# Fileset

[s41699-024-00503-7.pdf](https://mdr.nims.go.jp/filesets/2341988b-e1fb-4f07-9ee9-c889fd173c59/download)

## Creator

Harsimran Kaur Mann, Mainak Mondal, Vivek Sah, [Kenji Watanabe](https://orcid.org/0000-0003-3701-8119), [Takashi Taniguchi](https://orcid.org/0000-0002-1467-3105), [Akshay Singh](https://orcid.org/0000-0003-1059-065X), [Aveek Bid](https://orcid.org/0000-0002-2378-7980)

## Rights

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

## Other metadata

[Optical control of multiple resistance levels in graphene for memristic applications](https://mdr.nims.go.jp/datasets/ed2140c0-418f-4f2e-b376-406a53b61a0f)

## Fulltext

Optical control of multiple resistance levels in graphene for memristic applicationsnpj | 2D materials and applications ArticlePublished in partnership with FCT NOVA with the support of E-MRShttps://doi.org/10.1038/s41699-024-00503-7Optical control of multiple resistancelevels in graphene for memristicapplicationsCheck for updatesHarsimran Kaur Mann1, Mainak Mondal1, Vivek Sah1, Kenji Watanabe 2, Takashi Taniguchi 3,Akshay Singh 1 & Aveek Bid 1Neuromorphic computing has emphasized the need for memristors with non-volatile, multipleconductance levels. This paper demonstrates thepotential of hexagonal boron nitride (hBN)/grapheneheterostructures to act asmemristors withmultiple resistance states that can be optically tuned usingvisible light. The number of resistance levels in graphene can be controlled by modulating dopinglevels, achieved by varying the electric field strength or adjusting the duration of optical illumination.Our measurements show that this photodoping of graphene results from the optical excitation ofcharge carriers from the nitrogen-vacancy levels of hBN to its conduction band, with these carriersthen being transferred to graphene by the gate-induced electric field. We develop a qualitative modelto describe our observations. Additionally, utilizing our device architecture, we propose a memristivecrossbar array for vector-matrix multiplications.Neuromorphic computing (NC), inspired by the humanbrain, has emergedas a new paradigm beyond von Neumann computing. NC provides a low-energy alternative to the traditional vonNeumann architectures, promotingsustainable computation in our modern information age. The criticalhardware building blocks for NC are memristors (artificial synapse), neuralprocessing units, and threshold switches (artificial neurons)1–5. Amemristoris a fundamental electronic component whose resistance is dependent onthe charge that has flown through it; it is a resistor with a memory. Thischarge-history dependence of resistance can be exploited in machinelearning frameworks where computation can be carried out via cross-bararrays, and synaptic weights (such as weights for neural networks) can bemodified by a certain number of electrical pulses, leading to in-memorycomputing.A low-power, non-volatilememristor is essential to exploitNC’sbenefits fully.A prerequisite for memristic action is a platform with non-volatiledoping.Chemical and electrostatic doping are the twomost commonlyusedtechniques to induce charge carriers in 2-D channels. Chemical doping isdone by hetero-atom substitution and adsorption of molecular adsorbatesonto graphene and other 2D materials. There are several drawbacks of thisapproach, including limited control over doping concentration, introduc-tion of uncontrolled structural defects and lattice strains, unintentionalimpurity introduction, challenges in maintaining stability and diffusioncontrol, and the sensitivity of the sample to processing conditions6–14.Electrostatic doping involves dopingusing an external local gate.Whilethis technique avoids most of the disadvantages of chemical doping listedabove, the magnitude of doping attainable is limited by the breakdownvoltage of the gate dielectric15–18. Liquid ionic gating has the potential ofreaching higher doping levels than is achievable by dielectric gates but at thecost of device instability and non-scalability19,20.A viable alternative is optical doping of graphene on hexagonal boronnitride (hBN) substrates21–36. hBN has a band gap of 6 eV with severalintermediate defect states that can be optically excited37–40. This methodinvolves the controlledoptical excitationof charge carriers fromdefect statesof hBN and their transfer to graphene using an external electric field. Thistechnique is reversible without adversely affecting transport mobility anddefect density.In this article, we present an in-depth study of optical doping in hBN/graphene heterostructures using visible light. Our research reveals thatelectrons in thenitrogen-vacancydefect state inhBNcanbeoptically excitedwith violet light and transferred to graphene through electrostatic gating.This technique enables high-density doping of graphene, with the dopinglevel controlled by gate voltage or illumination time. The dynamics of thisdoping process, measured with millisecond temporal resolution, can beadjusted by varying the illumination wavelength and optical power. Fur-thermore, we demonstrate that a graphene/hBN heterostructure can func-tion as a memristor, exhibiting multiple resistance levels. The device’s1Department of Physics, Indian Institute of Science, Bangalore, 560012, India. 2ResearchCenter for Electronic andOpticalMaterials, National Institute forMaterialsScience, 1-1Namiki, Tsukuba, 305-0044, Japan. 3ResearchCenter forMaterialsNanoarchitectonics, National Institute forMaterials Science, 1-1Namiki, Tsukuba,305-0044, Japan. e-mail: aksy@iisc.ac.in; aveek@iisc.ac.innpj 2D Materials and Applications |            (2024) 8:69 11234567890():,;1234567890():,;http://crossmark.crossref.org/dialog/?doi=10.1038/s41699-024-00503-7&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1038/s41699-024-00503-7&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1038/s41699-024-00503-7&domain=pdfhttp://orcid.org/0000-0003-3701-8119http://orcid.org/0000-0003-3701-8119http://orcid.org/0000-0003-3701-8119http://orcid.org/0000-0003-3701-8119http://orcid.org/0000-0003-3701-8119http://orcid.org/0000-0002-1467-3105http://orcid.org/0000-0002-1467-3105http://orcid.org/0000-0002-1467-3105http://orcid.org/0000-0002-1467-3105http://orcid.org/0000-0002-1467-3105http://orcid.org/0000-0003-1059-065Xhttp://orcid.org/0000-0003-1059-065Xhttp://orcid.org/0000-0003-1059-065Xhttp://orcid.org/0000-0003-1059-065Xhttp://orcid.org/0000-0003-1059-065Xhttp://orcid.org/0000-0002-2378-7980http://orcid.org/0000-0002-2378-7980http://orcid.org/0000-0002-2378-7980http://orcid.org/0000-0002-2378-7980http://orcid.org/0000-0002-2378-7980mailto:aksy@iisc.ac.inmailto:aveek@iisc.ac.inwww.nature.com/npj2dmaterialscompact structure and room-temperature operation enhance its potentialfor scalable, room-temperature neuromorphic applications.Results and discussionSingle-layer graphene devices encapsulated between thin hBN crystals werefabricated using the dry transfer method. 1-D electrical contacts to thegraphene were achieved by lithography and dry etching, followed by Cr/Aumetallization (Fig. 1a)41. A back-gate voltage, Vg, tuned the charge carriernumber density. Electrical transportmeasurements were performed using alow-frequencyACmeasurement technique.TheDiracpoint (maxima in thedevice resistance R) is attained atVg = 0V (Fig. 1b), solid blue line, attestingto the absence of charged impurities in the graphene channel.Optoelectronic measurementsThe optoelectronic measurements were carried out at room temperature,and the sample was illuminated by using either an LED of wavelength,λ = 427m or a Ti Sapphire pulsed laser (80MHz repetition rate, 100 fspulse width). To photo-dope the device, we use the following protocol:The gate voltage is set to a desired value V�g , and the device is exposed tothe light of wavelength 427 nmof intensity I = 62Wm−2, till the resistancesaturates to its value at theDirac point,RDP (Fig. 1c). The light is turned offat this point, and the gate response of the device is measured. It was foundthat the entireR−Vg plot shifts with theDirac point atV�g (Fig. 1b—solidred line; in this example V�g ¼ �3 V), establishing that the device is nowelectron-doped. We refer to this step as the SET protocol wherein theDirac point can be set deterministically at any desired value of Vg. ToRESET the device, it is exposed to a higher light intensity I = 458Wm−2 atV�g ¼ 0 V until the Dirac point shifts toVg = 0 V. This protocol brings theDirac point of the device back toVg = 0 V (Fig. 1b—dotted green line). Asillustrated in Fig. 1b, the process can be repeated without degradation inthe device characteristics.Figure 1 (d) shows snapshots of theR−Vgdata in 10 s intervals duringthe SET process. During this measurement, the light was turned on for 10 swithV�g ¼ �2V.The illuminationwas turnedoff, and theR−Vg curvewasmeasured.Theprocess is repeatedmultiple times to generate theplots inFig.1d. The intensity of the light was kept very low (I = 4Wm−2) to allow for amuch slower rate of resistance change. One can see that the entire transfercurve shifts gradually to the left until the Dirac point reaches V�g ¼ �2 V.The results for similar measurements done with different values of V�g areshown in Fig. 1e. These measurements establish that the Dirac point can bemoveddeterministically to any value of the gate voltage eitherby controllingthe exposure time (Fig. 1d) or the value of V�g at which the device is illu-minated (Fig. 1e). The detailed temporal dynamics of R during the entire‘SET-RESET’ process are discussed in Supplementary Note 3.After setting up the SET/RESET protocol, we now show that theresistance of the sample at a specific gate voltage canbe repeatedly alternatedbetween twodistinct states (Fig. 2a).Thedata for a single lightpulse is shownin Fig. 2b, which shows the saturation in the two states, hence the stability ofthe SET and RESET states. Moreover, adjusting the light exposure timemakes switching betweenmultiple resistance values possible, as depicted inFig. 2c. In this measurement, the light was turned on for 10 s (blue shadedregion of the timeline), duringwhichR increased. The light was then turnedoff. The value of R was stable at the value it reached when the illuminationwas cut off. This process can be repeated to produce multiple stable resis-tance levels in graphene.Figure 2d, e shows controlled switching between multiple resistancestates in an arbitrary and on-demand manner, demonstrating stability andrepeatability in the resistance states. The maximum fluctuations measuredin the resistance around the resistance states for repeated switching wasfound to be ± 0.05 × R/RDP (Supplementary Note 6). This on-demandswitching, stability and repeatability is a key indicator for the robustness ofthe device for memristor application.h-BNSLGh-BNSiOSiVVgh�(a)-8 -6 -4 -2 0 2 4 6 8 100.00.20.40.60.81.0R/RDPVg(V)(c) (e)-8 -6 -4 -2 0 2 4 6 880160240320400V g(V)R( �)1230RunNumber(b)(d)  t1  t0I-8 -6 -4 -2 0 2 4 6 80.00.20.40.60.81.0R/RDPVg(V)PristineVg=-3VVg=-4VVg=-5V0 20 40 60 800.20.40.60.81.0R/RDPt(ms)t=t0 t=t1Fig. 1 |Device characteristics and optical doping. aOptical image of the device anda schematic of the device structure. b Longitudinal resistance as a function of gatevoltage for the pristine sample before illumination (blue solid line), after photo-doping at V�g ¼ �3 V (red solid line), after erasing the doping (green dotted line),and after photodoping again at V�g ¼ �3 V (pink dotted line). c Change in thelongitudinal R as a function of time for constant power and gate voltage on exposureto light. t = t0 marks when the light was turned on and t = t1 when the light wasturned off. d Snapshots of R− Vg plots as the Dirac point progressively shifts fromVg = 0 V at t = t0 (solid purple line) to Vg* =− 2 V at t = t1 (solid red line).eRepresentative plots of deterministic shifting of the Dirac point to desired values ofVg by photodoping. RDP is the four-probe resistance at the Dirac point.https://doi.org/10.1038/s41699-024-00503-7 Articlenpj 2D Materials and Applications |            (2024) 8:69 2www.nature.com/npj2dmaterialsApplication as memristorAs shown in Fig. 2c, our device has at least six stable resistance states, withmore resistance states also accessible by lower optical powers. Such an hBN/graphene heterostructure with multiple stable resistance values holds sig-nificant potential for developing memristor devices for vector-matrix mul-tiplication andmachine learning (ML) applications42–45. Several such devicescan be fabricated in a cross-bar array for a typical linear algebra calculation,with voltages as inputs and currents as outputs. A new vector-matrixmultiplication operation can be carried out bymodifying theweights of eachcross-bar intersection, i.e., by changing the resistance of the channels. Forhardware implementation of ML training, the synaptic weights of a neurallayer (each layer will be a separate cross-bar array) can be similarlymodified.We propose a cross-bar geometry schematically shown in Fig. 3a toachieve the above objectives. Its compact footprint offers distinct advantagesover other structures. Each device unit (shown schematically in Fig. 3b) isindividually gated; this architecture is conveniently achievable using0 2000 4000 60000.20.40.60.81.0R/RDPt(s)Vg=-5V Vg=-5VVg=0V0 1000 2000 30000.20.40.60.81.0R/RDPt(s)-3V -4V -4V -1V -2.5V -4.5V-2.5V0V0V-1V-2V-1V RESETSETMeasured at Vg=0V0 500 1000 1500 2000 25000.20.40.60.81.0R/RDPt(s)-4V -3V -2V -1VMeasured at Vg=0VRESETSET0V 0V 0V 0V0 2000 4000 6000 80000.40.60.81.0R/RDPt(s)0 50 100 150 200 250 300 3500.30.50.60.70.80.91.0R/RDPt(s)ON OFF OFF OFF OFFOFF OFF ON ON ON ON(c)(b)(a)(a)(d) (e)Fig. 2 | Reproducible multiple states. aOptical switching of the device’s resistancebetween two well-defined values at V�g ¼ �5V with LED light source of λ = 427 nmand I = 62Wm−2. b Resistance versus time graph for a single light pulse of SET andRESET. c Figure illustrating optical pinning of R to multiple values. The blue shadedregion marks when the light was switched on (hence R increased with t). The whiteregions show the time the light was off. LED light source of λ = 427 nm andI = 4Wm−2 was used with V�g ¼ �5 V. For a–c the R/RDP values are recorded atV�g ¼ �5 V. d Successive optical switching of the device’s resistance between twowell-defined levels; SET level at V�g ¼ �4 V, −3 V, −2 V, −1 V and RESET levelV�g ¼ 0 V. e Optical switching of the device’s resistance between different distinctresistance levels, defined by V�g . For both (d) and (e), the measurements were donewith an LED light source of λ = 427 nm. The intensity of illumination wasI = 62Wm−2 for SET and I = 458Wm−2 for RESET. The R/RDP value is measuredat V�g ¼ 0 V.Fig. 3 | Memristor. a Schematic of the proposedcross bar geometric memristor. b Schematic of theindividual device forming a cell of the memristor.I1I2I3V1V2V3Gate voltageshBNgate electrodegraphene(a) (b)https://doi.org/10.1038/s41699-024-00503-7 Articlenpj 2D Materials and Applications |            (2024) 8:69 3www.nature.com/npj2dmaterialsmodern nano-fabrication processes. Before each operation, the channelresistance of each device is SET by a global incident optical beam and theapplicationof distinct back-gate voltages todifferent devices. TheRESET forany desired device can done by setting its gate voltages to zero and illumi-nating with a global incident optical beam.Origin of the phenomenonPhotodoping of the graphene channel requires charge transfer from hBN tographene. The energy corresponding to violet light (λ = 427 nm) is 2.9 eV,much lower than the band gap of hBN (≈6 eV), precluding photo-excitationof carriers from the valence band of hBN. We also find that the graphenechannel remained undoped in a device with only the top hBN flake andwithout the bottom hBN flake upon using the same protocol describedabove (Supplementary Note 2). This study confirms that only the bottomhBN was responsible for the photodoping effect. Although the top hBN isnot essential for the operation of the device, it is crucial for having a high-quality, low-defect graphene channel with long-term stability.Based onour observations,we sketch out a possible scenario below thatexplains all our experimental findings. Figure 4a is a schematic of the energyalignment of the bottom hBN and the graphene without photo-excitationand at Vg = 0. Several mid-gap states in hBN can act as electron donors. Ofthese, the one most relevant for us is the defect state of nitrogen vacancies(marked asEN in Fig. 4). This level canhave stable charge states of+ 1or 038.A negative gate-voltage Vg ¼ V�g dopes the graphene channel with holesand creates an electric field Eext directed from graphene into the hBN(Fig. 4b) leading to band bending. Illuminating the device with violet lightexcites electrons from EN to the conduction band EC of hBN. These elec-trons are funneled to the graphene channel under the influence of theelectric field. The holes left at EN in the hBN generate an electric field Ei inthe direction opposite Eext (Fig. 4c). The electron transfer process continuesuntil the net electric field between graphene and hBN Enet = Eext− Eibecomes zero. Simple electrostatics arguments show that the effectivenumberdensity in graphenebecomes zero at this point,whichmanifests as ashift of the charge neutrality point to V�g (Fig. 4d).On reducing Vg to zero, graphene draws negative charges from themetal contacts, making the net device charge neutral (Fig. 4e). This chargeconfiguration generates an electric field Ei directed from the hBN to gra-phene. The consequent band bending and the fact that EN>ED forbidselectron transfer back from graphene to the hBN; graphene remainsnegatively charged with the Dirac point at Vg ¼ V�g , and hBN is positivelycharged (Fig. 4e). This constitutes the SET protocol. The energy barrier toback-transfer electrons from graphene to hBN explains the long-termcharge retention in graphene after the photodoping is completed.Exposing the device to light withV�g ¼ 0V erases the doping, bringingthe graphene’s Dirac point back to Vg = 0 V. For this to happen, excesselectrons need to be removed from graphene and transferred back to hBN.h-BN grapheneEextEext > EiEiECENEDEVh�h-BN grapheneEextECENEDEV6 eV4.5 eV2.8 eV1.1 eVh-BN grapheneECENEDEVhBNSLGSiSiO2h-BN graphene(e)ECENEDEVh-BN graphene(d)EextEiEext = EiECENEDEVEiEih-BN grapheneECENEDEVEi12Ei(f)(a) (b) (c)Fig. 4 | Schematic energy band alignment (top panels) and charge distribution(bottom panels) in the device under different conditions. EC , EN and EVrepresent the conduction band edge, N-vacancy mid-gap state and valence bandedge of hBN, respectively. a For V�g ¼ 0; hν ¼ 0, hBN and graphene are chargeneutral. bThe graphene channel acquires a positive charge forV�g<0 and hν = 0. Theresultant electric field Eext bends the bands of the hBN. c SET: For V�g<0 and hν ≠ 0,electrons are transferred from EN (orange dotted line) to EC (solid red line) of hBN.These electrons drift into graphene (represented by the dotted electrons) and reduceits net positive charge. d The doping process stops when Ei = Eext. The hBN bandsflatten out, graphene is charge neutral, and the hBN is left positively charged. eAfterthe gate voltage is set to zero, graphene draws a net negative charge from the contacts,making it electron-doped and the total system charge neutral. (f) RESET: ForV�g ¼ 0 V and hν ≠ 0, electrons are excited from Ev (blue solid line) of the hBN torecombine with the holes in EN (orange dotted line). The electrons from graphene(represented by the red dotted circles) drift into hBN Ev and recombine with theholes left behind.https://doi.org/10.1038/s41699-024-00503-7 Articlenpj 2D Materials and Applications |            (2024) 8:69 4www.nature.com/npj2dmaterialsNote, however, that this process is energetically unfavorable asEN>ED at theinterface. We propose a phenomenological scenario in which this RESETprocess is a two-step process involving (1) the transfer of electrons from thevalence band of hBN to its mid-gap states due to optical excitation and (2)subsequent electron transfer from graphene to the empty valence bandstates of hBNdue to electric field (Fig. 4f). Consequently, theRESETprocess(involving electron transfer from graphene to hBN) ismuch slower than theSET process for electron doping. Device-level simulations are required toverify if the above scenario correctly captures the doping erasure process.To experimentally verify the scenario proposed above, we used a tunablepulse laser to study the wavelength dependence of the ‘SET’ time τs. Figure 5aplots τs versus the laser photon energy for a constant laser intensity andV�g ¼ �3 V. It shows a minimum centered around E= 2.8 eV. This photonenergycorresponds to theoptical absorptionbyvalencenitrogendefect inhBN‘EN ,’ leading credence to our understandingof the photodoping process37. Thewavelengthdependenceof the time required for the ‘RESET’ τr is shown inFig.5b. In contrast to the SET process, τr reaches its minimum value at aroundE= 3.2 eV; this value matches exactly EN � EV ; the energy required to excitecharge carriers from the valence band to the nitrogen-vacancy state.We note that the optical power used for the RESET process is higherthan that used during the SET process. Thus, one must consider the pos-sibility of the formation of defect states in the hBN during this process. Werule out this scenario by noting the lack of long-term resistance drift, as seenfromFig. 2.Wehave also considered the possibility of the reversible creationof additional mid-gap states but ruled them out from energetic considera-tion. The maximum used power for the optical measurements was105Wm−2, which is significantly lower than the reported power needed tocreate defects in hBN46–48.Figure 5c shows the dependence of τs on the intensity of the light, I at afixed V�g and λ = 427 nm. We find that τs∝ I−1. This dependence isunderstandable, as an increase in the intensity of photons leads tomore freecharges being produced in hBN, reducing the time taken to dope(see Supplementary Note 1 for detailed derivation). For measurementsperformed at a fixed intensity of light, the time constant to dope shouldincrease with an increase in the magnitude of V�g (equivalently, of Eext);measurements confirm this (Fig. 5d) (See SupplementaryNote 1 for detailedderivation).To summarize, we demonstrate a reversible control of the Diracpoint in the graphene-hBN heterostructure to encode the resistancevalues via a combined optical-electrical stimulus. The device’s resistancecan be modified by varying the gate voltage in the presence of an opticalincident power or by fixing the gate voltage and illuminating it withmultiple optical pulses. The switching time can be tuned by the incidencelight wavelength (Fig. 5a), light intensity (Fig. 5c), or the gate voltage(Fig. 5d) providing tremendous tunability of the properties of the device.The time taken to electron dope is much less compared to hole dopingfor a fixed power and gate voltage, and further experiments need to bedone to make them comparable. It should be possible to control theswitching time by modifying the defect density in hBN using electronirradiation or annealing processes. The ability to photoelectrically SET,READ, and RESET multiple stable and non-volatile resistance states ofthe device makes it ideal for use as a memristor.MethodsDevice fabricationThe devices were fabricated using the dry transfer technique49. Single-layergraphene (SLG) and hexagonal boron nitride (hBN) flakes were mechani-cally exfoliated onto a Si/SiO2 substrates. The hBN flakes had a thickness of25–30 nm. Electron beam lithographywas used to define electrical contacts.This was followed by etching with a mixture of CHF3 (40 sscm) and O2(10 sscm). The metallization was done with Cr/Au (5 nm/60 nm) to formthe 1D electrical contacts with SLG.MeasurementsAll electrical transportmeasurements were performed at room temperatureusing a low-frequency AC measurement technique. For low-temperaturemeasurements, the sample was cooled down in a cryostat to 4.7 K. Foroptoelectronic measurements, the sample was illuminated using either anLED of wavelength, λ = 427 nm, or a Ti Sapphire pulsed laser (80MHzrepetition rate, 100 fs pulse width).Fig. 5 |Wavelength, power and gate dependence ofdoping dynamics. a Plot of time constant, τs versusthe laser photon energy for electron doping. Thedata were taken at a fixed gate voltage V�g ¼ �3 Vand laser intensity 103 Wm−2. The dashed gray linemarks EN . The dotted red line is a guide to the eye.b Plot of time constant, τr versus the laser photonenergy for hole doping. The data were taken at afixed gate voltage V�g ¼ 3 V and laser intensity105 Wm−2. The dotted red line is a guide to the eye.c Log-log plot of time constant τs versus the intensityof incident light. The data were taken at V�g ¼ �5 Vand λ = 427 nm using LED. The open symbols arethemeasured data points, and the dashed gray line isa linear fit to the data. d Plot of τs versus the V�g forelectron doping. The data was taken for λ = 427 nmand I = 62Wm-2 using LED.2.4 2.6 2.8 3.0 3.2 3.4110E (eV)(a) (b)2.8 3.0 3.201234�� s)E (eV)��s)s r0.1 11100 2 4 6 8 10 12 14 165101520I (nW/�m2)�� s)Vg(V)��s)(c) (d)s shttps://doi.org/10.1038/s41699-024-00503-7 Articlenpj 2D Materials and Applications |            (2024) 8:69 5www.nature.com/npj2dmaterialsData availabilityThe authors declare that the data supporting the findings of this study areavailable within the main text and its supplementary Information. Otherrelevant data are available from the corresponding author upon request.Received: 17 July 2024; Accepted: 19 September 2024;References1. Sangwan, V. K. & Hersam, M. C. Neuromorphic nanoelectronicmaterials. Nat. Nanotechnol. 15, 517–528 (2020).2. Walters, B., Jacob, M. V., Amirsoleimani, A. & Rahimi Azghadi, M. Areview of graphene-based memristive neuromorphic devices andcircuits. Adv. Intell. Syst. 5, 2300136 (2023).3. Lu, H. et al. Ferroelectric tunnel junctions with graphene electrodes.Nat. Commun. 5, 5518 (2014).4. Wu, J. et al. High tunnelling electroresistance in a ferroelectric van derWaals heterojunction via giant barrier height modulation. Nat.Electron. 3, 466–472 (2020).5. Yan, X. et al. Moiré synaptic transistor with room-temperatureneuromorphic functionality. Nature 624, 551–556 (2023).6. Liu,H., Liu, Y. &Zhu,D.Chemical dopingof graphene. J.Mater.Chem.21, 3335–3345 (2011).7. Wehling, T. O. et al. Molecular doping of graphene. Nano Lett. 8,173–177 (2008).8. Jung, N. et al. Charge transfer chemical doping of few layergraphenes: charge distribution and band gap formation.Nano Lett. 9,4133–4137 (2009).9. Bruna, M. & Borini, S. Observation of Raman $G $-band splitting intop-doped few-layer graphene. Phys. Rev. B 81, 125421 (2010).10. Zhan, D. et al. FeCl3-based few-layer graphene intercalationcompounds: single linear dispersion electronic band structure andstrong charge transfer doping. Adv. Funct. Mater. 20, 3504–3509(2010).11. Zhao, W., Tan, P. H., Liu, J. & Ferrari, A. C. Intercalation of few-layergraphite flakes with FeCl3: Raman determination of fermi level, layerby layer decoupling, and stability. J. Am. Chem. Soc. 133, 5941–5946(2011).12. Zhao, W., Tan, P., Zhang, J. & Liu, J. Charge transfer and opticalphonon mixing in few-layer graphene chemically doped with sulfuricacid. Phys. Rev. B 82, 245423 (2010).13. Singh, A. K. et al. Molecular n-doping of chemical vapor depositiongrown graphene. J. Mater. Chem. 22, 15168 (2012).14. Medina, H., Lin, Y.-C., Obergfell, D. & Chiu, P.-W. Tuning of chargedensities in graphene by molecule doping. Adv. Funct. Mater. 21,2687–2692 (2011).15. Ryu, S. et al. Atmospheric oxygen binding and hole doping indeformed graphene on a SiO2 substrate. Nano Lett. 10, 4944–4951(2010).16. Dean, C. R. et al. Boron nitride substrates for high-quality grapheneelectronics. Nat. Nanotechnol. 5, 722–726 (2010).17. Das, A. et al. Monitoring dopants by Raman scattering in anelectrochemically top-gated graphene transistor. Nat. Nanotechnol.3, 210–215 (2008).18. Yan, J., Zhang, Y., Kim, P. & Pinczuk, A. Electric field effect tuning ofelectron-phonon coupling in graphene. Phys. Rev. Lett. 98, 166802(2007).19. Chen, F., Qing, Q., Xia, J., Li, J. & Tao, N. Electrochemical gate-controlled charge transport in graphene in ionic liquid and aqueoussolution. J. Am. Chem. Soc. 131, 9908–9909 (2009).20. Uesugi, E., Goto, H., Eguchi, R., Fujiwara, A. & Kubozono, Y. Electricdouble-layer capacitance between an ionic liquid and few-layergraphene. Sci. Rep. 3, 1595 (2013).21. Tiberj, A. et al. Reversible optical doping of graphene. Sci. Rep. 3,2355 (2013).22. Aftab, S., Iqbal, M. Z. & Iqbal, M. W. Programmable photo-induceddoping in 2D materials. Adv. Mater. Interfaces 9, 2201219 (2022).23. Neumann, C. et al. Spatial control of laser-induced doping profiles ingraphene on hexagonal boron nitride. ACS Appl. Mater. Interfaces 8,9377–9383 (2016).24. Ju, L. et al. Photoinduced doping in heterostructures of graphene andboron nitride. Nat. Nanotechnol. 9, 348–352 (2014).25. Roy, K. et al. Graphene-MoS2 hybrid structures for multifunctionalphotoresponsivememorydevices.Nat.Nanotechnol.8, 826–830 (2013).26. Kim, Y. D. et al. Focused-laser-enabled p-n junctions in graphenefield-effect transistors. ACS Nano 7, 5850–5857 (2013).27. Song, S.-B. et al. Deep-ultraviolet electroluminescence andphotocurrent generation in graphene/hBN/grapheneheterostructures. Nat. Commun. 12, 7134 (2021).28. Velasco, J. J. et al. Nanoscale control of rewriteabledopingpatterns inpristine graphene/boron nitride heterostructures. Nano Lett. 16,1620–1625 (2016).29. Kim, S. H. et al. Multilevel MoS2 optical memory with photoresponsivetop floating gates.ACSAppl. Mater. Interfaces 11, 25306–25312 (2019).30. Liu, C.-H., Chang, Y.-C., Norris, T. B. & Zhong, Z. Graphenephotodetectors with ultra-broadband and high responsivity at roomtemperature. Nat. Nanotechnol. 9, 273–278 (2014).31. Xiang, D. et al. Two-dimensional multibit optoelectronic memory withbroadband spectrum distinction. Nat. Commun. 9, 2966 (2018).32. Tran, M. D. et al. Two-terminal multibit optical memory via van derWaals heterostructure. Adv. Mater. 31, 1807075 (2019).33. Lee, I. et al. Photoinduced tuning of Schottky barrier height ingraphene/MoS2 heterojunction for ultrahigh performance shortchannel phototransistor. ACS Nano 14, 7574–7580 (2020).34. Gorecki, J., Apostolopoulos, V.,Ou, J.-Y.,Mailis, S. &Papasimakis, N.Optical gating of graphene on photoconductive Fe:LiNbO3. ACSNano 12, 5940–5945 (2018).35. Seo, B. H., Youn, J. & Shim, M. Direct laser writing of air-stable p-njunctions in graphene. ACS Nano 8, 8831–8836 (2014).36. Miller, D., Blaikie, A. & Alemán, B. J. Nonvolatile rewritable frequencytuning of a nanoelectromechanical resonator using photoinduceddoping. Nano Lett. 20, 2378–2386 (2020).37. Attaccalite, C., Bockstedte, M., Marini, A., Rubio, A. & Wirtz, L.Coupling of excitons and defect states in boron-nitridenanostructures. Phys. Rev. B 83, 144115 (2011).38. Weston, L., Wickramaratne, D., Mackoit, M., Alkauskas, A. & Van DeWalle, C. G. Native point defects and impurities in hexagonal boronnitride. Phys. Rev. B 97, 214104 (2018).39. Sajid, A., Reimers, J. R. & Ford, M. J. Defect states in hexagonal boronnitride: assignments of observed properties and prediction of propertiesrelevant to quantum computation. Phys. Rev. B 97, 064101 (2018).40. Strand, J., Larcher, L. & Shluger, A. L. Properties of intrinsic pointdefects and dimers in hexagonal boron nitride. J. Phys. Condens.Matter 32, 055706 (2020).41. Amin, K. R., Nagarajan, R., Pandit, R. & Bid, A. Multifractalconductance fluctuations in high-mobility graphene in the integerquantum hall regime. Phys. Rev. Lett. 129, 186802 (2022).42. Ducry, F. et al. An ab initio study on resistance switching in hexagonalboron nitride. npj 2D Mater. Appl. 6, 1–10 (2022).43. Xie, J., Afshari, S. & Sanchez Esqueda, I. Hexagonal boron nitride (h-BN) memristor arrays for analog-based machine learning hardware.npj 2D Mater. Appl. 6, 1–7 (2022).44. Maier, P. et al. Electro-photo-sensitive memristor for neuromorphicand arithmetic computing. Phys. Rev. Appl. 5, 054011 (2016).45. Schranghamer, T. F., Oberoi, A. & Das, S. Graphene memristivesynapses for high precision neuromorphic computing.Nat. Commun.11, 5474 (2020).46. Gan, L. et al. Large-scale, high-yield laser fabrication of bright andpure single-photon emitters at room temperature in hexagonal boronnitride. ACS Nano 16, 14254–14261 (2022).https://doi.org/10.1038/s41699-024-00503-7 Articlenpj 2D Materials and Applications |            (2024) 8:69 6www.nature.com/npj2dmaterials47. Kianinia, M., White, S., Fröch, J. E., Bradac, C. & Aharonovich, I.Generation of spin defects in hexagonal boron nitride.ACSPhotonics7, 2147–2152 (2020).48. Yang, Y.-Z. et al. Laser direct writing of visible spin defects inhexagonal boron nitride for applications in spin-based technologies.ACS Appl. Nano Mater. 6, 6407–6414 (2023).49. Tiwari, P. et al. Observation of the time-reversal symmetric hall effectin graphene-WSe2 heterostructures at room temperature. Nano Lett.23, 6792–6798 (2023).AcknowledgementsA.B. acknowledges funding from the Department of Science &Technology FIST program and the U.S. Army DEVCOM Indo-Pacific(Project number: FA5209 22P0166). K.W. and T.T. acknowledge supportfrom JSPS KAKENHI (Grant Numbers 19H05790, 20H00354, and21H05233). A.S. acknowledges funding from the Indian Institute ofScience start-up grant, DST Nanomission CONCEPT (Consortium forCollective and Engineered Phenomena in Topology) grant and projectMoE-STARS-2/2023-0265. M.M. acknowledges the Prime Minister’sResearch Fellowship (PMRF).Author contributionsH.K.M, M.M., V.S., A.S. and A.B. conceptualized the study, performed themeasurements, and analyzed the data. K.W. and T.T. grew the hBN singlecrystals. All the authors contributed to preparing the manuscript.Competing interestsThe authors declare no competing interests.Additional informationSupplementary information The online version containssupplementary material available athttps://doi.org/10.1038/s41699-024-00503-7.Correspondence and requests for materials should be addressed toAkshay Singh or Aveek Bid.Reprints and permissions information is available athttp://www.nature.com/reprintsPublisher’s note Springer Nature remains neutral with regard tojurisdictional claims in published maps and institutional affiliations.Open Access This article is licensed under a Creative CommonsAttribution 4.0 International License, which permits use, sharing,adaptation, distribution and reproduction in anymedium or format, as longas you give appropriate credit to the original author(s) and the source,provide a link to the Creative Commons licence, and indicate if changeswere made. The images or other third party material in this article areincluded in the article’s Creative Commons licence, unless indicatedotherwise in a credit line to the material. If material is not included in thearticle’sCreativeCommons licence and your intended use is not permittedby statutory regulation or exceeds the permitted use, you will need toobtain permission directly from the copyright holder. To view a copy of thislicence, visit http://creativecommons.org/licenses/by/4.0/.© The Author(s) 2024https://doi.org/10.1038/s41699-024-00503-7 Articlenpj 2D Materials and Applications |            (2024) 8:69 7https://doi.org/10.1038/s41699-024-00503-7http://www.nature.com/reprintshttp://creativecommons.org/licenses/by/4.0/www.nature.com/npj2dmaterials Optical control of multiple resistance levels in graphene for memristic applications Results and discussion Optoelectronic measurements Application as memristor Origin of the phenomenon Methods Device fabrication Measurements Data availability References Acknowledgements Author contributions Competing interests Additional information