# Fileset

[s41467-022-32292-2.pdf](https://mdr.nims.go.jp/filesets/2b62b20a-7b80-4b08-8a91-0f712830123c/download)

## Creator

Juan Francisco Gonzalez Marin, Dmitrii Unuchek, Zhe Sun, Cheol Yeon Cheon, Fedele Tagarelli, [Kenji Watanabe](https://orcid.org/0000-0003-3701-8119), [Takashi Taniguchi](https://orcid.org/0000-0002-1467-3105), Andras Kis

## Rights

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

## Other metadata

[Room-temperature electrical control of polarization and emission angle in a cavity-integrated 2D pulsed LED](https://mdr.nims.go.jp/datasets/5c1a61a8-4544-4545-8da9-5c66045f6396)

## Fulltext

Room-temperature electrical control of polarization and emission angle in a cavity-integrated 2D pulsed LEDnature communicationsArticle https://doi.org/10.1038/s41467-022-32292-2Room-temperature electrical controlof polarization and emission angle in acavity-integrated 2D pulsed LEDJuan Francisco Gonzalez Marin 1,2, Dmitrii Unuchek1,2, Zhe Sun 1,2,Cheol Yeon Cheon1,2, Fedele Tagarelli 1,2, Kenji Watanabe 3,Takashi Taniguchi 4 & Andras Kis 1,2Devices based on two-dimensional (2D) semiconductors hold promise for therealization of compact and versatile on-chip interconnects between electricaland optical signals. Although light emitting diodes (LEDs) are fundamentalbuilding blocks for integrated photonics, the fabrication of light sourcesmadeof bulk materials on complementary metal-oxide-semiconductor (CMOS) cir-cuits is challenging. While LEDs based on van der Waals heterostructures havebeen realized, the control of the emissionproperties necessary for informationprocessing remains limited. Here, we show room-temperature electrical con-trol of the location, directionality and polarization of light emitted from a 2DLED operating at MHz frequencies. We integrate the LED in a planar cavity tocouple the polariton emission angle and polarization to the in-plane excitonmomentum, controlled by a lateral voltage. These findings demonstrate thepotential of TMDCs as fast, compact and tunable light sources, promising forthe realization of electrically driven polariton lasers.Two-dimensional direct bandgap semiconductors are promisingmaterials for on-chip integrated optoelectronic devices1. Since thedemonstration of the first ultrasensitive photodetectors2 and lightemitting diodes3,4, there has been a strong effort to improve thequantumefficiency, speed, energy consumption andwavelength rangeof 2D-based devices5,6. In recent years, integration with CMOS-compatible photonic platforms has proven the viability of the tech-nology for the development of optical interconnects7–9. Furthermore,the integration of 2D materials with photonic cavities has enabled thedemonstration of lasers10,11 and deterministic single photon emitters12,main building blocks for quantum information processing.The spin-valley properties of TMDCs have also been exploited as anew degree of freedom, with the purpose of overcoming some of thefundamental limits related to Joule heating, speed and coherence incharge-based devices. The K and K’ valleys in monolayer TMDCs arecoupled to circularly polarized light via the optical selection rules13,enabling the realization of polarization-sensitive photodetectors14 andlight-emitting diodes15,16. However, so far, these devices requiredcryogenic temperatures and complex fabrication.Although practical applications of LEDs for information proces-sing require fast electrical switching and low energy consumption17,most devices based on 2D materials operate with direct current (DC)and present very low values of radiant efficiency, due to the largecontact resistance and low quantum efficiency. The control of theemission pattern is also fundamental for the coupling to photonicwaveguides and detectors, as well as for illumination applications18.Yet, the location and directionality of light emitted from TMDCs couldonly be controlled by the coupling with photonic cavities and meta-surfaces, where the far field emission pattern is fixed by the geometryof the structure19.Here, we overcome these limitations by integrating a pulsed lightemitting diode (LED) based on monolayer WSe2 in a distributed BraggReceived: 20 January 2022Accepted: 25 July 2022Check for updates1Institute of Electrical and Microengineering, École Polytechnique Fédérale de Lausanne (EPFL), CH-, 1015 Lausanne, Switzerland. 2Institute of MaterialsScience and Engineering, École Polytechnique Fédérale de Lausanne (EPFL), CH-, 1015 Lausanne, Switzerland. 3Research Center for Functional Materials,National Institute for Materials Science, 1-1 Namiki, Tsukuba 305-0044, Japan. 4International Center for Materials Nanoarchitectonics, National Institute forMaterials Science, 1-1 Namiki, Tsukuba 305-0044, Japan. e-mail: andras.kis@epfl.chNature Communications |         (2022) 13:4884 11234567890():,;1234567890():,;http://orcid.org/0000-0001-5648-7493http://orcid.org/0000-0001-5648-7493http://orcid.org/0000-0001-5648-7493http://orcid.org/0000-0001-5648-7493http://orcid.org/0000-0001-5648-7493http://orcid.org/0000-0002-7460-2229http://orcid.org/0000-0002-7460-2229http://orcid.org/0000-0002-7460-2229http://orcid.org/0000-0002-7460-2229http://orcid.org/0000-0002-7460-2229http://orcid.org/0000-0003-2588-7405http://orcid.org/0000-0003-2588-7405http://orcid.org/0000-0003-2588-7405http://orcid.org/0000-0003-2588-7405http://orcid.org/0000-0003-2588-7405http://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/0000-0002-3426-7702http://orcid.org/0000-0002-3426-7702http://orcid.org/0000-0002-3426-7702http://orcid.org/0000-0002-3426-7702http://orcid.org/0000-0002-3426-7702http://crossmark.crossref.org/dialog/?doi=10.1038/s41467-022-32292-2&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1038/s41467-022-32292-2&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1038/s41467-022-32292-2&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1038/s41467-022-32292-2&domain=pdfmailto:andras.kis@epfl.chreflector (DBR) cavity. By applying a square train of pulses to a sourceelectrode20,21, we show MHz operation with the emission of ns pulses.In addition,wedemonstrate electrical control of the emission angle forvalley-polarized photons at room temperature. Together, our resultsestablish amethod to electrically modulate the location, directionalityand polarization of 2D LEDs,with switching times ultimately limited bythe pulse linewidth.ResultsDevice design and basic characterizationOur device consists of a WSe2 monolayer encapsulated with a thickbottom h-BN and top trilayer h-BN. Monolayer graphene is used as atransparent gate electrode and top Co/Ti contacts are employed forelectrical injection through the thin h-BN tunneling barrier (Fig. 1a).Figure 1b shows the optical image of the device after contact deposi-tion (See Methods). The uniform and high-quality factor λ/2 cavity(Supplementary Figs. 1, 2) is formed by a bottom mirror consisting of12 pairs of Ta2O5/SiO2 and a topmirror consisting of 9 pairs of SiO2/SiN(Fig. 1c). The optical properties of the mirrors are analyzed in detail inSupplementary Fig. 3. The cavity structure is designed such thatmonolayer WSe2 is located at the maximum of the electric field (Sup-plementary Fig. 4).Figure 1d demonstrates electron and hole transport for the cavityintegrated WSe2 at room temperature (RT). The efficient electrostaticgating is confirmed by the gate-dependent PL intensity (Supplemen-tary Figs. 5, 6). To achieve electrically driven light emission, we apply ahigh frequency square pulse (Vsq) with rise time τrise < 10 ns to thesource contact while we ground both the drain and the gate. Themechanism behind pulsed electroluminescence (EL) is depicted inFig. 1e. At time T1 (Fig. 1f, inset), the monolayer is initially n-doped.When Vsq switches to positive values with a rising time on the order ofns, the chemical potential of the monolayer cannot react as fast as theFermi level of the metal, leading to a steep band bending at theinterface21. Consequently, an electron tunneling current flows towardsthe contact while a hole tunneling current flows outwards from it. Thespatial overlapbetween electron and holewavefunctions at T2 leads tothe formation of excitons and subsequent light emission. Thismechanism produces the emission of short pulses with frequencies intheMHz range anddurationbelow τFWHM= 3 ns, as shown inFig. 1f. Themaximum operation speed can therefore reach 1=τFWHM =0: _3GHz,limited by the device geometry and rising time of the pulse generator.Exciton-photon couplingBefore top cavity growth, the high-quality hBN encapsulation allows usto achieve narrow excitonic emission at low temperature (Fig. 2a). AtT = 5 K, we identify neutral exciton (X0), charged exciton (X−/+), nega-tively charged biexciton (XX−) and bound exciton (L) emission20,whereas the neutral exciton dominates the spectrum at T = 300K.Similar results are obtained in PL measurements (SupplementaryFigs. 5, 6). Temperature also modifies the exciton oscillator strengthand exciton lifetime, crucial for reaching the strong couplingregime22,23. In Fig. 2c, we plot the temperature dependence of thenormalized reflectance at Vg = 0V. There is a clear correlation betweenreflectance and EL, with neutral and charged exciton absorption peaksappearing at low temperatures. The oscillator strength is proportionalto the reflection contrast and linewidth24, and remains high at roomtemperature.The top DBR growth results in a uniform cavity mode at thedevice25 (Supplementary Fig. 7). For the bare cavity with hBN, thequality factor isQ = 296.9 ± 0.5 (Supplementary Fig. 8). The addition ofthe monolayer graphene gate reduces the quality factor toQ = 179.2 ± 0.5, with a cavity linewidth γc = 9.132 ± 0.006meV (Fig. 2g).To quantify the coupling betweenWSe2 excitons and the cavity mode,the setup is modified to image the back focal plane of the objective26,as indicated in Fig. 2e. The energy dispersion of the heterostructureconsisting of DBR/Gr/hBN/WSe2/hBN/DBR is plotted in Fig. 2h. Thedata points are obtained from Lorentzian fits to the reflectance dis-persion (Supplementary Fig. 8). There is a clear Rabi splitting with theformation of a lower (LP) and upper (UP) polariton branches, comingfrom the strong coupling between the cavitymode andWSe2 excitons.Numerical simulations considering a coupled oscillator model27 (seeMethods) allow us to reproduce the experimental results and extract a-5  5V (V)10-2100T = 300 K V   = 3.500 VdsgV  = 8 VgV  (V)dsI   (nA)dsVsqadbechBNVsqGrVgDBRp++ SiSiO2WSe2WSe2hBNGrCo/TiVsqT20Time (ns)00.20.40.60.81EL Intensity (arb. units)10 20 30TimeVoltage       ELVsq0T2T1 T3 T4 IntensityfCo/TihBN VlatDBR VlatGNDVsqT1ExGND-11000 1103PECVD SiO2/SiNIBS Ta O5/SiO22Fig. 1 | Device structure and basic characterization. a Schematics of the devicestructure. Vsq: square voltage pulse. Vlat: lateral voltage. Vg: gate voltage. b Opticalimage of the completed device. Scale bar, 4 µm. c Scanning electron microscopy(SEM) image of the distributed Bragg reflector (DBR) cavity. Scale bar, 500nm.d Electrical transport properties of the device, showing ambipolar behavior. Insetshows the drain-source current Ids as a function of the drain-source bias voltage Vdse Schematicsof excitonic emissiondrivenbya high frequency square voltagepulse.Red (blue) circles correspond to electrons (holes). The red arrow indicates lightemission after exciton (dashed line) recombination. The black arrows correspondto the position (x) and energy (E) axis. f Time-resolved measurement of electro-luminescence (EL) intensity for a square voltage with peak-to-peak amplitudeVppsq = 15:5 V and frequency f = 8 MHz. Inset shows the correlation between ELemission and voltage applied to the contact.Article https://doi.org/10.1038/s41467-022-32292-2Nature Communications |         (2022) 13:4884 2coupling strength g = 13.4 ± 1.0meV, with a detuning between thecavity mode and the exciton energy Δ = Ec − EX = −21meV. The proxi-mity to metal contacts and the presence of the graphene gate reducesthe coupling strength compared to previous reports28,29. Supplemen-tary Fig. 11 shows further proof of the strong coupling by measuringthe Zeeman splitting under external magnetic fields.Electrical control of light emissionAfter characterizing the optical and electrical properties of the LED,wefurther apply a lateral voltage on the monolayer WSe2 to electricallycontrol the spatial, angular andpolarization properties of emitted lightat room temperature. With the application of a finite voltage on asecond contact deposited on the flake, the band alignment and thetotal external electric field aremodified. Both parameters have a directimpact on the velocity and location of carriers before recombination,which enables the electrical displacement of the polariton cloud, asshown in Fig. 3. The spatial emission profile for three representativelateral voltages Vlat is shown in Fig. 3a. The cross section along thepropagation direction is plotted in Fig. 3b, where d =0μmdenotes theposition of the source electrode. By performing a Gaussian fitting ofthe EL intensity profile, it is possible to extract the EL area andposition,which is linearly dependent on Vlat (Fig. 3c). We note that the con-formal growth of the top mirror ensures a high cavity Q-factor evenwhen emission occurs close to the contact edge (See SupplementaryFig. 7). The uniformity of the cavity mode and coupling strength isevidenced by measuring the polariton emission energy along thepropagation direction (Supplementary Fig. 10).The lateral voltage alsodetermines the resulting emission angle ofphotons out of the cavity, by modifying the momentum of electronsand holes leading to the formation of polaritons. Since the movementof carriers in WSe2 can be described by diffusive transport, theirvelocity is given by their mobility and the electric field:Energy (eV)aeb c dg h1.55 1.6024681Normalized EL1.65 1.7 1.75X0CCDObj TL BLBFP IP BFPkxλCavity WSe2, Kσ+gΓcavX-/+XX-LX0 X0300 K4 KEnergy (eV)yx110 K1.6 1.7 1.751.65Energy (eV)50100150200250 T (K) 5ELIntensity (cts)10310X0XX-LX-/+1.55 1.6 1.7 1.751.65Energy (eV)50100150200250 T (K).6X0X-/+1.551.1R/R01.64 1.68 1.72.81194 K4.8 K294 KX0X-/+R/R001Energy (eV)R/R01.6 1.7Q = 179.2CavityDevice1.65 -0.5 0.5Angle (rad)1.621.68Energy (eV)0g = 13.4 meV fX0 CUPLPFig. 2 | Excitonic light emission and strong coupling regime. a, b Temperaturedependence of EL at Vppsq = 13:5V and f = 9 MHz. X0: neutral exciton, X−/+: chargedexciton. XX−: negatively charged biexciton. L: bound exciton. Dashed lines areguides to the eye indicating the different exciton energies. c, d Temperaturedependence of reflectance before cavity growth. R0: background reflectancetaken on the heterostructure without WSe2. e Schematics of the setup used forback focal plane spectroscopy. BFPback focal plane, TL tube lens, IP image plane,BL Bertrand lens. The surface plots correspond to the reflected light intensitymeasured at the image plane and back focal plane. f Physical model forexciton–photon coupling. A cavity mode with decay rate Γcav interacts withcoupling strength g with an exciton in the K valley, generated by a circularlypolarized (σ+) laser. g Reflectance measured on the bare cavity (DBR/Gr/hBN/hBN/DBR) and device (DBR/Gr/hBN/WSe2/hBN/DBR). Continuous lines representLorentzian fits to the data, fromwherewe extract the linewidth γc = 9.132 ± 0.007meV and quality factorQ = 179.2 ± 0.5 of the cavity mode. h Energy dispersion onthe heterostructure with WSe2 together with fits to the data using a coupledoscillatormodel. The energies and standard errors are extracted fromLorentzianfits of the reflectance spectra. The upper (red, UP) and lower (blue, LP) polaritonbranches are clearly visible. The black and gray lines correspond to the energydispersion of the bare exciton and cavity modes.a b clatArea (μm )2-6 V3 V-1 V2.5 V-2d (μm)0 2 4 600.20.40.60.81V (V)d (μm)-6 -4 -20123450 240 42 44 46X (μm)30354045 Y (μm)EL IntensityV   = 3 Vlat.1140 42 44 46X (μm)V   = -1 Vlat40 42 44 46X (μm)V   = -6 VlatSLEL IntensitySL5152535SLSLFig. 3 | Electrical control of the spatial emission profile. a Spatial EL emissionpattern integrated over the whole spectral range at different lateral voltages Vlat.The source (S) and lateral (L) contacts are highlighted in gray. White dashed linesindicate the WSe2 edges. b Normalized EL intensity as a function of the distancefrom the source electrode (d), following the black dashed line in panel (a). c ELemission area at 1/e intensity and distance from the source contact, extracted fromGaussian fits to the data in panel (b). Error bars represent the standard error. Thegray rectangles indicate the position of the contacts.Article https://doi.org/10.1038/s41467-022-32292-2Nature Communications |         (2022) 13:4884 3ve,h = ± μe,hE = ±μe,hV lat=w, with w the distance between electrodes.Since the kinetic energy of single carriers is much lower than theexciton binding energy, excitons will be generated as long as electronsand holes spatially overlap. This remains valid despite the large lateralvoltages applied to the sample, since the electric field always remainslower than the dissociation field for excitons Emax < Ediss30. Consideringthat the scattering rate for electrons and holes is different, the excitonmomentum is pX =m*e ve +m*h vh≠0, withm* the carrier effective mass.Excitons with a finite in-plane momentum relax to the LP branch,where emission out of the cavity occurs through the photonic com-ponent, with pLP =pγ = _k==. The in-plane momentum k== is related tothe photon emission angle θ by k== =2πncλ0tanθc, with nc sin θc = sin θ,and θc the angle inside the cavity31. This implies that the polaritonemission angle is linearly proportional to the applied lateral voltage toa first order approximation (Supplementary Fig. 12). The EL dispersionas a function of θ is plotted in Fig. 4a–c for three representative valuesof the applied lateral voltage. The lateral voltage V0lat = �3V for whichEL is emitted normal to the 2Dplane of themonolayer is different fromzerobecause thedynamicsof carriers is also affectedby the voltageVsqapplied to the source. When V lat>V0lat, the electrically generated exci-tons acquire a net momentum in the −x direction, and photons areemitted with an angle θ < 0 with the direction perpendicular to themonolayer plane. Conversely, V lat <V0lat leads to photons emitted withangles θ >0. The energy and momentum conservation conditionsare relaxed at room temperature (kB T ~ 26meV) due to phononscattering32, which results in photons with a broad distribution ofenergy andmomentum for afixed valueof the lateral voltage, as shownin Fig. 4.Together with the photon emission angle, we can also electricallycontrol the EL polarization at room temperature. Electrical controlof EL polarization has so far only been demonstrated for electricdouble layer transistors based on WSe2. These devices, however,require an ionic liquid and can only be operated in DC at cryogenictemperatures15. Figure 4e–g shows the momentum and energyresolvedpolarization for EL at three representative values of the lateralvoltage. The circular polarization can be tuned from ρEL > 20% to ρEL <−20%. The origin of the observed polarization comes from the aniso-tropic band dispersion of carriers in WSe2 (see Methods). Under theapplication of a lateral electricfield, the electron andhole distributionsshift in momentum space. This shift induces an energy splittingbetween K and K’ excitons, which is proportional to the in-planemomentum. Exciton relaxation to the lowest energy state in one of thevalleys before scattering to the polariton can then explain theobserved valley polarization (Supplementary Fig. 15). The direction ofthe electric field with respect to the crystal lattice is crucial forachieving a high degree of polarization. In our case, the contacts aredeposited perpendicularly to the long straight edge of the WSe2 flake,which tends to have either armchair or zigzag edges33. SupplementaryFig. 13 shows the EL polarization as a function of the driving voltageamplitude Vsq before the cavity growth, which supports the trigonalwarping effect15,34 as the origin of the polarization change in our device(relevant for fields on the order of 1–10 V/μm) with negligible con-tribution from the nonlinear valley and spin currents (with a quadraticelectric field dependence). Further proof for the origin of the polar-ization is shown in Supplementary Fig. 14, where a second device isfabricated with contacts deposited along the armchair and zigzagdirections of the monolayer. In this device, the polarization is onlyobservable for the contact placed along the armchair direction, asexpected from the trigonal warp effect.Device operationFigure 5a shows the weighted arithmetic mean of the polarization foreach value of the lateral voltage (see Methods), which follows a linearbehavior formoderate values of the lateral voltage. The quasi-resonantelectrical injection (see Methods) together with a strong coupling tothe cavity mode can explain the large degree of circular polarizationshown in Fig. 535. In most experiments, non-resonant optical excita-tions are used to create a population imbalance in the K and K’ valleys.However, these high-energy excitons have a large center-of-mass wavevector.Under these conditions, there is a fast depolarization due to thelong-range Coulomb exchange interaction from electrons and holes36.Furthermore, the L-T energy splitting (for excitons with their dipolemoment parallel and perpendicular to the wave vector) increases lin-early with the wave vector, leading to a fast depolarization37. The lat-eral voltage, on the other hand, creates valley-polarized carriers with arelatively small energy andmomentumwith respect to the conductionIntensity (arb. units)a b c dAngle  (rad)1.621.641.661.68Energy (eV)1.60e f h.3.2.4.6.81V    = 1.5 VEL Intensity (arb. units)-0.25Angle  (rad) 0.250Angle  (rad)1.621.641.661.68Energy (eV)1.60-.20.2Angle  (rad) Angle  (rad)-0.2 0.2Angle (rad)-0.100.10.2PolarizationV   = 2.5lat0Angle (rad)0.20.40.60.81g-0.2 0.20lat.1-.1EL PolarizationV    = -3 V-0.25Angle  (rad) 0.250V    = -6 V-0.25  0.250V    = 1.5 V-0.25  0.250V    = -3 V-0.25  0.250Vlat = 1.5 V-0.25  0.250V    = -6 VV   = 2.5V   = -6 VlatV   = -6 Vlatlatlatlatlat lat latUPLPCX0UPLPCX0UPLPCX0UPLPCX0UPLPCX0UPLPCX0Fig. 4 | Electrical control of the emission angle and polarization. a–c EL dis-persion as a function of emission angle at three different lateral voltages.d Emission intensity along the calculated LP branch displayed as a function of theemission angle. White dashed lines correspond to the energy dispersion of theupper and lower polariton (UP, LP), cavity mode (C) and neutral exciton (X0).e–g Polarization-resolved EL dispersion as a function of emission angle at threedifferent lateral voltages. Data points with emission intensity comparable to thebackground counts are not shown. h Circular polarization along the calculated LPbranch displayed as a function of the emission angle. The data shown in (d, h) isaveraged over an energy ΔE = 1/2 kB T.Article https://doi.org/10.1038/s41467-022-32292-2Nature Communications |         (2022) 13:4884 4and valence band edges. In addition, the large spatial extent of thepolariton wavefunction38 reduces the intravalley scattering due to thedisorder potential39, resulting in a longer valley pseudospin relaxationtime compared to intralayer excitons37,40. Furthermore, the negativedetuning in our cavity reduces the intervalley scattering, which onlyoccurs through the excitonic component. To exclude the possibilitythat the observed polarization and directionality arises from the dis-placement of EL with respect to the focal point of the objective, weperform a calibration experiment (Supplementary Fig. 16), discussedin Methods. We further demonstrate that the spatial variability of thecavity coupled exciton emission angle and polarization is negligiblecompared to the modulation achieved by the lateral voltage (Supple-mentary Fig. 17).DiscussionThe successful integration of electrically tunable light emitting diodesin planar cavities achieved in this work (Fig. 5b) constitutes a crucialstep in the development of practical devices to study light-matterinteraction, where the frequency, position, directionality, and polar-ization can be tuned electrically at room temperature. This tunabilitywould facilitate the coupling of EL to other photonic structures suchasdiffraction gratings, waveguides and resonators, and can relax thealignment challenges of optical to electrical interconnects. The sim-plicity of the method for EL generation implies that large-area CVDgrown TMDCs with metal contact arrays and top cavity growth couldbe used to achieve high integration densities. Furthermore, the largetunneling current per voltage pulse together with the high frequencyoperation could lead to a transient population inversion and elec-trically driven lasing. In the regime of strong-coupling, polariton lasingcould be achieved at lower thresholds, with an electrically tunablechirality41.MethodsDevice fabricationThe substrate consists of a DBR mirror formed by 12 pairs of Ta2O5/SiO2 layers deposited by ion beam sputtering (IBS) on top of a dopedsilicon wafer. Single layer graphene flakes for the bottom gate wereobtained by mechanical exfoliation from graphite (NGS) on the DBRsubstrates and patterned to the final shape by electron beam litho-graphy and oxygen plasma etching. Few-layer h-BN and monolayerWSe2 (HQ Graphene) were stacked using the dry polymer-assistedtransfer42 after exfoliation on a viscoelastic stamp. The thin h-BNencapsulation and tunneling layer was transferred using a wetpolymer-assisted transfer after exfoliation on apolymer double layer43.The thickness of the bottom and top hBN is 29 nm and 1 nm, respec-tively. After the stack was completed, the heterostructure was ther-mally annealed at 340 °C under high vacuum (10−6 mbar) for 12 h. Co(35 nm)/Ti (10 nm) contactsweredefinedperpendicularly to oneof theWSe2 crystal axes by electron-beam lithography and deposited byelectron beam evaporation. The device was again annealed at 250 °Cunder high vacuum (10−6mbar) for 12 h. Finally, the topSiO2 spacer andnine pairs of SiO2/Si3N4 layers were deposited by plasma enhancedchemical vapor deposition (PECVD). The second device analyzed inSupplementary Fig. 11 was fabricated following the same steps as themain device, but Ti (2 nm)/Au (80nm) metal contacts were directlydeposited on top of WSe2 before the transfer of the hBNencapsulation layer.Optical and electrical measurementsAll measurements presented in this work were performed undervacuum at a temperature of 300K, unless specified otherwise.Reflectance measurements were taken by focusing a fiber- coupledwhite light source onto a 2.5 µm spot on the sample. The incidentpower was 90 µW. Photoluminescence was generated by excitationwith a continuous-wave 647 nm laser diode focused to a diffraction-limited spot size of about 1 µm. The incident power was 200 µW. Thespectral and spatial characteristics of the device emission were ana-lysed simultaneously. The emitted light was acquired using a spec-trometer (Andor Shamrock with Andor Newton CCD camera), and thelaser line was removed with a long-pass 650 nm edge filter. For spatialimaging, light was collected using an Andor Ixon CCD camera afterpassing through a long-pass 700nm edge filter. For polarization-resolved measurements, a rotator with a λ/2 plate was used to controlthe polarization incident on a λ/4 plate between the sample and theobjective. A calcite beam displacer was placed before the spectro-meter to separate the two measured polarizations. Angular resolvedmeasurementswere obtainedby imaging the Fourier plane on theCCDcamera after the spectrometer. Two convergent lenses are used toimage the back focal plane of the objective, with the spectrometer slitclosed to 10 µm. Electroluminescence was generated by applying asquare voltage wave using an Agilent 33520 function generator. Time-resolved EL measurements were obtained with a silicon avalanchephotodiode (APD) from Excelitas (SPCM-AQRH) and time correlatedsingle photon counting (TCSPC) system (Picoharp 300).Cavity reflectance and exciton-photon couplingThe normalized reflectance for the cavity mode is plotted in Fig. 2g.The back reflected intensity measured when exciting from the topmirror (Rt) follows the Airy distribution44:IrefI inc=ffiffiffiffiRtp�ffiffiffiffiRbp� �2+ 4ffiffiffiffiffiffiffiffiRtRbpsin2ϕ1�ffiffiffiffiffiffiffiffiRtRbp� �2+ 4ffiffiffiffiffiffiffiffiRtRbpsin2ϕð1ÞWith 2ϕ = 2πνtRT, ν the frequency and tRT the round-trip time inthe cavity. We use this equation to fit the reflectance spectra andextract the linewidth and resonant energy. The reflectance dispersiona bV  (V)latDBRSiOVsq VlatV  > 0latV  < 0latσ -σ+-6 -4 -2 0 20.51.01.52.02.5EL Intensity (cts/s)1050.0-10-50510EL Polarization (%)-10-505EL Angle (º)-152Fig. 5 | Device operation. a Weighted arithmetic mean of EL angle (gray), polar-ization (red) and integrated intensity (blue) extracted from the dispersionmaps inFig. 4. Error bars represent the standard error. The gray box covers the data pointsfor which the detected photon counts are of the same order of magnitude as thenoise. b Schematic representation of the device operation, where the properties ofemitted light can be electrically controlled.Article https://doi.org/10.1038/s41467-022-32292-2Nature Communications |         (2022) 13:4884 5for the bare cavity and device heterostructure is plotted in Supple-mentary Fig. 8. The cavity mode energy is calculated asEc = Eph=ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1� sin θ=nc� �2q, with Eph the photon energy andnc =nSiO2tSiO2 +nhBNthBNtSiO2 + thBN= 1:538 the effective refractive index based on thefilling fractions of SiO2 and hBN45.The reflectance dispersion in Supplementary Fig. 8 is fitted with adouble Lorentzian function to extract the energies of the polaritonbranches. The energy dispersion is then fitted according to equation27:EUP,LP =12 EX + EC � i γc + γX� �� �± 12ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi4g2 + EX � Ec + i γc � γX� �� �2qð2ÞFor thefitting, the exciton and cavity linewidths arefixed, and takevalues γc = 9.132 ± 0.007 meV (Lorentzian fits from Fig. 2g) and γX =27.9 ± 0.2 meV (Gaussian fits from Supplementary Fig. 6). The cou-pling strength extracted from the fitting is g = 13.4 ± 1.0meV. The Rabisplitting can then be calculated as ℏΩ = 19 ± 3meV, with_Ω= _ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi4g2 � γc � γX� �2q. The condition for strong couplingg > ∣γX�γc ∣2 = 9:3 ±0:1meV is therefore satisfied in our sample. In addi-tion, for the two resonances to be spectrally separable, the energysplitting needs to be larger than the sum of the half-linewidths:ℏΩ > (γX + γc)/2 = 18.5 ± 0.1meV. The fact that the Rabi splitting iscomparable to the half-linewidths makes the observation of strongcoupling difficult in our sample.The measured splitting in reflectance is usually smaller than theintrinsic energy splitting of polariton states. In the high reflectivitylimit, and assuming exciton-cavity resonance27:_Ωref = 2ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffig4 1 + 2γXγc� �2+ 2g2γ2X 1 + γXγc� �r� 2g2γXγc� γ2Xsð3ÞThis equation can be used to recognize the underestimation of gin reflectance measurements. Considering the observed Rabi splittingof ℏΩref = 19 ± 3meV and the experimental values of γX and γc, thecoupling strength can be derived from the above equation as gi =21.3 ± 0.6meV. The coupling strength best describing the EL energydispersion in Fig. 4 is gEL = 17.4 ± 0.9meV = (gref + gi)/2.Pulsed electrical injectionThe transient origin of current injection results in a localized emissionclose to the source electrode. However, as shown in Fig. 3a, theintensitymaximum is not centered at the contact, due to strong lateralelectric fields that deplete the electron and hole populations awayfrom the electrode. The initial tunneling of electrons and holes cangenerate hot carriers with a large wave-vector and energy above thebandgap. On the other hand, electrons and holes recombine at a dis-tance d > 500nm from the contact (Fig. 3a). This implies that carriersneed a travel time τt = d·w/(μe,h·Vlat) > 5 ps to reach the recombinationspot, with μe,h taken from46. Therefore they relax from the excitedinjection condition before exciton formation occurs (scattering rateτs≪ τt)47.Valley polarization of carriers under external electric fieldsLateral electric fields shift the distribution of electrons and holes in k-space, with a distribution function for carriers that can be phenom-enologically described by15:f c,v kð Þ= f 0c,v k� qτ_ E� � ð4Þwhere f 0c,v r,kð Þ=Nc,vexp � ∣ϵc,v kð Þ�μc,v rð Þ∣kBTeff� �is the distribution of electronsand holes at the band extrema. Nc,v is the normalization factor, ϵc,v (k)the banddispersion, μc,v (r) the chemical potential andTeff the effectivecarrier temperature. The relaxation time is given by τ = μc,vm*c,v∣q∣ , with μc,vthe carrier mobility and m*c,v the carrier effective mass.The energy dependent emission intensity resulting fromelectron-hole recombination is given by the Fermi golden rule, which for aconstant dipole moment takes the form:I Eð Þ / Rδ E � ϵc kð Þ+ ϵv kð Þð Þ2 + δ2h i�1f v kð Þf c kð Þd2k ð5Þwhere δ represents the broadening. Due to the anisotropy in thevalence band of TMDCs at the K and K’ valleys, the resulting lumi-nescence is circularly polarized: IK Eð Þ ≠ IK’ Eð Þ: This mechanism isresponsible for the valley polarization observed in the light emittingdiode without the cavity (Supplementary Figs. 13, 14). However, thismodel is not valid in the strong coupling regime. The exciton-polaritonvalley polarization requires a more complex kinetic equation con-sidering relaxation rates from the exciton reservoir to the polaritonstates, exciton-cavity detuning, coherence and valley lifetimes. Atheoretical model for the case of TMDCs integrated in optical cavitieshas beenpreviously developed38. Thismodel canbecombinedwith theanisotropic band dispersion of carriers to understand the observedvalley polarization. In Supplementary Fig. 15 we show a simplifiedmodel in one dimension in reciprocal space to explain the effect of alateral electric field on the exciton dispersion and polariton polariza-tion. First, we plot the energy dispersion for single carriers at the K andK’ valleys taking into account the trigonal warp effect by considering atwo-band k·pmodel up to thirdorder in the crystalmomentum48,49. Forsimplicity, we focus on the crystal direction kx, with x being the axisalong the zigzag direction of the lattice, where the electric field isapplied. The exciton dispersion for both valleys at zero electric field isthen simply calculated from the electron and hole energies. Even atzero field, there is a sizable splitting at large in-plane momentum.However, both K and K’ excitons are degenerate in energy at thebottom of the bands. When applying an in-plane electric field, theFermi levels of electrons and holes shift inmomentum space by qτ_ E, asdescribed before. For an applied field of E = −1 V/μm and relaxationrate50 τ = 150 fs, the exciton dispersion becomes highly anisotropic forthe K and K’ valleys, as shown in Supplementary Fig. 15c. Excitonsgenerated at kx >0 then relax to the K’ valley, which becomes thelowest energy state. These parameters represent the experimentalconditions in Fig. 4a, e. The magnitude of this effect is stronglydependent on the carrier relaxation rate. Taking τ = 400 fs, the excitonenergy splittingbetween theKandK’ valleys can reachΔE = 5meV. Thissimple model can qualitatively explain the observed valley polariza-tion, coming from exciton relaxation to one of the valleys beforescattering to the polariton state.Simulations of exciton–photon momentum and energyconservationTo theoretically compute the photon emission angle as a function ofthe lateral electric field, we apply momentum conservation to therecombination process (pγ =pLP=pX0 ). Here we neglect Coulombinteractions and phonon scattering. The photon momentum can becalculated from its emission angle with respect to the normal directionto the monolayer plane θ, as described in the main text. The excitonmomentum is calculated from the electron and hole velocities,pX0 =m*eve +m*hvh. Carrier velocities are limited by the mobility in thediffusive regime, with ve,h = μe,hVlat/w, where μe,h is the carriermobility,Vlat the lateral voltage and w is the width of the channel. The effectivemass for electrons and holes is taken as m*h =0:51me andm*e = 0:39me47. The dependence of the photon emission angle with thelateralfield is calculated in SupplementaryFig. 12 fordifferent valuesofthe electron and hole mobilities. It should be noted that theArticle https://doi.org/10.1038/s41467-022-32292-2Nature Communications |         (2022) 13:4884 6simulations can only qualitatively describe the observed change in thephoton emission angle. A more detailed description considering thecarrier dynamics at the metal–semiconductor junction would be nee-ded for a quantitative description but is outside of the scope ofthis work.Reference experimentTo prove that the observed polarization and directionality is not anartifact related to the spatial displacement of EL at different lateralvoltages, we perform an experiment where the objective position isscanned for fixed EL emission conditions. Supplementary Fig. 16demonstrates that themovement of the focal point with respect to theEL location cannot explain the large change in emission angle andpolarization observed in the main text.Magnetic field dependence of electroluminescenceFurther proof for the presenceof strong coupling in our device is givenby the external response of the LP state to an external magnetic field,which is shown in Supplementary Fig. 11. The exciton and photonfractions in the LP state are given by the Hopfield coefficients51, whichare calculated for our device in Supplementary Fig. 9. The negativedetuning Δ = −21meV implies that at θ =0 rad emission angle, thephoton fraction |C0|2 is much larger than the exciton fraction |X0|2,resulting in a weak interaction of the polariton state with themagneticfield. At θ =0 rad, the Zeeman splitting for the two circular polarizationstates is ΔER�L = gLPμBB, with gLP = 0:41 ± 1:19 the polariton g-factorand μB the Bohr magneton. This value is not statistically significantsince the noise is higher than the energy splitting with magnetic field.At larger emission angles, the Zeeman splitting is higher due to theincrease in the exciton fraction of the lower polariton. At θ =0.15 rad,the g-factor becomes gLP = 1.06 ± 0.28. The exciton g-factor can becalculated asgX =gLP∣X0:15 ∣2 = 3:53 ±0:93, in good agreementwithpreviousreports52. Exciton polaritons from the K and K’ valleys have the sameenergy and effective mass at B = 0 T, but they preserve the circularpolarization associated to the spin-orbit (SO) coupling of the baremonolayer exciton. Supplementary Fig. 11c demonstrates how EL cir-cular polarization can be tuned with external magnetic fields.Data analysisElectroluminescence polarization is defined asρEL = ðIK � IK’Þ=ðIK + IK’Þ � ρ0EL, where IK (IK’) corresponds to right (left)circularly polarized intensity and ρ0EL corresponds to the circularpolarization obtained when Vlat = 0 and Vsq = 16 V.The extracted polarization in Fig. 5 is the weighted arithmeticmean of the polarization of each pixel in the image of the Fourierplane, ρwEL =∑ρiEL IiT∑I iT, where I iT = IiK + IiK’ is the total intensity for pixel i.Pixels with intensity I < 0.35 Imax are not taken into account, to reducenoise. Similarly, the emission angle is calculated as θwEL =∑θiEL IiT∑I iT. The ELintensity in Fig. 5 corresponds to the total integrated intensity for allpixels in the detector, with the subtraction of the darkcounts ITEL =∑ðI iEL � IdarkEL Þ.For magnetic-field-dependent EL measurements, the degree ofcircular polarization ρEL is calculated by taking the average polariza-tion of theN = 130 values around the peak energy, corresponding to anenergy window of kBT/2, with kB the Boltzmann constant and T = 300K. The emission energy EEL is calculated as a weighted average oversame energy window, with EEL =∑E i I iT∑I iT, where Ei is the emission energyfor a point in the spectra and I iT = IiK + IiK’ the corresponding intensity.The standard errors are computed considering the error in theintensity of each pixel IK as ΔIK = 60 cts (intensity fluctuations), theerror in the spectral energy E as ΔE =0.002meV (spectrometerresolution) and the error in the measured angle θ as Δθ =0.03 rad(angular resolution).Data availabilityThe data generated in this study have been deposited in Zenododatabase at https://doi.org/10.5281/zenodo.6850668.References1. Mak, K. F. & Shan, J. Photonics and optoelectronics of 2D semi-conductor transition metal dichalcogenides. Nat. Photonics 10,216–226 (2016).2. Lopez-Sanchez, O., Lembke, D., Kayci, M., Radenovic, A. & Kis, A.Ultrasensitive photodetectors based on monolayer MoS2. Nat.Nanotechnol. 8, 497–501 (2013).3. Ross, J. S. et al. Electrically tunable excitonic light-emitting diodesbased on monolayer WSe2 p–n junctions. Nat. Nanotechnol. 9,268–272 (2014).4. Sundaram, R. S. et al. Electroluminescence in single layer MoS2.Nano Lett. 13, 1416–1421 (2013).5. Withers, F. et al. Light-emitting diodes by band-structure engi-neering in van der Waals heterostructures. Nat. Mater. 14,301–306 (2015).6. Mueller, T., Xia, F. & Avouris, P. Graphene photodetectors for high-speed optical communications. Nat. Photonics 4, 297–301 (2010).7. Bie, Y.-Q. et al. A MoTe 2 -based light-emitting diode and photo-detector for silicon photonic integrated circuits. Nat. Nanotechnol.12, 1124–1129 (2017).8. Youngblood, N., Chen, C., Koester, S. J. & Li, M. Waveguide-integrated black phosphorus photodetector with high responsivityand low dark current. Nat. Photonics 9, 247–252 (2015).9. Miller, D. A. B. Attojoule optoelectronics for low-energy informationprocessing and communications. J. Light. Technol. 35,346–396 (2017).10. Liu, Y. et al. Room temperature nanocavity laser with interlayerexcitons in 2D heterostructures. Sci. Adv. 5, eaav4506 (2019).11. Paik, E. Y. et al. Interlayer exciton laser of extended spatialcoherence in atomically thin heterostructures. Nature 576,80–84 (2019).12. Peyskens, F., Chakraborty, C., Muneeb, M., Van Thourhout, D. &Englund, D. Integration of single photon emitters in 2D layeredmaterials with a silicon nitride photonic chip. Nat. Commun. 10,4435 (2019).13. Xiao, D., Liu, G.-B., Feng, W., Xu, X. & Yao, W. Coupled spin andvalley physics in monolayers of MoS2 and other group-VI dichal-cogenides. Phys. Rev. Lett. 108, 196802 (2012).14. Avsar, A. et al. Optospintronics in graphene via proximity coupling.ACS Nano 11, 11678–11686 (2017).15. Zhang, Y. J., Oka, T., Suzuki, R., Ye, J. T. & Iwasa, Y. Electricallyswitchable chiral light-emitting transistor. Science 344,725–728 (2014).16. Ye, Y. et al. Electrical generation and control of the valley carriers ina monolayer transition metal dichalcogenide. Nat. Nanotechnol. 11,598–602 (2016).17. Shambat, G. et al. Ultrafast direct modulation of a single-modephotonic crystal nanocavity light-emitting diode. Nat. Commun. 2,539 (2011).18. Lozano, G., Rodriguez, S. R., Verschuuren, M. A. & Gómez Rivas, J.Metallic nanostructures for efficient LED lighting. Light Sci. Appl. 5,e16080–e16080 (2016).19. Wang, J. et al. Routing valley exciton emission of aWS 2monolayervia delocalized Bloch modes of in-plane inversion-symmetry-broken photonic crystal slabs. Light Sci. Appl. 9, 148 (2020).20. Paur, M. et al. Electroluminescence from multi-particle excitoncomplexes in transition metal dichalcogenide semiconductors.Nat. Commun. 10, 1709 (2019).Article https://doi.org/10.1038/s41467-022-32292-2Nature Communications |         (2022) 13:4884 7https://doi.org/10.5281/zenodo.685066821. Lien, D.-H. et al. Large-area and bright pulsed electroluminescencein monolayer semiconductors. Nat. Commun. 9, 1229 (2018).22. Robert, C. et al. Exciton radiative lifetime in transition metaldichalcogenide monolayers. Phys. Rev. B 93, 205423 (2016).23. Zhang, B., Kano, S. S., Shiraki, Y. & Ito, R. Reflectance study of theoscillator strength of excitons in semiconductor quantum wells.Phys. Rev. B 50, 7499–7508 (1994).24. Masselink, W. T. et al. Absorption coefficients and exciton oscillatorstrengths in AlGaAs-GaAs superlattices. Phys. Rev. B 32,8027–8034 (1985).25. Pozo-Zamudio, O. D. et al. Electrically pumped WSe2-based light-emitting van der Waals heterostructures embedded in monolithicdielectric microcavities. 2D Mater. 7, 031006 (2020).26. Richard, M., Kasprzak, J., Romestain, R., André, R. & Dang, L. S.Spontaneous coherent phase transition of polaritons in CdTemicrocavities. Phys. Rev. Lett. 94, 187401 (2005).27. Savona, V., Andreani, L. C., Schwendimann, P. & Quattropani, A.Quantum well excitons in semiconductor microcavities: unifiedtreatment of weak and strong coupling regimes. Solid State Com-mun. 93, 733–739 (1995).28. Schneider, C., Glazov, M. M., Korn, T., Höfling, S. & Urbaszek, B.Two-dimensional semiconductors in the regime of strong light-matter coupling. Nat. Commun. 9, 2695 (2018).29. Gu, J., Chakraborty, B., Khatoniar, M. & Menon, V. M. A room-temperature polariton light-emitting diodebased onmonolayerWS2. Nat. Nanotechnol. 14, 1024–1028 (2019).30. Massicotte, M. et al. Dissociation of two-dimensional excitons inmonolayer WSe2. Nat. Commun. 9, 1633 (2018).31. Schneider, L. M. et al. Shedding light on exciton’s nature inmonolayer quantum material by optical dispersion measurements.Opt. Express 27, 37131–37149 (2019).32. He, M. et al. Valley phonons and exciton complexes in a monolayersemiconductor. Nat. Commun. 11, 618 (2020).33. Geim, A. K. & Novoselov, K. S. The rise of graphene. Nat. Mater. 6,183–191 (2007).34. Kormányos, A. et al.MonolayerMoS2: Trigonalwarping, theGammavalley, and spin-orbit coupling effects. Phys. Rev. B 88,045416 (2013).35. Król, M. et al. Valley polarization of exciton–polaritons inmonolayerWSe2 in a tunable microcavity. Nanoscale 11, 9574–9579 (2019).36. Glazov, M. M. et al. Exciton fine structure and spin decoherence inmonolayers of transition metal dichalcogenides. Phys. Rev. B 89,201302 (2014).37. Maialle, M. Z., de Andrada e Silva, E. A. & Sham, L. J. Excitonspin dynamics in quantum wells. Phys. Rev. B 47,15776–15788 (1993).38. Dufferwiel, S. et al. Valley-addressable polaritons in atomically thinsemiconductors. Nat. Photonics 11, 497–501 (2017).39. Whittaker, D. M. et al. Motional narrowing in semiconductormicrocavities. Phys. Rev. Lett. 77, 4792–4795 (1996).40. Qiu, L., Chakraborty, C., Dhara, S. & Vamivakas, A. N. Room-temperature valley coherence in a polaritonic system. Nat. Com-mun. 10, 1513 (2019).41. Schneider, C. et al. An electrically pumped polariton laser. Nature497, 348–352 (2013).42. Castellanos-Gomez, A. et al. Deterministic transfer of two-dimensional materials by all-dry viscoelastic stamping. 2D Mater. 1,011002 (2014).43. Mayorov, A. S. et al. Micrometer-scale ballistic transport in encap-sulated graphene at room temperature. Nano Lett. 11,2396–2399 (2011).44. Ismail, N., Kores, C. C., Geskus, D. & Pollnau, M. Fabry-Pérotresonator: spectral line shapes, generic and related Airy dis-tributions, linewidths, finesses, and performance at low orfrequency-dependent reflectivity. Opt. Express 24,16366–16389 (2016).45. Liu, X. et al. Strong light–matter coupling in two-dimensionalatomic crystals. Nat. Photonics 9, 30–34 (2015).46. Allain, A. & Kis, A. Electron and holemobilities in single-layerWSe2.ACS Nano 8, 7180–7185 (2014).47. Jin, Z., Li, X., Mullen, J. T. & Kim, K. W. Intrinsic transport propertiesof electrons and holes in monolayer transition metal dichalcogen-ides. Phys. Rev. B 90, 045422 (2014).48. Liu, G.-B., Shan, W.-Y., Yao, Y., Yao, W. & Xiao, D. Three-band tight-binding model for monolayers of group-VIB transition metaldichalcogenides. Phys. Rev. B 88, 085433 (2013).49. Chen, W., Zhou, X., Liu, P., Xiao, X. & Zhou, G. Effect of trigonalwarping on the Berry curvature and valley/spin Hall effects inmonolayer MoS2. Phys. Lett. A 384, 126344 (2020).50. Yadav, D., Trushin, M. & Pauly, F. Thermalization of photoexcitedcarriers in two-dimensional transition metal dichalcogenides andinternal quantum efficiency of van der Waals heterostructures.Phys. Rev. Res. 2, 043051 (2020).51. Hopfield, J. J. Theory of the contribution of excitons to thecomplex dielectric constant of crystals. Phys. Rev. 112,1555–1567 (1958).52. Li, Z.,Wang, T., Miao, S., Lian, Z. & Shi, S.-F. Fine structures of valley-polarized excitonic states in monolayer transitional metal dichal-cogenides. Nanophotonics 9, 1811–1829 (2020).AcknowledgementsThis work was financially supported by the European Research Council(grant no. 682332), the Swiss National Science Foundation (grants no.175822, 177007 and 164015). This project has received funding from theEuropean Union’s Horizon 2020 research and innovation programmeunder grant agreement No 785219 and 881603 (Graphene FlagshipCore2 and Core 3 phases) as well as support from the CCMX MaterialsChallenge grant “Large area growth of 2D materials for device integra-tion”. K.W. and T.T. acknowledge support from JSPS KAKENHI (GrantNumbers 19H05790, 20H00354 and 21H05233).Author contributionsA.K. initiated and supervised the project. J.G. fabricated the device withinput from C.Y. J.G. performed the optical measurements with the helpofD.U. J.G. analysed thedataandperformed the simulations. F.T. helpedwith LabVIEW data acquisition. Z.S. helped with time resolved mea-surements. K.W. and T.T. grew the h-BN crystals. J.G. and A.K. wrote themanuscript with input from all authors.Competing interestsThe authors declare no competing interests.Additional informationSupplementary information The online version containssupplementary material available athttps://doi.org/10.1038/s41467-022-32292-2.Correspondence and requests for materials should be addressed toAndras Kis.Peer review information Nature Communications thanks the anon-ymous reviewer(s) for their contribution to the peer review of this work.Reprints and permission information is available athttp://www.nature.com/reprintsPublisher’s note Springer Nature remains neutral with regard tojurisdictional claims in published maps and institutional affiliations.Article https://doi.org/10.1038/s41467-022-32292-2Nature Communications |         (2022) 13:4884 8https://doi.org/10.1038/s41467-022-32292-2http://www.nature.com/reprintsOpen Access This article is licensed under a Creative CommonsAttribution 4.0 International License, which permits use, sharing,adaptation, distribution and reproduction in any medium or format, aslong as you give appropriate credit to the original author(s) and thesource, provide a link to the Creative Commons license, and indicate ifchanges were made. The images or other third party material in thisarticle are included in the article’s Creative Commons license, unlessindicated otherwise in a credit line to the material. If material is notincluded in the article’s Creative Commons license and your intendeduse is not permitted by statutory regulation or exceeds the permitteduse, you will need to obtain permission directly from the copyrightholder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.© The Author(s) 2022Article https://doi.org/10.1038/s41467-022-32292-2Nature Communications |         (2022) 13:4884 9http://creativecommons.org/licenses/by/4.0/http://creativecommons.org/licenses/by/4.0/ Room-temperature electrical control of�polarization and emission angle in a cavity-integrated 2D pulsed LED Results Device design and basic characterization Exciton-photon coupling Electrical control of light emission Device operation Discussion Methods Device fabrication Optical and electrical measurements Cavity reflectance and exciton-photon coupling Pulsed electrical injection Valley polarization of carriers under external electric fields Simulations of exciton&#x02013;nobreakphoton momentum and energy conservation Reference experiment Magnetic field dependence of electroluminescence Data analysis Data availability References Acknowledgements Author contributions Competing interests Additional information