# Fileset

[s41467-024-54870-2.pdf](https://mdr.nims.go.jp/filesets/9103d5e1-c3ed-4806-8641-2331de85e0a6/download)

## Creator

[Fengrui Yao](https://orcid.org/0000-0003-3754-0628), [Dario Rossi](https://orcid.org/0009-0004-6803-0940), Ivo A. Gabrovski, [Volodymyr Multian](https://orcid.org/0000-0003-2553-9275), [Nelson Hua](https://orcid.org/0000-0002-4642-3821), [Kenji Watanabe](https://orcid.org/0000-0003-3701-8119), [Takashi Taniguchi](https://orcid.org/0000-0002-1467-3105), [Marco Gibertini](https://orcid.org/0000-0003-3980-5319), [Ignacio Gutiérrez-Lezama](https://orcid.org/0000-0003-1280-915X), [Louk Rademaker](https://orcid.org/0000-0001-6053-8150), [Alberto F. Morpurgo](https://orcid.org/0000-0003-0974-3620)

## Rights

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

## Other metadata

[Moiré magnetism in CrBr3 multilayers emerging from differential strain](https://mdr.nims.go.jp/datasets/a3e98129-1566-4438-9b01-dbe37d9802cb)

## Fulltext

Moiré magnetism in CrBr3 multilayers emerging from differential strainArticle https://doi.org/10.1038/s41467-024-54870-2Moiré magnetism in CrBr3 multilayersemerging from differential strainFengrui Yao 1,2 , Dario Rossi 3, Ivo A. Gabrovski1, Volodymyr Multian 1,2,4,Nelson Hua 5, Kenji Watanabe 6, Takashi Taniguchi 7, Marco Gibertini 8,9,Ignacio Gutiérrez-Lezama 1,2, Louk Rademaker 1 &Alberto F. Morpurgo 1,2Interfaces between twisted 2D materials host a wealth of physical phenomenaoriginating from the long-scale periodicity associated with the resultingmoiréstructure. Besides twisting, an alternative route to create structures withcomparably long—or even longer—periodicities is inducing a differential strainbetween adjacent layers in a van der Waals (vdW) material. Despite recenttheoretical efforts analyzing its benefits, this route has not yet been imple-mented experimentally. Here we report evidence for the simultaneous pre-sence of ferromagnetic and antiferromagnetic regions in CrBr3—a hallmark ofmoiré magnetism—from the observation of an unexpected magnetoconduc-tance in CrBr3 tunnel barriers with ferromagnetic Fe3GeTe2 and grapheneelectrodes. The observedmagnetoconductance evolves with temperature andmagnetic field as the magnetoconductance measured in small-angle CrBr3twisted junctions, in which moiré magnetism occurs. Consistent with Ramanmeasurements and theoretical modeling, we attribute the phenomenon to thepresence of a differential strain in the CrBr3 multilayer, which locally modifiesthe stacking and the interlayer exchange between adjacent CrBr3 layers,resulting in spatially modulated spin textures. Our conclusions indicate thatinducing differential strain in vdW multilayers is a viable strategy to createmoiré-like superlattices, which in the future may offer in-situ continuoustunability even at low temperatures.Twisted stacks of 2D materials result in the formation of moiré struc-tures that exhibit fascinating emergent electronic phenomena. Well-known examples include flat-band superconductivity1 andmagnetism2,3, Mott–Hubbard states4, and spatially modulated non-collinear magnetic textures5–9. The physical properties of moiré vander Waals (vdW) structures depend sensitively on the twist angle(Fig. 1a), which is normally fixed at the assembly stage and cannot befurther changed. Ensuring the uniformity of the twist angle over a largearea, and developing strategies to continuously tune the moirésuperlattice, represent major experimental challenges10,11.To gain additional control, it has been proposed to exploitmoiré-like structures resulting from differential strain in theReceived: 12 May 2024Accepted: 22 November 2024Check for updates1Department of QuantumMatter Physics, University of Geneva, Geneva, Switzerland. 2Group of Applied Physics, University of Geneva, Geneva, Switzerland.3Department of Theoretical Physics, University of Geneva, Geneva, Switzerland. 4Advanced Materials Nonlinear Optical Diagnostics lab, Institute of Physics,NAS of Ukraine, Kyiv, Ukraine. 5Laboratory for X-ray Nanoscience and Technologies, Paul Scherrer Institut, Villigen PSI, Switzerland. 6Research Center forElectronic and Optical Materials, National Institute for Materials Science, Tsukuba, Japan. 7Research Center for Materials Nanoarchitectonics, NationalInstitute for Materials Science, Tsukuba, Japan. 8Dipartimento di Scienze Fisiche, Informatiche e Matematiche, University of Modena and Reggio Emilia,Modena, Italy. 9Centro S3, CNR-Istituto Nanoscienze, Modena, Italy. e-mail: fengrui.yao@unige.ch; louk.rademaker@unige.ch;alberto.morpurgo@unige.chNature Communications |        (2024) 15:10377 11234567890():,;1234567890():,;http://orcid.org/0000-0003-3754-0628http://orcid.org/0000-0003-3754-0628http://orcid.org/0000-0003-3754-0628http://orcid.org/0000-0003-3754-0628http://orcid.org/0000-0003-3754-0628http://orcid.org/0009-0004-6803-0940http://orcid.org/0009-0004-6803-0940http://orcid.org/0009-0004-6803-0940http://orcid.org/0009-0004-6803-0940http://orcid.org/0009-0004-6803-0940http://orcid.org/0000-0003-2553-9275http://orcid.org/0000-0003-2553-9275http://orcid.org/0000-0003-2553-9275http://orcid.org/0000-0003-2553-9275http://orcid.org/0000-0003-2553-9275http://orcid.org/0000-0002-4642-3821http://orcid.org/0000-0002-4642-3821http://orcid.org/0000-0002-4642-3821http://orcid.org/0000-0002-4642-3821http://orcid.org/0000-0002-4642-3821http://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/0000-0003-3980-5319http://orcid.org/0000-0003-3980-5319http://orcid.org/0000-0003-3980-5319http://orcid.org/0000-0003-3980-5319http://orcid.org/0000-0003-3980-5319http://orcid.org/0000-0003-1280-915Xhttp://orcid.org/0000-0003-1280-915Xhttp://orcid.org/0000-0003-1280-915Xhttp://orcid.org/0000-0003-1280-915Xhttp://orcid.org/0000-0003-1280-915Xhttp://orcid.org/0000-0001-6053-8150http://orcid.org/0000-0001-6053-8150http://orcid.org/0000-0001-6053-8150http://orcid.org/0000-0001-6053-8150http://orcid.org/0000-0001-6053-8150http://orcid.org/0000-0003-0974-3620http://orcid.org/0000-0003-0974-3620http://orcid.org/0000-0003-0974-3620http://orcid.org/0000-0003-0974-3620http://orcid.org/0000-0003-0974-3620http://crossmark.crossref.org/dialog/?doi=10.1038/s41467-024-54870-2&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1038/s41467-024-54870-2&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1038/s41467-024-54870-2&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1038/s41467-024-54870-2&domain=pdfmailto:fengrui.yao@unige.chmailto:louk.rademaker@unige.chmailto:alberto.morpurgo@unige.chwww.nature.com/naturecommunicationsdirection perpendicular to the layers12,13. The idea is to create a strainpattern in multilayers of vdW materials such that neighboring layersare strained differently, with the difference in lattice vectors inadjacent layers determining the resulting moiré pattern (Fig. 1b, c).This scheme mimics what happens in hetero-bilayers of semi-conducting transition metal dichalcogenides, where the moiré ori-ginates from the naturally occurring difference in latticeconstants14,15. The key advantage offered by differential strain is thatits strength can in principle be varied continuously, resulting in atunable moiré periodicity. Despite the timeliness of the subject,however, no experiments have been reported that show how thepresence of a strain gradient in vdW materials creates moiré-likestructures hosting phenomena analogous to those observed intwisted multilayers. (%)-2 20-2 20 -2 20Interlayer strain (Uniaxial)Interlayer strain (Biaxial)Interlayer twistingMAAABAAABM MAA ABAFM CrBr3FM CrBr3i j0H⊥ (T) 0H⊥ (T)0H⊥ (T) 0H⊥ (T)(%)(%)T(K)T(K)FM CrBr3 AFM CrBr3M-stacking AB-stacking AA-stackingAFMMFMFACrBrf245245-2 2005002cbag hd e015 (%)053Fig. 1 | Strain-inducedmoiré superlattices and stacking-dependentmagnetismof CrBr3. a Moiré superlattices commonly originate from a small twist angle θbetween identical vdW layers. They can also arise from interlayer biaxial (b) anduniaxial (c) differential strain, which cause adjacent layers to have slightly differentlattice vectors. Red and blue honeycomb lattices represent atoms in the two layers.d Top view of the lattice structure of monolayer CrBr3 and its unit cell, displayingthe honeycomb lattice formed by Cr atoms (blue balls) within the edge-sharingoctahedra of Br atoms (white balls). e Depending on how layers are stacked, threedistinct (meta)stable structures of CrBr3 are known, with different interlayerexchange coupling: theM (monoclinic) and AA stackings lead to antiferromagnetic(AFM) interlayer coupling; the rhombohedral (AB) stacking leads to interlayerferromagnetic (FM) ordering (only the Cr atoms are depicted; red and blue ballsrepresent Cr atoms in the two layers). f In differentially strained CrBr3 layers, thespatial dependence of the interlayer exchange spontaneously results in the for-mation of non-collinear spin textures (i.e., moiré magnetism). g the tunnelingmagnetoconductance δG(H, 2 K) of FM barriers (data measured on a four-layer ABstacked CrBr3 junction) is small (2%) at low temperature and exhibits characteristic“lobes” near TC as shown in h. In contrast, AFM barriers (data measured on a four-layer M stacked CrBr3 junction) exhibit a large low-T magnetoconductance (i) dueto the spin-flip transitions of the inner and outer layers, which is suppressed as T isincreased, and which vanishes above TN (j see ref. 28. for the analogous datameasured in an AA stacked AFM CrBr3 barrier and for more information about thedevices used to measure the data shown in this figure).Article https://doi.org/10.1038/s41467-024-54870-2Nature Communications |        (2024) 15:10377 2www.nature.com/naturecommunicationsHere, we demonstrate that differential strain gives rise to moirémagnetism5–9,16–24 in multilayers of an originally ferromagnetic system,resulting in the coexistence of ferromagnetic and antiferromagneticregions. Our experiments rely on tunneling magnetotransport mea-surements through CrBr3 barriers sandwiched between a Fe3GeTe2metallic ferromagnetic electrode and a graphene contact. The mag-netoconductance of such devices shows that the expected spin-valeeffect –determined by the relative orientation of the magnetization inthe Fe3GeTe2 electrode and the CrBr3 barriers—coexists with anunexpected, reproducible background. This background is virtuallyidentical to the tunneling magnetoconductance that we measure onsmall-angle twisted CrBr3 barriers contacted exclusively with (non-magnetic) graphene electrodes, from which we conclude moiré mag-netism is at the origin of the effect. To elucidate what causes theemergence of a moiré pattern in multilayers that are originally ferro-magnetic, we perform Raman measurements showing how at lowtemperatures under the Fe3GeTe2 electrode, the structure of the CrBr3multilayer breaks its rhombohedral symmetry , as expected in thepresenceof differential strain.We complement our experimentswith atheoretical analysis, which predicts that differentially strained CrBr3barriers should host a backgroundmagnetoconductance with a shapeand on a magnetic field scale compatible with our experimentalobservations. These results demonstrate the possibility to inducemoiré physics in the absence of twisting between layers, exclusivelyfrom differences in lattice parameters that originate from differentialstrain.ResultsDetecting moiré magnetism with magnetotransportA single CrBr3 layer (see Fig. 1d) is ferromagnetic with out-of-planemagnetic order, and Curie temperature near 30 K25,26. In the threeknown (meta)stable structures of the material27,28, the couplingbetween adjacent CrBr3 layers is either ferromagnetic –for rhombo-hedral (AB) stacking—or antiferromagnetic—for AA or Monoclinic (M)stacking (see Fig. 1e). As our investigations of Fe3GeTe2(FGT)/CrBr3structures rely on tunneling magnetotransport measurements, weillustrate the methodology by discussing the recently reported mag-netoconductance δG of these naturally occurring ferro and anti-ferromagnetic CrBr3 barriers with graphene (Gr) electrodes.Tunneling occurs in the Fowler-Nordheim regime and the mag-netoconductance is due to the alignment of the spins in the CrBr3barrier, with increasing spin alignment that lowers the barrierheight28–30. Accordingly, in ferromagnetic CrBr3 barriers the magne-toconductance is small at low T (Fig. 1g)—because the spins alreadyalign spontaneously in the absence of an applied field μ0H—and peaksnear the Curie temperature (Fig. 1h)—where the magnetic suscept-ibility tends to diverge30. In the antiferromagnetic phases of CrBr3,instead, the low-Tmagnetoconductance is large (see Fig. 1i, j), becausethe applied field flips the magnetization of individual layers and dras-tically improves spin alignment30–35. For both ferro and anti-ferromagnetic CrBr3 barriers, the evolution of the tunnelingmagnetoconductance with H and T correlates to the magnetization ofthebarrier, and for ferromagnetic CrBr3 barriers,δGhasbeen shown tobe a function of M (i.e, δG (H, T) = δG (M (H, T)))30.If one of the graphene electrodes is substituted with a FGTmultilayer36 (Fig. 2a, b), the behavior of the low-T magnetoconduc-tance changes qualitatively. When electrons are injected from FGT(Fig. 2d), hysteresis appears and the barrier conductance is smallerwhen the magnetization directions in CrBr3 and FGT are antiparallel(Fig. 2f and Supplementary Fig. 1). The phenomenon is the expectedspin-valve effect37, as the CrBr3 barrier spin-filters the spin-polarizedelectrons injected from the ferromagnetic contact38. Unexpectedly,however, the hysteretic contribution is superimposed onto a positivemagnetoconductance background absent in devices with only gra-phene contacts. The background (δGbg, bottom panel of Fig. 2f)resembles the magnetoconductance of antiferromagnetic CrBr3 bar-riers: it occurs on comparable magnetic field scales, has smaller butcomparable magnitude, and an identical temperature dependence(compare with Fig. 1i, j, and discussion of Fig. 4), albeit without equallysharp jumps.When electrons are injected from the graphene electrode(Fig. 2e), hysteresis is nearly absent, but the magnetoconductancebackground remains unchanged (Fig. 2g). Virtually identical behaviorhas been seen in all four FGT/CrBr3/Gr junctions that we have studiedexperimentally (see also Supplementary Fig. 4, which shows that atT = 2K, the background δGbg, is on average one order of magnitudelarger than the magnetoconductance of a ferromagnetic CrBr3barrier).The absence of hysteresis when electrons are injected from gra-phene is understandable, because in the Fowler-Nordheim regime theresistance is dominated by the electron injection process. Hysteresis istherefore not expected when injecting from graphene, as grapheneinjects spin-unpolarized electrons. Following the same logic, findingthat the background magnetoconductance is the same irrespective ofthe injecting electrode indicates that the phenomenon does not ori-ginate from injection at either contact, but is a manifestation of aproperty of the CrBr3 barrier itself, which is modified by the presenceof the FGT electrode.To confirm that the background magnetoconductance onlyoccurs in the presence of FGT electrodes we fabricated a pair of tunneljunctions on the sameCrBr3multilayer, separatedby only 2–3microns.In one of the junctions, the CrBr3 barrier is sandwiched between twographene contacts, and in the other junction, one of the electrodes isan FGT crystal (see Fig. 3a). As expected, the magnetoconductancemeasured on the junction with two graphene contacts shows thetypical behavior of ferromagnetic CrBr3 barriers: very small magne-toconductance at low-T, Fig. 3b, and “lobes” near TC, Fig. 3c (comparewith Fig. 1g,h), confirming that the multilayer is indeedferromagnetic28,30. The nearby junction realized with one FGT contact(Fig. 3d), instead, shows spin-valve effect when injecting electronsfrom FGT (evidence for ferromagnetism in CrBr3), coexisting with themagnetoconductance background described above, which persistswhen injecting electrons from graphene (Fig. 3e). Again, the tem-perature evolution of the magnetoconductance background (Fig. 3f)resembles that measured in antiferromagnetic CrBr3 tunnel barriers(compare with Fig. 1i, j), with all features shifting to lower field astemperature is increased and disappearing above TC. Note that thebackground coexists with the positive magnetoconductance “lobes”above TC typical of ferromagnetism30.These observations establish that whenever FGT contacts areused themagnetoconductance systematically exhibits amagneticfieldand temperature dependent background that is indicative of the pre-sence of antiferromagnetism, even if a purely ferromagnetic pristineCrBr3 multilayer is employed to realize the tunnel barrier. This isconfirmed by a second device with the same geometry, which exhibitsvirtually identical behavior (Supplementary Fig. 2). We therefore con-clude that bringing a CrBr3 ferromagnetic multilayer into contact witha FGT electrode induces antiferromagnetism in CrBr3. Ferro and anti-ferromagnetic regions are then simultaneously present, as expected inthe presence of a moiré, and such coexistence can account for all thedifferent aspects of the measured magnetoconductance.Magnetoconductance of twisted CrBr3 tunnel barriersTo confirm that the coexistence of ferromagnetism and anti-ferromagnetism measured in FGT/CrBr3/Gr tunneling junction comesfrom moiré magnetism, we have compared the behavior of thesedevices to that of small-angle (less than 3°) twisted CrBr3 barriers,similar to twisted CrI3 bilayers in which moiré magnetism isestablished5–8. Three twisted barriers were fabricated employing acommon tear-and-stack process (see Methods for detail), to assembletwo ferromagnetic CrBr3 multilayers (~10 nm thick) on top of eachArticle https://doi.org/10.1038/s41467-024-54870-2Nature Communications |        (2024) 15:10377 3www.nature.com/naturecommunicationsotherwith a nominal twist angle of 2.5°, 2°, and 1.5°, respectively. Thesetwist angles are within the range for which moiré magnetism isexpected for Chromium trihalides5–8. The twisted CrBr3multilayers aresandwiched between graphene contacts. In one device, the non-twisted region was also sandwiched by two graphene contacts (seeFig. 4a) to confirm that the constituent CrBr3 multilayers consist ofrhombohedral ferromagnetic stacking. The results of the low-temperature magnetoconductance measurements of non-twistedand twisted regions are shown in Fig. 4b, c, respectively.In all twisted multilayer devices, a positive magnetoconductancebackground nearly saturating at (or just below) 1 T is observed atT = 2K, whose shape is very similar to the magnetoconductancebackground seen in FGT/CrBr3/Gr devices (compare Fig. 4c withFig. 2f, g bottom panels and Fig. 3e). No sharp jumps but smoothshoulders are present, in contrast with the CrBr3 antiferromagneticbarriers, in which sharp jumps are in general seen at approximately0.2 T and 0.4T or at 0.55 and 1.1 T depending on the specific anti-ferromagnetic stacking28. The magnetoconductance background ofthe twisted CrBr3 devices is nearly symmetric upon reversing theapplied bias, analogous to the behavior of FGT/CrBr3/Gr devices.Magnetoconductance data measured upon increasing the tempera-ture for one device (plotted in Supplementary Fig. 3) show the coex-istence of features originating from ferromagnetism (lobes near Tc)and antiferromagnetism (all features shift to lower fields as T increasesand disappear as T reaches TC), closely matching the evolution seen inFGT/CrBr3/Gr device whose data are shown in Fig. 3f.To better compare the magnetoconductance curves of the fourdifferent FGT/CrBr3/Gr barriers with that of the twisted CrBr3 barriers,-2 0 2-20020-1001020-10010-2 -1 0 1 2010-1001020-10010-2 -1 0 1 20100H⊥ (T)V = 2.1V V = - 2.1V0H⊥ (T)V  0CrBr3Gr-FGTV  0↑- ↓(%)(%)(%)FN tunnellingCrBr3FGT-GrH H↑- ↓(%)(%)(%)V (V)I (nA)FGT-CrBr3 -GrFGTCrBr3AuAuAuAuAuAuAuAu2 μmedgfabcFig. 2 | Coexistence of ferro- and antiferromagnetism in Fe3GeTe2(FGT)/CrBr3barriers. Schematic structure (a) optical micrographs (b) and an example ofcurrent-voltage (I-V) characteristics (c) of a FGT/CrBr3/graphene (Gr) tunnel barrierdevice (data measured at T = 2 K). d, e for positive and negative bias, electronsinjected from the FGT or the Gr electrode respectively, tunnel through the CrBr3barrier (~8.5 nm), with transport occurring in the Fowler-Nordheim (FN) regime.f Tunneling magnetoconductance δG(H, 2 K) for electrons injected from the FGTcontact (top panel; V = 2.1 V; the blue and red curves are the magnetoconductanceδG" and δG# measuredwhen sweeping themagneticfield in the direction indicatedby the arrows). The hysteresis is amanifestation of the spin-valve effect, resulting ina larger (smaller) conductance when the magnetizations of FGT and CrBr3 areparallel (antiparallel) to each other. The spin-valve magnetoconductance (δG"-δG#, middle panel) is superimposed on a sizable magnetoconductance back-ground (δGbg = ((δG" + δG#)+( δG" � δG#�� ��))/2, bottom panel) that resembles themagnetoconductancemeasured in AFMCrBr3 barriers (compare to b).g Tunnelingmagnetoconductance measured with electrons injected from the graphene elec-trode (top panel; V = −2.1 V). The spin valve effect (middle panel) is absent but thebackground magnetoconductance (bottom panel) is virtually identical to thatmeasured when injecting electrons from the FGT contact. The observation of spin-valve effect and of the magnetoconductance background in a same device providedirect evidence for the coexistence of FM of AFM regions in the CrBr3 barrier. Thesame behavior has been observed in all tunnel barriers that we realized with FGTcontacts (in all measurements, the magnetic field is applied perpendicular to thelayers).Article https://doi.org/10.1038/s41467-024-54870-2Nature Communications |        (2024) 15:10377 4www.nature.com/naturecommunicationswe normalized the data to the value of the magnetoconductancemeasured at μ0H = 1 T and plotted all curves together (see Fig. 4d). Thedifferences in magnetoconductance between twisted CrBr3 and FGT/CrBr3/Grdevices iswithin the spreadof the curvesdue todifferences intwist angle or in strain orientation (i.e., the relative orientation of thecrystalline structures of the FGT and CrBr3 multilayers). Finding thatthe magnetoconductance of FGT/CrBr3/Gr devices exhibits trendsidentical to those of devices based on twisted CrBr3, whose magne-toconductance is due to moiré magnetism, confirms that –despite theabsence of any twist—moiré magnetism is present FGT/CrBr3/Grdevices.Probing the structure of CrBr3 under the FGT contactTo understand why FGT/CrBr3/Gr devices exhibit moiré magnetism inthe absence of any twist between the CrBr3 layers, we performedRaman spectroscopy measurements to probe the structure of theCrBr3multilayer under a FGT contact. In CrBr3 –as in all other commonChromium trihalides—Raman spectroscopy can discriminate betweenthe naturally occurring AB-stacking of the constituent layers (i.e.,rhombohedral, with three-fold rotation symmetry) leading to ferro-magnetism, from the monoclinic staking of the most common anti-ferromagnetic state, which breaks three-fold rotation symmetry.Indeed, Raman measurements are expected to exhibit a dependenceon the polarization of the incident and detected light in themonoclinicstructure22,39–41, absent in the rhombohedral one. The measurementsfocused on the modes within the 135 –165 cm−1 range, known to besensitive to the stacking configuration, and were performed in thecrossed (XY configuration, Fig. 5a) and parallel (XX configuration,Fig. 5b) polarization channels of the incident and detected light (see“Methods” Section for details).At 20K, Raman spectra of CrBr3 away from the FGT contact (illu-strated in Fig. 5a, b by three representative green and red curves010-2 -1 0 1 2010-10010-2 -1 0 1 2010FGT-CrBr3 –Gr -2 -1 0 1 204 Gr-CrBr3 -GrH(%)0H⊥ (T)T(K)0H⊥ (T)T(K)0H⊥ (T)T(K)(%)V  00H⊥ (T)(%)V  00H⊥ (T)Gr(%)(%)V  0V  0V  0V  0Gr-CrBr3 -GrFGT-CrBr3 –Gr 2410-1-2264842648040(%)040 (%)2410-1-22648fecbadFig. 3 | Magnetoconductance of nearby FGT/CrBr3 /Gr and Gr/CrBr3/Gr junc-tions. a Schematic view of a device consisting of a FGT/CrBr3/Gr and a Gr/CrBr3/Grtunnel barriers realized on a same CrBr3 multilayer (~3.4 nm), at a few microndistance from each other (h-BN encapsulating layers not shown). b, Tunnelingmagnetoconductance of the Gr/CrBr3/Gr barrier and (c) color plot of its tempera-ture dependence: the small low-temperaturemagnetoconductance and the “lobes”nearTc confirm that the CrBr3multilayer is fully ferromagneticwhen not in contactwith a FGT crystal (compare with Fig. 1g, h). d Tunneling magnetoconductance ofthe FGT/CrBr3/Gr junction with electrons injected from the FGT (V > 0, top panel)and the Gr (V <0, bottom panel) electrode (data taken at T = 2 K). Spin-valvemagnetoconductance is observed (onlywhen injecting fromFGT) and indicates thepresence of ferromagnetism in the CrBr3 multilayer. The magnetoconductancebackground—present irrespective of injecting electrode (see top and bottom plotsof the (e))—indicates the simultaneous presence of antiferromagnetism. Thecomparison of the magnetoconductance measured on the two nearby junctionstherefore confirm that the coexistence of ferro and antiferromagnetism occursexclusively in the CrBr3 multilayer under the FGT crystal. f The color plot of thetemperature-dependent magnetoconductance background extracted from theFGT/CrBr3/Gr junction (electrons injected from the FGT (top panel) and from Gr(bottom panel)) further confirms the coexistence of ferromagnetism: the “lobes”near Tc illustrate the presence of ferromagnetism, and the background shrinking inmagnetic field as T approaches Tc (and disappearing for T > Tc) originates from thepresence of antiferromagnetism.Article https://doi.org/10.1038/s41467-024-54870-2Nature Communications |        (2024) 15:10377 5www.nature.com/naturecommunicationsmeasured on each side of the FGT electrode, see Supplementary Fig. 5for detailed positions) reveal two peaks at ~142 cm−1 and 152 cm−1, cor-responding to twofold degenerate Eg modes42. The peak positions andintensities are the same in the twopolarization channels, as expected forthe rhombohedral (ferromagnetic) stacking of CrBr3. In contrast, underFGT, we observe a broadening of the peaks, whose shape suggests thepresence of overlapping peaks from multiple stackings (the Ramansignal is weaker—because the CrBr3 multilayer is located under themetallic FGT crystal—which makes it difficult to fully resolve the split-ting). More importantly, the peak positions of the two Raman modesexhibit an unambiguous dependence on the polarization channelemployed for the measurements, i.e., the peak positions differ formeasurements done in the XX and XY polarization (Fig. 5c, d). Both thesplitting of the peaks and the sensitivity to the polarization channel aredistinct signatures of monoclinic stacking in CrBr3. Their observationconfirms that under the FGT electrode the structure of the CrBr3 mul-tilayer is modified from the common rhombohedral (ferromagnetic)stacking of CrBr3 multilayers, as expected in the presence of a moiré.We attribute the presence of the moiré identified by the Ramanmeasurements—and responsible for the coexistence of ferro andantiferromagnetism in CrBr3 in contact with FGT—to differential strainin the CrBr3 multilayer. To explain its presence, we propose a scenarioin which differential strain originates from the different thermalexpansions of FGT and CrBr342,43. In simple terms, the FGT/CrBr3/Grstructures are assembled at room temperature, where couplingbetween FGT and CrBr3 is established. Upon cooling, the difference inthermal expansion of the two materials imposes a strain in the upperlayers of CrBr3 that propagates in the layers further away. As a result,differential strain appears in CrBr3, resulting in the appearance ofmoirémagnetismwith coexisting of FMandAFM regions. Consistentlywith this scenario, we find that at room temperature no Raman shiftdue to a splitting nor any polarization dependence is observed (seeSupplementary Fig. 5), as the shift and the polarization dependenceevolve gradually and continuously upon cooling, becoming sizableonly well below 100K (see Supplementary Fig. 6, the data show noindication of sharp changes associated to a structural or magneticphase transition).Theoretical analysis of strain-induced moiré magnetismHaving concluded experimentally that the magnetoconductancemeasured in FGT/CrBr3/Grdevices originates fromdifferential strain inCrBr3 induced by the contact with FGT, we analyze theoretically themagnetic states that are expected to emerge in the presence of astrain-induced moiré. The moiré pattern that appears when twoneighboring layers experience differential strain is illustrated inFig. 1b, c. The moiré causes the stacking to depend on position, which0100050-2 -1 0 1 2050100aGr0H⊥ (T)c(%)= 2.5 = 2 = 1.5 0110 # 1 /3.4 nm # 2 /4.0 nm # 3 /6.5 nm # 4 /8.5 nm twist 2.5° twist 2° twist 1.5°(arb. units)0H⊥ (T)FGT/CrBr3Twist CrBr3d-2 -1 0 1 204Gr/non-twisted CrBr3 /Gr(%)0H⊥ (T)bFig. 4 | Magnetoconductance of small-angle twisted multilayer CrBr3 devices.a Schematic of the device configuration, showing a Gr/twisted CrBr3/Gr junctionwith aGr/untwistedCrBr3/Gr tunnel barrier fabricatedon the sameCrBr3multilayer(approximately 10 nm thick) at a few microns’ distance from each other (h-BNencapsulating layers not shown). Three deviceswere realized using a tear-and-stacktechnique to control the relative angle (θ) of the two CrBr3 multilayers.b Magnetoconductance of the Gr/non-twisted CrBr3/Gr junction measured atT = 2 K, showing the behavior typical of ferromagnetic CrBr3 barriers.c, Magnetoconductance background of the three Gr/twisted CrBr3/Gr junctionswith twist angle 2.5° (top panel), 2° (middle panel), and 1.5° (bottom panel). Themagnetoconductance background is 2-to-4 times larger than that of FGT/CrBr3/Grbarriers but otherwise shows nearly identical behavior. d Plot of the magneto-conductance background δGbg of all the measured FGT/CrBr3/Gr devices (con-tinuous lines) compared to the magnetoconductance of the three twisted CrBr3devices (dashed lines). All curves are normalized to 1 at μ0H⊥ = 1 T for ease ofcomparison. The comparison illustrates the similarity between the curves mea-sured in twisted CrBr3 devices and FGT/CrBr3/Gr devices.Article https://doi.org/10.1038/s41467-024-54870-2Nature Communications |        (2024) 15:10377 6www.nature.com/naturecommunicationsin CrBr3 inevitably results in the simultaneous presence of interlayerferromagnetic and antiferromagnetic domains. In practice, in ourdevices, the layers in contact with the FGT crystal are under strain dueto the coupling between the twomaterials, with the strain that relaxesin the layers further away from FGT. Due to the relatively weak vdWinterlayer bonding in the CrBr3 multilayer, we expect that the com-bined elastic and stacking energy is lowered when all the differentialstrain is localized at a single bilayer moiré interface, whose exactlocation depends on microscopic details (see Supplementary Infor-mation Sec. 2.2 for details).We model the presence of AFM and FM domains at this moiréinterface to understand whether their coexistence can account for theexperimentally observed magnetoconductance. To this end, we cal-culate the dependence of the magnetization M on the applied mag-netic field μ0H in the presence of many different differential strainconfigurations, and search for similarities between the magneto-conductance background (δGbg) and the M(H) curve (as mentionedearlier, in CrBr3 barriers there is a close correspondence betweentunneling magnetoconductance and magnetization30). For the calcu-lations, we use a continuum field theory based on ref. 17. The mag-netization of a single layer is described by a spin stiffness and a single-ion anisotropy, while the interlayer coupling is modulated throughoutthe moiré unit cell. Taking into account experimentally relevantparameters, we find that a CrBr3 moiré bilayer in the absence of amagnetic field has a magnetic texture with locally c-axis aligned fer-romagnetic or antiferromagnetic order, separated by coplanar domainwalls (Fig. 6a). Upon the application of a magnetic field these domainwalls move, and the antiferromagnetic domains shrink, leading tosmooth changes in magnetization that indeed mimic the observedsmooth change in magnetoconductance (Fig. 6).At a first critical field –whose value depends on the strain patternand effective spin stiffness– the antiferromagnetic domain around theAA region disappears through a spin-flip transition (diagram I→ II;Fig. 6a, b), causing a kink in the magnetization curve (Fig. 6c, pinkshaded region). This is because dM/dH is determined by the shift ofdomain walls, and the removal of AFM domains changes this slope. Asecond kink at higher fields (Fig. 6c, blue shaded region) is associatedwith a spin-flop (diagram II→ III; Fig. 6 a, b) at a field comparable to—but somewhat larger than—the spin-flip field seen in M-stacked bilay-ers. This is consistent with the ab initio prediction that the strongestantiferromagnetic interlayer coupling does not occur for theM-stacked structures, but for a different monoclinic stacking (M’) thatdoes not correspond to a (meta)stable M stacking of CrBr3.The features that we find are robust, in the sense that the twokinks appear regardlessof thedetails suchas theprecise formof strain,stiffness, and spin anisotropy. The exact values of the spin flip and flopfields depend on details such as the thickness of the CrBr3 multilayerand the induced differential strain, as explained in SupplementaryInformation Sec. 2 and Sec. 4. As shown in Fig. 6c, a qualitativeagreement between the experimentally measured magnetoconduc-tance and the square of themagnetization is found for a realistic set ofmodel parameters (we compare to the square of the magnetization,because for ferromagnetic barriers the relation between magneto-conductance and magnetization is approximately quadratic30).140 160XY 140 160XX 0 4 8141142143 XY XXcRaman shift (cm-1) Distance (um)aRaman shift (cm-1) Raman shift (cm-1) Intensity (arb. units)bCrBr3FGT/CrBr3CrBr3CrBr3FGT/CrBr3CrBr30 4 8152.2152.5 XY XXdDistance (um)T ~ 20 KFig. 5 | Comparison of CrBr3 Raman spectra next to and under a FGT crystal.Togain information about the influence of an FGT contact on the structure of theunderlying CrBr3 multilayer (magnetoconductance in Fig. 3), the Raman spectra ofthe CrBr3 tunnel barrier next to and under the FGT electrode were compared. Wefocused on the two Eg modes near 140 cm−1 and 150 cm−1, known to be highlysensitive to the symmetry of the CrBr3 structure. Raman spectra of CrBr3measuredat different points of the CrBr3 tunnel barrier, under crossed (XY, a) and parallel(XX, b) polarization configuration of the incident and detected light (data mea-sured around 20K). The red and blue curves are the Raman spectra measured atpositions next to the FGT electrode (on opposite sides); the green curves aremeasured at different positions under the FGT electrode (see Supplementary Fig. 3for position details). Under the FGT electrode, the peaks are broader and exhibit ashoulder (suggesting that they in fact consist of distinct peaks, i.e., that they aresplit). More importantly, the vertical dotted lines mark the peak position in CrBr3away from the FGT contact. It is apparent that under FGT the peak position isshifted. c, d Extracted peak wavelength of the two individual modes, as a functionof position on the CrBr3 multilayer where the measurements are done (on the x-axis, Distance = 0μmcorresponds to a position to the left of the FGT electrode; theFGT crystal is located at distances between approximately 3μm and 5 μm). Apolarization-dependent shift of the peaks measured under the FGT electrodes isapparent.Article https://doi.org/10.1038/s41467-024-54870-2Nature Communications |        (2024) 15:10377 7www.nature.com/naturecommunicationsDiscussionThe background magnetoconductance of FGT/CrBr3/Gr barriers—which exhibits the same evolution (with magnetic field, temperatureand bias polarity) as the magnetoconductance of small-angle twistedCrBr3 barrier– and the direct signature in Raman spectroscopy of thepresence of monoclinic stacking in CrBr3 under FGT provide con-clusive evidence for the presence of moiré magnetism in CrBr3 underFGT. The Raman data—which show a different peak position andpolarization dependence in theCrBr3multilayer under andnext to FGTonly at low temperatures—indicate that CrBr3 under FGT experiencesstrain. Taken together, these two experimental findings indicate thatstrain causes a moiré in CrBr3, as expected in the presence of differ-ential strain.At this stage, not much can be quantitatively said about the spe-cific properties of the strain-induced moiré at interfaces, and why thedifferential strain is larger for some interfaces as compared to others(for instancewhy it is larger at the FGT/CrBr3 interface as compared tothe Gr/CrBr3 interface). Since strain is likely small (1% is normallyconsidered to be a sizable strain) and the differential strain can only bea smaller fraction of the strain of the individual layers, we expect themoiré wavelength to be much longer than in twisted structures. Thedata, however, does not give direct indications as to the periodicity ofthe strain-induced moiré. Similarly—even if Raman maps are ratherhomogeneous over the CrBr3 under the FGT crystal—inhomogeneity inthe moiré can be present (since the spatial resolution of Raman isdiffraction-limited). Even though technically challenging—because thestrained part of the structure is buried—finding ways to characterizethe structural properties of themoiré present in FGT/CrBr3/Gr devicesis highly desirable. As we mentioned earlier, however, we can never-theless establish from the temperature dependence of the Raman shift(see Supplementary Fig. 6) that strain emerges progressively as thetemperature is lowered, with the Raman shift in CrBr3 that increasesgradually upon cooling, and becomes sizable only well below 100K.That is why a scenario in which strain in CrBr3 originates from the0.0 0.5 1.00101     1-10MzM’ AAABM’ AA ABM’ AA ABM’ AA ABcbaIIIIIII II III0H⊥ (T)δ M2(arb. units)δGbg(arb. units)# 1# 2# 3# 4Biaxial # 1Biaxial # 2Uniaxial # 1Uniaxial # 2TheoryFGT/CrBr3Fig. 6 | Theoretical magnetic textures of CrBr3 multilayers. We analyze themeasured magnetoconductance background in terms of the theoretically pre-dicted evolution of magnetic textures in a CrBr3 moiré interface under an appliedout-of-plane magnetic field H. The magnetoconductance is expected to followclosely the magnetic field dependence of the square of the magnetization30,δG ~ (δM)2. a Visualization of magnetic textures at selected applied magnetic fields,labeled by I, II, and III. The textures are represented by the z-component of the localmagnetization (Mz); yellow dashed line represents the moiré unit cell; black circlesrepresent the boundary between ferromagnetic and antiferromagnetic interlayerHeisenberg exchange. b Visualization of the spin orientation in the two layers atthree points in the moiré unit cell AA, AB, and M’ (another monoclinic stacking atthe midpoint between two neighboring AA regions). c, Plot of themagnetoconductance background δGbg and of δM2, as a function of H (quantitiesare normalized to 1 at 1 T, to enable their comparison). Both δM and δGbg increasesmoothly at first, up to the critical field for the spin-flip transition at the AA-stackedregion at μ0H⊥ ~ 0.2 T (I→ II; pink shaded region). A second smooth increase thenoccurs with the antiferromagnetic domains near the M’-stacked region that arefurther reduced in size, up to a second criticalfieldμ0H⊥ ~ 0.5–0.7 T associatedwitha spin flop-transition (II→ III; blues shaded region). The two theoretical curves forbiaxial strain have 1% strain, spin stiffness is 1.4meV, and anisotropy is 0.01meVand 0.02meV, respectively. The two curves with uniaxial strain have 1% and 3%strain, respectively, the spin stiffness is 10meV, and anisotropy is 0.01meV. Theoverall evolution is the same irrespective of these details and reproduces qualita-tively the evolution of the background magnetoconductance.Article https://doi.org/10.1038/s41467-024-54870-2Nature Communications |        (2024) 15:10377 8www.nature.com/naturecommunicationsdifference in thermal expansion coefficients of FGT and CrBr3 appearsrealistic.Conceptually, thefindings reportedhereare relevant as they showexperimentally thatmoiré physics can indeed be accessed by inducingdifferential strain in multilayers of vdWmaterials, without the need torealize small-angle twisted structures. Creatingmoiré structures in thismanner can be advantageous, because strain can be controlled in avariety of ways. Examples are the use of piezo actuators44–46—whichwould eventually allow strain and moiré to be tuned in-situ—of sus-pended structures of vdW materials47, or possibly even of large-areasmultilayers grown on suitably chosen substrates, in which the latticemismatch determines the induced strain48. Inducing, controlling, andprobing strain in vdW materials is a very active field of research49,50,which will help the identification of the best-suited experimentalroutes to create controlled differential strain inmultilayers of interest.MethodsDevice fabrication and measurementThe h-BN/Gr/Fe3GeTe2(FGT)/CrBr3/Gr/h-BN and h-BN/Gr/twistedCrBr3/Gr/h-BN heterostructures were assembled by means of a drypick-up and transfer technique, employing PDMS-PC stampswithin thecontrolled inert environment of a N2-filled glove box (H20 <0.1 ppmand O2 <0.1 ppm). The FGT and CrBr3 multilayers used in the experi-ments were obtained viamicromechanical exfoliation (done inside theglove box) of bulk crystals purchased from HQ graphene. In theassembly process, a PDMS-PC stamp was used to pick up the top h-BNat 90 °C, followed by the top graphene, FGT, CrBr3 and bottom gra-phene, each at 70 °C, and the bottom h-BN at 90 °C. The PC with thewhole stack was finally released onto a SiO2/Si substrate at 160 °C.After transfer, the substrate was immersed in chloroform to dissolvethe PC, leaving the heterostructure on the substrate. As the FGTcrystals used as electrodes are typically 10 nm thick (or somewhatthicker), air canflowbetween the hBN encapsulating layer and the FGTcrystal edge if the edge is exposed to ambient. If so, air can reach theCrBr3 multilayer causing its degradation. To eliminate these problems,we avoided etching the hBN encapsulating layer to contact directly theFGTelectrode. Instead,weused separate graphite stripes connected tothe FGT crystal (as detailed in Supplementary Fig. 2) and formedelectrical contact to these stripes by edge contacts located far awayfrom the FGT crystal (edge contacts were realized using electron beamlithography, reactive-ion etching, electron-beam evaporation of 10 nmCr followed by 50 nm Au, and lift-off). For the twistedmultilayer CrBr3samples, we employed the so-called ‘tear-and-stack’ technique. Aportion of the CrBr3 multilayer was picked up and stacked onto alarger, remaining layer on the substrate at a targeted twist angle (θ).Designing the remaining layer larger than the picked-up portionallowed attaching graphene contacts to both the twisted and untwis-ted regions of the structure, which were encapsulated with h-BN onboth sides.Systematic transportmeasurementswere conducted in anOxfordInstruments cryostat equipped with a superconducting magnet and aheliox insert. Homemade low-noise voltage bias and current mea-surement modules coupled with digital multi-meters were employedfor the data acquisition.Raman MeasurementsRaman spectroscopy was conducted using a Horiba system (LabramHR evolution) equipped with a helium flow cryostat (Konti Micro fromCryoVac GMBH). A linearly polarized laser (532 nm, spot size ~1μm)was focused on the sample within the cryostat through a 50XOlympusobjective. The scattered light was captured by the same objective,passed through an analyzer, and directed to a Czerni–Turner spec-trometer equipped with a 1800 groovesmm−1 grating. Detection wascarried out using a liquid nitrogen-cooled CCD array. By varying thehalf-wave plate while keeping the analyzer on the detecting light pathfixed, measurements under either parallel (XX) or crossed (XY)polarization were performed. All measured Raman spectra were fittedwith a set of Voigt functions51 (Gaussian–Lorentzian convolution) toaccurately resolve the peakpositions. Similarly to previous studies39–41,the Raman tensors of the non-degenerate Ag and Bg modes (in stack-ings with broken three-fold rotation symmetry) and doubly degen-erate Eg1 and Eg2 modes (in the AB and AA stacking, with three-foldrotation symmetry) of CrBr3 multilayer can be written as:Ag =a 0 d0 c 0d 0 b0B@1CA,Bg =0 e 0e 0 f0 f 00B@1CA, Eg1 =m n pn �m qp q 00B@1CA,Eg2 =n �m �q�m �n p�q p 00B@1CA,Accordingly, for AB and AA stacked CrBr3 multilayers, the Ramanintensity for the Eg1 and Eg2 modes as a function of θ can be derived as:I(Eg1)∝ |m sin(θ)−n cos(θ)|2 and I(Eg2)∝ |m cos(2θ) +n sin(2θ) | 2, where θ isthe polarized direction of excitation light with respect to the analyzer.Thus, the dependence on the polarization angle cancels out when thetwo modes (Eg1 and Eg2) are degenerate, resulting in one single Eg peak(the total intensity of the degeneratemodes is the same under either XXconfiguration or XY configuration; observed in Fig. 4, CrBr3 away fromthe FGT contact). However, for stackingswith broken three-fold rotationsymmetry, thedegenerateEgmodes split into thenon-degenerateAg andBgmodes. Thus, thepositionofBgmode is distinct fromtheEgmodeandits Raman intensity as a function of θ can be expressed as:I(Bg)∝ e2cos2(θ), different intensities under the XX configuration and XYconfiguration are observed (observed in Fig. 4, CrBr3 under FGT flake).Theoretical calculationsWe compute the theoretical magnetization curves using the con-tinuous spin model17, with the inclusion of an out-of-plane magneticfield. The local magnetization is modeled as planar spins in the x–zplane with intra-layer spin stiffness, out-of-plane single-ion anisotropy,and inter-layer Heisenberg spin exchange. For the interlayer coupling,we consider data from first-principle calculations from ref. 27. andrescale it to match the experimentally measured spin-flip criticalmagnetic fields for the AA and M antiferromagnetic configurations inCrBr328. We consider different types of moiré lattices, derived fromstrain and relative rotation of the layers and extract the magnetizationcurves from the minimization solutions as a function of the magneticfield. Further details are provided in the Supplementary Information.Data availabilityThe data generated in this study have been deposited in the Yaretarepository of the University of Geneva. Source data file is provided athttps://doi.org/10.26037/yareta:ftcobqbmk5bh5glrdpukq3xmuu.Code availabilityThe code adopted for calculations in this study have been deposited inthe Yareta repository of the University of Geneva. Codes are providedat https://doi.org/10.26037/yareta:ftcobqbmk5bh5glrdpukq3xmuu.References1. Cao, Y. et al. Unconventional superconductivity in magic-anglegraphene superlattices. Nature 556, 43–50 (2018).2. Sharpe, A. L. et al. Emergent ferromagnetism near three-quartersfilling in twisted bilayer graphene. Science 365, 605–608 (2019).3. Ciorciaro, L. et al. Kinetic magnetism in triangular moiré materials.Nature 623, 509–513 (2023).4. Wang, L. et al. Correlated electronic phases in twisted bilayertransition metal dichalcogenides. Nat. Mater. 19, 861–866 (2020).Article https://doi.org/10.1038/s41467-024-54870-2Nature Communications |        (2024) 15:10377 9https://doi.org/10.26037/yareta:ftcobqbmk5bh5glrdpukq3xmuuhttps://doi.org/10.26037/yareta:ftcobqbmk5bh5glrdpukq3xmuuwww.nature.com/naturecommunications5. Song, T. et al. Direct visualization of magnetic domains and moirémagnetism in twisted 2D magnets. Science 374, 1140–1144 (2021).6. Xu, Y. et al. Coexisting ferromagnetic-antiferromagnetic state intwisted bilayer CrI3. Nat. Nanotechnol. 17, 143–147 (2022).7. Xie, H. et al. Evidence of non-collinear spin texture in magneticmoiré superlattices. Nat. Phys. 19, 1150–1155 (2023).8. Cheng, G. et al. Electrically tunable moiré magnetism in twisteddouble bilayers of chromium triiodide. Nat. Electron 6, 434–442(2023).9. Boix-Constant, C. et al. Multistep magnetization switching inorthogonally twisted ferromagnetic monolayers. Nat. Mater. 23,212–218 (2024).10. Lau,C.N., Bockrath,M.W.,Mak,K. F.&Zhang, F. Reproducibility in thefabrication and physics of moirématerials.Nature 602, 41–50 (2022).11. Kapfer, M. et al. Programming twist angle and strain profiles in 2Dmaterials. Science 381, 677–681 (2023).12. Li, L. &Wu, M. Binary compound bilayer andmultilayer with verticalpolarizations: two-dimensional ferroelectrics, multiferroics, andnanogenerators. ACS nano 11, 6382–6388 (2017).13. Escudero, F. et al. Designing moire patterns by strain. Phys. Rev.Research 6, 023203 (2024).14. Mak, K. F. & Shan, J. Semiconductor moiré materials. Nat. Nano-technol. 17, 686–695 (2022).15. Regan, E. C. et al. Emerging exciton physics in transition metaldichalcogenide heterobilayers. Nat. Rev. Mater. 7, 778–795 (2022).16. Tong,Q., Liu, F., Xiao, J. & Yao,W. Skyrmions in themoiré of van derWaals 2D magnets. Nano Lett. 18, 7194–7199 (2018).17. Hejazi, K., Luo, Z.-X. & Balents, L. Noncollinear phases in moirémagnets. PNAS 117, 10721–10726 (2020).18. Wang, C. et al. Stacking domain wall magnons in twisted van derwaals magnets. Phys. Rev. Lett. 125, 247201 (2020).19. Xiao, F., Chen, K. & Tong, Q. Magnetization textures in twistedbilayer CrX3 (X= Br, I). Phys. Rev. Res. 3, 013027 (2021).20. Akram, M. & Erten, O. Skyrmions in twisted van der Waals magnets.Phys. Rev. B 103, L140406 (2021).21. Akram, M. et al. Moiré skyrmions and chiral magnetic phases in twis-ted CrX3 (X= I, Br, and Cl) bilayers. Nano Lett. 21, 6633–6639 (2021).22. Xie, H. et al. Twist engineering of the two-dimensionalmagnetism indouble bilayer chromium triiodide homostructures. Nat. Phys. 18,30–36 (2022).23. Yang, B. et al. Moiré magnetic exchange interactions in twistedmagnets. Nat. Comput. Sci. 3, 314–320 (2023).24. Fumega, A. O. & Lado, J. L. Moiré-driven multiferroic order in twis-ted CrCl3, CrBr3 and CrI3 bilayers. 2D Mater. 10, 025026 (2023).25. Chen, W. et al. Direct observation of van der Waals stacking-dependent interlayer magnetism. Science 366, 983–987 (2019).26. Kim, M. et al. Micromagnetometry of two-dimensional ferro-magnets. Nat. Electron 2, 457–463 (2019).27. Gibertini, M. Magnetism and stability of all primitive stacking pat-terns in bilayer chromium trihalides. J. Phys. D 54, 064002 (2020).28. Yao, F. et al. Multiple antiferromagnetic phases and magneticanisotropy in exfoliated CrBr3 multilayers. Nat. Commun. 14,4969 (2023).29. Kim, H. H. et al. Evolution of interlayer and intralayer magnetism inthree atomically thin chromium trihalides. PNAS 116, 11131–11136(2019).30. Wang, Z. et al. Magnetization dependent tunneling conductance offerromagnetic barriers. Nat. Commun. 12, 6659 (2021).31. Klein, D. R. et al. Probingmagnetism in 2D van derWaals crystallineinsulators via electron tunneling. Science 360, 1218–1222 (2018).32. Kim, H. H. et al. One million percent tunnel magnetoresistance in amagnetic van derWaals heterostructure.Nano Lett. 18, 4885–4890(2018).33. Song, T. et al. Giant tunneling magnetoresistance in spin-filter vander Waals heterostructures. Science 360, 1214–1218 (2018).34. Wang, Z. et al. Very large tunneling magnetoresistance in layeredmagnetic semiconductor CrI3. Nat. Commun. 9, 2516 (2018).35. Mak, K. F., Shan, J. & Ralph, D. C. Probing and controlling magneticstates in 2D layered magnetic materials. Nat. Rev. Phys. 1, 646–661(2019).36. Tan, C. et al. Hard magnetic properties in nanoflake van der WaalsFe3GeTe2. Nat. Commun. 9, 1554 (2018).37. Coey, J. M. Magnetism and magnetic materials (Cambridge uni-versity press, 2010).38. Wang, Z. et al. Tunneling spin valves based on Fe3GeTe2/hBN/Fe3GeTe2 van der Waals heterostructures. Nano Lett. 18,4303–4308 (2018).39. Ubrig, N. et al. Low-temperature monoclinic layer stacking inatomically thin CrI3 crystals. 2D Mater. 7, 015007 (2019).40. Li, T. et al. Pressure-controlled interlayer magnetism in atomicallythin CrI3. Nat. Mater. 18, 1303–1308 (2019).41. Song, T. et al. Switching 2D magnetic states via pressure tuning oflayer stacking. Nat. Mater. 18, 1298–1302 (2019).42. Kozlenko, D. et al. Spin-induced negative thermal expansion andspin–phonon coupling in van derWaalsmaterial CrBr3.Npj ComputMater. 6, 1–5 (2021).43. Claro, M. S. et al. Temperature and thickness dependence of thethermal conductivity in 2D Ferromagnet Fe3GeTe2. ACS Appl.Mater. Interfaces 15, 49538–49544 (2023).44. Edelberg, D. et al. Tunable strain soliton networks confine electronsin van der Waals materials. Nat. Phys. 16, 1097–1102 (2020).45. Wang, Y. et al. Strain‐sensitive magnetization reversal of a van derWaals magnet. Adv. Mater. 32, 2004533 (2020).46. Cenker, J. et al. Reversible strain-inducedmagnetic phase transitionin a van der Waals magnet. Nat. Nanotechnol. 17, 256–261 (2022).47. Šiškins, M. et al. Nanomechanical probing and strain tuning of theCurie temperature in suspended Cr2Ge2Te6-based hetero-structures. npj 2D Mater. Appl. 6, 41 (2022).48. Merino, P. et al. Strain-driven moiré superstructures of epitaxial gra-phene on transition metal surfaces. Acs Nano 5, 5627–5634 (2011).49. Shen, T., Penumatcha, A. V. & Appenzeller, J. Strain engineering fortransitionmetal dichalcogenides based field effect transistors.ACSnano 10, 4712–4718 (2016).50. Dai, Z., Liu, L. & Zhang, Z. Strain engineering of 2Dmaterials: issuesand opportunities at the interface. Adv. Mater. 31, 1805417 (2019).51. Zakaraya, M. G., Maisuradze, G. G. & Ulstrup, J. Theory of inhomo-geneous environmental Gaussian broadening of resonance Ramanexcitation profiles for polyatomic molecules in solution. J. RamanSpectrosc. 20, 359–365 (1989).AcknowledgementsThe authors gratefully acknowledge Alexandre Ferreira for technicalsupport anduseful discussionswithFranciscoGuinea, ZheWang, Zhu-XiLuo, Jérémie Teyssier and Menghan Liao. A.F.M. gratefully acknowl-edges the Swiss National Science Foundation (Division II, project#200020_178891) and the EU Graphene Flagship project for support.M.G. acknowledges support from the Italian Ministry for University andResearch through the PNRR project ECS_00000033_ECOSISTER andthe PRIN2022 project E53D23001700006. L.R. and I.A.G. acknowledgethe Swiss National Science Foundation (Starting grant TMSGI2_211296).K.W. and T.T. acknowledge support from the JSPS KAKENHI (GrantNumbers 21H05233 and 23H02052) the CREST (JPMJCR24A5), andWorld Premier International Research Center Initiative (WPI),MEXT, Japan.Author contributionsF.Y. and A.F.M. conceived the project. F.Y. fabricated the devices andperformed the transport measurements, assisted by I.G.L; D.R., I.A.G.,and L.R. performed the theoretical calculations; V.M. and F.Y. performedoptical measurements; T.T. and K.W. grew and provided the h-BNArticle https://doi.org/10.1038/s41467-024-54870-2Nature Communications |        (2024) 15:10377 10www.nature.com/naturecommunicationscrystals. A.F.M. and I.G.L. supervised the research. F.Y., D.R., I.A.G., V.M.,N.H., M.G., I.G.L., L.R., and A.F.M. analyzed the data and wrote themanuscriptwith input fromall authors. All authors discussed the results.Competing interestsThe authors declare no competing interests.Additional informationSupplementary information The online version containssupplementary material available athttps://doi.org/10.1038/s41467-024-54870-2.Correspondence and requests for materials should be addressed toFengrui Yao, Louk Rademaker or Alberto F. Morpurgo.Peer review information Nature Communications thanks Weibo Gao,Dahlia Klein and the other, anonymous, reviewer(s) for their contributionto the peer review of this work. A peer review file is available.Reprints and permissions 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.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 licence, and indicate ifchanges were made. The images or other third party material in thisarticle are included in the article’s Creative Commons licence, unlessindicated otherwise in a credit line to the material. If material is notincluded in the article’s Creative Commons licence 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 licence, visit http://creativecommons.org/licenses/by/4.0/.© The Author(s) 2024Article https://doi.org/10.1038/s41467-024-54870-2Nature Communications |        (2024) 15:10377 11https://doi.org/10.1038/s41467-024-54870-2http://www.nature.com/reprintshttp://creativecommons.org/licenses/by/4.0/http://creativecommons.org/licenses/by/4.0/www.nature.com/naturecommunications Moiré magnetism in CrBr3 multilayers emerging from differential strain Results Detecting moiré magnetism with magnetotransport Magnetoconductance of twisted CrBr3 tunnel barriers Probing the structure of CrBr3 under the FGT contact Theoretical analysis of strain-induced moiré magnetism Discussion Methods Device fabrication and measurement Raman Measurements Theoretical calculations Data availability Code availability References Acknowledgements Author contributions Competing interests Additional information