# Fileset

[s41377-024-01491-5.pdf](https://mdr.nims.go.jp/filesets/f6093482-181b-4b0a-aa17-db61cb74d8f6/download)

## Creator

Magdalena Grzeszczyk, Kristina Vaklinova, [Kenji Watanabe](https://orcid.org/0000-0003-3701-8119), [Takashi Taniguchi](https://orcid.org/0000-0002-1467-3105), Konstantin S. Novoselov, Maciej Koperski

## Rights

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

## Other metadata

[Electroluminescence from pure resonant states in hBN-based vertical tunneling junctions](https://mdr.nims.go.jp/datasets/4e1d8ccd-ac03-4797-b3fa-2af0ca391f4b)

## Fulltext

Electroluminescence from pure resonant states in hBN-based vertical tunneling junctionsGrzeszczyk et al. Light: Science & Applications          (2024) 13:155 Official journal of the CIOMP 2047-7538https://doi.org/10.1038/s41377-024-01491-5 www.nature.com/lsaART ICLE Open Ac ce s sElectroluminescence from pure resonant states inhBN-based vertical tunneling junctionsMagdalena Grzeszczyk1✉, Kristina Vaklinova1, Kenji Watanabe 2, Takashi Taniguchi 3,Konstantin S. Novoselov 1,4 and Maciej Koperski 1,4✉AbstractDefect centers in wide-band-gap crystals have garnered interest for their potential in applications amongoptoelectronic and sensor technologies. However, defects embedded in highly insulating crystals, like diamond, siliconcarbide, or aluminum oxide, have been notoriously difficult to excite electrically due to their large internal resistance.To address this challenge, we realized a new paradigm of exciting defects in vertical tunneling junctions based oncarbon centers in hexagonal boron nitride (hBN). The rational design of the devices via van der Waals technologyenabled us to raise and control optical processes related to defect-to-band and intradefect electroluminescence. Thefundamental understanding of the tunneling events was based on the transfer of the electronic wave functionamplitude between resonant defect states in hBN to the metallic state in graphene, which leads to dramatic changesin the characteristics of electrons due to different band structures of constituent materials. In our devices, the decay ofelectrons via tunneling pathways competed with radiative recombination, resulting in an unprecedented degree oftuneability of carrier dynamics due to the significant sensitivity of the characteristic tunneling times on the thicknessand structure of the barrier. This enabled us to achieve a high-efficiency electrical excitation of intradefect transitions,exceeding by several orders of magnitude the efficiency of optical excitation in the sub-band-gap regime. This workrepresents a significant advancement towards a universal and scalable platform for electrically driven devices utilizingdefect centers in wide-band-gap crystals with properties modulated via activation of different tunneling mechanismsat a level of device engineering.IntroductionThe functionalities of crystals at the atomic scale canbe activated by the systematic, controllable, and preciseformation of defect centers. From this perspective, wide-band-gap materials are particularly attractive1–3, as themidgap defect levels are often decoupled from the fun-damental electronic bands, consequently becoming purequantum systems4–7. When properly controlled, defectcenters act as qubits8,9, single photon emitters10–15,sensors of pressure16,17, magnetic fields18–20, electricfields21, thermal conductivity22, or dielectric constant23at the ultimate limit of miniaturization. They alsoactivate the optical response of materials, enabling therealization of lasers24 and photodetectors25. Defect levelslocated deep within the band gap26 enable room tem-perature operation of devices, making research in suchcenters highly technological. However, all of thesefunctionalities suffer from a significant bottleneckresulting from the insulating character of the hostcrystals. Large internal resistance prevents, in mostcases, the development of electrically driven devi-ces27–29. Notably, in rare cases when the n-type andp-type doping of the insulators can be controllably andefficiently introduced, intradefect electroluminescence(EL) could be realized in three-dimensional materi-als30,31 through the creation of p–n junctions.In this study, we explored an alternative path towardelectrical excitation of defect centers realized with thinfilms of hexagonal boron nitride (hBN) integrated intovan der Waals (vdW) vertical tunneling junctions.© The Author(s) 2024OpenAccessThis article is licensedunder aCreativeCommonsAttribution 4.0 International License,whichpermits use, sharing, adaptation, distribution and reproductionin any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate ifchangesweremade. The images or other third partymaterial in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to thematerial. Ifmaterial is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtainpermission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.Correspondence: Magdalena Grzeszczyk (magda@nus.edu.sg) orMaciej Koperski (msemaci@nus.edu.sg)1Institute for Functional Intelligent Materials, National University of Singapore,Singapore 117544, Singapore2Research Center for Functional Materials, National Institute for MaterialsScience, Tsukuba 305-0044, JapanFull list of author information is available at the end of the article1234567890():,;1234567890():,;1234567890():,;1234567890():,;www.nature.com/lsahttp://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/0000-0003-4972-5371http://orcid.org/0000-0003-4972-5371http://orcid.org/0000-0003-4972-5371http://orcid.org/0000-0003-4972-5371http://orcid.org/0000-0003-4972-5371http://orcid.org/0000-0002-8301-914Xhttp://orcid.org/0000-0002-8301-914Xhttp://orcid.org/0000-0002-8301-914Xhttp://orcid.org/0000-0002-8301-914Xhttp://orcid.org/0000-0002-8301-914Xhttp://creativecommons.org/licenses/by/4.0/mailto:magda@nus.edu.sgmailto:msemaci@nus.edu.sgConceptually, the design of the devices was adapted fromthe semiconductor technology32, however, their operationwas based on fundamentally different principles. The keyenabler of the transition from semiconducting to insu-lating defect-based light-emitting diodes (LEDs) camefrom realizing specific radiative centers in hBN throughcarbon-doping33. The emerging defects led to homo-geneous photoluminescence (PL) signals resembling theoptical response of pristine excitonic semiconductorsrather than disordered defect-based systems. This allowedus to design devices where the defect levels acted as purelyresonant states, providing dominant tunneling pathwaysbetween two graphene (Gr) electrodes. As the timescalesof the tunneling processes are strongly dependent on thebarrier width and structure, we demonstrated a transitionfrom defect-to-band to intradefect EL and electron-to-photon conversion modulated by almost two orders ofmagnitude via rational engineering of the devicearchitecture.Our findings are relevant for the advancements ininsulating defect-based LEDs, providing general guide-lines for the transition from p–n junction to verticaltunneling junction geometries. The latter structures canachieve significant levels of tunability of the optoelec-tronic properties of devices based on the understanding ofthe tunneling processes mediated by resonant defectstates.ResultsSamplesThe creation of hBN-based LEDs was enabled throughthe introduction of well-defined, homogeneously dis-tributed, and reproducible radiative defect centersthrough carbon doping34. The crystallographic quality ofhBN must be high for thin films to fulfill the role of atunneling barrier35. To that end, we grew the bulk hBNcrystals via high pressure, temperature-gradientmethod36. From a single growth, we annealed a frac-tion of the crystals in a graphite furnace at the tem-perature of 2000 °C for a period of 1–5 h to achievevaried concentrations of carbon impurities. The carbon-doped hBN (hBN:C) became yellow to blackish,depending on the annealing time, contrary to the pris-tine transparent hBN (see Fig.1a, b for comparison ofthe optical images between bulk hBN and hBN:Ccrystals).The carbon doping procedure activated multipleresonances in the PL spectra in the ultraviolet, visible,and near-infrared spectral regions33,37,38. The transla-tion from optical to electrical excitation requiresunderstanding and controlling the dynamics of chargecarriers participating in the tunneling and radiativerecombination processes. Therefore, we have fabricatedtwo types of devices with different architectures of thetunneling barrier, which are schematically depicted inSihBN:CGrhBNhBNGrCr/AuCr/AuSiSi02 Si02hBN:CGrCr/AuGrCr/AuGrhBNhBNhBN:CGrGrhBN:Cabc de fhBNhBN:CFig. 1 The architecture of the vertical tunneling devices with hBN:C as an energy barrier. The optical images of the hBN (a) and hBN:C bulkcrystals (b). Schematic representation of the light emitting diode structures with the layer sequence Gr/hBN:C/Gr (c) and Gr/hBN/hBN:C/hBN/Gr (d).The devices were created by stacking mechanically exfoliated graphene, hBN, and hBN:C layers. The optical images of example devices with Gr/hBN:C/Gr and Gr/hBN/hBN:C/hBN/Gr architectures are presented in (e) and (f), respectively. The thickness of hBN:C was ~20 nm, while the hBNbarriers were ~3–5 nm thick. The scale bar corresponds to 10 μmGrzeszczyk et al. Light: Science & Applications          (2024) 13:155 Page 2 of 8Fig. 1c, d. A Si/(290 nm) SiO2 wafer was used as thesubstrate for the vdW heterostructure with the fol-lowing layer sequence: (1) Gr/hBN:C/Gr and (2) Gr/hBN/hBN:C/hBN/Gr. The optically active hBN:C layerwas ~20 nm thick and the additional hBN spacer was3–5 nm thick for the devices discussed herein. Opticalmicroscope images of the representatives of the twoclasses of devices are shown in Fig. 1e, f. Details aboutdevice fabrication can be found in the “Materials andmethods” section.We begin with the characterization of the Gr/hBN:C/Gr device, whose IV characteristics are demonstrated inFig. 2a. The zero-bias resistance yielded 0.1 GΩ, whichindicated that the direct Gr–Gr tunneling through thebarrier defined by the hBN band gap, illustrated sche-matically in Fig. 2c, was inefficient. The conductancea01.2 1.5 1.8 2.1 2.4 2.7 3.3 3.6 3.93.05101520EL intensity (counts/s)Energy (eV)1.2 1.5 1.8 2.1 2.4 2.7 3.3 3.6 3.93.0Energy (eV)VT = +6.0 VVT = +5.6 VVT = +5.4 VVT = +5.2 VVT = +5.0 VVT = –6.0 VVT = –5.6 VVT = –5.4 VVT = –5.2 VVT = –5.0 VGr/hBN:C/Gr Gr/hBN:C/Grbce fg0100200300400500600Tunnelling bias (V)III IIIII IIIIIIIIIIIIIIIV IVIV IVTunnelling current (A)Total EL intensity (counts/s)10–610–710–810–910–1010–11GrhBN:CI: Tunnelling via empty gape ede eGr GrhBN:CeII: Tunnelling via single defect levelhBN:CeeIII : Injection of e–h pair for intradefect electroluminescencephotonhBN:CeeIV: Fowler-Nordheim tunellingfor defect-band electroluminescence–6 –4 –2 0 2 4 6–6 –4 –2 0 2 4 6Tunnelling bias (V)hGrGr Gr Gr GrFig. 2 Electroluminescence from Gr/hBN:C/Gr devices. Tunneling IV curve for a Gr/hBN:C/Gr (a) demonstrates activation of novel tunnelingpaths. The device displays electroluminescence at a higher bias threshold, which can be seen from the dependence of the integrated lightemission intensity on the tunneling bias (b). The integration is done in the spectral region 1.2–3.9 eV. Schematic depiction of the tunnelingprocesses based on the band structure of the heterostructure and the Fermi level alignment with bias is presented in various regimes:tunneling via an empty gap for small bias voltage (c), tunneling via a single defect level with an increased bias voltage (d), formation ofelectron–hole pairs activating intradefect electroluminescence (e), defect-to-band electroluminescence in the sup-band-gap bias regime (f).The electroluminescence spectra for different bias voltages are presented under the forward (g) and reverse (h) biasing direction. Allmeasurements were made at T= 5 KGrzeszczyk et al. Light: Science & Applications          (2024) 13:155 Page 3 of 8related to such a tunneling process39 is given by G /exp(�2dffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2m�hBNϕBp_ Þ, where d is the thickness of thetunneling barrier, m*hBN is the effective mass of anelectron in hBN, ϕB is the barrier height and _ is thereduced Planck’s constant. The utilization of a relativelythick hBN barrier (20 nm) characterized by a 6 eV bandgap enabled us to quench the direct Gr–Gr tunnelingmechanism. The first onset of tunneling was observedat the bias of about ±0.5 V, which was associated withthe threshold related to the resonant alignment of theFermi level in graphene with a defect state energy inhBN as shown in Fig. 2d. This constitutes a qualitativelydifferent tunneling mechanism driven by the transfer ofthe amplitude of the electron’s wave function betweenthe metallic state in graphene to the resonant defectstate in the hBN barrier. This process resembles tun-neling through a bound resonant state in a semi-conductor quantum well40 with the characteristictunneling time given by τtunnel= τ0 expð2dffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2m�ðeϕB�EresÞp_ Þ,where τ0 is a normalization factor, m* is the effectivemass of the electron in the quantum well, e is the ele-mentary charge, Eres is the energy of the resonant state.τtunnel is indicative of the decay of the wave functionamplitude from the resonant defect state and becomesthe key factor in controlling the optoelectronic prop-erties of the tunneling junctions. In the Gr/hBN:C/Grdevice, the interpretation of τtunnel is more intricate, asthe tunneling process implies the gradual transforma-tion of the characteristics of an electron from a masslessparticle in graphene to a resonant dispersionless defectstate in hBN. Such a process cannot be captured by theeffective mass approximation, therefore it is challengingto make quantitative predictions for τtunnel in differentdevice architectures. In typical semiconductor tunnel-ing junctions, the contribution of the resonant state tothe wave function of the electron remains very smallthroughout the entire tunneling event due to lowerbarrier heights40. In contrast, the defect states in hBNexhibited purely resonant characteristics, evidenced bythe negligible current due to direct Gr–Gr tunneling atthe onset of the activation of defect-mediated tunnel-ing. The resonant characteristics are strengthened inhBN due to the large value of the band gap comparedwith the low onset of defect-mediated tunneling atabout 0.5 V, which demonstrated that the defect levelsare decoupled from the electronic bands. Moreover, thewave functions of electrons occupying the defect levelin hBN are typically localized within volumes char-acterized by dimensions below several nanometers37,which facilitates their confinement in the sizable bar-riers. In the regime marked by the first tunneling onset,EL was not allowed due to the absence of an empty finalstate of the recombination process if we assume themore likely case of tunneling via a donor-like defectlevel below the conduction band. A second tunnelingonset related to the injection of the hole into anacceptor-like defect level (see Fig. 2e) was observed at5 V, which coincided with the detection of light emittedby the device.DiscussionThe inspection of the EL spectra, which are demon-strated in Fig. 2g, h, granted further insight into thecompetition between the tunneling processes and radia-tive recombination. Two types of qualitatively differentcontributions to the EL spectra could be identified:(1) broadband features with a width of about 1 eV, whichcould be attributed to transitions between the mid-gapdefect level and electronic sub-bands41 and (2) narrowresonances characteristic of intradefect excitations. Theactivation of the defect-to-band EL requires a differenttunneling process via the Fowler–Nordheim mechan-ism42,43. Either an electron can be injected into the con-duction band (Fig. 2f), recombining with the underlyingempty defect level, or a hole can be injected into thevalence band, enabling the recombination of an electronoccupying a defect level. In general terms, the biasthreshold for the defect-to-band and intradefect opticaltransitions is determined by the electronic structure of themid-gap levels in combination with the position of theFermi level in graphene in relation to the band edges ofhBN forming a triangular tunneling barrier. In the Gr/hBN:C/Gr device, the threshold for both types of ELcoincided at 5 V when the tunneling current was 400 nA.The low contribution from the intradefect transitions tothe EL spectra indicated, that the Gr/hBN:C/Gr deviceoperated in the regime τtunnel « τrad, where τrad is theradiative lifetime of the electron determining the decay ofphotoluminescence intensity IPL(t)= I0 expð�t=τrad),where I0 is a normalization constant, t is a time typicallymeasured with respect to a short laser pulse, and τrad isthe radiative lifetime. τrad was previously measured to be1.0 ± 0.1 ns for the intradefect transition at 1.54 eV and1.3 ± 0.2 ns for the intradefect transition at 1.99 eV inhBN:C films33. These values act as an estimation of theupper limit for τtunnel in the Gr/hBN:C/Gr devices.In order to access τtunnel » τrad regime favoring theefficiency of intradefect radiative transitions, we fabricatedGr/hBN/hBN:C/hBN/Gr devices with enhanced barrierwidth and additional non-aligned hBN/hBN:C interfaces.The zero-bias resistance of these devices increased bymore than two orders of magnitude to the value of~20 GΩ. The sequential tunneling onsets were not visiblein the IV curves due to the large resistance (Fig. 3a). Insuch a device, the onset of EL was driven by the currentGrzeszczyk et al. Light: Science & Applications          (2024) 13:155 Page 4 of 8rather than the voltage threshold. The light emissionappeared at the bias of 8 V, which corresponded to thetunneling current of 10 nA (Fig. 3b). Although the voltagethreshold increased, the EL required a 40-times smallercurrent to be activated. The reduction in the thresholdtunneling current coincided with a qualitative modifica-tion of the EL spectra, as demonstrated in Fig. 3d, e. Inthis device architecture, the intradefect transitions char-acterized by narrow linewidth resonances dominated thebroadband contribution associated with defect-to-bandtransitions. The total emission intensity, integrated overthe spectral region 1.2–3.3 eV increased significantly. Toprovide a rational comparison, the integrated EL signalnormalized by the area of the device was about 80 timesstronger for the Gr/hBN/hBN:C/hBN/Gr device than forGr/hBN:C/Gr device for the same current of 1 μA, as canbe seen in Fig. S11.We interpret the barrier-induced modifications of theoptoelectronic properties of our tunneling devices interms of the band alignment schematically illustrated inFig. 3c. The enhanced efficiency of EL originated from theincrease in τtunnel affected by the introduction ofadditional pristine hBN barriers. Concurrently, theoperation of the Gr/hBN/hBN:C/hBN/Gr device in theFowler–Nordheim tunneling regime led to quenching ofthe efficiency of the defect-to-band transition, as thedirect Gr–Gr tunneling directly through the conductionand/or valence band became a dominant tunnelingmechanism for electrons occupying states close to theFermi levels in graphene electrodes due to the reductionof the effective barrier width with increasing voltage in thepresence of a triangular tunneling barrier. Notably, ourdevices often exhibited an asymmetry of the IV char-acteristics and the EL intensity dependence on the tun-neling bias. The origin of the asymmetry could beattributed to a variation in the number of hBN layers and/or the varied quality of the interfaces determined by thestacking process, as the symmetric design of our struc-tures implies that the charge carrier dynamics should notdepend on the sign of the applied bias. From the practicalperspective, our devices exhibited remarkable stability, asdemonstrated by the consistent emission in terms ofintensity and energy of the optical resonances tested overa 10-minute duration, as shown in the SupplementaryadTunnelling bias (V)Tunnelling current (A)1010–710–810–910–1010–11–12 –8 –4 0 4 8 12 –12 –8 –4 0 4 8 12ee IV: Fowler–Nordheim tunellingfor defect-band electroluminescenceTunnelling bias (V)e1.2 1.5 1.8 2.1 2.4 2.7 3.0 3.3 3.6 3.9050100150200EL intensity (counts/s)Energy (eV)1.2 1.5 1.8 2.1 2.4 2.7 3.0 3.3 3.6 3.9Energy (eV)Gr/hBN/hBN:C/hBN/Gr Gr/hBN/hBN:C/hBN/GrTotal EL intensity (counts/s)0200040006000b chBNGrhBNGrhBN:CNon-radiative tunnelling EL EL Non-radiative tunnelling EL ELVT = +14 VVT = +13 VVT = +12 VVT = +11 VVT = +10 VVT = –14 VVT = –13 VVT = –12 VVT = –11 VVT = –10 VFig. 3 Electroluminescence from Gr/hBN/hBN:C/hBN/Gr devices. Tunneling IV curves (a) and the integrated EL intensity dependence on thetunneling bias (b) demonstrate the increase of the tunneling and electroluminescence voltage threshold induced by additional pristine hBN barriers.Schematic depiction of the charge tunneling processes in the sup-band-gap voltage regime enabling non-radiative Gr–Gr tunneling (includingtunneling mediated by electronic bands via Fowler–Nordheim mechanism), defect-to-band electroluminescence and intradefectelectroluminescence (c). The evolution of the electroluminescence spectra with bias voltage under forward (d) and reverse (e) directiondemonstrates that the electroluminescence spectra are dominated by intradefect optical transitions. All the measurements were done at T= 5 KGrzeszczyk et al. Light: Science & Applications          (2024) 13:155 Page 5 of 8Information (SI) in Fig. S3. Furthermore, we conductedtemperature-dependent measurements, revealing that theemission remained distinguishable with characteristicresonances even at room temperature, underlining theversatility of these structures for practical applications.The temperature-dependent EL spectra are presented inSI in Fig. S4.Our work provides general guidelines that can be uni-versally applied to transition from understanding thephysical mechanisms of the tunneling processes to theactivation/quenching of specific radiative pathways andmodulation of the light emission efficiency of optoelec-tronic devices. In order to provide a quantitative contextfor this analysis, we compared the PL and EL spectra froma single device under identical physical conditions, whichare presented in Fig. 4a. The presence of multiple reso-nances near the excitation energy in the PL spectra,notably absent in the EL response, originated fromphonon modes associated with the constituent materialsof the device (Si, Gr, and hBN). We identified three typesof defects contributing identical signatures of intradefectresonances to PL and EL spectra, which we labeledA1–A3. Their origin was related to intradefect transitionscharacterized by varied degrees of electron–phononcoupling quantified by Huang–Rhys factors, as thoroughlydiscussed in ref. 38. A comparable emission intensity of allthree resonances was observed for optical excitation at2.56 eV with the power of 1 mW and for the electricalexcitation with the current of 2 μA. These values corre-spond to excitations with 2.4 × 1015 photons per secondand 1.2 × 1013 electrons per second, which was furtherreduced to 2.5 × 1010 electrons per second if we considertunneling only through the area from which we collectlight in the confocal geometry. Such interpolation wasjustified by the homogenous EL signal, as illustrated inFig. S2. These considerations imply that the photon-to-050100150200250EL/PL intensity (counts/s)Energy (eV)PL ExcitationSi phononhBN phononGr phononEL signalPL signalGr/hBN/hBN:C/hBN/GrA1A2A3Energy (eV)0 9 10 11 12 13 141.9411.9441.9891.992Voltage (V)1.3801.3831.386Energy (eV)1.5421.5451.5481.551Energy (eV)1.90 1.95 2.001.35 1.400.00.51.01.50 1.55Energy (eV) Energy (eV)Normalized intensity (a. u.)A1 A2 A3ab c defgA1A2A35.0 meV7.2 meV4.6 meV4.9 meV2 0 1.2 1.4 1.6 1.8 2.0 2.2 2.6 2.8 3.0 3.2 3.4 3.62.4Energy (eV)Fig. 4 Electrical and optical excitation of carbon centers in hBN and the bias-driven control of the emission energy. The comparison of thephotoluminescence and electroluminescence spectra for the Gr/hBN/hBN:C/hBN/Gr device (a). The photoluminescence spectrum was collected with2.56 eV laser excitation. The laser was focused to a spot of 1 μm diameter with a power of 1 mW. Electroluminescence was measured with −14 V biasvoltage, which corresponded to IT=−2 μA. Photoluminescence and electroluminescence spectra were collected via the same microscope objectivefrom a region of the sample with a size close to the diffraction limit for the adequate comparison of emission intensity. All individual emissioncenters, A1 (b), A2 (c), and A3 (d), are present in both optically and electrically excited spectra. The attribution to a specific defect complex of eachemission resonance was previously reported in ref. 38. The energy of emission for defects A1 (e), A2 (f), and A3 (g) changes with the tunneling bias inforward (red dots) and reverse (blue dots) directions. The extrapolated zero-bias energy is offset by a few meV from the energy of the resonance inthe photoluminescence spectrum (black dot in e–g). All measurements were made at T= 5 KGrzeszczyk et al. Light: Science & Applications          (2024) 13:155 Page 6 of 8photon conversion was about five orders of magnitudesmaller than the electron-to-photon conversion. Thesignificantly enhanced quantum efficiency for electricalexcitation can be understood from geometrical and bandstructure considerations. The optical excitation was rea-lized in the sub-band-gap regime, in which photons areabsorbed exclusively at the defect sites in resonant con-ditions33 (i.e., the host crystal is optically transparent).The IV characteristics of our devices demonstrated thatthe tunneling current in the sub-band-gap bias regime inthe absence of available defect states was negligible (i.e.,the host crystal is not electrically conductive). Hence, in asimple view, the hBN crystal is transparent for photonsbut opaque for electrons in the mid-gap regime.The realization of the electrical excitation enabledcontrol of the energy of the optical resonances, asillustrated in Fig. 4b–d. In the Fowler–Nordheimtunneling regime, we observed a redshift of the emis-sion energy for defects A1–A3 upon increasing thetunneling bias. We attributed this observation to aneffect of screening by electrons injected into the con-duction band of hBN. The increase in effectivedielectric constant leads to a reduction in the energy ofthe intradefect optical transitions related to carbon andvacancy centers in hBN38. In the mid-gap regime forthe Gr/hBN:C/Gr devices, the evolution of the emis-sion energy was qualitatively different, in agreementwith a Stark effect44,45, as illustrated in Fig. S15.Therefore, the energy tunability in our devices wasdriven by a competition between the dielectricscreening and the Stark effect, with the dominantcontribution determined by the tunneling regime.While single-state considerations are sufficient toaccount for experimental observations, we expandedour discussion to include many-body states in SI (seeFig. S16).In conclusion, we have successfully developed LEDsbased on vertical tunneling junctions utilizing hBN:C asthe barrier material. We have achieved a unique regimewhere mid-gap defect levels acted as resonant states,resulting in highly tunable electron tunneling mechan-isms. Through optoelectronic characterization of thesedevices, we have uncovered universal principles con-necting the electrical performance of the device with itsoptical response, governed by electron dynamics con-trollable via band structure engineering. Despite thetechnological challenges posed by LEDs based on insu-lating materials, they offer access to purely resonant statesand unprecedented control over the efficiency of radiativetransitions, owing to high tunneling barriers. Our findingspave the way for optoelectronic applications leveragingwide-band-gap 2D materials, which have the potential tofully exploit the advantages of miniaturized tunnelingdevices.Materials and methodsSample fabricationGraphite, hBN, and hBN:C crystals were mechanicallycleaved onto silicon wafers with 300 and 90 nm layers ofSiO2, respectively. Thin graphite films were selected to actas electrodes for hBN:C of ~20 nm thickness, identified byoptical contrast and atomic force microscopy.Using a polydimethylsiloxane/polycarbonate stamp, theassembly of graphite/hBN/hBN:C/hBN or graphite/hBN:C/ was lifted from the Si/ SiO2 wafer at 100 °C.Subsequently, the stacks were released onto the pre‐exfoliated graphite flake together with the polycarbonatefilm at 180 °C, which was washed away thereafter by usingdichloromethane, acetone, and isopropanol to removepolymer residues. The electrical contacts to the top andbottom graphite electrodes were patterned using electronbeam lithography followed by evaporation of 5 nm‐Cr/60 nm‐Au layer and finalized by a lift‐off process.Experimental setupThe optical spectra were measured in dry cryogenicsystems with a base temperature of 4.2 K. The sample wascooled down via thermal contact with a cold finger. Thelaser light was focalized on the surface of the device, andthe PL/EL signal was collected through an in‐situ objec-tive with a numerical aperture of 0.82. The sample wasmounted on a chip carrier positioned on a set of x/ypiezo‐positioners that allow alignment, while the objectivewas fixed on a piezo positioner z that allows focalizationof laser light. The optical signal was dispersed by a 0.75 mspectrometer equipped with a 150 g/mm grating. Thelight was detected by a liquid nitrogen-cooled charge-coupled device camera. For PL excitation λ= 488 nm(2.56 eV) or λ= 514 nm (2.41 eV) continuous wave (CW)laser diode was used. The excitation power focused on thesample was kept at 1 mW. The current–voltage char-acteristics were measured with a Keithley 24XX sourcemeter.AcknowledgementsThis project was supported by the Ministry of Education (Singapore) throughthe Research Centre of Excellence program (grant EDUN C‐33‐18‐279‐V12,I‐FIM), AcRF Tier 3 (MOE2018-T3-1-005). This research is supported by theMinistry of Education, Singapore, under its Academic Research Fund Tier 2(MOE-T2EP50122-0012). This material is based upon work supported by the AirForce Office of Scientific Research and the Office of Naval Research Globalunder award number FA8655-21-1-7026. K.W. and T.T. acknowledge supportfrom JSPS KAKENHI (Grant Numbers 19H05790, 20H00354, and 21H05233).Author details1Institute for Functional Intelligent Materials, National University of Singapore,Singapore 117544, Singapore. 2Research Center for Functional Materials,National Institute for Materials Science, Tsukuba 305-0044, Japan. 3InternationalCenter for Materials Nanoarchitectonics, National Institute for MaterialsScience, Tsukuba 305-0044, Japan. 4Department of Materials Science andEngineering, National University of Singapore, Singapore 117575, SingaporeGrzeszczyk et al. Light: Science & Applications          (2024) 13:155 Page 7 of 8Author contributionsM.K. and K.S.N. conceived the project. M.G. conducted optical characterizationand analyzed the data. K.W. and T.W. grew bulk carbon-doped hBN crystals. K.V.fabricated the samples. M.G. and M.K. wrote the manuscript with inputs fromall co-authors. All co-authors contributed to this work, read the manuscript,discussed the results, and agreed on the included contents.Data availabilityThe data that support the findings of this study are openly available at thefollowing https://doi.org/10.58132/B4QQ5E.Conflict of interestThe authors declare no competing interests.Supplementary information The online version contains supplementarymaterial available at https://doi.org/10.1038/s41377-024-01491-5.Received: 4 November 2023 Revised: 8 May 2024 Accepted: 21 May 2024References1. Heremans, F. J., Yale, C. G. & Awschalom, D. D. Control of spin defects in wide-bandgap semiconductors for quantum technologies. Proc. IEEE 104,2009–2023 (2016). 10.1109/JPROC.2016.2561274.2. Castelletto, S. & Boretti, A. Silicon carbide color centers for quantum applica-tions. J. Phys.: Photonics 2, 022001 (2020). 10.1088/2515-7647/ab77a2.3. Diler, B. et al. Coherent control and high-fidelity readout of chromium ions incommercial silicon carbide. npj Quantum Inf. 6, 11 (2020). 10.1038/s41534-020-0247-7.4. Redman, D. A. et al. Spin dynamics and electronic states of N-V centers indiamond by EPR and four-wave-mixing spectroscopy. Phys. Rev. Lett. 67,3420–3423 (1991). 10.1103/PhysRevLett.67.3420.5. Fuchs, G. D. et al. Excited-state spectroscopy using single spin manipulation indiamond. Phys. Rev. Lett. 101, 117601 (2008). 10.1103/PhysRevLett.101.117601.6. Hanson, R., Gywat, O. & Awschalom, D. D. Room-temperature manipulationand decoherence of a single spin in diamond. Phys. Rev. B 74, 161203 (2006).10.1103/PhysRevB.74.161203.7. Childress, L. et al. Coherent dynamics of coupled electron and nuclear spinqubits in diamond. Science 314, 281–285 (2006). 10.1126/science.1131871.8. Gordon, L. et al. Quantum computing with defects. MRS Bull. 38, 802–807(2013). 10.1557/mrs.2013.206.9. Dutt, M. V. G. et al. Quantum register based on individual electronic andnuclear spin qubits in diamond. Science 316, 1312–1316 (2007). 10.1126/science.1139831.10. Aharonovich, I., Tetienne, J. P. & Toth, M. Quantum emitters in hexagonalboron nitride. Nano Lett. 22, 9227–9235 (2022). 10.1021/acs.nanolett.2c03743.11. Tran, T. T. et al. Quantum emission from hexagonal boron nitride monolayers.Nat. Nanotechnol. 11, 37–41 (2016). 10.1038/nnano.2015.242.12. Koperski, M. et al. Towards practical applications of quantum emitters in boronnitride. Sci. Rep. 11, 15506 (2021). 10.1038/s41598-021-93802-8.13. Koperski, M. et al. Optical properties of atomically thin transition metaldichalcogenides: observations and puzzles. Nanophotonics 6, 1289–1308(2017). 10.1515/nanoph-2016-0165.14. Koperski, M., Nogajewski, K. & Potemski, M. Single photon emitters in boronnitride: more than a supplementary material. Opt. Commun. 411, 158–165(2018). pii/S0030401817310039.15. Chen, Y. L. et al. Annealing of blue quantum emitters in carbon-doped hex-agonal boron nitride. Appl. Phys. Lett. 123, 041902 (2023). 10.1063/5.0155311.16. Chijioke, A. D. et al. The ruby pressure standard to 150 GPa. J. Appl. Phys. 98,114905 (2005). 10.1063/1.2135877.17. Gottscholl, A. et al. Spin defects in hBN as promising temperature, pressureand magnetic field quantum sensors. Nat. Commun. 12, 4480 (2021). 10.1038/s41467-021-24725-1.18. Gruber, A. et al. Scanning confocal optical microscopy and magnetic reso-nance on single defect centers. Science 276, 2012–2014 (1997). 10.1126/science.276.5321.2012.19. Maze, J. R. et al. Nanoscale magnetic sensing with an individual electronic spinin diamond. Nature 455, 644–647 (2008). 10.1038/nature07279.20. Degen, C. L., Reinhard, F. & Cappellaro, P. Quantum sensing. Rev. Mod. Phys. 89,035002 (2017). 10.1103/RevModPhys.89.035002.21. Dolde, F. et al. Electric-field sensing using single diamond spins. Nat. Phys. 7,459–463 (2011). 10.1038/nphys1969.22. Laraoui, A. et al. Imaging thermal conductivity with nanoscale reso-lution using a scanning spin probe. Nat. Commun. 6, 8954 (2015).10.1038/ncomms9954.23. Singh, A., Manjanath, A. & Singh, A. K. Engineering defect transition-levelsthrough the van der Waals heterostructure. J. Phys. Chem. C. 122,24475–24480 (2018). 10.1021/acs.jpcc.8b08082.24. Watanabe, K., Taniguchi, T. & Kanda, H. Direct-bandgap properties and evi-dence for ultraviolet lasing of hexagonal boron nitride single crystal. Nat.Mater. 3, 404–409 (2004). 10.1038/nmat1134.25. Wu, D. et al. Ultrabroadband and high-detectivity photodetector based onWS2/Ge heterojunction through defect engineering and interface passivation.ACS Nano 15, 10119–10129 (2021). 10.1021/acsnano.1c02007.26. Huang, P. et al. Carbon and vacancy centers in hexagonal boron nitride. Phys.Rev. B 106, 014107 (2022). 10.1103/PhysRevB.106.014107.27. Song, S. B. et al. Deep-ultraviolet electroluminescence and photocurrentgeneration in graphene/hBN/graphene heterostructures. Nat. Commun. 12,7134 (2021). 10.1038/s41467-021-27524-w.28. Moon, S. et al. Hexagonal boron nitride for next-generation photonicsand electronics. Adv. Mater. 35, 2204161 (2023). 10.1002/adma.202204161.29. Liu, Y. et al. Recent progress of heterostructures based on two dimensionalmaterials and wide bandgap semiconductors. J. Phys.: Condens. Matter 34,183001 (2022). 10.1088/1361-648X/ac5310.30. Lohrmann, A. et al. Single-photon emitting diode in silicon carbide. Nat.Commun. 6, 7783 (2015). 10.1038/ncomms8783.31. Khramtsov, I. A., Vyshnevyy, A. A. & Fedyanin, D. Y. Enhancing the brightness ofelectrically driven single-photon sources using color centers in silicon carbide.npj Quantum Inf. 4, 15 (2018). 10.1038/s41534-018-0066-2.32. Chang, L. L., Esaki, L. & Tsu, R. Resonant tunneling in semiconductor doublebarriers. Appl. Phys. Lett. 24, 593–595 (1974). 10.1063/1.1655067.33. Koperski, M. et al. Midgap radiative centers in carbon-enriched hexagonalboron nitride. Proc. Natl Acad. Sci. USA 117, 13214–13219 (2020). 10.1073/pnas.2003895117.34. Onodera, M. et al. Carbon annealed HPHT-hexagonal boron nitride: exploringdefect levels using 2D materials combined through van der Waals interface.Carbon 167, 785–791 (2020). pii/S0008622320304735.35. Britnell, L. et al. Electron tunneling through ultrathin boron nitride crystallinebarriers. Nano Lett. 12, 1707–1710 (2012). pii/S0008622320304735.36. Kubota, Y. et al. Deep ultraviolet light-emitting hexagonal boron nitride syn-thesized at atmospheric pressure. Science 317, 932–934 (2007). 10.1126/science.1144216.37. Qiu, Z. et al. Atomic structure of carbon centres in hBN: towards engineeringof single photon sources. arXiv preprint arXiv: 2110.07842 https://doi.org/10.48550/arXiv.2110.07842 (2021)38. Badrtdinov, D. I. et al. Dielectric environment sensitivity of carboncenters in hexagonal boron nitride. Small 19, 2300144 (2023). 10.1002/smll.202300144.39. Simmons, J. G. Generalized formula for the electric tunnel effect betweensimilar electrodes separated by a thin insulating film. J. Appl. Phys. 34,1793–1803 (1963). 10.1063/1.1702682.40. Harada, N. & Kuroda, S. Lifetime of resonant state in a resonant tunnelingsystem. Jpn. J. Appl. Phys. 25, L871 (1986). 10.1143/JJAP.25.L871.41. Rieger, W. et al. Defect-related optical transitions in GaN. Phys. Rev. B 54,17596–17602 (1996). 10.1103/PhysRevB.54.17596.42. Fowler, R. H. & Nordheim, L. Electron emission in intense electric fields. Proc. R.Soc. A Math. Phys. 119, 173–181 (1928). 10.1098/rspa.1928.0091.43. Hattori, Y. et al. Determination of carrier polarity in Fowler–Nordheim tun-neling and evidence of fermi level pinning at the hexagonal boron nitride/metal interface. ACS Appl. Mater. Interfaces 10, 11732–11738 (2018). 10.1021/acsami.7b18454.44. Tamarat, P. et al. Stark shift control of single optical centers in diamond. Phys.Rev. Lett. 97, 083002 (2006). 10.1103/PhysRevLett.97.083002.45. Noh, G. et al. Stark tuning of single-photon emitters in hexagonal boronnitride. Nano Lett. 18, 4710–4715 (2018). 10.1021/acs.nanolett.8b01030.Grzeszczyk et al. Light: Science & Applications          (2024) 13:155 Page 8 of 8https://doi.org/10.58132/B4QQ5Ehttps://doi.org/10.1038/s41377-024-01491-5https://doi.org/10.48550/arXiv.2110.07842https://doi.org/10.48550/arXiv.2110.07842 Electroluminescence from pure resonant states in hBN-based vertical tunneling junctions Introduction Results Samples Discussion Materials and methods Sample fabrication Experimental setup Acknowledgements Acknowledgements