# Fileset

[s41467-022-34104-z.pdf](https://mdr.nims.go.jp/filesets/f8109ccf-67b7-4917-b92e-c06d8601d2c8/download)

## Creator

Ruirui Niu, Zhuoxian Li, Xiangyan Han, Zhuangzhuang Qu, Dongdong Ding, Zhiyu Wang, Qianling Liu, Tianyao Liu, Chunrui Han, [Kenji Watanabe](https://orcid.org/0000-0003-3701-8119), [Takashi Taniguchi](https://orcid.org/0000-0002-1467-3105), Menghao Wu, Qi Ren, Xueyun Wang, Jiawang Hong, Jinhai Mao, Zheng Han, Kaihui Liu, Zizhao Gan, Jianming Lu

## Rights

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

## Other metadata

[Giant ferroelectric polarization in a bilayer graphene heterostructure](https://mdr.nims.go.jp/datasets/463d46d3-88cd-404c-9026-9cb5ea0e80b0)

## Fulltext

Giant ferroelectric polarization in a bilayer graphene heterostructureArticle https://doi.org/10.1038/s41467-022-34104-zGiant ferroelectric polarization in a bilayergraphene heterostructureRuirui Niu1, Zhuoxian Li1, Xiangyan Han1, Zhuangzhuang Qu1, Dongdong Ding1,Zhiyu Wang1, Qianling Liu1, Tianyao Liu1, Chunrui Han 2,3, Kenji Watanabe 4,Takashi Taniguchi 4, Menghao Wu5, Qi Ren6, Xueyun Wang 6,Jiawang Hong 6, Jinhai Mao 7, Zheng Han 8,9, Kaihui Liu 1, Zizhao Gan1 &Jianming Lu 1At the interface of van der Waals heterostructures, the crystal symmetry andthe electronic structure can be reconstructed, giving rise to physical proper-ties superior to or absent in parentmaterials. Here by studying a Bernal bilayergraphenemoiré superlattice encapsulated by 30°-twisted boron nitride flakes,we report an unprecedented ferroelectric polarization with the areal chargedensity up to 1013cm−2, which is far beyond the capacity of a moiré band. Thetranslated polarization ~5 pCm−1 is among the highest interfacial ferroelectricsengineered by artificially stacking van der Waals crystals. The gate-specificferroelectricity and co-occurring anomalous screening are further visualizedvia Landau levels, and remain robust for Fermi surfaces outside moiré bands,confirming their independence on correlated electrons. We also find that thegate-specific resistance hysteresis loops could be turned off by the other gate,providing an additional control knob. Furthermore, the ferroelectric switchingcan be applied to intrinsic properties such as topological valley current.Overall, the gate-specific ferroelectricity with strongly enhanced chargepolarization may encourage more explorations to optimize and enrich thisnovel class of ferroelectricity, and promote device applications for ferro-electric switching of various quantum phenomena.A ferroelectric endowed with electrically switchable dipoles is pro-mising in the application of nonvolatile electronics with fast switchingspeed, non-destructive readout, and capability of high-densityintegration1–3. The device miniaturization towards the two-dimensional limit, however, has long been hindered by the incom-plete screening of depolarization fields at the surface4. Van der Waals(vdW) materials and their heterostructures with chemically inertinterfaces open more opportunities to realize atomically thinferroelectrics5–9. Various types of ferroelectricity have been discoveredin the exfoliated mono- or few-layer flakes from bulk polar vdWcrystals10–23, e.g., out-of-plane ferroelectricity in CuInP2S6, d1T-MoTe2and Bi2O2Se, in-plane ferroelectricity in SnS, SnSe, and β-In2Se3, andintercorrelated ferroelectricity in α-In2Se3. More interestingly, inter-facial ferroelectricity could emerge by stacking non-polar materials ina symmetry-breaking way24–32, e.g., 1T’-WTe2, parallel-hexagonal boronnitride (hBN), R-type transition metal dichalcogenides (TMDs) andReceived: 12 July 2022Accepted: 14 October 2022Check for updates1State Key Laboratory for Mesoscopic Physics, School of Physics, Peking University, Beijing, China. 2Institute of Microelectronics, Chinese Academy ofSciences, Beijing, China. 3University of Chinese Academy of Sciences, Beijing, China. 4National Institute for Materials Science, Tsukuba, Japan. 5School ofPhysics, Huazhong University of Science and Technology, Wuhan, Hubei, China. 6School of Aerospace Engineering, Beijing Institute of Technology,Beijing, China. 7School of Physical Sciences, University of Chinese Academy of Sciences, Beijing, China. 8State Key Laboratory of Quantum Optics andQuantum Optics Devices, Institute of Opto-Electronics, Shanxi University, Taiyuan, China. 9Collaborative Innovation Center of Extreme Optics, ShanxiUniversity, Taiyuan, China. e-mail: jmlu@pku.edu.cnNature Communications |         (2022) 13:6241 11234567890():,;1234567890():,;http://orcid.org/0000-0002-6257-1103http://orcid.org/0000-0002-6257-1103http://orcid.org/0000-0002-6257-1103http://orcid.org/0000-0002-6257-1103http://orcid.org/0000-0002-6257-1103http://orcid.org/0000-0003-3701-8119http://orcid.org/0000-0003-3701-8119http://orcid.org/0000-0003-3701-8119http://orcid.org/0000-0003-3701-8119http://orcid.org/0000-0003-3701-8119http://orcid.org/0000-0002-1467-3105http://orcid.org/0000-0002-1467-3105http://orcid.org/0000-0002-1467-3105http://orcid.org/0000-0002-1467-3105http://orcid.org/0000-0002-1467-3105http://orcid.org/0000-0001-5264-9539http://orcid.org/0000-0001-5264-9539http://orcid.org/0000-0001-5264-9539http://orcid.org/0000-0001-5264-9539http://orcid.org/0000-0001-5264-9539http://orcid.org/0000-0002-9915-8072http://orcid.org/0000-0002-9915-8072http://orcid.org/0000-0002-9915-8072http://orcid.org/0000-0002-9915-8072http://orcid.org/0000-0002-9915-8072http://orcid.org/0000-0002-9034-3642http://orcid.org/0000-0002-9034-3642http://orcid.org/0000-0002-9034-3642http://orcid.org/0000-0002-9034-3642http://orcid.org/0000-0002-9034-3642http://orcid.org/0000-0001-5721-6206http://orcid.org/0000-0001-5721-6206http://orcid.org/0000-0001-5721-6206http://orcid.org/0000-0001-5721-6206http://orcid.org/0000-0001-5721-6206http://orcid.org/0000-0002-8781-2495http://orcid.org/0000-0002-8781-2495http://orcid.org/0000-0002-8781-2495http://orcid.org/0000-0002-8781-2495http://orcid.org/0000-0002-8781-2495http://orcid.org/0000-0002-1558-4040http://orcid.org/0000-0002-1558-4040http://orcid.org/0000-0002-1558-4040http://orcid.org/0000-0002-1558-4040http://orcid.org/0000-0002-1558-4040http://crossmark.crossref.org/dialog/?doi=10.1038/s41467-022-34104-z&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1038/s41467-022-34104-z&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1038/s41467-022-34104-z&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1038/s41467-022-34104-z&domain=pdfmailto:jmlu@pku.edu.cnTMDs heterobilayers, greatly expanding the scope of ferroelectriccandidates.In particular, the most thoroughly studied non-polar carbonmaterial – graphene – was also found to be ferroelectric:33,34 a bilayergraphene aligned with hBN on both sides exhibits a (special) gate-specific ferroelectricity (GSFE) and concomitant anomalous screeningascribed to electron dynamics rather than purely ionic displacement.According to the interlayer charge transfer (ICT) model33, the two-dimensional polarization P2D depends on the density of correlatedelectrons accommodated by an engineered moiré band, a flexiblycontrolled degree of freedom. However, the lack of moiré features inexperimental samples so far has prevented quantitative comparisonwith the theoretical model, leaving open questions to the validity ofthe physical mechanism that is essential for extensive explorations ofthis novel family of interfacial ferroelectrics.In this work, we take advantages of clear moiré peaks and Landaulevels (LL) to quantitatively explore ferroelectricity and anomalousscreening. Surprisingly, they could be observed even when the Fermisurface is deep into the dispersive bands that are beyondmoiré bands,contradicting the prevailing ICTmodelwhere a half filling splits the flatmoiré band and induces interlayer charge transfer. The polarizedcharge density –without the stringent constraint ofmoiré bands –wasfound to reach 1013cm−2. Opposite to previous observation, GSFE canbe also switched off by the normal gate in a broad parameter space,while the anomalous screening keeps to be stable. Specifically, whenthe normal gate was set in a high-voltage range, the hysteresis loopobtained by scanning the special gate tends to collapse into the samestates with anomalous screening. For device applications, GSFE is alsoproved to act as a nonvolatile switch of intrinsic quantum propertiessuch as topological valley current35,36. Our results have clarifiedimportant characteristics of GSFE, which are essential to unveil thephysical mechanisms and promote potential applications of ferro-electric vdW devices with appreciable charge polarization.ResultsUntying the ferroelectricity and correlated electrons inmoiré bandsThe deviceD1 consists of a Bernal bilayer graphene sandwiched by twohBN flakes and graphite gates (Fig. 1a). Straight edges of graphene andhBN are used for crystallographic alignment. The resulting anglesbetween graphene and top/bottom hBN, derived from optical andelectrical characterizations (Supplementary Note 1), are respectively~30° and ~0°, where the smaller angle enables the observation of clearmoiré features acting as a marker to locate the Fermi surface duringthe ferroelectric hysteresis.Figure 1bdisplays anormal transfer characteristic by sweeping theback gate Vb, i.e., the forward and backward curves coincide with eachother. In contrast, the top gate Vt is a special one that gives rise toferroelectric hysteresis (Fig. 1c), where the resistive peaks at chargeneutral point (CNP) and full filling point (FFP, see SupplementaryNote 1.1 for the assignment) of amoiré band aredelayed to appear. Thedelay is ~0.2 V nm−1, slightly larger than the separation between CNPand FFP. Such a gate-specific hysteresis resembles that in ref.33 andD2in this study (Supplementary Note 2). Throughout the manuscript, wedefine Vb as a normal gate and Vt a special gate. To be complete, wemeasured the Vt-Vb phase diagramby sweeping Vt fast while increasinggraphiteh-BNh-BNBernal BLGVbVtgraphite~0◦~30◦h-BNGraphenea b cRxx (kΩ)Rxx (kΩ)-0.4 -0.2 0 0.2 0.4-0.2 -0.1 0 0.1 0.20.11100.1110100Vt/dt (V nm-1)Vb/db (V nm-1)Vb/db  (V nm-1)-0.200.2-0.2 0 0.2 -0.2 0 0.2Vt/dt (V nm-1)-0.4 -0.2 0 0.2 0.4Vt/dt (V nm-1)Vb/db  (V nm-1)-0.1-0.200.10.2d e -2 250 0∆Rxx (kΩ) ∆Rxx (kΩ)Fig. 1 | Ferroelectricity in abilayergraphenemoiré superlattice. aSchematicof avan der Waals heterostructure and the device measurement set-up, in which thegraphene is roughly twisted by 0° and 30° with two hBN flakes, respectively. Amoiré superlattice is thus formed at one interface and a close-quasicrystal at theother. Vt: the top gate voltage; Vb: the back gate voltage. b The transfer curves haveno hysteresis between the forward and backward scanning of Vb. The second Diracpeak indicates the moiré superlattice is of 11.5 nm, corresponding to an angle of0.69°betweenBLGandhBN.cA ferroelectric hysteresis is observedby sweepingVt.dA complete phase diagram of resistance is constructed by sweeping Vt forth (left)and back (right) while increasing Vb in a step-by-step fashion. Anomalous screeningat both CNP and FFP could be identified by horizontal lines (red and white in color,respectively). e Difference in resistance between the two panels in d. Multiplehysteresis loops are formed by three pairs of blue and red peak lines at CNP andFFP. Note that the loops for electron and hole FFP (highlighted by dashed lines) areonly half of a closed loop.Article https://doi.org/10.1038/s41467-022-34104-zNature Communications |         (2022) 13:6241 2Vb step by step from −0.2 to 0.2 V nm−1. Throughout the manuscript,the fast-scan gate is plotted in the x axis and the slow one in the y axis;the scanningdirections are denoted by arrows in each graph. As shownin Fig. 1d, striking hysteresis (and anomalous screening manifested ashorizontal ridges) are observed at both CNP (red) and two FFP (white).For better visualization, subtraction between the left and right panelsis performed in Fig. 1e, where the central parallelogram composed ofred and blue resistance peaks represents the hysteresis loop at CNP.There are another two replicas in light red and blue (highlighted bydashed lines) for two FFP, though only the lower (upper) half is formedon the electron (hole) side. It’s surprising that the shift between theparallelogram and two replicas is basically along the vertical direction,i.e., the turning into anomalous screening states depends merely onthe special gateVt/dt rather than a total displacement fieldD=(Db-Dt)/2,where Db(t)=ε0εrVb(t)/db(t). More importantly, the hysteretic FFP on thehole side indicates that GSFE and anomalous screening exist evenwhen the (possibly correlated) moiré band is empty. In the followingwe check their existence for dispersive bands at higher energy levels.Figure 2a shows the phase diagram of a backward sweep, wherefive typical linecuts (colored triangles) correspond to representativeFermi surfaces (inset of Fig. 2a): higher than the electron moiré band(Fig. 2c), inside the electron moiré gap (Fig. 2d), within the electronmoiré band (Fig. 2e), at CNP (Fig. 2f), or within the hole moiré band(Fig. 2g). All these states are driven out of the anomalous screeningstatus (shaded region) by nearly the same Vt/dt, consistent with theobservation in Fig. 1e. To better present the behavior of low-resistancestates, perpendicularmagnetic fields were applied to show the Landaulevels. In Fig. 2h, the horizontal LL could be clearly identified above 5 T.Since the flat LL is ended close to CNP, the Fermi surface in theanomalous screening regime must stay within the electron moiréband. For Fermi surfaces beyond electron and hole moiré bands, LLfans are shown in Fig. 2i, j, respectively. Then, we fix B as 10 T andmeasure the Vb-Vt phase diagram (Fig. 2b). Multiple hysteresis loops,both within and outside moiré bands, could be observed owing to theenhanced visibility by LL (See raw data in Fig. S10). To conclude, theaccess to the anomalous screening status does not rely on a specificFermi surface within a moiré band.Actually, there have been already some signatures for the absenceof correlation, e.g., no correlated resistive peaks at half filling could beobserved in the transfer curves shown in Fig. 1b. In another sample D2with nearly identical properties (Supplementary Note 2), we evencannot see FFP peaks because of relatively larger twist angle, neitherdid previous reports of GSFE33. In addition, themoiré bandmay not beisolated from dispersive bands. As shown in Fig. S2a, b, we can onlyobserve the sign reversal of carriers in LL fans of longitudinal resis-tance but not in Hall signals, which indicates the energy gap at FFP isnot complete. The absence of a full gap also explains the relativelyweak FFP peak in Fig. 1b.Strongly enhanced charge polarizationBefore presenting experimental results, we first make a simple esti-mation ofpolarized charge carriers. Following the ICTmodel33, amoiré∆Rxx(kΩ)Vb/db=-0.12 V nm-1Vb/db=-0.12 V nm-1Vb/db=0.12 V nm-1-0.1 0 0.1-0.2 0 0.2-0.2 0 0.2-0.2 0 0.2-0.3 -0.2 -0.1 0 0.1 0.2 0.3Vt/dt (V nm-1)0.20-0.2 30020040050010002505006000510051005100.20-0.2Vt/dt (V nm-1)Vt/dt (V nm-1) Vt/dt (V nm-1)Vb/db (V nm-1)Vb/db (V nm-1)0 50 510-100Rxx (kΩ) Rxx (kΩ)0.17 V nm-10 V nm-1-0.07V nm-1-0.15V nm-1-0.22 V nm-1-0.2 0 0.20.1-0.1-0.3 0.3102104a c hijdefgbRxx (Ω)Rxx (Ω)Rxx (Ω) B (T)B (T)B (T)Rxx (Ω)Rxx (Ω)EnergyDOSFig. 2 | Ferroelectricity and anomalous screening in magnetic fields. a Theanomalous screening at CNP and FFP is highlighted in a Vt-Vb phase diagram, wherefive linecuts are denoted by colored triangles. Inset: Their correspondences in theband structure. b In perpendicular magnetic fields (10 T), anomalous screeningbehavior and the associated hysteresis loops become visible in many more elec-tronic states, two of which are highlighted by the dashed curves. This is due toprominent Landau levels that enhance the visibility of low-resistance states.c–gCareful examination reveals that, a flat resistance curve independent on Vt (i.e.,anomalous screening) indeed exists in electronic states with Fermi surfaces higherthan the electron moiré band c, at FFP d, within the electron moiré band e, at CNPf, andwithin the holemoiré band g. Importantly, these anomalous curves spanoveralmost the same Vt regimes highlighted by the grey shadow, despite their distinctdisplacement fields. h–j Landau fans are characterized by three typical electronicstates, whose Fermi surfaces (when the anomalous screening occurs) are aboveh and within i the electron moiré band, and below the hole moiré bandj, respectively.Article https://doi.org/10.1038/s41467-022-34104-zNature Communications |         (2022) 13:6241 3band would accommodate strongly correlated electrons37,38, half ofwhich transfer to another spatially separate graphene layer and even-tually induce interlayer electric dipoles33. Take the twist angle of theD1 sample (0.69°) as an example, a half moiré band will accommodateelectrons ~1.7 × 1012cm−2. Assuming the band polarization in the bilayergraphene is perfect, then the interlayer charge transferΔn is just equalto the half moiré band. Since the interlayer distance ddipole is 0.34 nm,the polarization would be P2D~eΔn × ddipole~0.9 pCm−1. We note that inref. 33. the difference in displacement fields between two gaplesscharge neutral points P3D = ε0×Δ(D/ε0) was taken as an independentway to determine the remnant polarization P2D = P3D× ddipole=ΔD×ddipole. However, according to the standard parallelogram depic-ted in the inset of Fig. 3a, one gets ΔD = eΔn. Consequently, these twomethods are not independent but rather exactly the same.Experimentally, the charge polarization 2Δn in graphene can beconveniently measured from the shift of CNP between back and forthsweeping (the inset of Fig. 3a). During several thermal cycling of D1, wefound the ferroelectric hysteresismay change itsmagnitude by severaltimes (labeled as States 1–4 in Fig. 3a), but are qualitatively similar (Seeadditional data for State 1 in Figs. S17, 18). Note that the moiré featureremains the same in all states. It’s well known that thermal cyclingmaychange the heterostructures via strain and twist angle39,40, however,the twist angle of 0.69° responsible for the moiré superlattice seemsnot influenced. We then compare 2Δn for all the four states (filledcircles) with the electron density accommodated by the moiré band(dashed horizontal lines) in Fig. 3b. Distinct from similar devices inprevious reports where Δn (open symbols) are only a small part of amoiré band, Δn in State 1 (red circle) is in approximate to the halffilling, and States 2–4 have much larger polarized charge density thanthe full filling (See Supplementary Note 1.1 for the identification ofFFP). As for themost striking State 4,Δn approaches 1013cm−2, which istranslated as P2D~5 pC m−1, significantly larger than the calculated 0.9pC m−1. The exceptionally high charge polarization questions the roleof moiré bands in the ICT model33.Also plotted in Fig. 3b are ferroelectric van der Waals materials,including α-In2Se3 that is arguably the strongest monolayerferroelectric8, and typical interfacial ferroelectricity24,27,30 (1T′-WTe2,parallel-hBN and R-TMDs). Note that here the Δn for these systems arenot intrinsically polarized charge within ferroelectrics but are sensedby a conduction layer, where Δn = P2D/edwith d as the thickness of thetotal gating dielectric27. Were the Δn observed in GSFE caused byparallel-stacking (i.e., rhombohedral) of hBN layers, the large polar-ization would require each interface inside the hBN flake of the specialgate to be such aligned, which is energetically unfavorable.Discussion about the physical mechanismThe origin of GSFE can be either electron dynamics (the ICT model inthe bilayer graphene) or lattice distortion (sliding ferroelectricitywithin hBN flakes), both of which cannot explain all of the essentialfeatures observed in experiments. As discussed for the ICTmodel, theamount of transferred electrons can well exceed the capacity of amoiré band (Fig. 3) and the concomitant anomalous screening canoccur outside the moiré bands, i.e., within dispersive bands (Fig. 2),both suggesting that the mechanism does not rely on electroniccorrelation.The sliding ferroelectricity, although excluded when confinedwithin hBN flakes, was also proposed to take place at the interfacebetween crystallographically aligned graphene and hBN7. Two polarstates correspond to carbon atoms overlapping with either boron ornitrogen atoms; by expanding the area of one polar state (andshrinking theother correspondingly), the net polarization over amoirésupercell can be switched upwards or downwards (Figs. S20, S21 andassociated discussions in Supplementary Information, section 4). Thesame order of P2D in GSFE (~5 pCm−1) and prototypical sliding ferro-electrics (~2 pCm−1) encourages further exploring the relation betweenthe two. The drawback, however, lies in that (1) anomalous screening ismore likely to be driven by electron dynamics; (2) the polar states areexpected to survive only under perpendicular electric fields in ideal2∆n (10¹² cm-2)d (nm)θ=0.69°Moiré band θ=0°)State 1State 2State 3State 4Parallel-BNP2D~2.25 pC·m-1 (exp.)3R-TMDsP2D~2 pC·m-1 (exp.)Monolayer �-In SeP2D~12 pC·m-1 (theory)1T'-WTeP2D~0.16 pC·m-1 (exp.)D2D10.010.1101 100110010Rxx (kΩ)0.11100.10100.11100.1110Vt/dt  (V nm-1)0-0.4 0.80.4a b0-6 6next (10¹² cm-2)D/ε  (V nm-1)-0.600.60 5∆Rxx (kΩ) -5P3D2∆nP2D~5 pC·m-1Untwisted MoS /WSP2D~1.45 pC·m-1 (exp.)Fig. 3 | Enhanced charge polarization. a Four states with distinct magnitudes ofhysteresis. Note the satellite peaks at FFP exhibit a constant shift from CNP in allstates, indicating the moiré superlattice keeps unchanged. The sweeping curves ofState 2 is also denoted as the black cutting line in the inset, where the polarizedcharge Δn and the internal electric dipole P3D are also illustrated. b Summary ofpolarized charge densities in various ferroelectric systems. The 2Δn for D1 and D2are depicted by filled circles, which are close to or well above the capability oftypical moiré bands (the red dashed line: the twisted angle of 0.69° as in D1). GSFEdevices from literatures are plotted as open circles33 and squares34. The horizontalgrey background emphasizes that their polarization Δn actually do not depend onthe thickness of gating dielectrics d. However, for others integrated in the dielectricgate, the induced charge polarization in the conduction channel is inversely pro-portional to the total thickness of the gating dielectrics. 1T’-WTe2, parallel-hBN andR-TMDs are taken from experiments, while monolayer α-In2Se3 from theoreticalcalculations.Article https://doi.org/10.1038/s41467-022-34104-zNature Communications |         (2022) 13:6241 4cases; for practical samples with pinning centers the states may exist,but probably coexist during switching, contrasting a uniform ferro-electric polarization in GSFE (Fig. S1c).Overall, to reconcile all the experimental findings, e.g., the giantcharge polarization and anomalous screening, it may be necessary totake into account both the electron-driven ICT model and the lattice-driven sliding ferroelectricity. A convincing mechanism calls for moreconclusive experimental features.The normal gate as a switch knob of the special gateFigure 4a summarizes the scan ranges performed. In brief, within theregime of Vb/db < 0.15 V nm−1 (highlighted in blue), a stable ferroelec-tricity could be observed. However, above this threshold value(determined in Fig. S7), the hysteresis loop is metastable: the forwardand backward scan tend to collapse into the same stable states inFig. 4b (see the vanishing hysteresis in Fig. S3). The above stabilitycriterion seems not to depend on Vt (Figs. S8, S9). To quantify thedynamic process, a relaxation time τ is derived by monitoring theevolution of CNP as a function of time that is fitted by an exponentialdecay function. As shown in Figs. 4c and S4, τ is around 7.1 and0.2 hours for Vt > 0 and Vt < 0, respectively.The strong dependence of relaxation time on the normal gateprovides a convenient way to switch ferroelectricity. Firstly, whenVb/db is relatively small, the resistance ridge of anomalous screen-ing is strictly in parallel with the Vt axis, forming a standard paral-lelogram (Fig. 4d with raw data in Fig. S6). Upon increasing thescanning range of Vb, e.g., 0.21 V nm−1 in Fig. 4e, the relaxation to thestable states in Fig. 4b becomes so significant that the CNP peakdeviates from a vertical line gradually. The relaxation is much moreserious for Vt < 0 (τ~12minutes), so the hysteresis almost dis-appears. To suppress the relaxation, in Fig. 4f the scanning rate isaccelerated by a particular gate sweeping scheme (Fig. S5). Asexpected, a full hysteresis loop could be restored, resembling thatin Fig. 1e.Ferroelectric switching of intrinsic quantum propertiesWe firstly examine the thermal stability. At various temperatures upto 300 K, typical phase diagrams are measured (Figs. S11–14), fromeach of which the magnitude of ferroelectric polarization could beextracted. As shown in Fig. 5a, the polarization remains stable below100 K, but decays gradually at higher temperatures. The robustnessin a wide range of temperature allows for investigating manyquantum phenomena, e.g., the topological valley current. Followingwell-established measurement schemes34,35, we obtain the local andnonlocal signals in Fig. 5b, c, respectively (See raw data in Fig. S15).To confirm the nonlocal results indeed originate from valleytransport, we extract the resistance of CNP and compare theirrelationship with RNL~RL3 in Fig. 5d. Apart from the low-resistanceregime, the fitting is quite good over two decades, indicating thatthe bulk valley transport starts to dominate once the energy gap isopened at CNP. The results also suggest that the ferroelectricity iscompatible with intrinsic properties of a bilayer graphene, openingmore opportunities to manipulate these quantum phenomena withadditional degree of freedom.0-0.1 0.1 0-0.1 0.1-5 0 5Vt/dt  (V nm-1)Vb/db (V nm-1)∆Rxx (kΩ)ON ON0-0.10.1Time (h)40 2 8 10-0.14-0.16=7.1-0.18-0.260.20.10Vt   backwardVt   forward�=0.2�ad0-0.2 -0.1 0.20.10 1-1Vt/ dt  (V nm-1)Vb/db (V nm-1)∆Rxx (kΩ)0-0.4-0.20.20.4f0-0.2 -0.1 0.20.1-2 0 2Vt/dt  (V nm-1)Vb/db (V nm-1)∆Rxx (kΩ)0-0.4-0.20.20.4ec0-0.2 -0.1 0.20.10 5Vt/dt  (V nm-1)Vb/db (V nm-1)Rxx (kΩ)0-0.3-0.2-0.10.10.30.2bVb/ db (V nm-1)Vb/db (V nm-1)-0.3 0.30-0.15 0.15Vb/db (V nm-1)Vt/dt  (V nm-1)-0.2-0.40.20.40bde&f Fig.S6ON ONOFFFig. 4 | Gate-specific ferroelectricity controlledby theother gate. aTwo regimesare identified according to the stability of the hysteretic behavior. Below thethreshold of Vb/db ~0.15 V nm−1 the ferroelectricity is stable (highlighted in blue),whereas it becomes more and more fragile when exceeding the threshold. Thescanning ranges used in other panels and Fig. S6 are denoted. b Take the scanningrange of Vb/db as above 0.2 V nm−1, one obtains a phase diagram without ferro-electric hysteresis by scanning Vt/dt slowly (see its hysteresis-free counterpart inFig. S3). However, the anomalous screening remains robust. c The relaxation to thestable states shown in b is quantified by a temporal characterization. Two repre-sentative points are selected: one is for Vt > 0 in the backward scan and the other isVt < 0 in the forward scan. The characteristic time fitted by an exponential decayfunction are 7.1 and 0.2 h, respectively. Measurement details and raw data can befound in Fig. S4. d Below the threshold, a regular parallelogram can be formed. Anobvious criterion is that the resistance ridge at CNP in the anomalous screeningregime is strictly parallel with the y axis. Here the scanning rangeofVb/db is 0.1 (left)and 0.15 (right) V nm−1. The label ON/OFF means hysteresis loops are switched on/off. e In a larger scanning range of Vb ~0.21 V nm−1, the relaxation distorts theparallelogram in the upper-left corner of the Vt-Vb phase diagram, i.e., the ridge isno more in parallel with the y axis; furthermore, it eliminates completely the hys-teresis in the lower-right corner. f By increasing the scanning rate through a par-ticular scheme (see details in Fig. S5), the relaxation could be overcome andhenceafull hysteresis loop is restored.Article https://doi.org/10.1038/s41467-022-34104-zNature Communications |         (2022) 13:6241 5DiscussionTo conclude, by using moiré features and Landau levels as indicators,wemanage to prove that the gate-specific ferroelectricity does not relyon correlated electrons. Naturally, the charge polarization has beenfound to exceed the half filling of a moiré band. The maximum chargepolarization achieved so far is approximately 1013cm−2, which may befurther improved by optimizing twist angles and constituentmaterials.For device applications, we find the normal gate actually acts as aswitching knob of the functioning of the special gate; the wide com-patibility allows for ferroelectric switching of various quantum phe-nomena including topological valley transport as demonstrated.Overall, our results disclose sophisticated functionalities of the gate-specific ferroelectricity and provide key clues of the underlyingmechanism that may be essential to the discovery of a large family offerroelectrics, promoting potential applications of van der Waals fer-roelectric devices with a large charge polarization.MethodsSample fabricationThe bilayer graphene crystals, graphite gates, and boron nitride flakeswere allmechanically exfoliatedon to siliconwafers and identifiedwithoptical microscopy. The multilayer heterostructure was then fabri-cated following the standard dry transfer method41. Then reactive ionetching was used to pattern the Hall bar geometry. At last, one-dimensional edge contacts42 were prepared following the standarde-beam lithography and e-beam evaporation of Cr (1 nm)/Au (50 nm).The device geometries are detailed in Fig. S1.Characterization of crystallographic directionFor graphene, samples with a pair of long and straight edges whoseangles are 30° or 90° were selected and then 532 nm laser withpolarization aligned in parallel with edges was used to obtain Ramanspectra at D band (~1350 cm−1). The intensity for zigzag edges is neg-ligible, whereas that for an armchair edge ismore significant43. For hBNflakes, the crystallographic axes were identified by Second HarmonicGeneration (SHG)44. SHG signals were collected by WiTec UHTS300with incident lightwavelength of 1064 nmand afixed excitation powerof 20mW. It is minimized when the polarization of incidence is alongthe zigzag and maximized for the armchair direction.Electrical measurementAll the transport measurements were carried out in a cryostat withbase temperature of 1.5 K and a superconducting magnet up to 14 T.Unless specified otherwise, the sample temperature was at base tem-perature. A standard four-probe method of constant current wasperformed. The AC current was supplied by Stanford Research Sys-tems SR830 lock-in amplifiers with a working frequency of 17.777Hz.TheDCgate voltageswereoutput by twoKeithley 2400 SourceMeters(See device configurations in Fig. S1). The nonlocal measurement inFig. 5 in the main text was carried out at 50 K following previousworks35,36.Reporting summaryFurther information on experimental design is available in the NatureResearch Reporting Summary linked to this paper.0.10-0.10.4-0.4 0.2-0.2 0-5 0 5-0.5 0 0.5RL (kΩ)∆RNL (kΩ)Vt/dt (V nm-1)Vb/db (V nm-1)0.10-0.10.4-0.4 0.2-0.2 0Vt/dt (V nm-1)Vb/db (V nm-1)∆RL (kΩ)cd0.1110100101RNL (kΩ)RNL~R3LVIVIa bState 2P  (V nm-1)01.51.10.20.42∆n (10¹² cm-2)016122450 100 1500T  (K)Full filling ofMoiré band117101717107Vt   backwardVt   forwardFig. 5 | Ferroelectric switching of topological valley current. a The internalelectrical polarization P3D (State 2) keeps almost the same below 100K, but gra-dually decays at higher temperature. b, c Local b and nonlocal c signals arecompared at 50K. Only at CNP the nonlocal resistance could be observed. Inset:Measurement schemes. d The scaling of RNL~RL3 is consistent with the topologicalvalley transport.Article https://doi.org/10.1038/s41467-022-34104-zNature Communications |         (2022) 13:6241 6Data availabilityAll figures are provided in Source Data file. All other data that supportthe findings of this study are available from the corresponding authorupon reasonable request. Source data are provided with this paper.References1. Martin, L. W. & Rappe, A. M. Thin-film ferroelectric materials andtheir applications. Nat. Rev. Mater. 2, 1–14 (2016).2. Khan, A. I., Keshavarzi, A. & Datta, S. The future of ferroelectric field-effect transistor technology. Nat. Electron. 3, 588–597 (2020).3. Bertolazzi, S. et al. Nonvolatile memories based on graphene andrelated 2D materials. Adv. Mater. 31, 1806663 (2019).4. Dawber, M., Rabe, K. M. & Scott, J. F. Physics of thin-film ferro-electric oxides. Rev. Mod. Phys. 77, 1083–1130 (2005).5. Wu, M. & Jena, P. The rise of two-dimensional van der Waals ferro-electrics. Wiley Interdiscip. Rev. Comput. Mol. Sci. 8, e1365 (2018).6. Wu, M. & Li, J. Sliding ferroelectricity in 2D van der Waals materials:Related physics and future opportunities. Proc. Natl Acad. Sci. USA118, e2115703118 (2021).7. Li, L. &Wu,M. Binary Compoundbilayer andmultilayer with verticalpolarizations: Two-dimensional ferroelectrics, multiferroics, andnanogenerators. ACS Nano 11, 6382–6388 (2017).8. Ding, W. et al. Prediction of intrinsic two-dimensional ferroelectricsin In2Se3 and other III2-VI3 van der Waals materials. Nat. Commun.8, 14956 (2017).9. Gong, C., Kim, E. M., Wang, Y., Lee, G. & Zhang, X. Multiferroicity inatomic van der Waals heterostructures. Nat. Commun. 10,2657 (2019).10. Higashitarumizu, N. et al. Purely in-plane ferroelectricity in mono-layer SnS at room temperature. Nat. Commun. 11, 2428 (2020).11. Chang, K. et al. Microscopic manipulation of ferroelectric domainsin SnSe monolayers at room temperature. Nano Lett. 20,6590–6597 (2020).12. Chang, K. et al. Discovery of robust in-plane ferroelectricity inatomic-thick SnTe. Science 353, 274–278 (2016).13. Zheng, C. et al. Room temperature in-plane ferroelectricity in vander Waals In2Se3. Sci. Adv. 4, eaar7720 (2018).14. Zhou, Y. et al. Out-of-plane piezoelectricity and ferroelectricity inlayered α-In2Se3 nanoflakes. Nano Lett. 17, 5508–5513 (2017).15. Si, M. et al. A ferroelectric semiconductor field-effect transistor.Nat. Electron. 2, 580–586 (2019).16. Si, M. et al. Asymmetric metal/α-In2Se3/Si crossbar ferroelectricsemiconductor junction. ACS Nano 15, 5689–5695 (2021).17. Cui, C. et al. Intercorrelated in-plane and out-of-plane ferroelec-tricity in ultrathin two-dimensional layered semiconductor In2Se3.Nano Lett. 18, 1253–1258 (2018).18. Liu, F. et al. Room-temperature ferroelectricity in CuInP2S6 ultrathinflakes. Nat. Commun. 7, 12357 (2016).19. Belianinov, A. et al. CuInP2S6 room temperature layered ferro-electric. Nano Lett. 15, 3808–3814 (2015).20. Si, M., Liao, P.-Y., Qiu, G., Duan, Y. & Ye, P. D. Ferroelectric field-effect transistors based on MoS2 and CuInP2S6 two-dimensionalvan der Waals heterostructure. ACS Nano 12, 6700–6705 (2018).21. Jiang, X. et al. Manipulation of current rectification in van derWaalsferroionic CuInP2S6. Nat. Commun. 13, 574 (2022).22. Yuan, S. et al. Room-temperature ferroelectricity in MoTe2 down tothe atomic monolayer limit. Nat. Commun. 10, 1775 (2019).23. Ghosh, T. et al. Ultrathin free-standing nanosheets of Bi2O2Se:Room temperature ferroelectricity in self-assembled chargedlayered heterostructure. Nano Lett. 19, 5703–5709 (2019).24. Fei, Z. et al. Ferroelectric switching of a two-dimensional metal.Nature 560, 336 (2018).25. Sharma, P. et al. A room-temperature ferroelectric semimetal. Sci.Adv. 5, eaax5080.26. de la Barrera, S. C. et al. Direct measurement of ferroelectricpolarization in a tunable semimetal. Nat. Commun. 12, 5298(2021).27. Yasuda, K., Wang, X., Watanabe, K., Taniguchi, T. & Jarillo-Herrero,P. Stacking-engineered ferroelectricity in bilayer boron nitride.Science 372, 1458–1462 (2021).28. Vizner Stern, M. et al. Interfacial ferroelectricity by van der Waalssliding. Science 372, 1462–1466 (2021).29. Woods, C. R. et al. Charge-polarized interfacial superlattices inmarginally twisted hexagonal boron nitride. Nat. Commun. 12,347 (2021).30. Wang, X. et al. Interfacial ferroelectricity in rhombohedral-stackedbilayer transition metal dichalcogenides. Nat. Nanotechnol. 17,367–371 (2022).31. Weston, A. et al. Interfacial ferroelectricity in marginally twisted 2Dsemiconductors. Nat. Nanotechnol. 17, 390–395 (2022).32. Rogée, L. et al. Ferroelectricity in untwisted heterobilayers of tran-sition metal dichalcogenides. Science 376, 973–978 (2022).33. Zheng, Z. et al. Unconventional ferroelectricity in moiré hetero-structures. Nature 588, 71–76 (2020).34. Wang, Y. et al. Tunable ferroelectricity in hBN intercalated twisteddouble-layer graphene. ArXiv210212398 Cond-Mat (2021).35. Sui, M. et al. Gate-tunable topological valley transport in bilayergraphene. Nat. Phys. 11, 1027–1031 (2015).36. Shimazaki, Y. et al. Generation and detection of pure valley currentby electrically induced Berry curvature in bilayer graphene. Nat.Phys. 11, 1032–1036 (2015).37. Andrei, E. Y. &MacDonald, A. H. Graphene bilayers with a twist.Nat.Mater. 19, 1265–1275 (2020).38. Balents, L., Dean, C. R., Efetov, D. K. & Young, A. F. Super-conductivity and strong correlations in moiré flat bands. Nat. Phys.16, 725–733 (2020).39. Cao, Y. et al. Correlated insulator behaviour at half-filling in magic-angle graphene superlattices. Nature 556, 80–84 (2018).40. Wang, D. et al. Thermally induced graphene rotation on hexagonalboron nitride. Phys. Rev. Lett. 116, 126101 (2016).41. Kim, K. et al. Tunable moiré bands and strong correlations in small-twist-angle bilayer graphene. Proc. Natl Acad. Sci. USA 114,3364–3369 (2017).42. Wang, L. et al. One-dimensional electrical contact to a two-dimensional material. Science 342, 614–617 (2013).43. Cançado, L. G., Pimenta, M. A., Neves, B. R. A., Dantas, M. S. S. &Jorio, A. Influence of the atomic structure on the Raman spectra ofgraphite edges. Phys. Rev. Lett. 93, 247401 (2004).44. Li, Y. et al. Probing symmetryproperties of few-layerMoS2 andh-BNby optical second-harmonic generation. Nano Lett. 13,3329–3333 (2013).AcknowledgementsJ.L. acknowledges support from the National Key R&D Program of China(Grant nos. 2019YFA0307800 and 2021YFA1400100), Beijing NaturalScience Foundation (grant no. Z190011), NSF of China (grant no.11974027). C.H. acknowledges support from Beijing Natural ScienceFoundation (grant no. 4222084) and CAS Pioneer Hundred TalentsProgram.Author contributionsJ.L. conceived the project. R.N. fabricated devices and performedtransport measurements with assistance from Z.L., X.H., Z.Q., and Z.Y.Crystallographic characterization was performed by R.N., Z.L., Q.L., andT.L.; K.W. and T.T. synthesized boron nitride crystals. M.W., X.W., Q.R.,and J.H. provided theoretical support. J.L., C.H., J.M., Z.H., K.L., Z.G.supervised the project. All authors contribute to the data analysis. R.N.,and J.L. wrote the paper with input from all authors.Article https://doi.org/10.1038/s41467-022-34104-zNature Communications |         (2022) 13:6241 7Competing interestsThe authors declare no competing interests.Additional informationSupplementary information The online version containssupplementary material available athttps://doi.org/10.1038/s41467-022-34104-z.Correspondence and requests for materials should be addressed toJianming Lu.Peer review information Nature Communications thanks David Ruiz-Tijerina and theother anonymous reviewer(s) for their contribution to thepeer review of this work. Peer reviewer reports are available.Reprints and permissions information is available athttp://www.nature.com/reprintsPublisher’s note Springer Nature remains neutral with regard tojurisdictional claims in published maps and institutional affiliations.Open 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/s41467-022-34104-zNature Communications |         (2022) 13:6241 8https://doi.org/10.1038/s41467-022-34104-zhttp://www.nature.com/reprintshttp://creativecommons.org/licenses/by/4.0/http://creativecommons.org/licenses/by/4.0/ Giant ferroelectric polarization in a bilayer graphene heterostructure Results Untying the ferroelectricity and correlated electrons in moiré bands Strongly enhanced charge polarization Discussion about the physical mechanism The normal gate as a switch knob of the special gate Ferroelectric switching of intrinsic quantum properties Discussion Methods Sample fabrication Characterization of crystallographic direction Electrical measurement Reporting summary Data availability References Acknowledgements Author contributions Competing interests Additional information