# Fileset

[s41467-022-33965-8.pdf](https://mdr.nims.go.jp/filesets/4f8fb606-a366-4753-b688-821b2bf582b1/download)

## Creator

Matthew Klein, Rolf Binder, Michael R. Koehler, David G. Mandrus, [Takashi Taniguchi](https://orcid.org/0000-0002-1467-3105), [Kenji Watanabe](https://orcid.org/0000-0003-3701-8119), John R. Schaibley

## Rights

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

## Other metadata

[Slow light in a 2D semiconductor plasmonic structure](https://mdr.nims.go.jp/datasets/62addc1e-461c-4bfc-9a40-f0e995b9a016)

## Fulltext

Slow light in a 2D semiconductor plasmonic structureArticle https://doi.org/10.1038/s41467-022-33965-8Slow light in a 2D semiconductor plasmonicstructureMatthew Klein1, Rolf Binder1,2, Michael R. Koehler3, David G. Mandrus3,4,5,Takashi Taniguchi 6, Kenji Watanabe 7 & John R. Schaibley 1Spectrally narrow optical resonances can be used to generate slow light, i.e., alarge reduction in the group velocity. In a previous work, we developed hybrid2D semiconductor plasmonic structures, which consist of propagating opticalfrequency surface-plasmon polaritons interacting with excitons in a semi-conductor monolayer. Here, we use coupled exciton-surface plasmon polar-itons (E-SPPs) in monolayer WSe2 to demonstrate slow light with a 1300 folddecrease of the SPP group velocity. Specifically, we use a high resolution two-color laser technique where the nonlinear E-SPP response gives rise to ultra-narrow coherent population oscillation (CPO) resonances, resulting in a groupvelocity on order of 105m/s. Our work paves the way toward on-chip activelyswitched delay lines and optical buffers that utilize 2D semiconductors asactive elements.Coherent population oscillations (CPOs) have been used in atomicvapors and III–V quantum well semiconductor structures to generatespectrally narrow features in a medium’s index of refraction thatresults in slow light1–5. CPOs originate from the interference of twodriving fields (lasers) acting on an optical transition that gives rise to amodulation of the excited and ground state populations at the opticaldifference frequency between the pump and probe fields. The spectralwidth of the CPO resonance in a two-level system is given by theexcited state lifetime. In a solid state system, such as a transitionmetaldichalcogenide (TMD) monolayer studied here, the spectral width ofthe resonance of the CPO is usually determined by the longest lifetimestate that is coupled to the optical transition6, which can be orders ofmagnitude longer lifetime (narrower line width) than the excitondephasing time4,6–8. Monolayer TMD semiconductors host excitonswhich interact strongly with light and have been the focus of intensestudyover thepast decade9–14, leading to the riseof optoelectronic andplasmonic devices at the atomically thin limit15–20. It was previouslyshown in far-field optical measurements, that excitons in monolayerTMDs exhibit CPOs with narrow (few µeV) linewidths6 offering thepossibility to realize CPO-induced slow light via monolayer TMDs.However, the atomically thin nature of monolayer TMDs limits theirapplications to optical propagation effects such as slow light, since inthe typical optical configuration, the propagation vector is perpendi-cular to the two-dimensional (2D) layer, resulting in effectively zero(0.7 nm) interaction length. To overcome this limitation, we make useof a hybrid 2D material plasmonic structure consisting of a hexagonalboron nitride (hBN)-encapsulated WSe2 monolayer transferred on topof a metallic waveguide that supports the propagation of surfaceplasmon polaritons16 (SPPs) along the layer.Recent measurements on 2D semiconductor plasmonic struc-tures demonstrated coupling to the dark (out-of-plane) exciton21–23,electrically tunable exciton–plasmon coupling19,24, and nonlinearplasmonic modulation16,25. Previously, we developed a coupledexciton–SPP (E-SPP) model to explain the nonlinear response of theE-SPP in non-degenerate pump probe spectroscopy16. We emphasizethat the E-SPP is a hybrid mode coupling 2D excitons to propagatingSPPs over a relatively long (several micron) interaction length which isnot possible in traditional far-field opticalmeasurements, where a lightbeam is perpendicular to the 0.7 nm thick TMD monolayer. Specifi-cally, we use a 2D semiconductor plasmonic structure to realize anReceived: 31 May 2022Accepted: 10 October 2022Check for updates1Department of Physics, University of Arizona, Tucson, AZ 85721, USA. 2Wyant College of Optical Sciences, University of Arizona, Tucson, AZ 85721, USA.3Department of Materials Science and Engineering, University of Tennessee, Knoxville, TN 37996, USA. 4Materials Science and Technology Division, OakRidge National Laboratory, Oak Ridge, TN 37831, USA. 5Department of Physics and Astronomy, University of Tennessee, Knoxville, TN 37996, USA. 6Inter-national Center for Materials Nanoarchitectonics, National Institute for Materials Science, 1-1 Namiki, Tsukuba 305-0044, Japan. 7Research Center forFunctional Materials, National Institute for Materials Science, 1-1 Namiki, Tsukuba 305-0044, Japan. e-mail: johnschaibley@arizona.eduNature Communications |         (2022) 13:6216 11234567890():,;1234567890():,;http://orcid.org/0000-0002-1467-3105http://orcid.org/0000-0002-1467-3105http://orcid.org/0000-0002-1467-3105http://orcid.org/0000-0002-1467-3105http://orcid.org/0000-0002-1467-3105http://orcid.org/0000-0003-3701-8119http://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/0000-0002-8024-9193http://orcid.org/0000-0002-8024-9193http://orcid.org/0000-0002-8024-9193http://orcid.org/0000-0002-8024-9193http://orcid.org/0000-0002-8024-9193http://crossmark.crossref.org/dialog/?doi=10.1038/s41467-022-33965-8&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1038/s41467-022-33965-8&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1038/s41467-022-33965-8&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1038/s41467-022-33965-8&domain=pdfmailto:johnschaibley@arizona.eduSPP–TMD interaction lengthof L = 3μmanddemonstrate slow light vianonlinear plasmonic population oscillations. Our structure designconsists of an hBNencapsulatedWSe2monolayer transferredon topofan optically thick gold waveguide in a “cross” shape with input andoutput grating couplers allowing efficient coupling (~10% per coupler)between free space photons and SPPs (see “Methods”). We use a“cross”waveguide structure in this workbecause it allows us to controlthe propagation directions of the pump and probe SPPs. Recall thatSPPs have both an out-of-plane and in-plane polarization component,i.e.: x̂ei kSPPx�ωtð Þ + ẑei kSPPx�ωtð Þ for the x-propagating SPPs, andŷei kSPPy�ωtð Þ + ẑei kSPPy�ωtð Þ for the y-propagating SPPs where kSPP is theSPP wave vector and ω is the angular frequency. Here, we investigatethe nonlinear response of the WSe2 bright exciton whose opticaldipole is in-plane and therefore couples to the “x” and “y” componentsof the SPPs. Therefore, by studying co-propagating pump and probeSPPs, we measure a signal analogous to the co-linearly polarized far-field response, and cross-propagating SPPsmeasures the cross-linearlypolarized response.ResultsCPOs of E-SPPsThe 2D material plasmonic structures (depicted in Fig. 1a) were mea-sured with linear and nonlinear two-color continuous wave (CW)pump-probe spectroscopy (see “Methods”). Figure 1b shows an opticalmicroscope image of the structure with 2D material layers outlined.Both CW lasers were broadly tunable Ti:sapphire lasers with narrowline widths on order of 50 kHz. The pump and probe lasers werecoupled into the waveguide generating SPPs that propagated throughthe TMD layer region. SPPs were then coupled back to free spacephotons using the output gratings. A confocal pinholewas used beforedetection with a silicon photodiode to isolate light coupled out from aspecific output grating. The linear transmission spectrum of thestructure is shown in Supplementary Fig. 1. To characterize the slow-light structure, we maximized the nonlinear response by first mea-suring the differential transmission (DT) spectrum (Fig. 1c), i.e., thepump induced change to the probe transmission, with a pump laserphoton energy of 1.725 eV. The raw DT signal was recorded using alock-in amplifier andwas then divided by the linear transmission signal(T) to yield the dimensionless DT/T response. The DT/T spectrumshown in Fig. 1c was recorded using 400μW pump and probe powers(free space laser power incident on the coupler), in the co-propagatingconfiguration as depicted in the Fig. 1c inset. We attribute the primaryresonance near 1.73 eV with DT/T magnitude of ~20% to the neutralexciton (X0)26. The weaker signals near 1.71 and 1.70 eV are identified asthe charged exciton resonances27,28, and the signal at 1.68 eV is the darkexciton29. In this work, we focus on the nonlinear response of theneutral exciton (near 1.73 eV) since it exhibits the largest nonlinearresponse, which is advantageous for large slow light effects via CPO4.Nearly degenerate two-color nonlinear spectroscopy was thenperformed by fixing the pump photon energy and scanning the probephoton energy through resonance with the pump at high resolutionover a scan range of ~30 µeV (~8GHz). Figure 1d shows the high-resolution DT/T response as a function of pump–probe detuning δ(probe energy—ωpr minus pump energy—ωpp), recorded on the lowenergy side of the neutral exciton resonance (1.725 eV), for co-propagating SPPs. The DT/T response shows a narrow 6.8μeV (fullwidth half maximum) resonance with an additional amplitude of ~10%sitting “on top” of the broader DT/T resonance (shown in Fig. 1c)a bDT/T (%)DT/T (%)Probe Energy (eV) Detuning ( eV)dc-10 1002025301.7501.7251.6751.65001020 PumpProbeOutputWSe2hBNTop hBNBot hBNWSe2Fig. 1 | Coherent population oscillations of excitons in 2D plasmonic struc-tures. a Depiction of the 2D material plasmonic structure. Surface plasmonpolaritons (SPPs) are launched at the input of the device by focusing a free spacelaser onto either/both input couplers. The SPPs propagate through the waveguidewhere they interactwith excitons in theWSe2monolayer, encapsulated in hBN. TheSPPs are coupled back to free space photons by the output gratings. The upperinset depicts the hBN-WSe2-hBNheterostructure on topof thewaveguide. The axesused for the theory and simulations are shown. bOptical image of the main deviceused in the experiments. Scale bar 5 µm.The reddotted lines show the active regionof the WSe2 and blue dashed and dotted lines show the outlines of the top andbottom hBN, respectively. c Broad normalized differential transmission spectra(DT/T) for co-propagating SPP excitation. The green line denotes where theWSe2 ispumped for d. The inset shows the schematic for co-propagating SPPs. d High-resolution DT/T spectra for co-propagating SPPs being pumped at 1.731 eV(716 nm). The inset depicts the pump and probe coupling to the X0 neutral excitonstate where ωpr is the probe frequency, ωpp is the pump frequency, and δ is thedetuning between the pump and probe frequencies.Article https://doi.org/10.1038/s41467-022-33965-8Nature Communications |         (2022) 13:6216 2around zero pump–probe detuning. This narrow spectral resonance isthe signature of CPO of E-SPPs, similar to previous far-field measure-ments on excitons in monolayer MoSe26.The high-resolutionmeasurements were repeated with differentpump and probe configurations to explore the dependence of theCPOs on the pump polarization. Figure 2a, d show the experimentaldepictions for SPP pump and SPP probe as well as optical pump andoptical probe respectively. Figure 2b (c) shows the experimental(theoretical) SPP pump, SPP probe case where the blue curves cor-respond to co-propagating pump and probe, and the black is cross-propagating pump and probe. Figure 2e (f) shows the experimental(theoretical) optical pump, optical probe case where the blue curvescorrespond to co-linearly polarized pump and probe, and the black iscross-linearly polarized pump and probe. In both cases, there is anarrow CPO resonance in the co-propagation/polarized (blue) casesand no significant resonance for the cross-propagation/polarizedcases (black). The resonances in Fig. 2b, e were fitted to a weightedsum of the real and imaginary parts of a Lorentzian function with alinear offset yielding linewidths of 6.8 and 8.3μeV for Fig. 2b, e,respectively. The similar linewidths and amplitudes (within a factorof 2) of the CPO resonances for both experiment and theory showgood agreement. Similar data on a different device is shown in Sup-plementary Figs. 2a–f.E-SPP CPO theoryIn order to explain the polarization dependence of the observed CPOresonances, we developed a theoretical model based on previousoptical work on monolayer TMDs6 and E-SPP16. To do this, we modelthe TMD excitons as Lorentz oscillators a small distance above a goldsubstrate to find the E-SPP dispersion relation in the linear andnonlinear regime. The excitonicnonlinearity, including its dependenceon the pump and probe polarizations, is computed in third-orderperturbation theory along the lines of ref. 30, here assumed to bedominated by phase-space filling. The detailed expressions for thenonlinear susceptibility are given in the Supplementary Note 1. Fig-ure 2c(f) show the theoretical CPO response corresponding to probingwith SPP (optical) and pumping with SPPs (optical). In this configura-tion, the co-polarized theory shows a CPO resonance, which dis-appears in the cross-polarized case (black), consistent with ourexperimental results. We note that the theoretical results of Fig. 2ccould also be compared to an optical pumpand SPP probe experimentsince (within our approximation of neglecting out-of-plane excitons)only the squared magnitude of the in-plane electric field, present inboth optical pump and SPP pump, contributes to the nonlinearsusceptibility.Slow light from CPOsWe now turn to the slow light effect. The narrow CPO resonanceresults in a highly frequency dependent index of refraction andsignificant reduction in the E-SPP group velocity. To demonstrateslow light, we followed the technique of the Ku et al.4, and wemeasured the group velocity using a heterodyne Mach–Zehnderinterferometer (MZI) set-up depicted in Fig. 3a. Here, the 2D semi-conductor plasmonic structure is in one arm of the interferometerand the probe reference with a piezotranslator (PZT) in the otherallowing for the relative phase to be controlled. By mixing theprobe beam with a probe reference, we are able to measure thefrequency dependent phase shift. It is shown that the differencesignal between photodiode 1 (D1) and photodiode 2 (D2) is pro-portional to the cosine of the phase delay of the probe path2 Co Cross Co Cross Co Cross Co CrossDR/R (%)Detuning (μeV)DT/T (%)bad eCo-PumpPumpProbeDetuning (μeV)cDT/T (%) DR/R (%) f-10 100 -10 100-10 100 -10 10020253051015-15-10-28-26-24-20ProbeCross-PumpDetuning (μeV) Detuning (μeV)Fig. 2 | Polarization dependence of the coherent population oscillation effect.a Depiction of the surface plasmon polariton (SPP) pump and SPP probe config-uration. b High-resolution normalized differential transmission spectra (DT/T) forSPP pump and SPP probe. The blue points correspond to co-propagating which arewith to a Lorentzian (blue curve), and the black correspond to cross-propagating.cTheoretical DT/T for SPPpump and SPP probe. The blue curve corresponds to co-propagating, and the black corresponds to cross-propagating. d Depiction of theoptical pump and optical probe configuration. e High-resolution normalized dif-ferential reflection spectra (DR/R) for optical pump-optical probe. The blue pointscorrespond to co-polarizedwhich are fit with a Lorentzian (blue line), and the blackcorrespond to cross-polarized. f Theoretical DR/R for optical pump and opticalprobe. The blue curve corresponds to co-polarized and the black curve for cross-polarized.Article https://doi.org/10.1038/s41467-022-33965-8Nature Communications |         (2022) 13:6216 3(Supplementary Note 2):D1 � D2 / e�αL=2cos ϕMZ � ωLcnðωÞ� �ð1Þwhere n(ω) is the index of refractive, e�αL=2 is the absorption of theTMD layer where α is the absorption coefficient and L is its length(~3μm), and ϕMZ is a phase difference in the interferometer, con-trolled by the PZT (see Supplementary Note 2).ϕMZ is set such that atzero detuning theD1 −D2 signal is zero. The phase delay as a functionof ϕMZ is shown in Supplementary Fig. 3a, b. Figure 3b shows themeasured population pulsation signal (blue) and the phase delay thatcan be inferred from the normalized MZI signal D1 −D2 for pump onminus pump off (black). We note that the signal is scaled by 4 toaccount for our modulation scheme (see Supplementary Note 3).These lineshapes also compare favorably to what our theory predictsfor the population pulsation (blue) and the normalized D1 −D2 signal(black) in Fig. 3d. For small detunings, we can use the small angleapproximation to take the slope of the phase delay signal and esti-mate the group velocity of thematerial at different pumppowers.Wethen take c/vg − c/vg,0 as a figure of merit for the slowdown of oursystem and plot it in Fig. 3c, where c/vg,0 is the inverse group velocitywithout pump (in Supplementary Fig. 4, c/vg,0 is approximately −0.4at 0 detuning). For our highest available (vacuum equivalent) pumpintensity (450W/cm2 in Fig. 3c), we measure a group velocity of2.3 × 105m/s, which corresponds to a slowdown factor of ~1300.Using our theoretical approach, we obtain the E-SPP wave vector kxas a function of frequency directly and can thus calculate the groupvelocity as dkx=dω directly. We then show the slowdown as afunction of the pump power in Fig. 3e. The detuning dependence ofthe group velocity in the absence of the pump for the E-SPP is shownin Supplementary Fig. 4 where the magnitude of the SPP or E-SPPgroup velocity never drops below 0.9c. This implies that thesignificant decrease in the group velocity of this system is entirely Phase Delay Diff. T Phase Delay Diff. TDetuning (μeV)D1-D2c/ν g–c/vg,0DT/T (%)b caPump Intensity (W/cm2)Detuning (μeV)d e2.50-2.5-1015000-10 10002040100-22350 400 450250 50050010001075015Pump Intensity (W/cm2)10005005DT/T (%)1500D3 44mod.(%)D1-D2D3 44mod.(%)c/ν g–c/vg,0Fig. 3 | Slow light via coherent population oscillations (CPOs). a Depiction ofthe slow-light measurement. The probe is split off before the sample to create areference beam path for one arm of the Mach–Zehnder interferometer (MZI). Thereference beam path has a piezo translator (PZT) to control the phase of the het-erodyne signalmeasured fromphotodiodes 1 and 2 (D1,D2).b Pump-induced phasedelayMZI signal (black) from theCPO for 100μWprobe and400 μWpumppowers.The raw signal is scaled by a factor of 4 due to the modulation (subscript “mod”)schemeused (see SupplementaryNote 3). The normalized differential transmission(Diff. T) shown in blue shows the corresponding CPO DT/T resonance. c Groupvelocity slowdown (c/vg − c/vg,0) calculated from the phase delay for 100μWprobepower as a function of pump electric field strength at theWSe2 layer (Ep), where c—speed of light, vg—group velocity with the pump, and vg,0—group velocity inabsence of the pump. c/vg,0 is approximately−0.4. The equivalent vacuum intensityis calculated via I = 12 ϵ0c∣Ep∣2 (ϵ0 =permittivity of free space) and is used as thehorizontal axis in c and e. Error bars denote one standard deviation. d Theoreticalplots of the normalized D1 −D2 signal (black) and DT/T (blue) that correspond tothe data shown in b. e Theoretical intensity-dependent slowdown c/vg − c/vg,0.Article https://doi.org/10.1038/s41467-022-33965-8Nature Communications |         (2022) 13:6216 4due to the CPO effect, and only occurs in a narrow ~1 µeV bandwidtharound the CPO resonance (Supplementary Fig. 5).We find an (approximately) linear relationship between theslowdown and the pump power, both in theory within the χ(3)approximation (third-order nonlinear response) and in the experi-ment. This suggests that the slowdown can be further increased, atleast as long as the system is in the χ(3) regime (the experimental valuefor the slowdown (c/vg) per pump intensity, given by the slope inFig. 3c, is 7.8 cm2/W).We note that the asymmetry in Fig. 2c is due to the relativeposition of the exciton energy with respect to the SPP energy. In CPOmeasurements on other devices, the SPP probe signals were moreasymmetric. There are even small changes to the symmetry of thesignal based on the alignment of the beams into the gratings. It is verylikely that the nuances of the device fabrication, material defects (oursample is slightly doped), and alignment cause the disagreementbetween the lineshapes of the experiment and the theory. We alsoexpect that the quantitative agreement of the theory with the experi-ment could be further improved if in the theory deviations from Lor-entzian lineshapeswere taken into account, and other contributions tothe nonlinear optical response, in addition to phase-space filling, wereincluded.DiscussionIn this work, we have demonstrated a significant slow light effect with2D semiconductor excitons using an on-chip SPP waveguide. We useda nonlinear CPO resonance to demonstrate a slowdown of c/vg ~1300,limited by the available pump power in our experiment. We find thatour slow down factor compares favorably to other reports includinghBN phonons31,32 (c/vg ~500), photonic crystals33 (c/vg ~100), carbonnanotubes34 (c/vg ~200), and stack of 15 GaAs/AlGaAs quantum wells4(c/vg ~30,000). We emphasize that the 2D semiconductor plasmonicstructure used here exhibits a significant slow-down factor at opticalfrequencies and using only a single layer of a 2D semiconductor. Wenote that there are several possibilities available to increase the mag-nitude of the slow-down effect in 2D semiconductor plasmonicstructures. As previously discussed16, the nonlinear response of theE-SPP could also be enhanced by selecting materials whose excitonresonance is closer to the SPP resonance, or by using multi-layer TMDmonolayer structures separated by hBN, which could increase thenonlinear response and resulting slow-down factor by orders of mag-nitude. There are also multiple solutions to improving the overalldevice quality by increasing the transmission of the waveguide such asby using directional grating couplers35 or by using bow-tie couplers36,which could increase the overall transmission to over 80%. Our workdemonstrates the potential applications of TMD–plasmonic structuresfor optical buffers and other on-chip optical information processingapplications that require control of the group velocity of light.MethodsOptical measurementsThe hybrid hBN-WSe2-hBN/plasmonic structures were measured at4.5 K in a closed-cycle optical cryostat (Montana Instruments) toreduce thermal broadening effects and enhance the nonlinearresponse. The transmission spectra, broad nonlinear, and high reso-lution measurements were performed using tunable CW Ti:sapphirelasers (MSquared SolsTiS). The lasers were focused to a spot onto therequisite input grating coupler. Light scattered from the probe outputgrating coupler was isolated using a spatial filter and detected with asilicon photodiode. In the linear transmission measurements, theprobe laser was modulated for lock-in detection. In the nonlinearspectroscopy measurements, pump and probe beams were bothmodulated at different frequencies near 500 kHz to allow for lock-indetection at the modulation difference frequency. Due to our methodof lock-in detection, we directly measure one fourth of the totalnonlinear response, which we correct for (see Supplementary Note 3).In the high resolutionmeasurements, the pump and probe lasers werelocked to external reference cavities, and the probe laser was finescanned while maintaining reference cavity lock.In the phase delay measurements, the probe signal was split intoreference and probe beams where the reference beam path includes amirror on a PZT to control the relative phase (ϕMZ) between probe andreference. The probe beam was transmitted through the waveguidestructure, and the referencebeamwas reflectedoff a separate goldpadbefore being interfered together. The interference signal was mea-sured from both outputs of the MZI using a lock-in amplifier sub-tracting the D1 and D2 signals. In order to set ϕMZ such that the cosinefunction of Eq. 1 could be approximated by its argument, a voltagewasapplied to the PZT such that, at zero detuning between the pump andprobe energies, the measured D1 −D2 signal was zero. The detuningspectra were then obtained as described above in the high-resolutionmeasurements.Device fabricationThe gold waveguide was fabricated on 285 nm SiO2/Si using a multi-step lithography and etching process. The substrate is spun with S1813photoresist and exposed using a maskless photolithography systemand developed usingMF-319. After the photolithography step, 200nmgold was thermally evaporated onto the substrate using 10 nm tita-nium sticking layer. In the second lithography step, poly(methylmethacrylate) was spun and the grating pattern was written anddeveloped using electron beam lithography. We used an Ar-basedreactive-ion etching process to etch the grating couplers into thewaveguide. The arms of the waveguide are 5 µm× 5.5 µm with a5 µm× 5 µm central region. The grating couplers are composed of 5grooves that are 60 nm deep with a width of 110 nm and period of570 nm. The bare waveguide was characterized using atomic forcemicroscopy and optical spectroscopy. The waveguide and gratingcoupler designs were optimized using a finite-difference time-domain(Lumerical) model. 2D materials were obtained via Scotch tape exfo-liation from bulk crystals. The 2D heterostructure was fabricated andtransferred onto the waveguide using a polymer-based technique37.Data availabilityThe data that support the findings of this study are available in theFigshare database at the following link: https://figshare.com/projects/Slow_Light_in_a_2D_Semiconductor_Plasmonic_Structure/150330.Code availabilityCodes used in this paper may be requested from the correspondingauthor.References1. Wang, H., Jiang, M. & Steel, D. Measurement of phonon-assistedmigration of localized excitons in GaAs/AlGaAs multiple-quantum-well structures. Phys. Rev. Lett. 65, 1255–1258 (1990).2. Bigelow, M. S., Lepeshkin, N. N. & Boyd, R. W. Observation ofultraslow light propagation in a ruby crystal at room temperature.Phys. Rev. Lett. 90, 113903 (2003).3. Ulbrich, R. G. & Fehrenbach, G. W. Polariton wave packet propa-gation in the exciton resonance of a semiconductor. Phys. Rev. Lett.43, 963–966 (1979).4. Ku, P.-C. et al. Slow light in semiconductor quantum wells. Opt.Lett. 29, 2291 (2004).5. Piredda, G. & Boyd, R. W. Slow light by means of coherent popu-lation oscillations: laser linewidth effects. J. Eur. Opt. Soc RapidPubl. 2, 07004 (2007).6. Schaibley, J. R. et al. Population pulsation resonances of excitons inmonolayer MoSe2 with sub-1 μeV linewidths. Phys. Rev. Lett. 114,137402 (2015).Article https://doi.org/10.1038/s41467-022-33965-8Nature Communications |         (2022) 13:6216 5https://figshare.com/projects/Slow_Light_in_a_2D_Semiconductor_Plasmonic_Structure/150330https://figshare.com/projects/Slow_Light_in_a_2D_Semiconductor_Plasmonic_Structure/1503307. Chang, S.-W. et al. Slow light using excitonic population oscillation.Phys. Rev. B 70, 235333 (2004).8. Palinginis, P. et al. Ultraslow light (<200m/s) propagation in asemiconductor nanostructure. Appl. Phys. Lett. 87, 171102 (2005).9. Dias, A. C., Qu, F., Azevedo, D. L. & Fu, J. Band structure of mono-layer transition-metal dichalcogenides and topological propertiesof their nanoribbons: next-nearest-neighbor hopping. Phys. Rev. B98, 075202 (2018).10. Finteis, T. et al. Occupied and unoccupied electronic band struc-ture of WSe. Phys. Rev. B 55, 10400–10411 (1997).11. Chernikov, A., Ruppert, C., Hill, H. M., Rigosi, A. F. & Heinz, T. F.Population inversion and giant bandgap renormalization in atom-ically thin WS2 layers. Nat. Photonics 9, 466–470 (2015).12. Hong, X. et al. Structuring nonlinear wavefront emitted frommonolayer transition-metal dichalcogenides. Research 2020,9085782 (2020).13. Kośmider, K., González, J. W. & Fernández-Rossier, J. Large spinsplitting in the conduction band of transition metal dichalcogenidemonolayers. Phys. Rev. B 88, 245436 (2013).14. Mak, K. F., Lee, C., Hone, J., Shan, J. & Heinz, T. F. Atomically thinMoS2: a new direct-gap semiconductor. Phys. Rev. Lett. 105,136805 (2010).15. Chang, D. E., Sørensen, A. S., Hemmer, P. R. & Lukin, M. D.Quantum optics with surface plasmons. Phys. Rev. Lett. 97,053002 (2006).16. Klein, M. et al. 2D semiconductor nonlinear plasmonic modulators.Nat. Commun. 10, 3264 (2019).17. Sun, Z., Martinez, A. &Wang, F. Optical modulators with 2D layeredmaterials. Nat. Photonics 10, 227–238 (2016).18. Shi, J. et al. Plasmonic enhancement and manipulation of opticalnonlinearity in monolayer tungsten disulfide. Laser Photonics Rev.12, 1800188 (2018).19. Dibos, A. M. et al. Electrically tunable exciton–plasmon coupling ina WSe2 monolayer embedded in a plasmonic crystal cavity. NanoLett. 19, 3543–3547 (2019).20. Ono, M. et al. Ultrafast and energy-efficient all-optical modulatorbased on deep-subwavelength graphene-loaded plasmonic wave-guides. In Conference on Lasers and Electro-Optics FF2L.4(IEEE, 2018).21. Zhou, Y. et al. Probing dark excitons in atomically thin semi-conductors via near-field coupling to surface plasmon polaritons.Nat. Nanotechnol. 12, 856–860 (2017).22. Arora, A. et al. Plasmon induced brightening of dark exciton inmonolayerWSe2 for quantumoptoelectronics.Appl. Phys. Lett. 114,201101 (2019).23. Park, K.-D., Jiang, T., Clark, G., Xu, X. & Raschke, M. B. Radiativecontrol of dark excitons at room temperature by nano-opticalantenna-tip Purcell effect. Nat. Nanotechnol. 13, 59–64 (2018).24. Jiang, P. et al. Tunable strong exciton–plasmon–exciton coupling inWS2–J-aggregates–plasmonic nanocavity. Opt. Express 27,16613 (2019).25. Li, J., Tao, J., Chen, Z. H. &Huang, X. G. All-optical controlling basedon nonlinear graphene plasmonic waveguides. Opt. Express 24,22169 (2016).26. Jones, A. M. et al. Optical generation of excitonic valley coherencein monolayer WSe2. Nat. Nanotechnol. 8, 634–638 (2013).27. Rivera, P. et al. Intrinsic donor-bound excitons in ultraclean mono-layer semiconductors. Nat. Commun. 12, 871 (2021).28. Wang, G. et al. Valley dynamics probed through charged andneutral exciton emission in monolayer WSe2. Phys. Rev. B 90,075413 (2014).29. Zhang, X.-X., You, Y., Zhao, S. Y. F. & Heinz, T. F. Experimentalevidence for dark excitons in monolayer WSe2. Phys. Rev. Lett. 115,257403 (2015).30. H. Kwong, N., Rumyantsev, I., Binder, R. & Smirl, A. L. Relationbetween phenomenological few-level models and microscopictheories of the nonlinear optical response of semiconductorquantum wells. Phys. Rev. B 72, 235312 (2005).31. Ambrosio, A. et al. Selective excitation and imaging of ultraslowphonon polaritons in thin hexagonal boron nitride crystals. Light.Sci. Appl. 7, 27 (2018).32. Yoxall, E. et al. Direct observation of ultraslow hyperbolic polaritonpropagation with negative phase velocity. Nat. Photonics 9,674–678 (2015).33. Baba, T. Slow light in photonic crystals. Nat. Photonics 2,465–473 (2008).34. Li, J.-J. & Zhu, K.-D. Tunable slow and fast light device based ona carbon nanotube resonator. Opt. Express 20, 5840 (2012).35. Liu, W., Wang, G., Wen, K., Hu, X. & Qin, Y. Efficient unidirectionalSPP launcher: coupling the SPP to a smooth surface for propaga-tion. Opt. Lett. 47, 621 (2022).36. Fang, Z. et al. Plasmonic coupling of bow tie antennas with Agnanowire. Nano Lett. 11, 1676–1680 (2011).37. Elías, A. L. et al. Controlled synthesis and transfer of large-areaWS2sheets: from single layer to few layers. ACS Nano 7,5235–5242 (2013).AcknowledgementsThis work is primarily supported by AFOSR (Grant No. FA9550-20-1-0217). J.R.S. acknowledges additional support from the U.S. NationalScience Foundation (Grant Nos. ECCS-2054572 and DMR- 2054572),ARO (Grant No. W911NF2010215), and AFOSR (Grant No. FA9550-21-1-0219). R.B. acknowledges support from the U.S. National ScienceFoundation (Grant No. DMR-1839570). D.G.M. acknowledges supportfrom the Gordon and Betty Moore Foundation’s Epics Initiative, GrantGBMF9069. K.W. and T.T. acknowledge support from the ElementalStrategy Initiative conducted by the MEXT, Japan (Grant No.JPMXP0112101001) and JSPS KAKENHI (Grant Nos. 19H05790 andJP20H00354).Author contributionsJ.R.S. conceived and supervised the project. M.K. fabricated the devicesand performed the experiments. M.K. and J.R.S. analyzed the data withinput from R.B. M.R.K. and D.G.M. provided and characterized the bulkWSe2 crystals. T.T. and K.W. provided hBN crystals. R.B. developed andevaluated the theory. M.K., R.B., and J.R.S. wrote the paper. All authorsdiscussed the results.Competing interestsThe authors declare no competing interests.Additional informationSupplementary information The online version containssupplementary material available athttps://doi.org/10.1038/s41467-022-33965-8.Correspondence and requests for materials should be addressed toJohn R. Schaibley.Peer review information Nature Communications thanks the anon-ymous reviewer(s) for their contribution to the peer review of thiswork. Peer reviewer reports are available.Reprints and permission information is available athttp://www.nature.com/reprintsPublisher’s note Springer Nature remains neutral with regard to jur-isdictional claims in published maps and institutional affiliations.Article https://doi.org/10.1038/s41467-022-33965-8Nature Communications |         (2022) 13:6216 6https://doi.org/10.1038/s41467-022-33965-8http://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/s41467-022-33965-8Nature Communications |         (2022) 13:6216 7http://creativecommons.org/licenses/by/4.0/http://creativecommons.org/licenses/by/4.0/ Slow light in a 2D semiconductor plasmonic structure Results CPOs of E-nobreakSPPs E-nobreakSPP CPO theory Slow light from CPOs Discussion Methods Optical measurements Device fabrication Data availability Code availability References Acknowledgements Author contributions Competing interests Additional information