# Fileset

[[21928614 - Nanophotonics] Asymmetric photoelectric effect Auger-assisted hot hole photocurrents in transition metal dichalcogenides.pdf](https://mdr.nims.go.jp/filesets/b97c50c1-ab85-472b-b5e4-40206911fed8/download)

## Creator

Andrey Sushko, Kristiaan De Greve, Madeleine Phillips, Bernhard Urbaszek, Andrew Y. Joe, [Kenji Watanabe](https://orcid.org/0000-0003-3701-8119), [Takashi Taniguchi](https://orcid.org/0000-0002-1467-3105), Alexander L. Efros, C. Stephen Hellberg, Hongkun Park, Philip Kim, Mikhail D. Lukin

## Rights

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

## Other metadata

[Asymmetric photoelectric effect: Auger-assisted hot hole photocurrents in transition metal dichalcogenides](https://mdr.nims.go.jp/datasets/2cfea359-7b62-4415-b589-00586b6de019)

## Fulltext

NanophotonicsResearch articleAndrey Sushko*, Kristiaan De Greve, Madeleine Phillips, Bernhard Urbaszek, Andrew Y. Joe,Kenji Watanabe, Takashi Taniguchi, Alexander L. Efros, C. Stephen Hellberg, Hongkun Park,Philip Kim and Mikhail D. LukinAsymmetric photoelectric effect: Auger-assistedhot hole photocurrents in transition metaldichalcogenideshttps://doi.org/10.1515/nanoph-2020-0397Received July 17, 2020; acceptedSeptember 3, 2020; publishedonlineSeptember 25, 2020Abstract: Transition metal dichalcogenide (TMD) semi-conductor heterostructures are actively explored as a newplatform for quantum optoelectronic systems. Most state ofthe art devices make use of insulating hexagonal boronnitride (hBN) that acts as a wide-bandgap dielectricencapsulating layer that also provides an atomicallysmooth and clean interface that is paramount for properdevice operation. We report the observation of large,through-hBN photocurrents that are generated upon opti-cal excitation of hBN encapsulated MoSe2 and WSe2monolayer devices. We attribute these effects to Augerrecombination in the TMDs, in combination with anasymmetric band offset between the TMD and the hBN. Wepresent experimental investigation of these effects andcompare our observations with detailed, ab-initiomodeling. Our observations have important implicationsfor the design of optoelectronic devices based on encap-sulated TMDdevices. In systemswhere precise charge-statecontrol is desired, the out-of-plane current path presentsboth a challenge and an opportunity for optical dopingcontrol. Since the current directly depends on Augerrecombination, it can act as a local, direct probe of both theefficiency of the Auger process aswell as its dependence onthe local density of states in integrated devices.Keywords:Auger excitation; 2Dmaterials; optoelectronics;transition metal dichalcogenides.1 IntroductionTransition metal dichalcogenides (TMDs) [1] have recentlyattracted significant interest for their optoelectronicproperties [2], which are dominated by strongly boundexcitons. As van der Waals (vdW) 2D materials, TMDs canbe incorporated into complex, high cleanliness vdW het-erostructures tailored to a myriad of possible applications[3]. In particular, such systems can be used to isolate andmanipulate electronic and excitonic excitations whichallow the creation of engineered, controlled quantumsystems [4, 5]. In general, such heterostructures rely onhexagonal boron-nitride (hBN) [6] as an atomically cleandielectric encapsulation layer to separate the active ma-terials from each other, surrounding electrostatic gates,*Corresponding author: Andrey Sushko, Department of Physics,Harvard University, Cambridge, Massachusetts 02138, USA,E-mail: asushko@g.harvard.edu. https://orcid.org/0000-0002-2756-9753Kristiaan De Greve, Department of Physics, Harvard University,Cambridge,Massachusetts 02138, USA; andDepartment of Chemistryand Chemical Biology, Harvard University, Cambridge,Massachusetts02138, USA; and Currently at Imec, Kapeldreef 75, Leuven, Belgium,E-mail: kristiaan.degreve@gmail.comMadeleine Phillips, Alexander L. Efros and C. Stephen Hellberg,NavalResearch Laboratory (NRL), Washington, DC 20375, USA,E-mail: madeleine.phillips.ctr@nrl.navy.mil (M. Phillips),sasha.efros@nrl.navy.mil (A.L. Efros), steve.hellberg@nrl.navy.mil(C.S. Hellberg)Bernhard Urbaszek, Université de Toulouse, INSA-CNRS-UPS, LPCNO,135 AvenueRangueil, 31077 Toulouse, France, E-mail: urbaszek@insa-toulouse.frAndrew Y. Joe, Philip Kim and Mikhail D. Lukin, Department ofPhysics, Harvard University, Cambridge, Massachusetts 02138, USA,E-mail: andrewjoe@g.harvard.edu (A.Y. Joe),pkim@physics.harvard.edu (P. Kim), lukin@physics.harvard.edu(M.D. Lukin)Kenji Watanabe, Research Center for Functional Materials, NationalInstitute for Materials Science, 1-1 Namiki, Tsukuba, 305-0044, Japan,E-mail: watanabe.kenji.aml@nims.go.jp. https://orcid.org/0000-0003-3701-8119Takashi Taniguchi, International Center for MaterialsNanoarchitectonics, National Institute for Materials Science, 1-1Namiki, Tsukuba, 305-0044, Japan,E-mail: TANIGUCHI.Takashi@nims.go.jpHongkun Park, Department of Physics, Harvard University,Cambridge,Massachusetts 02138, USA; andDepartment of Chemistryand Chemical Biology, Harvard University, Cambridge,Massachusetts02138, USA, E-mail: Hongkun_Park@harvard.eduNanophotonics 2021; 10(1): 105–113Open Access. © 2020 Andrey Sushko et al., published by De Gruyter. This work is licensed under the Creative Commons Attribution 4.0International License.https://doi.org/10.1515/nanoph-2020-0397mailto:asushko@g.harvard.eduhttps://orcid.org/0000-0002-2756-9753https://orcid.org/0000-0002-2756-9753mailto:kristiaan.degreve@gmail.commailto:madeleine.phillips.ctr@nrl.navy.milmailto:sasha.efros@nrl.navy.milmailto:steve.hellberg@nrl.navy.milmailto:urbaszek@insa-toulouse.frmailto:urbaszek@insa-toulouse.frmailto:andrewjoe@g.harvard.edumailto:pkim@physics.harvard.edumailto:lukin@physics.harvard.edumailto:watanabe.kenji.aml@nims.go.jphttps://orcid.org/0000-0003-3701-8119https://orcid.org/0000-0003-3701-8119mailto:TANIGUCHI.Takashi@nims.go.jpmailto:Hongkun_Park@harvard.eduand the environment [7]. Typically, these hBN layers aretreated as an inert buffer whose wide 6 eV bandgap [8]allows it to serve as both a physical and electronic barrierbetween different parts of the heterostructure device.Deviations from this simplified picture are mostlyconsidered in the context of trapped charge defects ordielectric breakdown.In this article, we describe an experimental observa-tion of a novel optoelectronic effect that challenges thissimple physical picture of perfectly insulating hBNencapsulation. Specifically, we report the robust observa-tion of a reversible, photoinduced current that appearsacross the thick, dielectric hBN layer. The current isobserved in dozens of devices with varying geometries andhBN thicknesses, consistent with other recent reports[9, 10]. We report on a systematic doping, wavelength,electric field and thickness dependent study of this effect intwo different TMD materials, which allows us to unam-biguously point to Auger recombination as the centralmechanism involved. Our evidence is multifold. Firstly, weobserve photocurrents over a wide range of hBN thick-nesses (3–90 nm in our devices) and, secondly, whendeconvolved from optical doping (Supplementary Fig-ures 4–6), it is spatially uniform throughout all devices– asverified via spatially scanning the excitation beam and byverification using split gate devices. Such uniformity andthickness independence make alternative explanationssuch as dielectric breakdown or tunneling via in-gap defectstates in the hBN unlikely.On the other hand, our systematic doping-, field- andwavelength dependence studies, corroborated by theoret-icalmodeling of relevant barrier heights in the twomaterialsystems, allow us to extract a Fowler–Nordheim tunnelingpicture that is activated by an Auger process involvingholes and excitons, which differs from previous pictures[9, 11] yet has some similarities with hot-carrier effectsas previously reported in graphene-hBN heterostructures[12–14]. Crucially, this picture explains the substantialdifferences in photocurrent efficiency between the twomaterials directly from computed band alignments,without postulating significant material-dependent varia-tion in the efficiency of Auger excitation. This process addsan important element to the physics of two-dimensionalTMD devices by introducing an optically controlled trans-port path outside the material. Potentially, it can beleveraged to locally sink unipolar currents from an opti-cally defined “contact” that can be arbitrarily swept over adevice structure. In addition, we show that the variation incurrent generation efficiency can provide insight into thedynamics, transitions, and relaxation pathways of stateswithin the TMD.2 Photocurrents in encapsulatedTMDsFigure 1a shows the schematic of a typical device structure,consisting in this case of a MoSe2 monolayer encapsulatedin hBN over a metal gate electrode. Upon off-resonant op-tical excitation at 660 nm, a substantial current ismeasured at the bottom gate, sourced from the MoSe2 viaits electrical contact. This photocurrent is substantial inmagnitude – up to 6 nA for an excitation power of 15 µW –and appears only in the hole doped regime, as inferredfrom photoluminescence (PL) emission (Figure 1b).We first investigate the effect in detail using a dual-gated device structure in which the TMD is groundedthrough a side contact and the field to the top and bottomgates can be independently varied (Figure 2a). The goal ofsuch a structure is to be able to decouple the dopingconditions from the electric field, and deduce the depen-dence of the effect on either of them independently.Figure 2b outlines the general band alignment of the TMD/hBN/gate electrode system along with the direction of theobserved current. Plotting external quantum efficiencyFigure 1: Consistent cross-hexagonal boron nitride (hBN)photocurrent.a) Gate-voltage dependence of through-hBN photocurrent for theinset device structure (device A) under 15 µW, 660 nm opticalexcitation. This behavior is spatially uniform and is reproduced in aseries of devices of varying hBN thickness. b) Photoluminescentemission from the same device showing presence of photocurrent inthe hole doped regime.106 A. Sushko et al.: Asymmetric photoelectric effect(EQE) curves as a function of bottom gate for a MoSe2device with no top gate (Figure 2c), we first observe linearoptical power dependence across three orders of magni-tude, consistent with a single-photon excitation process.Here, EQE is given as the number of carriers injected intothe gate per photon incident on the device structure. EQEis notably lower for the dual gate MoSe2 device (Figure 2d)due to absorption of incoming photons in the metal topgate. Figure 2d, e shows the top gate current as a functionof both gates for MoSe2 andWSe2, respectively. Cuts alongthe labeled lines are presented and analyzed in Figure 4.By fixing the potential difference between the TMD andthe gate into which current flow is being measured, whilevarying the potential at the other, we can examine thedoping dependence of the photocurrent process at fixedfield. At a field of 0.15 V/nm (lines (1) in Figure 2d, e), bothMoSe2 and WSe2 exhibit rapid onset of current upon holedoping, and subsequent saturation with increasingcarrier density. Similarly, by incrementing the potentialon one gate, in opposition to the other, by a ratio pro-portional to their relative capacitance, one can maintainconstant doping of the sample while varying the field(lines (2) in Figure 2d, e), thereby yielding the pure fielddependence, independent of doping effects. While thedoping dependences of MoSe2 and WSe2 appear verysimilar, the two materials exhibit a qualitative differencein electric field dependence, with the MoSe2 currentswitching on rapidly at negative field while the WSe2current remains negligible until reaching a field of around0.1 V/nm. Furthermore, even at high field, the EQE of thephotocurrent in WSe2 is an order of magnitude lower thanwhat is observed in MoSe2, a distinction which persistsunder resonant excitation (see Supplementary Figure 11).The decrease in top gate photocurrent at decreasing bot-tom gate voltage is attributed to competition between thetwo gates (see Supplementary Figure 3).Figure 2: Power, field, and doping dependence in MoSe2 and WSe2.a) Schematic of the dual-gated, hexagonal boron nitride (hBN) encapsulated transition metal dichalcogenide (TMD) structure designed toenable independent control of doping and electric field between the TMDandmetal gates (deviceB). b) Band structure schematic for half of thedevice in (a) illustrating the hole-side photocurrent into one of the top/bottom electrodes when the TMD is hole-doped and electric fieldoriented toward the electrode. c) Quantum efficiency curves for a single-gated configuration (device A) show linear dependence on opticalpower. Here, external quantumefficiency (EQE) is the ratio of carriers through the hBN to photons incident on the heterostructure. Dependenceof total current on gate conditions for MoSe2 (d) and WSe2 (e) shows qualitatively distinct characteristics. While both systems require holedoping, WSe2 also exhibits minimal current below a field of 0.1 V/nm and much lower overall quantum efficiency relative to MoSe2.A. Sushko et al.: Asymmetric photoelectric effect 107Figure 3a shows a spatial map of the photocurrentgenerated by scanning a diffraction limited excitation spot(see Figure 3b for the sample geometry). A graphite backgate extends under part of a MoSe2 flake, vertically sepa-rated from the graphite by 3 nm hBN. This geometry pro-duces a lateral potential step when the gate potential isadjusted relative to the TMD, as schematically shown inFigure 3c. Tuning to the hole side results in enhancedphotocurrent generation along the edges of the gate. Weattribute this effect to details of the hole-doping mecha-nism for MoSe2. While the Cr/Au edge contacts are trans-parent to electrons, they are inefficient at injecting holesinto the monolayer. The presence of an in-plane potentialstep, however, provides an alternative doping mechanismvia exciton dissociation upon optical excitation. Indeed,dissociation is energetically favorable for a sufficientlysharp step as long as the in-plane step exceeds the bindingenergy of the exciton (Eb ∼200 meV [15]) as shown inFigure 3c. This process is similar to the well-known, junc-tion-induced exciton dissociation process in organic pho-todetectors and solar cells, and is common to allsemiconductors with tightly bound excitons [16]. Thedissociated electron, subsequently, is able to leave via thecontact resulting in a net hole photodoping process. As thephotocurrent is a sensitive function of the doping in view ofthe above observations, for sufficient optical power, thesteady-state photocurrent will be limited by the rate atwhich holes can be replenished –which is governed by thisvery edge-gate photodoping process. This becomes moreevident at increasing optical power levels, when otherdoping mechanisms become comparatively negligible –we refer to Supplementary Figures 4–6 for further detailsand dual beam, power-dependent measurements con-firming this picture. In addition, the doping of the samplecan also be inferred from PL emission. As shown inFigure 3d, placing a strong excitation laser at the gate edgewhile collecting PL from the center results in onset of trionemission whenever the gate potential is greater than Ebfrom charge neutrality (located at 0.27 V in this device). Incontrast, in the absence of edge illumination, the devicenever accumulates holes or emits any hole-trions(Figure 3e). Interestingly, the onset of electron doping isFigure 3: Photodoping via exciton dissociation at lateral potential steps.a) Spatial photocurrent map of an MoSe2 device (device C) with a local graphite back gate, schematic in (b), for off-resonant excitation at660 nm. Enhanced current is seen when the excitation laser is located near an edge of the gated region, corresponding to a lateral potentialstep. c) Schematic of exciton dissociation at a potential step, when the step height exceeds binding energy. Due to the limited ability of Cr/Auedge contacts to inject holes into MoSe2, a hole photocurrent is maintained through neutral-exciton dissociation followed by an electroncurrent into the contacts and a hole current into the gated region. d, e) Photoluminescence (PL) spectra taken from the center of the gatedregionwith andwithout 660 nmexcitation at the gate edge, respectively. The onset of hole doping once the potential step exceeds Eb and lackof trion oscillator strengthwithout edge excitation indicates that photodoping is the primarymechanism for hole-doping theMoSe2 structure.108 A. Sushko et al.: Asymmetric photoelectric effectunaffected as efficient charge injection can still occur viathe contacts, independently of any photodoping effects.While the former clearly illustrates the role ofdoping in the process, it does not yet elucidate by whichmechanism holes are able to escape the TMD andpenetrate or bypass the hBN barrier. For an interfacebetween bulk crystals, a calculation of the bandoffset along with information about the momentum inthe direction perpendicular to the interface wouldprovide the necessary electronic information to under-stand the transport across the junction. However, for aninterface between a monolayer and a bulk crystal, thenotion of perpendicular momentum in the monolayer ismeaningless in view of the absence of periodicity in thisdirection. Instead, the relevant picture is one where theentire system is considered as one interface between themonolayer and the bulk. By calculating the properties ofthis interface, we can then obtain the relevant transportFigure 4: Band structure calculations and kinetics.Density functional theory (DFT) calculations of hybridization between transition metal dichalcogenide (TMD) and hexagonal boron nitride(hBN) states for MoSe2 (a), and WSe2 (b) indicate a valence band offset between TMD states and layer-hybridized states on the order of theexciton energy, Exh. c, d) Dual-gate doping dependence of photocurrent from device B fitted to extract hot hole generation rate and relativerates of tunneling and thermalization using a kinetic model. While both materials show a comparable hot hole generation, the lower internalquantumefficiency (IQE) inWSe2 is explained by substantially slower tunneling relative to thermalization. IQE is obtained from the EQEdata inFigure 2, after compensating for absorptivity of the TMD and photon losses in the top gate. Fitting the field-dependence of photocurrent to aFowler–Nordheim tunneling process gives effective barriers, Φ, of 50 meV in MoSe2 (e) and 300 meV in WSe2 (f).A. Sushko et al.: Asymmetric photoelectric effect 109behavior. While the exact values depend on details ofthe layer interface such as the orientation of the layers,interfacial reconstruction and the exact hBN layerthickness, the trends are clear and explain the observedbehavior well – especially, the notable difference be-tween MoSe2 and WSe2.3 Physics of photocurrentsTo capture the interface physics and explain the observedphotocurrents, we start by using first principlesmethods tocompute the density of states (DOS) of a bilayer system thatconsists of one monolayer of TMD and one monolayer ofhBN, after which we color each state according to its layerhybridization (Figure 4a, b). The cyan states are unhybri-dized, i.e., the state is localized either entirely in the hBNlayer or entirely in the TMD layer. The pure cyan states nearthe band gap in Figure 4a, b should thus be understood asTMD states, as the TMD band gap is much smaller than thehBN band gap. Since the layers form an interface andinteract, we now also need to consider hybridization be-tween them, which leads to delocalization across theinterface. For example, the magenta states have equalweight in the TMD and hBN layer. The colors in betweenpure cyan and magenta indicate a state with some weightin each layer, with ratios given according to the scale bar.To relate these DOS plots to the transmission of chargecarriers from the TMD to hBN, we note that hybridizedstates (denoted by any color other than pure cyan) areeffectively delocalized across the interface and thereforerepresent a pathway for a charge carrier to move betweenthe TMD and the hBN layers. Somewhat similar to the bandoffset picture in bulk heterojunctions, the distance in en-ergy from the TMD band edge (the cyan states near the gap)to the first hybridized (non-cyan) states can now beconsidered the effective band offset. In agreement withother studies of MoSe2/hBN systems, the valence bandoffset is much smaller than the conduction band offset,indicating that holes canmuchmore readily travel from theTMD to the hBN than electrons [9, 10].These DOS plots elucidate qualitatively why the photo-current in theMoSe2 system is larger than thephotocurrent intheWSe2 system. They also indicate effective band offsets onthe order of the excitonic energy (∼1.6–1.7 eV [17, 18], seeFigure 4a, b). Since our experiments take place at cryogenictemperatures (6 K), such energies are orders of magnitudehigher than the thermal energy in our system. The onlyprocess capable of providing such energies is Auger recom-bination, where the energy of an exciton is transferred non-radiatively to a resident carrier – in our case, a hole. In thepresence of free holes, the exciton can undergo Augerrecombination, during which the exciton annihilation en-ergy is completely transferred to the hole. The hole can thenscatter with a phonon in a very fast broadening process,which allows it to access hybridized states away from the Kpoint. Consistentwith thedata inFigure 2c, the probability ofthe Auger process at low excitation intensity is linear inoptical powerbecause it requires just one exciton, in contrastto the commonly studied exciton-exciton Auger recombi-nation, which has a quadratic dependence on excitationintensity. Due to the larger conduction band offsets betweenthe TMD and hBN, hot electrons are unable to transfer intothe hBN before thermalizing – in line with the observedelectron–hole asymmetry in our experiments. In addition,Auger recombination shifts a hole to an energy with a highdensity of hybridized states inMoSe2,while the sameprocessin WSe2 leaves the hole at an energy with a much lowerdensity of hybridized states, as indicated by the dashed linesin Figure 4a, b. Thus, Auger recombination is much morelikely to result in a hole transmitted to the hBN in the MoSe2system than in the WSe2 system. This trend should persistregardless of the exact details of the junction such as theexact hBN thickness. For instance, for the sake of computa-tional efficiency, we model a monolayer of hBN instead ofthe many different film thicknesses of hBN used in the ex-periments (see DFT methods for details). Including manylayers of hBN should systematically lower the valence bandmaximum of the hBN in each system [19], which may intro-duce an additional tunneling barrier the hole must over-come.Yet, a hole in theMoSe2 systemwould still encounter asmaller tunnel barrier than a hole in the WSe2 system.We examine the underlying Auger mechanism ingreater detail by deriving the dependence of photocurrentquantum efficiency on doping from a simple kineticmodel.In this model, Auger excited holes can either tunnelthrough an effective barrier (the aforementioned barrierminus the hot hole energy), or thermalize. By properlyaccounting for charge replenishment and competing (non-Auger) exciton decay paths, we can obtain approximatevalues for the rates of Auger recombination, hot holethermalization and tunneling under detailed balanceconditions – we refer to the Supplementary materials fordetails. We can use this model in combination with thedeconvolved field- and doping dependence in our devicesto extract relevant values. For example, fitting ourmodel tothe doping-dependence at fixed field allows us to extractapproximate values for the relative times of hot-holetunneling to thermalization (τtun/τter) and Auger recombi-nation relative to all exciton decay paths (τTA/τ) for eachsystem (Figure 4c, d; Supplementary material for details).We find a comparable τTA/τ of 103 for both systems.110 A. Sushko et al.: Asymmetric photoelectric effectHowever, the relative thermalization rate (compared totunneling) inWSe2 (104) significantly exceeds that ofMoSe2(102). This again reflects the higher effective tunnel barrierfor WSe2 as suggested by DFT. For the tunneling processitself, we consider thefield dependence of the photocurrentand model it as a Fowler–Nordheim tunneling process(Figure 4e, f). Themodel reproduces the observed data verywell, and allows us to extract effective (net) barrier heightsfor hot hole tunneling of 50meV forMoSe2 and 300meV forWSe2, consistent with the relative difference seen in DFTcalculations.We next consider the wavelength dependence of thephotocurrent under resonant excitation, which clearlyconfirms the essential role of excitons, consistent with ourAuger picture. Figure 5a shows the variation in gate currentfrom the MoSe2 device in Figure 3. We ensure reliable holedoping by photodoping through a strong (50 µW) above-band laser. This value exceeds that of the other rates in oursystem, and ensures barrier limiting (as opposed to charge-replenishment limited) behavior. When subsequentlysweeping a variable wavelength laser (5 µW), we observe apronounced set of resonances. Comparing those againstthe photoluminescence emission spectrum (Figure 5b) weobserve photocurrent emission coinciding with the excitonand hole-trion resonances – as expected from theperspective of an Auger picture involving excitons andholes. The substantial photocurrent from the neutralexciton in the hole-doped regime, despite low populationin PL, suggests an interesting interplay between the hole-exciton scatteringmechanisms that create hole-trions (alsoreferred to as attractive polarons [20]) andAuger processes.These observations imply that photocurrent may providean interesting probe of exciton dynamics, and could beused to shed light on varying decay mechanisms in TMDsas well as novel thermalization physics – as already sug-gested by our kinetic model.Finally, we consider similarities between the photo-current mechanism and a previously documented Augerexciton upconversion process [11]. When a TMD wasexcited on resonance with the lowest energy 1s exciton, PLemission was observed at higher energy (Rydberg) states[15], including the 2s exciton and B exciton from a higherFigure 5: Photocurrent under resonant excitation and competition with exciton upconversion.a) MoSe2 (device C) photocurrent differential for 50 µW off-resonant gate-edge excitation and 5 µW variable-wavelength center excitationshows current following the exciton and trion resonances seen in photoluminescence (PL) in (b). c) Upconverted photoluminescence spectrataken at 759 nm excitation, documented in literature to arise from Auger excitation of the 1s A exciton, yielding emission of higher Rydbergstates, alongwith the B exciton. d) Reflectance spectra indicating that the B exciton state persists in the hole-doped regime. The onset of hole-doping, however, corresponds to a loss of photoluminescence from the B exciton state and a corresponding onset of photocurrent, suggestinga competition between exciton–exciton Auger and exciton–hole Auger, with the latter dominating in the doped regime.A. Sushko et al.: Asymmetric photoelectric effect 111conduction band. In [11], this phenomenon was attributedto exciton–exciton annihilation, a related Auger process inwhich one exciton non-radiatively transfers its energy toanother, hot exciton. We indeed observe this phenomenonin our MoSe2 devices (Figure 5c), allowing us to examinethe relationship between these processes. In reflectancemeasurements (Figure 5d), we observe that the B excitonstate exists in both the neutral and the hole-doped regime.However, in PL, which measures population, we observe arapid suppression of the B exciton upconverted emissionwith hole doping (Figure 5c), while observing the presenceof a pronounced photocurrent in this regime. These ob-servations suggest a competition between the hole-Augerprocess and the upconversion process: the exciton–excitonAuger process necessary to create the hot excitons thatultimately relax into the B state appears to compete withthe hole-exciton Auger process. From a microscopicperspective, these observations are consistent with therelative densities of holes and excitons in our system. Tofirst order, from a gate capacitancemodel, we expect a holedensity of ∼7 × 1012cm−2V−1 which exceeds the approximateexciton density of ∼1010 cm−2 at only a few megavolt pastthe onset of hole doping. Assuming somewhat similarexciton–hole and exciton–exciton Auger recombinationrates would then indeed suggest a significant suppressionof upconversion upon doping due to simple competitionbetween the two processes. More detailed analyses, withindependently calibrated carrier and exciton densities,could therefore be used in combination with photocurrentand upconversion measurements to bound the ratio be-tween these respective rates more tightly.4 OutlookIn conclusion, we have shown that photoexcitation of ex-citons in hBN-encapsulated TMDs can give rise to a form of“photoelectric effect” for holes that results in a net andsubstantial current across the nominal hBN dielectricbarrier. We attribute this effect due to Auger-generated hotholes being swept through the barrier by the electric field ina tunneling process,whichwe substantiate by careful field,doping, wavelength and power dependencies. We furthersupport our claimswith detailed, ab initio calculations thatmatchwellwith our observations of a systematically highereffective hole tunnel barrier for WSe2 as compared toMoSe2. In addition to shedding light on the intrinsic Augereffects in TMDs, which are important to evaluate their de-vice performance as photodetectors and other optoelec-tronic devices, our studies also demonstrate thespectroscopic potential of photocurrent studies, andprovide a novel probe to study non-radiative effects such ascarrier thermalization in 2D semiconductors. Intriguingly,if further studies confirm a certain degree of coherence inthe photoelectric effect, our findings may open the doorfor on-chip, integrated probing of the local density ofstates – in a way similar to advanced spectroscopic tech-niques such as ARPES, but with greatly reducedcomplexity and with nanoscale resolution.Acknowledgments: The authors wish to acknowledge FalkoPientka and Richard Schmidt formany insightful discussionsconcerning this research. This work was supported by theNSF, CUA, DOE and Vannevar Bush Faculty FellowshipProgram. A.S. Acknowledges support from the Fannie andJohn Hertz Fellowship and Paul and Daisy Soros Fellowshipsfor New Americans. Al.L.E. and C.S.H acknowledge supportfrom the US Office of Naval Research. Al.L.E. also acknowl-edges support from the Laboratory-University CollaborationInitiative (LUCI) program of the DoD Basic Research Office.B.U. acknowledges aNanoX/NEXT travel grant. This researchwas performed while M.P. held a National Research Councilassociateship at NRL. Computational work was supported bya grant of computer time from the DoD High PerformanceComputing Modernization Program at the U.S. ArmyResearch Laboratory and the U.S. Air Force Research Labo-ratory Supercomputing Resource Centers (NRLDC04123333).K.W. and T.T. acknowledge support from the ElementalStrategy Initiative conducted by the MEXT, Japan, GrantNumber JPMXP0112101001, JSPS KAKENHI Grant NumberJP20H00354 and the CREST (JPMJCR15F3), JST.Author contribution: All the authors have acceptedresponsibility for the entire content of this submittedmanuscript and approved submission.Research funding: This work was supported by the NSF,CUA, DO and Vannevar Bush Faculty Fellowship Program.A.S. Acknowledges support from the Fannie and JohnHertzFellowship and Paul and Daisy Soros Fellowships for NewAmericans. Al.L.E. and C.S.H acknowledge support fromthe US Office of Naval Research. Al.L.E. also acknowledgessupport from the Laboratory-University CollaborationInitiative (LUCI) program of the DoD Basic ResearchOffice. B.U. acknowledges a NanoX/NEXT travel grant.This research was performed while M.P. held a NationalResearch Council associateship at NRL. Computationalwork was supported by a grant of computer time from theDoDHigh Performance ComputingModernization Programat the U.S. Army Research Laboratory and the U.S. AirForce Research Laboratory Supercomputing ResourceCenters (NRLDC04123333). K.W. and T.T. acknowledgesupport from the Elemental Strategy Initiative conductedby the MEXT, Japan, Grant Number JPMXP0112101001,112 A. Sushko et al.: Asymmetric photoelectric effectJSPS KAKENHI Grant Number JP20H00354 and the CREST(JPMJCR15F3), JST.Conflict of interest statement: The authors declare noconflicts of interest regarding this article.References[1] S. Manzeli, D. Ovchinnikov, D. Pasquier, O. V. Yaziev, and A. Kis,“2D transition metal dichalcogenides,” Nat. Rev. Mater., vol. 2,p. 17033, 2017.[2] K. F. Mak and J. Shan, “Photonics and optoelectronics of 2Dsemiconductor transition metal dichalcogenides,” Nat.Photonics, vol. 10, p. 216, 2016.[3] A. K. Geim and I. V. Grigorieva, “Van der Waals heterostructures,”Nature, vol. 499, p. 419, 2013.[4] R. Bekenstein, I. Pikovski, H. Pichler, E. Shahmoon, S. F. Yelin, andM. D. Lukin, “Quantum metasurfaces with atom arrays,” Nat.Phys., vol. 16, p. 676, 2020.[5] G. Scuri, Y. Zhou, A. High, et al., “Large excitonic reflectivity ofmonolayer MoSe2 encapsulated in hexagonal boron nitride,”Phys. Rev. Lett., vol. 120, p. 037402, 2018.[6] C. R. Dean, A. F. Young, I. Meric, et al., “Boron nitride substratesfor high-quality graphene electronics,” Nat. Nanotechnol., vol. 5,p. 722, 2010.[7] M. Yankowitz, Q. Ma, P. Jarillo-Herrero, and B. J. LeRoy, “van derWaals heterostructures combining graphene and hexagonalboron nitride,” Nat. Rev. Phys., vol. 1, p. 112, 2019.[8] G. Cassabois, P. Valvin, and B. Gil, “Hexagonal boron nitride is anindirect bandgap semiconductor,” Nat. Photonics, vol. 10, p. 262,2016.[9] E. Linardy, D. Yadav, D. Vella, et al., “Harnessing exciton–excitonannihilation in two-dimensional semiconductors,”Nano Lett., vol.20, p. 1647, 2020.[10] C. M. E. Chow, H. Yu, J. R. Schaibley, et al., “Monolayersemiconductor Auger detector,” Nano Lett., vol. 20, no. 7,pp. 5538–5543, 2020.[11] B. Han, C. Robert, E. Courtade, et al., “Exciton states inmonolayer MoSe2 and MoTe2 probed by upconversionspectroscopy,” Phys. Rev. X, vol. 8, p. 031073, 2018.[12] N. M. Gabor, J. C. W. Song, Q. Ma, et al., “Hot carrier–assistedintrinsic photoresponse in graphene,” Science, vol. 334, p. 648,2011.[13] Q. Ma, T. I. Andersen, N. Nair, et al., “Tuning ultrafast electronthermalization pathways in a van der Waals heterostructure,”Nat. Phys., vol. 12, p. 455, 2016.[14] K. J. Tielrooij, J. C. W. Song, S. A. Jensen, et al., “Photoexcitationcascade and multiple hot-carrier generation in graphene,” Nat.Phys., vol. 9, p. 248, 2013.[15] A. Chernikov, T. C. Berkelbach, H. M. Hill, et al., Phys. Rev. Lett.,vol. 113, p. 076802, 2014.[16] M. Massicotte, F. Vialla, P. Schmidt, et al., “Dissociation of two-dimensional excitons inmonolayerWSe2,”Nat. Commun., vol. 9,p. 1633, 2018.[17] S. Shree, M. Semina, C. Robert, et al., “Observation of exciton-phonon coupling in MoSe2 monolayers,” Phys. Rev. B, vol. 98,p. 035302, 2018.[18] E. Courtade, M. Semina, M. Manca, et al., “Charged excitons inmonolayer WSe2: Experiment and theory,” Phys. Rev. B, vol. 96,p. 085302, 2017.[19] D. Wickramaratne, L. Weston, and C. G. Van de Walle,“Monolayer to bulk properties of hexagonal boron nitride,” J.Phys. Chem. C, vol. 122, p. 25524, 2018.[20] M. Sidler, P. Back, O. Cotlet, et al., “Fermi polaron-polaritons incharge-tunable atomically thin semiconductors,”Nat. Phys., vol.13, p. 255, 2016.Supplementary material: The online version of this article offers sup-plementary material (https://doi.org/10.1515/nanoph-2020-0397).A. Sushko et al.: Asymmetric photoelectric effect 113https://doi.org/10.1515/nanoph-2020-0397 Asymmetric photoelectric effect: Auger-assisted hot hole photocurrents in transition metal dichalcogenides 1 Introduction 2 Photocurrents in encapsulated TMDs 3 Physics of photocurrents 4 Outlook Acknowledgments References