# Fileset

[PhysRevX.15.021081.pdf](https://mdr.nims.go.jp/filesets/5e6bf592-9a2e-40ff-8295-ce188e533845/download)

## Creator

Jing Liang, Yuan Xie, Dongyang Yang, Shangyi Guo, [Kenji Watanabe](https://orcid.org/0000-0003-3701-8119), [Takashi Taniguchi](https://orcid.org/0000-0002-1467-3105), Jerry I. Dadap, David Jones, Ziliang Ye

## Rights

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

## Other metadata

[Nanosecond Ferroelectric Switching of Intralayer Excitons in Bilayer <math display="inline">  <mrow>    <mn>3</mn>    <mi>R</mi>    <mtext>−</mtext>    <msub>      <mrow>        <mi>MoS</mi>      </mrow>      <mrow>        <mn>2</mn>      </mrow>    </msub>  </mrow></math> through Coulomb Engineering](https://mdr.nims.go.jp/datasets/67fc60e7-ad8c-4c77-831d-b5402b0d8e41)

## Fulltext

Nanosecond Ferroelectric Switching of Intralayer Excitons in Bilayer 3R-MoS2 through Coulomb EngineeringNanosecond Ferroelectric Switching of Intralayer Excitons in Bilayer 3R-MoS2through Coulomb EngineeringJing Liang ,1,2 Yuan Xie ,1,2 Dongyang Yang ,1,2 Shangyi Guo ,1,2 Kenji Watanabe ,3Takashi Taniguchi ,4 Jerry I. Dadap ,1,2 David Jones ,1,2 and Ziliang Ye 1,2,*1Quantum Matter Institute, The University of British Columbia,Vancouver, British Columbia V6T 1Z4, Canada2Department of Physics and Astronomy, The University of British Columbia,Vancouver, British Columbia V6T 1Z1, Canada3Research Center for Functional Materials, National Institute for Materials Science,1-1 Namiki, Tsukuba 305-0044, Japan4International Center for Materials Nanoarchitectonics, National Institute for Materials Science,1-1 Namiki, Tsukuba 305-0044, Japan(Received 18 October 2024; revised 11 February 2025; accepted 22 April 2025; published 4 June 2025)High-speed, nonvolatile tunability is critical for advancing reconfigurable photonic devices used inneuromorphic information processing, sensing, and communication. Despite significant progress indeveloping phase-change and ferroelectric materials, achieving highly efficient, reversible, rapid switchingof optical properties has remained a challenge. Recently, sliding ferroelectricity has been discovered in 2Dsemiconductors, which also host strong excitonic effects. Here, we demonstrate that these materials enablenanosecond ferroelectric switching in the complex refractive index, substantially modulating their linearoptical responses. The maximum index modulation reaches about 4, resulting in a relative reflectancechange exceeding 85%. Both on and off switching occur within 2.5 ns, with switching energy at femtojoulelevels. The switching mechanism is driven by tuning the excitonic peak splitting of a rhombohedralmolybdenum disulfide bilayer in an engineered Coulomb screening environment. This new switchingmechanism establishes a new direction for developing high-speed, nonvolatile optical memories and highlyefficient, compact reconfigurable photonic devices. Additionally, the demonstrated imaging techniqueoffers a rapid method to characterize domains and domain walls in 2D semiconductors with rhombohedralstacking.DOI: 10.1103/PhysRevX.15.021081 Subject Areas: Condensed Matter Physics,OptoelectronicsI. INTRODUCTIONSwitchable optical materials can enable nonvolatiletunability in photonic devices with applications that includeneuromorphic computing and artificial intelligence, quan-tum information processing, optical communications, andoptical sensing [1–6]. Unlike volatile tuning, no power isneeded to maintain a switched state in a nonvolatile device,which can therefore serve as an optical memory ormemristor [7,8]. To this end, much effort has been focusedon the large refractive-index tuning in chalcogenide-basedphase-change materials, which requires pico- to nanojoulesof thermal energy [9,10] to switch. Recently, bariumtitanate thin films have been developed as a promisingferroelectric material to achieve a switchable Pockels effectwith record-low switching energy and a refractive-indexmodulation up to 10−3 under a constant electric field [11].Here we present a new scheme for nonvolatile switching ofthe complex refractive index associated with the excitonicresonances in two-dimensional (2D) semiconductors. Themodulation in both the refractive index and extinctioncoefficient is about 4, resulting in a relative reflectancechange exceeding 85%. Since no heating is involved in ourswitching mechanism, the switching energy can be reducedto femtojoule levels, with both on and off switchingoccurring in less than 2.5 ns, making it the fastestnonvolatile tunability reported for optical materials [7].Our scheme relies on the dynamic tuning of the Coulombscreening of strong excitonic effects in 2D transition-metaldichalcogenides (TMDs) that are rhombohedrally stackedto enable sliding ferroelectricity.Sliding ferroelectricity is a hysteretic phenomenon in 2Dvan der Waals materials with specific stacking orders,*Contact author: zlye@phas.ubc.caPublished by the American Physical Society under the terms ofthe Creative Commons Attribution 4.0 International license.Further distribution of this work must maintain attribution tothe author(s) and the published article’s title, journal citation,and DOI.PHYSICAL REVIEW X 15, 021081 (2025)2160-3308=25=15(2)=021081(9) 021081-1 Published by the American Physical Societyhttps://orcid.org/0000-0001-6348-2068https://orcid.org/0000-0002-9788-3473https://orcid.org/0000-0002-0151-3102https://orcid.org/0009-0000-7558-3341https://orcid.org/0000-0003-3701-8119https://orcid.org/0000-0002-1467-3105https://orcid.org/0000-0001-5100-9396https://orcid.org/0000-0002-0654-9075https://orcid.org/0000-0001-8314-6977https://ror.org/03rmrcq20https://ror.org/03rmrcq20https://ror.org/026v1ze26https://ror.org/026v1ze26https://crossmark.crossref.org/dialog/?doi=10.1103/PhysRevX.15.021081&domain=pdf&date_stamp=2025-06-04https://doi.org/10.1103/PhysRevX.15.021081https://doi.org/10.1103/PhysRevX.15.021081https://doi.org/10.1103/PhysRevX.15.021081https://doi.org/10.1103/PhysRevX.15.021081https://creativecommons.org/licenses/by/4.0/https://creativecommons.org/licenses/by/4.0/where an electric field induces one layer of the material toslide relative to the other due to an interfacial polarizationarising from the interlayer coupling [12–15]. This effectcan occur in traditionally nonferroelectric materials and hasbeen electrically probed [16–21] with atomic structurechanges confirmed by scanning probe microscopy [22]and electron microscopy [23,24]. Initially found in artifi-cially stacked 2D materials with marginal twists, interfacialpolarization and its switching have recently also beenobserved in chemically synthesized rhombohedral (3R)TMDs [25–30], where preexisting domain walls play a keyrole in lowering the coercive field. Since atomically thinTMDs are semiconductors with much reduced Coulombscreening and extraordinary excitonic effects [31–34], theswitch in the stacking configuration can also be reflected inthe excitonic properties as optical contrasts, which havebeen observed between intermediate states in thick layerswith multiple interfaces [26,35]. Nevertheless, in an intrin-sic 3R bilayer with only one interface, the stable stackingsbefore and after the switch form a mirror image pair (i.e.,AB and BA), rendering them indistinguishable with con-ventional linear optical probes. Optically, the AB and BAstackings have therefore been resolved only by photo-luminescence spectroscopy under an external electricfield [26,36], which are incompatible with the applicationsdiscussed above.On the other hand, the mirror symmetry between AB andBA stacking configurations can be broken by introducingan asymmetric dielectric environment. Previous work hasshown that the Coulomb interaction in atomically thinTMDs can be modified by engineering the localenvironment [37–40]. Large Coulomb screening can sig-nificantly reduce the quasiparticle band gap and excitonbinding energy, and such a screening effect has a distancesensitivity down to a single monolayer [37]. Consequently,when a TMD bilayer is placed on a substrate, the lowerlayer will experience greater screening than the upper layerthat is not in direct contact with the dielectric substrate[Fig. 1(a)]. If the intrinsic excitonic responses in two layersare identical, such asymmetric screening will result in nonet contrast between the two stackings. However, theinterlayer coupling in a bilayer 3R TMD naturally breaksthe layer symmetry, enabling significant nonvolatile tuningof the complex refractive index near the excitonicresonance.II. STACKING-DEPENDENTEXCITONIC RESONANCESIn a rhombohedral bilayer, the two layers experiencedifferent chemical environments and can have distinctexciton peaks [36,41]. Here we define the A layer ashaving the molybdenum atom aligned with the sulfide atomin the B layer [Fig. 1(b)]. Under this definition, themolybdenum atom in the B layer is not aligned with thesulfide atom in the A layer. Consequently, the lowestexcitonic resonance in the monolayer splits into two peaksin a rhombohedral bilayer, each peak originating from anintralayer exciton composed of an electron and a holelocated within the same layer. We have previously con-firmed that the high-energy peak of this pair originates fromtheK=K0 point transition in the A layer (XA), while the low-energy peak arises from the B layer (XB) [41]. In asymmetric dielectric environment, the energy splittingbetween the two exciton peaks is determined by theinterlayer coupling strength, independent of the stackingconfiguration. However, this degeneracy is lifted when thedielectric environments of the upper and lower layersdiffer. For example, Coulomb screening from a substrate,as depicted in Fig. 1(b), modifies the energy splitting.BABASubstrategnidnib noti cxELowerUpper(a) (b)Photon energySide viewFIG. 1. Tuning the excitonic peak splitting through asymmetric dielectric screening. (a) Schematic illustration of the screening effectof a dielectric substrate on the intralayer exciton in the lower layer of a TMD bilayer. The exciton binding energy is lowered due to thereduced Coulomb interaction. (b) Schematic of the stacking-dependent excitonic peak splitting in a bilayer on a dielectric substrate.Black arrows indicate the direction of spontaneous polarization. The left and right halves illustrate the AB and BA stackings,respectively. The intrinsic peak splitting and screening-induced redshift are denoted by δ0 and Δ, respectively.JING LIANG et al. PHYS. REV. X 15, 021081 (2025)021081-2In one stacking configuration where the B layer is at thebottom of the A layer and is on top of the substrate, the low-energy XB experiences a redshift (Δ), and the peak splittingincreases (δAB ¼ δ0 þ Δ). Conversely, when the A layer isat the bottom of the B layer, the high-energy XA experi-ences a redshift, resulting in a decrease in energy splitting(δBA ¼ δ0 − Δ). As a result, when the stacking is switchedthrough sliding, the energy splitting between the twoexcitons changes accordingly. In real materials, the toplayer may also experience some degree of screening fromthe substrate, but the peak separation contrast between thetwo stackings should persist.Following this picture, we studied a 3R-MoS2 bilayerdirectly exfoliated onto a SiO2=Si substrate. The coexist-ence of both stacking configurations is reflected in thesurface potential contrast mapped by electrostatic forcemicroscopy (EFM) [Fig. 2(a)]. Because of the polarization-induced interlayer potential, the surface potential in the ABdomain is higher than that in the BA domain [42].Subsequently, we conducted optical reflectance contrast(RC) spectroscopy at 1.6 K to investigate the excitonicresponses of the two differently stacked domains, markedby orange and blue dots. Figure 2(b) presents the RCspectra and their second energy derivatives taken at thesetwo locations, respectively. The AB domain clearly exhibitstwo distinguishable peaks, while the BA domain displaysan effective single peak due to the unresolved peaksplitting, as illustrated in the lower right panel of Fig. 1(b).To quantify the asymmetric screening effect, we per-formed a mapping of the RC across the sample [Fig. 2(c)].Although optical imaging has lower resolution due to thediffraction limit, the reflectance at 1.919-eV photon energyshows a contrast similar to the EFM map. The statistics ofthe excitonic peak splitting among AB domains yield anaverage of δAB ¼ 35� 3 meV [Figs. 2(d) and 2(e)]. In aseparate experiment, we fabricated a switchable 3R-MoS2bilayer device in a symmetric dielectric environment (seeSupplemental Material Fig. S1 [43]). The excitonic peaksplitting in both stackings is about 15 meV, which is theintrinsic energy splitting (δ0) of the rhombohedral bilayer.Consequently, we conclude that the extra redshift Δ causedby the SiO2 substrate is approximately 20 meV. This valuesuggests δBA is about 5 meV smaller than the excitoniclinewidth in our unencapsulated sample (approximately40 meV), which explains the unresolved peak splitting inFig. 2(b). We found that the excitonic contrast induced bythe asymmetric Coulomb screening persists up to roomtemperature, enabling the direct visualization of the AB andBA domains using an optical microscope with a suitablebandpass filter tuned to the excitonic resonance energy[Fig. 2(f) and Supplemental Material Fig. S2 [43]]. Such anoptical contrast can be used to rapidly screen the domainFIG. 2. Excitonic contrast between AB and BA domains. (a) EFM map of a 3R-MoS2 bilayer with mixed AB and BA domains. Theflake is directly exfoliated onto an SiO2=Si substrate. The EFM contrast is caused by the surface potential difference between AB and BAstackings. Scale bar: 4 μm. (b) RC spectra and their second energy derivatives for the BA and AB domains. The measurement spots aremarked by orange and blue dots in panel (a). Two distinct exciton peaks are observed in the AB domain (XA and XB), while the BAdomain exhibits only one unresolved peak. Optical measurements are taken at 1.6 K. (c) Optical reflectance mapping (hν ¼ 1.919 eV)shows a similar contrast as in the red box in panel (a). Scale bar: 2 μm. (d),(e) Peak energy statistics of XA and XB indicate a splittingenergy of 35� 3 meV. (f) Optical images of the 3R-MoS2 bilayer captured without any filter (top), and with a band-pass filter centeredat 670 nm with a bandwidth of 10 nm (bottom). Scale bar: 4 μm.NANOSECOND FERROELECTRIC SWITCHING OF INTRALAYER … PHYS. REV. X 15, 021081 (2025)021081-3structure in chemically synthesized films on a wafer scale,benefiting the fast-growing 3R-TMD crystal-growthcommunity [44,45].III. DYNAMIC TUNING OF EXCITONICPEAK SPLITTINGNext, we show that the excitonic peak splitting can bedynamically tuned in a device that can switch the stackingconfiguration through sliding ferroelectricity [Fig. 3(a)].The device consists of an MoS2 bilayer, with the lowerlayer adjacent to a few-layer graphene (FLG) and the upperlayer next to an h-BN layer, forming an asymmetricCoulomb screening environment (see the optical imageof the device in Supplemental Material Fig. S3 [43]). Anadditional FLG is placed on top to serve as an electrode forapplying a vertical electric field. The entire stack is placedon an SiO2=Si substrate. The bilayer contains a preexistingdomain wall (DW), and as we have previously demon-strated, an out-of-plane electric field can induce an in-planepressure on this DW since the neighboring domains haveopposite interfacial polarizations [26]. When the inducedfree-energy difference across the DW exceeds the localpinning potential, the DW can be released and sweepsthrough a large area until it is trapped by another pinningcenter. This stacking configuration switching is analogousto polarization poling in conventional ferroelectrics, exceptthat the generation of DWs is challenging in slidingferroelectrics, and most switching relies on the motionof preexisting DWs [26]. Since excitonic peak splitting isenhanced in one stacking and reduced in the other, we canoptically probe the stacking configuration by measuring theRC spectrum after electrical poling. Here, the RC spec-troscopy measurements are used to characterize the steadyconfigurations. The switching dynamics are measuredusing a single-wavelength laser resonant with the excitonenergy as discussed in the next section.The electric-poling-field (Ep) -dependent RC spectra aremeasured at the center of the device using a broadbandwhite-light source focused down to a micron-sized spot[Fig. 3(b)]. When the poling field is scanned from positiveto negative, an unresolved single peak at approximately1.909 eV is observed at the positive limit, indicating theinitial stacking is BA. When Ep exceeds the coercive fieldin the negative direction (E−c ¼ −0.042 V=nm), a transi-tion in the spectrum occurs, revealing two excitonic peakswith a 22-meV splitting. Such a transition is consistent withthe substrate-only experiment and indicates that the stack-ing configuration has switched to AB. The energy splitting1.80 1.84 1.880%1%2%3%Photon energy (eV)1.88 1.92 1.960%5%10%Photon energy (eV)(a) (b)-0.1 0.0 0.10%5%10%Ep (V/nm)Reflectance1.88 1.961.922-20Gate voltage (V)1.88 1.961.92Ep(V/nm)0.1-0.10Photon energy (eV)0 0.12(c)ecnatcelfeRBAABFitsRT ReflectanceBAAB(d) (f)FLGh-BNABBAFLG1.88 1.92 1.960361.88 1.92 1.96048Photon energy (eV)(e)BAABFIG. 3. Nonvolatile switching of excitonic resonance. (a) Schematic of a 3R-MoS2 bilayer screened asymmetrically by FLG belowand an h-BN layer above. The top FLG serves as an electrode to apply a vertical electric field. The bilayer MoS2 has a preexisting DWthat can be released when the electric field surpasses the pinning threshold. Black arrows indicate the direction of spontaneouspolarization. (b) Local reflectance spectrum under a broadband white-light source focused on a micrometer-sized spot, as a function ofthe poling field in both negative (left) and positive (right) scan directions. An electric poling field is applied for one second and thenturned off, followed by a reflectance measurement. All measurements are performed at 4.5 K unless specified otherwise. (c) Reflectanceat 1.909 eV as the poling field is scanned in negative (blue) and positive (red) directions. A hysteretic loop with a large reflectancecontrast is observed, and the origin of the intermediate steps is discussed in the main text. (d),(e) The refractive index (n) and extinctioncoefficient (k) for AB and BA stackings near the excitonic resonances are extracted by fitting the reflectance spectra. The maximummodulations in both optical constants are about 4. (f) At room temperature, the excitonic peak splitting is not resolvable due to thermalbroadening, but the reflectance at the resonance peak can be switched between 1.42% and 1.95%, giving rise to a modulationdepth of 27%.JING LIANG et al. PHYS. REV. X 15, 021081 (2025)021081-4differences between the two types of asymmetric dielectricenvironments are attributed to variations in Coulombscreening, which depend on both the absolute dielectricconstant values and the contrasts between those of thesubstrate and superstrate [37]. Furthermore, polarization-switching-induced free-carrier-density changes in the FLGlayer are estimated to have a negligible impact (approx-imately 1 meV; see details in Supplemental Material [43])on δAB and δBA [46]. In contrast, no spectral change isobserved in the device with symmetric h-BN encapsulation(Supplemental Material Fig. S1 [43]). Similarly, a forwardEp scan shows an opposite stacking switch from AB to BAwhen the external field exceeds the positive coercive field(Eþc ¼ 0.038 V=nm). The coercive fields are not symmet-ric since they are mostly determined by the trappingpotential of the initial and final pinning centers. As thereflectance mapping below will show, the intermediate stateis caused by the pinning centers within the focus spot,which can trap DWs under a small poling field withvariations across devices (Supplemental MaterialFigs. S4 and S5 [43]). These intermediate states can beengineered for multibit optical memristors in the future [7].Compared to the case of the SiO2 substrate, δAB is smallerin the encapsulated device because the peak separation isdetermined by the relative screening strength between thesubstrate and the capping layer.At the photon energy of 1.909 eV, which is near thecenter of the unresolved peak and the middle of the doublepeak, a large reflectance contrast is observed before andafter the switch. The BA state yields a reflectance RBA of8.56%, which is over 6 times higher than that of the ABstate (RAB ¼ 1.23%) corresponding to a relative changeðRBA − RABÞ=RBA of 86%. (The unnormalized differ-ence spectrum is plotted in Fig. S6 of SupplementalMaterial [43]). The switch is nonvolatile with a clearhysteretic dependence on the poling field [Fig. 3(c)].Given the device geometry and the observed coercive field,the electrostatic energy needed to switch the stacking in ournonoptimized device is less than 1 pJ, indicating itspotential for reconfigurable photonic applications. Theswitching energy can be further lowered by reducing thedevice capacitance in future designs.Additionally, we retrieved the complex refractive-indexchange in the 3R-MoS2 bilayer by fitting the reflectancespectra in the AB and BA stackings [Fig. 3(d)]. The spectranear the excitonic resonance are fitted with a dielectricfunction model composed of two Lorentz peaks and abackground dielectric constant, while the multilayeredenvironment is taken into account via a transfer-matrixmodel [47–49] (see details in Supplemental Material [43]).The retrieved refractive index (n) and extinction coefficient(k) are plotted in Fig. 3(e). Near the excitonic resonance, amaximummodulation of about 4 is observed in both opticalconstants (Supplemental Material Fig. S6 [43]). The sum ofthe two oscillator strengths remains nearly constant duringthe switching event. To explore the compatibility of ourscheme with various applications, we also investigated theswitchability of the device at room temperature (Fig. 3(f)and Supplemental Material Fig. S7 [43]), where the peaksplitting is no longer resolvable due to thermal broadening.Since the merged peaks have very different effectivelinewidths, a sharp change can still be observed in thereflectance spectra when the poling field exceeds thecoercive field. At room temperature, the reflectancechanges from 1.42% to 1.95% at the resonance peak,resulting in a relative change of 27%. With the lightcurrently propagating vertically through a 1.4-nm thinactive layer, we expect that the modulation depth can befurther improved by integrating the material into anoptimized photonic device. Recent studies have shownthat sliding ferroelectricity-based memory can be retainedfor months and endure over 1011 switching cycles [20,21],making it suitable for most reconfigurable photonicdevices.IV. SWITCHING SPEEDThe large reflectance contrast between the AB and BAstackings can also reveal how the domain and DW evolvewith the poling field [Fig. 4(a)]. We focused a 648.2-nmcontinuous-wave laser on a diffraction-limited spot on thedevice and measured its reflected intensity using a high-speed avalanche photodiode (APD) while the sample israster scanned. Initially, the device is poled with a largenegative field, and most areas exhibit low reflectivity,indicating a uniform AB stacking, except for some highlyreflective regions attributed to bubbles, wrinkles, and flakeedges (see the optical image in Supplemental MaterialFig. S3 [43]). After a poling voltage of 0.55 V is applied tothe device, a small BA domain with higher reflectanceemerged near the edge of the FLG electrode. Since it is verychallenging to generate new DWs at such a low field, weattribute the new domain to the release and leftwardmovement of a preexisting DW pinned at the edge. Asthe poling field increased, the BA domain expanded further,eventually reaching the bubble region. Finally, at 0.9 V, theDW moved downward and stopped at the lowest point,converting half of the sample into a homogeneous BAdomain. The spectra in Fig. 3 measured in the middle ofthis switched BA domain (marked in Fig. 4(a) andSupplemental Material Fig. S3 [43]) reflects the propaga-tion of this single domain wall. When a negative polingfield exceeding the negative coercive field was applied, theBA domain shrank as the DW returned to its initial edge.We note that all states remained nonvolatile as the reflec-tance mapping was carried out with the electric poling fieldturned off.Finally, we performed the first real-time optical meas-urement of the switching dynamics in sliding ferroelectrics.We continuously monitored the reflected laser beamintensity [top blue panel in Fig. 4(b)] while applying aNANOSECOND FERROELECTRIC SWITCHING OF INTRALAYER … PHYS. REV. X 15, 021081 (2025)021081-5bipolar square-wave voltage to the device through anarbitrary waveform generator (AWG) (bottom red panel).The measurement spot is the same as marked in Fig. 4(a).The electric-field level was chosen to be well above thecoercive field but small enough to keep the Fermi level inthe MoS2 band gap; finite doping is unavoidable in asingle-gate device (see Supplemental Material Fig. S8[43]). In addition, since the electric field used in the speedmeasurement does not match the resonance condition, theK-K interlayer excitons do not affect the transient meas-urement results (Supplemental Material Fig. S8 [43]).Although the square wave voltage is not optimized due toa limited AWG bandwidth as well as impedance mismatch,the two signals displayed clear correlation with distincthigh and low contrast. The measured dynamic contrast issimilar to the contrast between the steady on and off statesat zero electric field, indicating a negligible contributionfrom the field-induced doping. Both rise and fall times inthe optical reflectance signal are about 2.5 ns [shown in theinsets of Fig. 4(b)], limited by the bandwidth of the APD—oscilloscope system (see Supplemental Material [43]).Interestingly, the step in the optical reflectance is signifi-cantly shorter and cleaner than that in the electric field,indicating that the abrupt change was caused by a fast-moving DW released by an electric field that exceeded thepinning threshold, and the duration spent by the DWwithinthe probing spot was very short. The intrinsic switchingspeed is thus determined only by how quickly the DWsweeps across the focus spot. Our real-time measurementresults therefore set a lower bound on the DW velocity ofapproximately 500 m=s. Assuming a free DW can propa-gate at the speed of sound, the intrinsic switching time for adiffraction-limited focus spot is expected to be on the orderof 100s of ps.V. CONCLUSIONIn summary, we have shown that nanosecond ferroelec-tric tuning of the complex refractive index with a magni-tude of approximately 4 is achievable near the excitonicresonance of a 3R-MoS2 bilayer by introducing an asym-metric Coulomb screening environment. Without furtheroptimization of the photonic environment, a large non-volatile reflectance change was observed in our device witha high energy efficiency. Compared to the contrastobserved between intermediate states in thick layers withmultiple interfaces [26,35], the engineered optical contrastbetween AB and BA stackings in the bilayer with a singleinterface is larger and more robust. In the future, one canimprove the scalability of the device by engineering thedomain walls and pinning center in chemically synthesizedfilms. The integration of this phenomenon into photonicFIG. 4. Nanosecond reflectance switching via domain-wall propagation. (a) Optical reflectance mapping (hν ¼ 1.904 eV) afterdifferent poling fields were applied shows that the DW was initially pinned at the upper right edge and propagated leftward anddownward after release. The initial, intermediate, and final states are repeatable and nonvolatile, as the DW is trapped by the same localpinning centers. Other strongly reflective features in the map match the locations of the bubbles, wrinkles, and flake edges in the device.The circle marks the measurement spot for data presented in panel (b) and Fig. 3. Scale bar: 3 μm. (b) A clear correlation is observed inthe time domain between the optical reflectance and the applied electric field at both the falling and rising edges. The insets show atransient time of about 2.5 ns, limited by the measurement instruments. The abrupt step in the APD signal is significantly shorter than theelectric-field transition, confirming that the switching is caused by the sudden release and rapid propagation of the single domain asseen in panel (a).JING LIANG et al. PHYS. REV. X 15, 021081 (2025)021081-6waveguides can also bring about new functionalities forintegrated photonics. Finally, our optical imaging methodenables the exploration of domain-wall dynamics and rapidcharacterization of the domain distribution in 3R-TMDswith a high throughput.ACKNOWLEDGMENTSWe acknowledge support from the Natural Sciences andEngineering Research Council of Canada, CanadaFoundation for Innovation, New Frontiers in ResearchFund, Canada First Research Excellence Fund, and MaxPlanck-UBC-UTokyo Centre for Quantum Materials. Z. Y.is also supported by the Canada Research Chairs Program.K.W. and T. T. acknowledge support from JSPSKAKENHI (Grants No. 19H05790, No. 20H00354, andNo. 21H05233).Z. Y. conceived and supervised the project. J. L., S. G.,Y. X., D. Y., K.W., and T. T. prepared the materials.J. L. conducted the EFM characterization. J. L. andS. G. performed the optical spectroscopy in bare flakes.J. L. and Y. X. fabricated the switchable device and per-formed the measurements with assistance from D. Y. andD. J. Data analysis was carried out by J. L., Y. X., D. Y., andZ. Y. J. L., J. D., D. J., and Z. Y. wrote the manuscript withinput fromall co-authors. J. L.,Y. X., andD. Y. contributed tothis work equally.The authors declare no competing interests.APPENDIX: METHODS1. Material and device fabricationThe 3R-MoS2 bilayer device featuring an asymmetricscreening environment was fabricated using a layer-by-layer dry-transfer technique under ambient conditions.Single crystals of 3R-MoS2 were purchased from HQGraphene.2. EFM characterizationEFM was conducted using a Molecular Vista atomicforce microscope (AFM) in the tapping mode [41]. Gold-coated AFM tips (Tap300GB-G) with a first mechanicalresonance of approximately 300 kHz, a second resonanceof approximately 1500 kHz, and a force constant ofapproximately 40 N=m were used. For topographic infor-mation, the cantilever was driven by the piezoelectricactuator at its second resonance with an oscillation ampli-tude close to 2 nm. To measure the surface potentialcontrast, the AFM tip was grounded and an ac voltageoscillating at the fundamental resonance with an amplitudeof 0.2 V was applied to the highly conductive Si substrate.The resulting electrostatic force was measured by theoscillation amplitude of the cantilever at its fundamentalresonance.3. Optical spectroscopyReflection spectroscopy was measured at the basetemperature of either an Attodry-2100 (1.6 K) orCryoAdvance-50 (4.5 K) closed-cycle optical cryostat. Aspatially filtered broadband tungsten halogen lamp wasfocused on a micrometer-sized spot on the device via amicroscope objective (NA was 0.65 for Attodry-2100 and0.5 for CryoAdvance-50). The reflected light was spectrallyresolved using a spectrometer (Princeton Instruments)equipped with a thermoelectric-cooled CCD camera. Thereflectance contrast spectra in the bare flake were normal-ized to the reference spectrum collected outside the flake.The reflectance spectra in the encapsulated device werenormalized to the reference spectrum from a silver mirror.4. Switching speed measurementThe switching speed was measured in the reflectiongeometry shown in Fig. S9 of Supplemental Material [43].The emission from a temperature-controlled diode laserwith a wavelength of 648.2 nm was focused onto adiffraction-limited spot of about 1.3 μm diameter on thedevice, which was cooled to the base temperature of theCryoAdvance-50. The focus intensity was kept below100 kW=cm2 to avoid saturation. The reflected light inten-sity was detected by a silicon avalanche photodetector witha 400-MHz bandwidth (Thorlabs APD430A). The APDvoltage was recorded by a 4-GHz digital oscilloscope(Tektronix MSO64B) using the single trigger mode. Thesquare wave output by an arbitrary waveform generator(Stanford Research DS345) was used to switch the deviceand trigger the oscilloscope. The APD-oscilloscope com-bination provides a time resolution of about 2.5 ns, asreflected in the calibration of the instrument responsefunction using a femtosecond laser (see SupplementalMaterial Fig. S10 [43]). The lower bound of thedomain-wall velocity was estimated from the ratio of thefocus diameter to the instrument response time.[1] W. Bogaerts, D. Pérez, J. Capmany, D. A. B. Miller, J. Poon,D. Englund, F. Morichetti, and A. Melloni, Programmablephotonic circuits, Nature (London) 586, 207 (2020).[2] Y. Shen, N. C. Harris, S. Skirlo, M. Prabhu, T. Baehr-Jones,M. Hochberg, X. Sun, S. Zhao, H. Larochelle, D. Englundet al., Deep learning with coherent nanophotonic circuits,Nat. Photonics 11, 441 (2017).[3] Y. Zhang, C. Fowler, J. Liang, B. Azhar, M. Y. Shalaginov,S. Deckoff-Jones, S. An, J. B. Chou, C. M. Roberts, V.Liberman et al., Electrically reconfigurable non-volatilemetasurface using low-loss optical phase-change material,Nat. Nanotechnol. 16, 661 (2021).[4] Y. Wang, P. Landreman, D. Schoen, K. Okabe, A. Marshall,U. Celano, H. S. P. Wong, J. Park, and M. L. Brongersma,Electrical tuning of phase-change antennas and metasur-faces, Nat. Nanotechnol. 16, 667 (2021).NANOSECOND FERROELECTRIC SWITCHING OF INTRALAYER … PHYS. REV. X 15, 021081 (2025)021081-7https://doi.org/10.1038/s41586-020-2764-0https://doi.org/10.1038/nphoton.2017.93https://doi.org/10.1038/s41565-021-00881-9https://doi.org/10.1038/s41565-021-00882-8[5] B. J. Shastri, A. N. Tait, T. Ferreira de Lima, W. H. P.Pernice, H. Bhaskaran, C. D. Wright, and P. R. Prucnal,Photonics for artificial intelligence and neuromorphiccomputing, Nat. Photonics 15, 102 (2021).[6] C. Wu, H. Yu, S. Lee, R. Peng, I. Takeuchi, and M. Li,Programmable phase-change metasurfaces on waveguidesfor multimode photonic convolutional neural network, Nat.Commun. 12, 96 (2021).[7] N. Youngblood, C. A. Ríos Ocampo, W. H. P. Pernice, andH. Bhaskaran, Integrated optical memristors, Nat. Photon-ics 17, 561 (2023).[8] P. L. McMahon, The physics of optical computing, Nat. Rev.Phys. 5, 717 (2023).[9] M. Wuttig, H. Bhaskaran, and T. Taubner, Phase-changematerials for non-volatile photonic applications, Nat. Pho-tonics 11, 465 (2017).[10] Y. Zhang, C. Ríos, M. Y. Shalaginov, M. Li, A. Majumdar,T. Gu, and J. Hu, Myths and truths about optical phasechange materials: A perspective, Appl. Phys. Lett. 118,210501 (2021).[11] J. Geler-Kremer, F. Eltes, P. Stark, D. Stark, D. Caimi, H.Siegwart, B. Jan Offrein, J. Fompeyrine, and S. Abel, Aferroelectric multilevel non-volatile photonic phase shifter,Nat. Photonics 16, 491 (2022).[12] M. Wu and J. Li, Sliding ferroelectricity in 2D van derWaals materials: Related physics and future opportunities,Proc. Natl. Acad. Sci. U.S.A. 118, e2115703118 (2021).[13] L.Li andM.Wu,Binary compoundbilayer andmultilayerwithvertical polarizations: Two-dimensional ferroelectrics, multi-ferroics, and nanogenerators, ACS Nano 11, 6382 (2017).[14] J. Ji, G. Yu, C. Xu, and H. J. Xiang, General theory forbilayer stacking ferroelectricity, Phys. Rev. Lett. 130,146801 (2023).[15] B. T. Zhou, V. Pathak, and M. Franz, Quantum-geometricorigin of out-of-plane stacking ferroelectricity, Phys. Rev.Lett. 132, 196801 (2024).[16] K. Yasuda, X. Wang, K. Watanabe, T. Taniguchi, and P.Jarillo-Herrero, Stacking-engineered ferroelectricity in bi-layer boron nitride, Science 372, 1458 (2021).[17] M. Vizner Stern, Y. Waschitz, W. Cao, I. Nevo, K.Watanabe, T. Taniguchi, E. Sela, M. Urbakh, O. Hod,and M. Ben, Shalom, Interfacial ferroelectricity by vander Waals sliding, Science 372, 1462 (2021).[18] X. Wang, K. Yasuda, Y. Zhang, S. Liu, K. Watanabe, T.Taniguchi, J. Hone, L. Fu, and P. Jarillo-Herrero, Interfacialferroelectricity in rhombohedral-stacked bilayer transitionmetal dichalcogenides, Nat. Nanotechnol. 17, 367 (2022).[19] S. Deb, W. Cao, N. Raab, K. Watanabe, T. Taniguchi, M.Goldstein, L. Kronik, M. Urbakh, O. Hod, and M. Ben,Shalom, Cumulative polarization in conductive interfacialferroelectrics, Nature (London) 612, 465 (2022).[20] K. Yasuda, E. Zalys-Geller, X. Wang, D. Bennett, S. S.Cheema, K. Watanabe, T. Taniguchi, E. Kaxiras, P. Jarillo-Herrero, and R. Ashoori, Ultrafast high-endurance memorybased on sliding ferroelectrics, Science 385, 53 (2024).[21] R. Bian, R. He, E. Pan, Z. Li, G. Cao, P. Meng, J. Chen, Q.Liu, Z. Zhong, W. Li et al., Developing fatigue-resistantferroelectrics using interlayer sliding switching, Science385, 57 (2024).[22] L. Molino, L. Aggarwal, V. Enaldiev, R. Plumadore, I.Fal’ko, and V. A. Luican-Mayer, Ferroelectric switching atsymmetry-broken interfaces by local control of dislocationsnetworks, Adv. Mater. 35, 2207816 (2023).[23] A. Weston, E. G. Castanon, V. Enaldiev, F. Ferreira, S.Bhattacharjee, S. Xu, H. Corte-León, Z. Wu, N. Clark, A.Summerfield et al., Interfacial ferroelectricity in marginallytwisted 2D semiconductors, Nat. Nanotechnol. 17, 390(2022).[24] K. Ko, A. Yuk, R. Engelke, S. Carr, J. Kim, D. Park, H. Heo,H.-M. Kim, S.-G. Kim, H. Kim et al., Operando electronmicroscopy investigation of polar domain dynamics intwisted van der Waals homobilayers, Nat. Mater. 22, 992(2023).[25] P. Meng, Y. Wu, R. Bian, E. Pan, B. Dong, X. Zhao, J. Chen,L. Wu, Y. Sun, Q. Fu et al., Sliding induced multiplepolarization states in two-dimensional ferroelectrics, Nat.Commun. 13, 7696 (2022).[26] D. Yang, J. Liang, J. Wu, Y. Xiao, J. I. Dadap, K. Watanabe,T. Taniguchi, and Z. Ye, Non-volatile electrical polarizationswitching via domain wall release in 3R-MoS2 bilayer, Nat.Commun. 15, 1389 (2024).[27] D. Yang, J. Wu, B. T. Zhou, J. Liang, T. Ideue, T. Siu, K. M.Awan, K. Watanabe, T. Taniguchi, Y. Iwasa et al., Sponta-neous-polarization-induced photovoltaic effect in rhombo-hedrally stacked, Nat. Photonics 16, 469 (2022).[28] J. Wu, D. Yang, J. Liang, M. Werner, E. Ostroumov, Y.Xiao, K. Watanabe, T. Taniguchi, J. I. Dadap, D. Jones et al.,Ultrafast response of spontaneous photovoltaic effect in3R-MoS2-based heterostructures, Sci. Adv. 8, eade3759(2022).[29] T. H. Yang, B.-W. Liang, H.-C. Hu, F.-X. Chen, S.-Z. Ho,W.-H. Chang, L.Yang,H.-C. Lo, T.-H. Kuo, J.-H. Chen et al.,Ferroelectric transistors based on shear-transformation-mediated rhombohedral-stacked molybdenum disulfide,Nat. Electron. Rev. 7, 29 (2024).[30] S. Roux, J. Fraunié, K. Watanabe, T. Taniguchi, B.Lassagne, and C. Robert, Optical detection of slidingferroelectric switching in hBN with a monolayer, NanoLett. 25, 321 (2025).[31] Z. Ye, T. Cao, K. O’Brien, H. Zhu, X. Yin, Y. Wang, S. G.Louie, and X. Zhang, Probing excitonic dark states insingle-layer tungsten disulphide, Nature (London) 513, 214(2014).[32] M.M. Ugeda, A. J. Bradley, S.-F. Shi, F. H. da Jornada, Y.Zhang, D. Y. Qiu, W. Ruan, S.-K. Mo, Z. Hussain, Z.-X.Shen et al., Giant bandgap renormalization and excitoniceffects in a monolayer transition metal dichalcogenidesemiconductor, Nat. Mater. 13, 1091 (2014).[33] A. Chernikov, T. C. Berkelbach, H. M. Hill, A. Rigosi, Y. Li,B. Aslan, D. R. Reichman, M. S. Hybertsen, and T. F. Heinz,Exciton binding energy and nonhydrogenic Rydberg seriesin monolayer, Phys. Rev. Lett. 113, 076802 (2014).[34] G. Wang, A. Chernikov, M. M. Glazov, T. F. Heinz, X.Marie, T. Amand, and B. Urbaszek, Colloquium: Excitonsin atomically thin transition metal dichalcogenides, Rev.Mod. Phys. 90, 021001 (2018).[35] J. Liang, D. Yang, J. Wu, Y. Xiao, K. Watanabe, T.Taniguchi, J. I. Dadap, and Z. Ye, Resolving polarizationJING LIANG et al. PHYS. REV. X 15, 021081 (2025)021081-8https://doi.org/10.1038/s41566-020-00754-yhttps://doi.org/10.1038/s41467-020-20365-zhttps://doi.org/10.1038/s41467-020-20365-zhttps://doi.org/10.1038/s41566-023-01217-whttps://doi.org/10.1038/s41566-023-01217-whttps://doi.org/10.1038/s42254-023-00645-5https://doi.org/10.1038/s42254-023-00645-5https://doi.org/10.1038/nphoton.2017.126https://doi.org/10.1038/nphoton.2017.126https://doi.org/10.1063/5.0054114https://doi.org/10.1063/5.0054114https://doi.org/10.1038/s41566-022-01003-0https://doi.org/10.1073/pnas.2115703118https://doi.org/10.1021/acsnano.7b02756https://doi.org/10.1103/PhysRevLett.130.146801https://doi.org/10.1103/PhysRevLett.130.146801https://doi.org/10.1103/PhysRevLett.132.196801https://doi.org/10.1103/PhysRevLett.132.196801https://doi.org/10.1126/science.abd3230https://doi.org/10.1126/science.abe8177https://doi.org/10.1038/s41565-021-01059-zhttps://doi.org/10.1038/s41586-022-05341-5https://doi.org/10.1126/science.adp3575https://doi.org/10.1126/science.ado1744https://doi.org/10.1126/science.ado1744https://doi.org/10.1002/adma.202207816https://doi.org/10.1038/s41565-022-01072-whttps://doi.org/10.1038/s41565-022-01072-whttps://doi.org/10.1038/s41563-023-01595-0https://doi.org/10.1038/s41563-023-01595-0https://doi.org/10.1038/s41467-022-35339-6https://doi.org/10.1038/s41467-022-35339-6https://doi.org/10.1038/s41467-024-45709-xhttps://doi.org/10.1038/s41467-024-45709-xhttps://doi.org/10.1038/s41566-022-01008-9https://doi.org/10.1126/sciadv.ade3759https://doi.org/10.1126/sciadv.ade3759https://doi.org/10.1038/s41928-023-01073-0https://doi.org/10.1021/acs.nanolett.4c05062https://doi.org/10.1021/acs.nanolett.4c05062https://doi.org/10.1038/nature13734https://doi.org/10.1038/nature13734https://doi.org/10.1038/nmat4061https://doi.org/10.1103/PhysRevLett.113.076802https://doi.org/10.1103/RevModPhys.90.021001https://doi.org/10.1103/RevModPhys.90.021001switching pathways of sliding ferroelectricity in trilayer3R −MoS2, Nat. Nanotechnol. 20, 500 (2025).[36] J. Sung, Y. Zhou, G. Scuri, V. Zólyomi, T. I. Andersen, H.Yoo, D. S. Wild, A. Y. Joe, R. J. Gelly, H. Heo et al., Brokenmirror symmetry in excitonic response of reconstructeddomains in twisted bilayers, Nat. Nanotechnol. 15, 750(2020).[37] A. Raja, A. Chaves, J. Yu, G. Arefe, H. M. Hill, A. F. Rigosi,T. C. Berkelbach, P. Nagler, C. Schuller, T. Korn et al.,Coulomb engineering of the bandgap and excitons in two-dimensional materials, Nat. Commun. 8, 15251 (2017).[38] A. Raja, L. Waldecker, J. Zipfel, Y. Cho, S. Brem, J. D.Ziegler, M. Kulig, T. Taniguchi, K. Watanabe, E. Malicet al., Dielectric disorder in two-dimensional materials,Nat. Nanotechnol. 14, 832 (2019).[39] M. He, J. Cai, H. Zheng, E. Seewald, T. Taniguchi, K.Watanabe, J. Yan, M. Yankowitz, A. Pasupathy, W. Yaoet al., Dynamically tunable moiré exciton Rydberg states ina monolayer semiconductor on twisted bilayer grapheme,Nat. Mater. 23, 224 (2024).[40] A. Chaves, J. G. Azadani, H. Alsalman, D. R. da Costa, R.Frisenda, A. J. Chaves, S. H. Song, Y. D. Kim, D. He, J.Zhou et al., Bandgap engineering of two-dimensionalsemiconductor materials, npj 2D Mater. Appl. 4, 29 (2020).[41] J. Liang, D. Yang, J. Wu, J. I. Dadap, K. Watanabe, T.Taniguchi, and Z. Ye, Optically probing the asymmetricinterlayer coupling in rhombohedral-stacked bilayer, Phys.Rev. X 12, 041005 (2022).[42] J. Liang, D. Yang, Y. Xiao, S. Chen, J. I. Dadap, J. Rottler,and Z. Ye, Shear strain-induced two-dimensional slipavalanches in rhombohedral, Nano Lett. 23, 7228(2023).[43] See Supplemental Material at http://link.aps.org/supplemental/10.1103/PhysRevX.15.021081 for estimationof polarization-induced free carrier screening effect ingraphene, transfer-matrix model, dielectric function model,and supplementary figures.[44] B. Qin, C. Ma, Q. Guo, X. Li, W. Wei, C. Ma, Q. Wang, F.Liu, M. Zhao, G. Xue et al., Interfacial epitaxy of multilayerrhombohedral transition-metal dichalcogenide single crys-tals, Science 385, 99 (2024).[45] Z. Zhang, M. Hocking, Z. Peng, M. Pendharkar, E. D. S.Courtney, J. Hu, M. A. Kastner, D. Goldhaber-Gordon, T. F.Heinz, and A. J. Mannix, Phase-selective synthesis of rhom-bohedral multilayers by confined-space hybrid metal–or-ganic chemical vapor deposition, Nano Lett. 24, 12775(2024).[46] D. Tebbe, M. Schütte, K. Watanabe, T. Taniguchi, C.Stampfer, B. Beschoten, and L. Waldecker, Tailoring thedielectric screening in–graphene heterostructures, npj 2DMater. Appl. 7, 29 (2023).[47] Y. Li and T. F. Heinz, Two-dimensional models for theoptical response of thin films, 2D Mater. 5, 025021 (2018).[48] V. G. Kravets, F. Wu, G. H. Auton, T. Yu, S. Imaizumi, andA. N. Grigorenko, Measurements of electrically tunablerefractive index of monolayer and its usage in opticalmodulators, npj 2D Mater. Appl. 3, 36 (2019).[49] M. Li, S. Biswas, C. U. Hail, and H. A. Atwater, Refractiveindex modulation in monolayer molybdenum diselenide,Nano Lett. 21, 7602 (2021).NANOSECOND FERROELECTRIC SWITCHING OF INTRALAYER … PHYS. REV. X 15, 021081 (2025)021081-9https://doi.org/10.1038/s41565-025-01862-yhttps://doi.org/10.1038/s41565-020-0728-zhttps://doi.org/10.1038/s41565-020-0728-zhttps://doi.org/10.1038/ncomms15251https://doi.org/10.1038/s41565-019-0520-0https://doi.org/10.1038/s41563-023-01713-yhttps://doi.org/10.1038/s41699-020-00162-4https://doi.org/10.1103/PhysRevX.12.041005https://doi.org/10.1103/PhysRevX.12.041005https://doi.org/10.1021/acs.nanolett.3c01487https://doi.org/10.1021/acs.nanolett.3c01487http://link.aps.org/supplemental/10.1103/PhysRevX.15.021081http://link.aps.org/supplemental/10.1103/PhysRevX.15.021081http://link.aps.org/supplemental/10.1103/PhysRevX.15.021081http://link.aps.org/supplemental/10.1103/PhysRevX.15.021081http://link.aps.org/supplemental/10.1103/PhysRevX.15.021081http://link.aps.org/supplemental/10.1103/PhysRevX.15.021081http://link.aps.org/supplemental/10.1103/PhysRevX.15.021081https://doi.org/10.1126/science.ado6038https://doi.org/10.1021/acs.nanolett.4c02766https://doi.org/10.1021/acs.nanolett.4c02766https://doi.org/10.1038/s41699-023-00394-0https://doi.org/10.1038/s41699-023-00394-0https://doi.org/10.1088/2053-1583/aab0cfhttps://doi.org/10.1038/s41699-019-0119-1https://doi.org/10.1021/acs.nanolett.1c02199 Nanosecond Ferroelectric Switching of Intralayer Excitons in Bilayer 3R-MoS2 through Coulomb Engineering I. INTRODUCTION II. STACKING-DEPENDENT EXCITONIC RESONANCES III. DYNAMIC TUNING OF EXCITONIC PEAK SPLITTING IV. SWITCHING SPEED V. CONCLUSION ACKNOWLEDGMENTS APPENDIX: METHODS 1. Material and device fabrication 2. EFM characterization 3. Optical spectroscopy 4. Switching speed measurement References