2023, 620,768–775.7 S. Chen, Nature, 2025, 639, 22–24.8 Q. Chen, J. Wang and J. Lin, Science, 2025, 387, 587.9 K. Yamamoto, A. Uno, H. Murai1, T. Tsukamoto, F. Shoji,S. Matsui, R. Sekizawa, F. Sueyasu, H. Uchiyama,M. Okamoto, N. Ohgushi, K. Takashina, D. Wakabayashi,Y. Taguchi and M. Yokokawa, Procedia Comput. Sci., 2014,29, 576–585.10 S. Pazos, K. Zhu, M. A. Villena, O. Alharbi, W. Zheng,Y. Shen, Y. Yuan, Y. Ping and M. Lanza, Nature, 2025, 640,69–76.11 G. M. Marega, Y. Zhao, A. Avsar, Z. Wang, M. Tripathi,A. Radenovic and A. Kis, Nature, 2020, 587, 72–77.12 Y. Wang, H. Tang, Y. Xie, X. Chen, S. Ma, Z. Sun, Q. Sun,L. Chen, H. Zhu, J. Wan, Z. Xu, D. W. Zhang, P. Zhou andW. Bao, Nat. Commun., 2021, 12, 3347.13 P. A. Merolla, J. V. Arthur, R. Alvarez-Icaza, A. S. Cassidy,J. Sawada, F. Akopyan, B. L. Jackson, N. Imam, C. Guo,Y. Nakamura, B. Brezzo, I. Vo, S. K. Esser, R. Appuswamy,B. Taba, A. Amir, M. D. Flickner, W. P. Risk, R. Manohar andD. S. Modha, Science, 2014, 345, 668–673.14 L. Danial, E. Pikhay, E. Herbelin, N. Wainstein, V. Gupta,N. Wald, Y. Roizin, R. Daniel and S. Kvatinsky, Nat. Elec-tron., 2019, 2, 596–605.15 S. Jung, H. Lee, S. Myung, H. Kim, S. K. Yoon, S.-W. Kwon,Y. Ju, M. Kim, W. Yi, S. Han, B. Kwon, B. Seo, K. Lee,G.-H. Koh, K. Lee, Y. Song, C. Choi, D. Ham and S. J. Kim,Nature, 2022, 601, 211–216.16 W.-X. You, B.-K. Huang and P. Su, IEEE Trans. ElectronDevices, 2022, 69, 444–446.17 T. Ohno, T. Hasegawa, T. Tsuruoka, K. Terabe, J. K.Gimzewski and M. Aono, Nat. Mater., 2011, 10, 591–595.18 A. Stanojevic, S. Woźniak, G. Bellec, G. Cherubini, A. Pantaziand W. Gerstner, Nat. Commun., 2024, 15, 6793.19 S. S. Radhakrishnan, A. Sebastian, A. Oberoi, S. Das andS. Das, Nat. Commun., 2021, 12, 2143.20 J. B. Roldan, D. Maldonado, C. Aguilera-Pedregosa,E. Moreno, F. Aguirre, R. Romero-Zaliz, A. M. Garcı́a-Vico,Y. Shen and M. Lanza, npj 2D Mater. Appl., 2022, 6, 63.21 Y. van de Burgt, A. Melianas, S. T. Keene, G. Malliaras andA. Salleo, Nat. Electron., 2018, 1, 386–397.22 H. Tanaka, M. Akai-Kasaya, A. T. Yousefi, L. Hong, L. Fu,H. Tamukoh, D. Tanaka, T. Asai and T. Ogawa, Nat. Com-mun., 2018, 9, 2693.23 D. Kireev, S. Liu, H. Jin, T. P. Xiao, C. H. Bennett,D. Akinwande and J. A. C. Incorvia, Nat. Commun., 2022,13, 4386.24 S. Kazemzadeh, L. Dodsworth, I. F. Pereira and Y. van deBurgt, Adv. Electron. Mater., 2022, 9, 2200427.25 E. R. W. van Doremaele, P. Gkoupidenis and Y. van deBurgt, J. Mater. Chem. C, 2019, 7, 12754–12760.26 Y. Wei, Y. Liu, Q. Lin, T. Liu, S. Wang, H. Chen, C. Li,X. Gu, X. Zhang and H. Huang, Nano-Micro Lett., 2023,15, 133.27 R. D. Hawkins, T. W. Abrams, T. J. Carew and E. R. Kandel,Science, 1983, 219, 400–405.28 I. Antonov, I. Antonova, E. R. Kandel and R. D. Hawkins,Neuron, 2003, 37, 135–147.29 M. Letellier, Y. K. Park, T. E. Chater, P. H. Chipman,S. G. Gautam, T. Oshima-Takago and Y. Goda, Proc. Natl.Acad. Sci. U. S. A., 2016, 113, E2685–2694.30 C. H. Bailey, M. Giustetto, Y. Y. Huang, R. D. Hawkins andE. R. Kandel, Nat. Rev. Neurosci., 2000, 1, 11–20.31 H. Yoo, S. On, S. B. Lee, K. Cho and J.-J. Kim, Adv. Mater.,2019, 31, 1808265.32 C. Lee, J. Choi, H. Park, C. Lee, C.-H. Kim, H. Yoo andS. G. Im, Small, 2021, 17, 2103365.33 K. Kobashi, R. Hayakawa, T. Chikyow and Y. Wakayama,Adv. Electron. Mater., 2017, 3, 1700106.34 K. Kobashi, R. Hayakawa, T. Chikyow and Y. Wakayama,J. Phys. Chem. C, 2018, 122, 6943–6946.35 K. Kobashi, R. Hayakawa, T. Chikyow and Y. Wakayama,ACS Appl. Mater. Interfaces, 2018, 10, 2762–2767.36 Y. Wakayama and R. Hayakawa, Adv. Funct. Mater., 2019,30, 1903724.Journal of Materials Chemistry C PaperOpen Access Article. Published on 12 June 2025. Downloaded on 10/20/2025 10:53:01 AM.  This article is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported Licence.View Article Onlinehttp://creativecommons.org/licenses/by-nc/3.0/http://creativecommons.org/licenses/by-nc/3.0/https://doi.org/10.1039/d5tc01712bThis journal is © The Royal Society of Chemistry 2025 J. Mater. Chem. C, 2025, 13, 14234–14241 |  1424137 Y. Wakayama, C.-H. Kim, D. Panigrahi and R. Hayakawa,Mater. Adv., 2022, 3, 5260–5273.38 K. Kobashi, R. Hayakawa, T. Chikyow and Y. Wakayama,Nano Lett., 2018, 18, 4355–4359.39 D. Panigrahi, R. Hayakawa and Y. Wakayama, J. Mater.Chem. C, 2022, 10, 5559–5566.40 D. Panigrahi, R. Hayakawa, K. Honma, K. Kanai andY. Wakayama, Appl. Phys. Express, 2021, 14, 081004.41 R. Hayakawa, K. Honma, S. Nakaharai, K. Kanai andY. Wakayama, Adv. Mater., 2022, 34, 2109491.42 D. Panigrahi, R. Hayakawa, X. Zhong, J. Aimi andY. Wakayama, Nano Lett., 2023, 23, 319–352.43 R. Hayakawa, K. Takahashi, X. Zhong, K. Honma,D. Panigrahi, J. Aimi, K. Kanai and Y. Wakayama, NanoLett., 2023, 23, 8339–8347.44 R. Hayakawa, S. Takeiri, Y. Yamada, Y. Wakayama andK. Fukumoto, Adv. Mater., 2022, 34, 2201277.45 R. Hayakawa, S. Takeiri, Y. Yamada and Y. Wakayama, Adv.Mater. Interfaces, 2023, 10, 2201857.46 C.-H. Kim and H. Yoo, Adv. Electron. Mater., 2021, 7, 2100167.47 J. Zhu and T. Mori, Adv. Electron. Mater., 2022, 9, 2200783.48 Y. Shingaya, A. Zulkefli, T. Iwasaki, R. Hayakawa, S. Nakaharai,K. Watanabe, T. Taniguchi and Y. Wakayama, Adv. Electron.Mater., 2023, 9, 2200704.Paper Journal of Materials Chemistry COpen Access Article. Published on 12 June 2025. Downloaded on 10/20/2025 10:53:01 AM.  This article is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported Licence.View Article Onlinehttp://creativecommons.org/licenses/by-nc/3.0/http://creativecommons.org/licenses/by-nc/3.0/https://doi.org/10.1039/d5tc01712b