# Fileset

[ncomms13747.pdf](https://mdr.nims.go.jp/filesets/309d77f2-5235-47c2-91af-00880636e251/download)

## Creator

John R. Schaibley, Pasqual Rivera, Hongyi Yu, Kyle L. Seyler, Jiaqiang Yan, David G. Mandrus, [Takashi Taniguchi](https://orcid.org/0000-0002-1467-3105), [Kenji Watanabe](https://orcid.org/0000-0003-3701-8119), Wang Yao, Xiaodong Xu

## Rights

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

## Other metadata

[Directional interlayer spin-valley transfer in two-dimensional heterostructures](https://mdr.nims.go.jp/datasets/22c57fc2-488c-4822-a5a8-b6fd66676d7b)

## Fulltext

Directional interlayer spin-valley transfer in two-dimensional heterostructuresARTICLEReceived 20 Aug 2016 | Accepted 25 Oct 2016 | Published 14 Dec 2016Directional interlayer spin-valley transferin two-dimensional heterostructuresJohn R. Schaibley1, Pasqual Rivera1, Hongyi Yu2, Kyle L. Seyler1, Jiaqiang Yan3,4, David G. Mandrus3,4,5,Takashi Taniguchi6, Kenji Watanabe6, Wang Yao2 & Xiaodong Xu1,7Van der Waals heterostructures formed by two different monolayer semiconductors haveemerged as a promising platform for new optoelectronic and spin/valleytronic applications. Inaddition to its atomically thin nature, a two-dimensional semiconductor heterostructure isdistinct from its three-dimensional counterparts due to the unique coupled spin-valleyphysics of its constituent monolayers. Here, we report the direct observation that an opticallygenerated spin-valley polarization in one monolayer can be transferred between layers of atwo-dimensional MoSe2–WSe2 heterostructure. Using non-degenerate optical circulardichroism spectroscopy, we show that charge transfer between two monolayers conservesspin-valley polarization and is only weakly dependent on the twist angle between layers.Our work points to a new spin-valley pumping scheme in nanoscale devices, provides afundamental understanding of spin-valley transfer across the two-dimensional interface,and shows the potential use of two-dimensional semiconductors as a spin-valley generator intwo-dimensional spin/valleytronic devices for storing and processing information.DOI: 10.1038/ncomms13747 OPEN1 Department of Physics, University of Washington, Seattle, Washington 98195, USA. 2 Department of Physics and Center of Theoretical and ComputationalPhysics, University of Hong Kong, Hong Kong, China. 3 Materials Science and Technology Division, Oak Ridge National Laboratory, Oak Ridge, Tennessee37831, USA. 4 Department of Materials Science and Engineering, University of Tennessee, Knoxville, Tennessee 37996, USA. 5 Department of Physics andAstronomy, University of Tennessee, Knoxville, Tennessee 37996, USA. 6 Advanced Materials Laboratory, National Institute for Materials Science, Tsukuba,Ibaraki 305- 0044, Japan. 7 Department of Materials Science and Engineering, University of Washington, Seattle, Washington 98195, USA. Correspondenceand requests for materials should be addressed to J.R.S. (email: johnschaibley@email.arizona.edu) or to X.X. (email: xuxd@uw.edu).NATURE COMMUNICATIONS | 7:13747 | DOI: 10.1038/ncomms13747 | www.nature.com/naturecommunications 1mailto:johnschaibley@email.arizona.edumailto:xuxd@uw.eduhttp://www.nature.com/naturecommunicationsSpin initialization is a crucial operation for spintronicdevices which require a net spin polarization for reading,writing and transferring information1. Two-dimensionalsemiconductors, such as monolayer MoSe2 and WSe2, haverecently emerged as a new spin/valleytronic platform2–5. Theirinversion-asymmetric honeycomb lattice structures give rise totwo energy degenerate but inequivalent (þK and �K)momentum-space valleys, forming a pseudospin systemanalogous to real electron spin6. Due to strong spin-orbitcoupling, the valley pseudospin is locked to the real spinorientation6. Since flipping an electron spin requires asimultaneous flip of a valley pseudospin, free carrier spin-valleypolarization at the band edge is expected to be robust and longlived, which has recently been measured to be on order of1–100 ns (refs 7,8). Large spin-valley polarizations associatedwith excitons have been generated by circularly polarizedlight excitation through a valley dependent optical selectionrule3,6,9–11. However, excitonic spin-valley polarization in amonolayer does not last long compared with free carriers dueto the picosecond timescale of the valley exciton depolarizationtime, which arises from the electron–hole exchangeinteraction3,12–14 and the ultrafast decay time of the excitonitself15,16. In addition, it is not clear how to exploit a monolayersystem as a spin generator to supply optically generatedspin-valley polarization to a different physical system.2D semiconductor heterostructures formed by stacking twomonolayers on top of each other can be designed to realize newspin-valley systems with important advantages over individualmonolayers. It has been established that WX2–MoX2 (whereX¼ S, Se) heterostructures have a type-II band alignment17,18,which leads to ultrafast charge transfer between layers andtunable photodetectors19,20. Such spatial separation of electronsand holes suppresses ultrafast electron–hole recombination21–24and their exchange interaction21–26, both of which limitthe practical application of optical spin-valley orientationin monolayers3,27,28. Very recently, helicity-dependent photo-luminescence (PL) measurements of interlayer excitonsrevealed spin-valley polarization lifetimes exceeding tens ofnanoseconds26, showing that the spatial separation of electronsand holes indeed provides a powerful approach towards practicalspin-valleytronics. However, interlayer exciton effects wereaccompanied by complicated electron–hole relaxation pathwaysand the effect of the twist angle between the two layers25, whichcomplicate the quantitative analysis of spin-valley polarizationfrom the polarization resolved interlayer exciton PL. In addition,interlayer exciton PL studies were limited to small twist anglesamples only, because the electron–hole momentum mismatch inlarge twist angle heterostructures strongly suppresses interlayerexciton light emission. All of these limitations obscured a clearunderstanding of the unique spin-valley properties of 2Dsemiconductor heterostructures, especially the transport ofspin-valley polarized free carriers across the 2D layer interface.In this work, by applying polarization resolved non-degeneratenonlinear optical spectroscopy, we provide a direct probe ofinterlayer spin-valley polarization transfer in a model 2Dheterostructure with varying twist angles formed by monolayerMoSe2 and WSe2. By optically exciting an intralayer excitonspin-valley polarization in one layer and probing the intralayerneutral and charge excitons in different layers, we demonstratethat the subsequent interlayer charge transfer is directional andconserves spin, that is, spin polarization transfer leads topolarized hole spins in WSe2 and electron spins in MoSe2(Fig. 1a). We find that the spin-valley polarization transfer hasonly a weak dependence on twist angles in the heterobilayer.Our results realize directional pumping of spin-valley polarizedcarrier spins into individual layers of a 2D heterostructure byharnessing the coupled spin-valley physics of the constituentmonolayers6.ResultsSample fabrication and electronic structure. The MoSe2–WSe2heterostructures were fabricated from independently isolated,exfoliated monolayers (see Fig. 1b). To investigate the effect ofheterostructure twist angle, we first measured the crystalaxes of individual monolayers by polarization resolved andphase-sensitive second-harmonic generation spectroscopy29–32(see Supplementary Fig. 1 and Supplementary Note 1). Themonolayers were then assembled into heterostructures using adry transfer stamping technique33 with known twist angle.Results from heterostructures with non-zero twist angels arepresented in Supplementary Note 2 and Supplementary Figs 2and 3. The sample in the main text has a twist angle near 0�,where the valleys from the different layers are nearly aligned inmomentum space (Fig. 1c). The lowest conduction band islocated in the MoSe2 and the highest valence band in WSe2.Within each monolayer, s± circularly polarized light couples totransitions in the ±K valley only. The high quality of ourheterostructure was confirmed by observing a strong PLquenching of the intralayer excitons, and the observation ofinterlayer excitons (see Supplementary Fig. 4), where Coulomb-bound electrons and holes are localized in opposite layers24.Nonlinear excitonic response of the heterostructure. We firstdetermined the energy position of intralayer excitons byperforming energy resolved continuous-wave differentialtransmission (DT) or differential reflection (DR) spectroscopy34.This is a two beam pump-probe technique which measuresthe difference of the probe transmission or reflection whenthe pump is on and off. The experiments in the main text wereall performed on the same heterostructure mounted onsapphire. Additional measurements were performed on differentheterostructures on SiO2 substrates and are in qualitativeagreement with the data presented in the main text (seeSupplementary Note 2). The experiments were performed at30 K, unless otherwise specified.The degenerate DT spectrum of a heterostructure is shownin Fig. 1d with cross-circularly polarized pump and probe.Compared with the DT spectrum from individual monolayers, wesee the intralayer exciton resonances in the heterostructure areconsistent with the spectral positions of isolated monolayers witha B20 meV redshift and broader linewidth. We attribute theB20 meV redshift to a reduction in the intralayer excitonbandgaps due to the coupling between layers. The linewidthbroadening is attributed to the charge transfer between the layers,which leads to an extra relaxation channel for the intralayerexcitons35. The resonance line shapes of the degenerate DTspectrum consist of a pump-induced increase to the probetransmission at high energy and a pump-induced absorption atlow energy. Note that the low-energy pump-induced absorptionfeature is stronger for the MoSe2 layer compared with WSe2. Weattribute this difference to the different oscillator strengths ofdifferent charged exciton species in each layer (see SupplementaryNote 3).Demonstration of interlayer charge transfer. To establishinterlayer carrier transfer, we performed two-colour non-degenerate DR and DT measurements. Both types of measure-ments were performed on the same sample and the data arequalitatively similar. We use the DT data exclusively in curvefitting to avoid the interference effects that arise from thesubstrate reflection in the DR measurements. Figure 2a shows theARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms137472 NATURE COMMUNICATIONS | 7:13747 | DOI: 10.1038/ncomms13747 | www.nature.com/naturecommunicationshttp://www.nature.com/naturecommunicationsDR spectrum with co-circularly polarized pump and probe, wherethe pump is resonant with the lower energy MoSe2 exciton at1.621 eV while the probe laser scans over the WSe2 excitonresonance near 1.68 eV. The green curve shows an enhanced DRresponse from the heterostructure region. In comparison, theblack curve shows the DR response when both pump and probeare focused on an isolated monolayer WSe2 region which shows anegligible DR response when the pump energy is fixed at theMoSe2 exciton resonance. In the heterostructure, since the MoSe2exciton has lower energy than WSe2, the observed DR responsenear the WSe2 exciton when pumping the MoSe2 excitonresonance is unlikely from the energy transfer from MoSe2exciton. Rather, it is a result of charge transfer from MoSe2 toWSe2. Specifically, the hole is transferred from the MoSe2 valenceband to the WSe2 valence band due to the type-II bandalignment.Demonstration of interlayer spin-valley polarization transfer.Interlayer spin-valley transfer was then investigated by perform-ing polarization resolved DT experiments which measure thepump-induced circular dichroism (CD). The pump laser polar-ization and energy were chosen to only excite valley polarizedexcitons in the MoSe2 layer. The DT spectrum was measured forboth co- (burgundy curve) and cross- (green curve) circularlypolarized configurations for the probe scanning through theWSe2 excitons (Fig. 2b). The CD can be defined as the differencebetween the cross- and co-polarized DT spectra for either fixedpump or fixed probe polarization. Both yield similar results. Forthe convenience of our experimental configuration, we choose tofix the probe helicity while switching the pump helicity, that is,CD ¼ DTT ðs�pumpÞ� DTT ðsþpumpÞ, where the subscript denotes thepump beam, and T is the probe transmission. As shown in Fig. 2c,the sign of the pump-induced CD response reverses for oppositeprobe helicities. The observed CD demonstrates a valleypopulation imbalance, that is, the creation of spin-valley polar-ization in WSe2. We attribute this population imbalance to thepumping of polarized hole spins as depicted in Fig. 1a. Circularlypolarized excitation resonantly pumps spin-valley polarizedexcitons in the MoSe2 layer, about 60 meV below the WSe2exciton energy. The spin polarized hole then transfers to theWSe2 þK valence band, which gives rise to hole spin-valleypolarization in WSe2 and electron spin-valley polarization inMoSe2. The observation of the CD response supports this picture.We also demonstrate electron spin transfer from the WSe2 tothe MoSe2 layer by resonantly pumping a WSe2 spin-valleypolarization and probing the MoSe2 excitons (Fig. 3). Similarphenomena, including the pump-induced CD is observed, whosesign depends on probe helicity. Since the WSe2 exciton has higherenergy than MoSe2, the observed CD will have two contributions.One is due to the electron spin transfer from the WSe2 to theMoSe2 conduction band. The other is from the above resonanceoptical excitation of valley-polarized excitons directly in theMoSe2. To distinguish these two effects, we measured the DRresponse on the heterostructure when pumping at the WSe2WSe2MoSe2Het.MoSe2d�+�–MoSe2WSe2MoSe2WSe2–KMoSe2WSe2Het.b1.5 1.6 1.7 1.8Energy (eV)DT/T (norm.)10 µmh+KWSe2ehcaFigure 1 | Interlayer spin-valley physics. (a) Depiction of the experiment. Spin-valley polarized excitons are resonantly injected in the MoSe2 layer with apolarized laser (red). The hole transfers to the WSe2 layer where its spin-valley polarization is measured with another polarized laser (blue), resonant withthe WSe2 excitons. The black arrows depict the real spin of the electrons and holes. (b) Optical microscope image of a MoSe2–WSe2 heterostructure (Het.)on SiO2, showing the different sample regions. (c) The 8-band model of the þK and –K valleys for a nearly aligned MoSe2–WSe2 heterostructure, showingthe valley dependent optical selection rules (s± for ±K valley) and real spins (black arrows) for electrons and holes. (d) Degenerate DT spectra fromdifferent sample regions for a MoSe2–WSe2 heterostructure on sapphire, which are normalized and stacked for comparison. The dashed lines correspond toDT/T¼0 for each spectrum. Due to the small isolated WSe2 area used in the DT study, the laser beam could not completely avoid the heterobilayer region,which results in the artifact of small positive signal at MoSe2 exciton energy on the WSe2 sample region.NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13747 ARTICLENATURE COMMUNICATIONS | 7:13747 | DOI: 10.1038/ncomms13747 | www.nature.com/naturecommunications 3http://www.nature.com/naturecommunicationsresonance, and then repeated the measurement on the isolatedmonolayer MoSe2 region of the same sample (SupplementaryFig. 5). We observe a threefold enhancement of the DR responseon the heterostructure region, which shows that electron transferfrom the WSe2 to the MoSe2 dominates the DR response. Wenote that since the electron and hole spin are separated inopposite layers, the exchange interaction between the electronand hole spins is strongly suppressed. This will give rise to a longpolarization lifetime26 and contributes to the enhanced DRresponse.Origin of the DT line shapes. We now turn to the discussion ofthe line shapes in the non-degenerate DT measurements (Figs 2band 3a), which further support the picture of directional spintransfer. For simplicity, we focus on the explanation of data inFig. 3a. Figure 3c–f illustrate the origins of the line shapes bypumping at the WSe2 exciton resonance while probing the MoSe2excitons. The DT spectra can be understood by taking thedifference between the probe transmission spectrum with thepump on and off (solid orange and dashed blue curve ofFig. 3c,e). The co-polarized pump and probe (burgundy data)laser configuration is shown in the left inset of Fig. 3a. The insetdepicts the pump (solid blue line) injecting þK polarized carriersin the WSe2 layer and the consequent electron transfer to the þKconduction band valley in the MoSe2 monolayer. The probe(dashed red line) measures the changes in transmission spectrumof the þK MoSe2 excitons. Figure 3c,d depict the effects thatdominate the co-polarized DT response. Because the conductionband is partially filled, phase-space filling leads to a blue shiftof the transmission resonance, and the neutral exciton (X�)oscillator strength is reduced (Fig. 3c). The inset to Fig. 3c depictsthe DT signal calculated by taking the difference between theorange and dashed blue curves.The cross-polarized pump and probe configuration (greendata) is depicted in the right inset of Fig. 3a. Here, the pump(solid blue line) injects carriers into the �K valley of the WSe2layer, and the subsequent electron transfer to the –K valley of theMoSe2. With this –K valley electron population, when the probebeam (dashed red line) excites electron–hole pairs in þK valley,negatively charged excitons (X� ) can form (Fig. 3e,f). The cross-polarized DT spectrum can be understood by examining Fig. 3e,which shows the pump-induced changes to the cross-polarizedprobe transmission spectrum. Relative to the pump off case,a population of electrons in the –K valley decreases the þKcross-polarized probe transmission at X� resonance due to theincreases of X� oscillator strength, and increases the transmis-sion at the X� resonance due to the decrease of X� oscillatorstrength. The inset to Fig. 3e shows the corresponding cross-polarized DT spectrum. We note that the 30 meV energyseparation between the peak and dip in both the cross-polarizedDT spectrum (green curve of Fig. 3a) and the CD spectra (Fig. 3b)is consistent with the binding energy of X� , and therefore furthersupports the picture of directional electron spin-valley transferfrom WSe2 to MoSe2.DiscussionWe estimate the resulting spin-valley polarization of electrons inthe MoSe2 layer by pumping the WSe2 resonance and comparingthe relative co- and cross-circularly polarized DT responsesof the X� in the MoSe2 layer. It has been demonstrated bothexperimentally4 and theoretically36 that the X� in MoSe2 isdominantly an intervalley charged exciton with the extra electronbacMoSe2WSe2WSe2Energy (eV)Energy (eV)Energy (eV)–0.10.10.01.65 1.70 1.75 1.80–0.150.150.000.301.65 1.70 1.75 1.801.65 1.70 1.75–0.150.150.000.30DR/R (× 10–3)DT/T (× 10–3)CD (× 10–3)–0.30CoWSe2MoSe2+K–KCross+K–KProbe �+ Probe �–PumpProbePumpProbeProbeProbePumpPumpFigure 2 | Interlayer hole spin-valley polarization transfer. (a) Non-degenerate DR of a MoSe2–WSe2 heterostructure, and an isolated WSe2 regionon the sample. When pumping on the lower energy MoSe2 exciton resonance (1.621 eV), there is a strong DR response corresponding to the WSe2exciton (dark cyan), whereas the isolated WSe2 monolayer shows negligible DR response (black). Co-circularly polarized pump and probe is shown.The insets depict the pump-probe scheme. The pump is shown as a solid red line, and the probe is the dashed blue line. DR data were measured at 50 K.(b) Co- (burgundy) and cross- (green) circularly polarized DT spectra of the WSe2 exciton resonances, when pumping the low-energy MoSe2 excitonresonance at 1.621 eV. The insets show the pump and probe scheme, where the band filling of the WSe2 valence is shown. The line shapes are discussedin the text. (c) Pump-induced CD of the WSe2 exciton resonances when pumping MoSe2 at 1.621 eV. CD highlights the differences between co- andcross-polarized DT responses. As expected, the sign of the CD response flips with probe (or pump) helicity.ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms137474 NATURE COMMUNICATIONS | 7:13747 | DOI: 10.1038/ncomms13747 | www.nature.com/naturecommunicationshttp://www.nature.com/naturecommunicationslocated in the lower conduction band of the opposite valley(see Supplementary Note 4 and Supplementary Fig. 6). The X�formation and the magnitude of its corresponding DT signalmeasures the population of polarized electrons in the valleyopposite the one being probed. Therefore, spin-valley polarization(r) in the MoSe2 layer resulting from interlayer spin-valleytransfer can be estimated by r¼ DT=T crossð Þ�DT=T coð ÞDT=T crossð ÞþDT=T coð Þ. Since thenegative MoSe2 charged exciton resonance is spectrally isolated, itis not significantly influenced by signals arising from the otherresonances (Fig. 3a). We fit single Lorentzians to the MoSe2 DTresponse near X� (1.589 eV) for both co- and cross-circularlypolarized pump and probe, and find that the ratio of the dip areasfor the co-polarized response is 37% of the cross-polarizedresponse (See Supplementary Fig. 7). This gives a 46% electronspin-valley polarization in the MoSe2 layer. This estimation isconsistent with previous measurements of the interlayer excitonin helicity-dependent PL measurements, which was reported to belimited by depolarization of the monolayer exciton beforeinterlayer transfer26.When pumping the MoSe2 and probing the WSe2 excitons, thecharged exciton feature is also clear in the CD response (Fig. 2c).Fitting the CD spectra with a difference of two Lorentzians,we find that the energy separation between the peak and dip isapproximately 19 meV, consistent with the binding energy ofpositively charged excitons (Xþ ) in WSe2 (refs 2,3). Thisobservation supports the picture of directional polarized holespin transfer from MoSe2 to WSe2. However, due to the overlapof spectral features near the WSe2 positively charged excitonpeak, we cannot accurately compare the co- and cross-circularDT responses of Xþ to estimate a hole spin-valley polarization inthe WSe2 layer.We also performed measurements on additional samples withvarying twist angles (Supplementary Fig. 2). There are finespectral features distinct from near zero twist angle samples,which require a future systematic study. However, both the signand signal amplitude of the CD spectra are consistent for all twistangles, which implies that spin-valley conserved interlayer chargetransport is robust for different twist angles.Our results demonstrate that spin-valley polarized carrierscan be efficiently transferred between layers, providing a novelmethod for optically injecting long-lived and spin-valleypolarized carriers in either layer of heterostructures with arbitrarytwist angles. We expect this scheme could be especially useful inrecent proposals that seek to use atomically thin-bilayer systemsfor spintronic or valleytronic applications25, or as a platform toinvestigate bosonic quasiparticle effects with spin structures37.MethodsSample fabrication. The heterostructures were assembled using a polycarbonatefilm dry transfer technique. Supplementary Note 1 contains the methods used todetermine the crystal axes. The sample in the main text was encapsulated in5–10 nm thick hexagonal boron nitride and mounted on a c-axis sapphire substrateto allow for optical transmission measurements.Pump onehX–  Incr. Abs.X0  Decr. Abs.ProbeTrion formationeBlue Shift+Decr. Abs.ProbeBand fillingeehbadcfeCo CrossEnergy (eV)DT/T (× 10–3)CD (× 10–3)–0.20.20.00.4Transmission (theory)Transmission (theory)Energy (eV)–0.20.20.0Pump off Pump offPump onEnergy (eV)DT (theory)1.60 1.70 1.60 1.70Energy (eV)DT (theory)WSe2MoSe2+K–K +K–KCross-polarizedCo-polarized–K +K –K +KProbe �+ Probe �–ProbePumpProbePump1.55 1.60 1.651.55 1.60 1.65 1.70Energy (eV)1.55 1.60 1.65 1.70Energy (eV)1.55 1.60 1.65 1.70Figure 3 | Interlayer electron spin-valley polarization transfer. (a) Co- (burgundy) and cross- (green) circularly polarized DT spectra of the MoSe2exciton resonances, when pumping the higher energy WSe2 exciton resonance at 1.710 eV. The insets show the pump and probe scheme, where the pumpis the solid blue line, and the probe is the dashed red line. Following the interlayer transfer of photo-excited electrons from WSe2 to MoSe2, spin-valleypolarized electrons are pumped into the MoSe2 layer. The band filling of the MoSe2 conduction band is shown. (b) The pump-induced CD of theMoSe2 exciton resonances when pumping WSe2 at 1.710 eV flips sign with probe (or pump) helicity. (c–f) Theoretical explanations of the DT line shapes.(c,d) For co-polarized pump and probe, the polarized electrons populate the same valley that the probe measures. The dominant effect is a band fillingeffect, so that when the pump is on (orange curve in c), the resonance is blue shifted and the exciton absorption is partially saturated, relative to the pumpoff case (blue dashed curve in c). In this co-polarized configuration, a charged exciton cannot form due to Pauli blocking. (e,f) For cross-polarized pump andprobe, the polarized electrons populate the opposite valley that the probe measures. The dominant effect is charged exciton (X-) formation, so that whenthe pump is on (orange curve in e), the transmission is decreased at the X- resonance, and increased at the neutral exciton (Xo) resonance, relative to thepump off case (blue dashed curve in e). The insets of c,e show the difference between the modelled pump on (orange) and pump off (blue dashed) curves,corresponding to the theoretical DT spectra.NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13747 ARTICLENATURE COMMUNICATIONS | 7:13747 | DOI: 10.1038/ncomms13747 | www.nature.com/naturecommunications 5http://www.nature.com/naturecommunicationsNonlinear optical measurements. The data shown in the main text weremeasured in a cold-finger cryostat. Two continuous-wave tunable Ti:sapphirelasers (M2 SolsTiS) provided the pump and probe beams, which were eachamplitude modulated with acousto-optic modulators at frequencies near 700 kHz.Both beams were actively intensity stabilized. A probe of 20 mW and a pump of40mW average power were used for all spectra in the main text. Polarizers andbroadband waveplates were used to set the polarization of pump and probe, whichwere focused onto the sample with a microscope objective to a beam spot ofB1 mm. The transmitted light was collected by a 15 mm spherical lens that wasmounted in the cold finger of the cryostat. In the DR measurements, the reflectedprobe was collected with the objective. The pump beam was rejected with a cross-polarized set-up, or with a short or long pass filter. The probe was detected with anamplified silicon photodiode. The DT or DR signal was then measured with aphase-sensitive lock-in amplifier which was locked to the difference between thepump and probe modulation frequencies. The transmitted (T) or reflected (R)probe power was measured simultaneously with the DT or DR signal while thepump was modulated and used to normalize the DT/T or DR/R response.Data availability. The authors declare that all of the data supporting the findingsof this study are available within the article and its Supplementary Information file.References1. Wolf, S. et al. Spintronics: a spin-based electronics vision for the future. Science294, 1488–1495 (2001).2. Ross, J. S. et al. Electrical control of neutral and charged excitons in amonolayer semiconductor. Nat. Commun. 4, 1474 (2013).3. Jones, A. M. et al. Optical generation of excitonic valley coherence inmonolayer WSe2. Nat. Nanotech. 8, 634–638 (2013).4. Li, Y. et al. Valley splitting and polarization by the Zeeman effect in monolayerMoSe2. Phys. Rev. Lett. 113, 266804 (2014).5. Aivazian, G. et al. Magnetic control of valley pseudospin in monolayer WSe2.Nat. Phys. 11, 148–152 (2015).6. Xiao, D., Liu, G.-B., Feng, W., Xu, X. & Yao, W. Coupled spin and valleyphysics in monolayers of MoS2 and other group-VI dichalcogenides. Phys. Rev.Lett. 108, 196802 (2012).7. Hsu, W.-T. et al. Optically initialized robust valley-polarized holes inmonolayer WSe2. Nat. Commun. 6, 8963 (2015).8. Xie, L. & Cui, X. Manipulating spin-polarized photocurrents in 2D transitionmetal dichalcogenides. Proc. Natl Acad. Sci. USA 113, 3746–3750 (2016).9. Cao, T. et al. Valley-selective circular dichroism of monolayer molybdenumdisulphide. Nat. Commun. 3, 887 (2012).10. Zeng, H., Dai, J., Yao, W., Xiao, D. & Cui, X. Valley polarization in MoS2monolayers by optical pumping. Nat. Nanotech. 7, 490–493 (2012).11. Mak, K. F., He, K. L., Shan, J. & Heinz, T. F. Control of valley polarization inmonolayer MoS2 by optical helicity. Nat. Nanotech. 7, 494–498 (2012).12. Yu, H., Liu, G.-B., Gong, P., Xu, X. & Yao, W. Dirac cones and Dirac saddlepoints of bright excitons in monolayer transition metal dichalcogenides. Nat.Commun. 5, 3876 (2014).13. Zhu, C. et al. Exciton valley dynamics probed by Kerr rotation in WSe2monolayers. Phys. Rev. B 90, 161302 (2014).14. Wang, G. et al. Valley dynamics probed through charged and neutral excitonemission in monolayer WSe2. Phys. Rev. B 90, 075413 (2014).15. Moody, G. et al. Intrinsic homogeneous linewidth and broadening mechanismsof excitons in monolayer transition metal dichalcogenides. Nat. Commun. 6,8315 (2015).16. Wang, H. et al. Radiative lifetimes of excitons and trions in monolayers of themetal dichalcogenide MoS2. Phys. Rev. 93, 045407 (2016).17. Kang, J., Tongay, S., Zhou, J., Li, J. & Wu, J. Band offsets and heterostructuresof two-dimensional semiconductors. Appl. Phys. Lett. 102, 012111 (2013).18. Kośmider, K. & Fernández-Rossier, J. Electronic properties of the MoS2-WS2heterojunction. Phys. Rev. B 87, 075451 (2013).19. Hong, X. et al. Ultrafast charge transfer in atomically thin MoS2/WS2heterostructures. Nat. Nanotech. 9, 682–686 (2014).20. Ceballos, F., Bellus, M. Z., Chiu, H.-Y. & Zhao, H. Ultrafast charge separationand indirect exciton formation in a MoS2–MoSe2 van der Waalsheterostructure. ACS Nano 8, 12717–12724 (2014).21. Furchi, M. M., Pospischil, A., Libisch, F., Burgdörfer, J. & Mueller, T.Photovoltaic effect in an electrically tunable van der Waals heterojunction.Nano Lett. 14, 4785–4791 (2014).22. Lee, C.-H. et al. Atomically thin p–n junctions with van der Waalsheterointerfaces. Nat. Nanotech. 9, 676–681 (2014).23. Huo, N. et al. Novel and enhanced optoelectronic performances of multilayerMoS2–WS2 heterostructure transistors. Adv. Funct. Mater. 24, 7025–7031(2014).24. Rivera, P. et al. Observation of long-lived interlayer excitons in monolayerMoSe2–WSe2 heterostructures. Nat. Commun. 6, 6242 (2015).25. Yu, H., Wang, Y., Tong, Q., Xu, X. & Yao, W. Anomalous light cones and valleyoptical selection rules of interlayer excitons in twisted heterobilayers. Phys. Rev.Lett. 115, 187002 (2015).26. Rivera, P. et al. Valley-polarized exciton dynamics in a 2D semiconductorheterostructure. Science 351, 688–691 (2016).27. Maialle, M. Z., de Andrada e Silva, E. A. & Sham, L. J. Exciton spin dynamics inquantum wells. Phys. Rev. B 47, 15776–15788 (1993).28. Moody, G., Schaibley, J. & Xu, X. Exciton dynamics in monolayer transitionmetal dichalcogenides. JOSA B 33, C39–C49 (2016).29. Wang, G. et al. Giant enhancement of the optical second-harmonic emission ofWSe2 monolayers by laser excitation at exciton resonances. Phys. Rev. Lett. 114,097403 (2015).30. Seyler, K. L. et al. Electrical control of second-harmonic generation in a WSe2monolayer transistor. Nat. Nanotech. 10, 407–411 (2015).31. Li, Y. et al. Probing symmetry properties of few-layer MoS2 and h-BN byoptical second-harmonic generation. Nano Lett. 13, 3329–3333 (2013).32. Kumar, N. et al. Second harmonic microscopy of monolayer MoS2. Phys. Rev. B87, 161403 (2013).33. Zomer, P., Guimarães, M., Brant, J., Tombros, N. & van Wees, B. Fast pick uptechnique for high quality heterostructures of bilayer graphene and hexagonalboron nitride. Appl. Phys. Lett. 105, 013101 (2014).34. Schaibley, J. R. et al. Population pulsation resonances of excitons in monolayerMoSe2 with sub-1 meV linewidths. Phys. Rev. Lett. 114, 137402 (2015).35. Rigosi, A. F., Hill, H. M., Li, Y., Chernikov, A. & Heinz, T. F. Probing interlayerinteractions in transition metal dichalcogenide heterostructures by opticalspectroscopy: MoS2/WS2 and MoSe2/WSe2. Nano Lett. 15, 5033–5038 (2015).36. Yu, H., Cui, X., Xu, X. & Yao, W. Valley excitons in two-dimensionalsemiconductors. Natl Sci. Rev. 2, 57–70 (2015).37. Fogler, M., Butov, L. & Novoselov, K. High-temperature superfluidity withindirect excitons in van der Waals heterostructures. Nat. Commun. 5, 4555(2014).AcknowledgementsThis work is mainly supported by the Department of Energy, Basic Energy Sciences,Materials Sciences and Engineering Division (DE-SC0008145 and SC0012509). H.Y. andW.Y. are supported by the Croucher Foundation (Croucher Innovation Award), the RGCof Hong Kong (HKU17305914P), and the HKU ORA. J.Y. and D.M. are supported by USDoE, BES, Materials Sciences and Engineering Division. K.W. and T.T. acknowledgesupport from the Elemental Strategy Initiative performed by the MEXT, Japan and aGrant-in-Aid for Scientific Research on Innovative Areas ‘Science of Atomic Layers’ fromJSPS. P.R. and X.X. acknowledge support from the State of Washington funded CleanEnergy Institute. X.X. also acknowledges a Cottrell Scholar Award, and support from theBoeing Distinguished Professorship in Physics.Author contributionsX.X. and W.Y. conceived and supervised the project. J.S., P.R. and K.S. fabricated thedevices. J.S. performed measurements, assisted by P.R. K.S. performed and analysed thepolarization and phase-sensitive second-harmonic generation measurements. J.S., H.Y.,X.X. and W.Y. analysed the data. J.Y. and D.G.M. provided and characterized the bulkMoSe2 and WSe2 crystals. T.T. and K.W. provided B.N. crystals. J.S., X.X., W.Y. and H.Y.wrote the paper. All authors discussed the results.Additional informationSupplementary Information accompanies this paper at http://www.nature.com/naturecommunicationsCompeting financial interests: The authors declare no competing financial interests.Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions/How to cite this article: Schaibley, J. R. et al. Directional interlayer spin-valley transfer intwo-dimensional heterostructures. Nat. Commun. 7, 13747 doi: 10.1038/ncomms13747(2016).Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims inpublished maps and institutional affiliations.This work is licensed under a Creative Commons Attribution 4.0International License. The images or other third party material in thisarticle are included in the article’s Creative Commons license, unless indicated otherwisein the credit line; if the material is not included under the Creative Commons license,users will need to obtain permission from the license holder to reproduce the material.To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/r The Author(s) 2016ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms137476 NATURE COMMUNICATIONS | 7:13747 | DOI: 10.1038/ncomms13747 | www.nature.com/naturecommunicationshttp://www.nature.com/naturecommunicationshttp://www.nature.com/naturecommunicationshttp://npg.nature.com/reprintsandpermissions/http://npg.nature.com/reprintsandpermissions/http://creativecommons.org/licenses/by/4.0/http://www.nature.com/naturecommunications title_link Results Sample fabrication and electronic structure Nonlinear excitonic response of the heterostructure Demonstration of interlayer charge transfer Demonstration of interlayer spin-valley polarization transfer Figure™1Interlayer spin-valley physics.(a) Depiction of the experiment. Spin-valley polarized excitons are resonantly injected in the MoSe2 layer with a polarized laser (red). The hole transfers to the WSe2 layer where its spin-valley polarization is meas Origin of the DT line shapes Discussion Figure™2Interlayer hole spin-valley polarization transfer.(a) Non-degenerate DR of a MoSe2-WSe2 heterostructure, and an isolated WSe2 region on™the sample. When pumping on the lower energy MoSe2 exciton resonance (1.621thinspeV), there is a strong DR resp Methods Sample fabrication Figure™3Interlayer electron spin-valley polarization transfer.(a) Co- (burgundy) and cross- (green) circularly polarized DT spectra of the MoSe2 exciton resonances, when pumping the higher energy WSe2 exciton resonance at 1.710thinspeV. The insets show th Nonlinear optical measurements Data availability WolfS.Spintronics: a spin-based electronics vision for the futureScience294148814952001RossJ. S.Electrical control of neutral and charged excitons in a monolayer semiconductorNat. Commun.414742013JonesA. M.Optical generation of excitonic valley coherence  This work is mainly supported by the Department of Energy, Basic Energy Sciences, Materials Sciences and Engineering Division (DE-SC0008145 and SC0012509). H.Y. and W.Y. are supported by the Croucher Foundation (Croucher Innovation Award), the RGC of Hong ACKNOWLEDGEMENTS Author contributions Additional information