# Fileset

[s41467-025-56804-y.pdf](https://mdr.nims.go.jp/filesets/f0602eb2-7379-496f-ba8e-ca599cfaa25b/download)

## Creator

Daniel J. Rizzo, Eric Seewald, Fangzhou Zhao, Jordan Cox, Kaichen Xie, Rocco A. Vitalone, Francesco L. Ruta, Daniel G. Chica, Yinming Shao, Sara Shabani, Evan J. Telford, Matthew C. Strasbourg, Thomas P. Darlington, Suheng Xu, Siyuan Qiu, Aravind Devarakonda, [Takashi Taniguchi](https://orcid.org/0000-0002-1467-3105), [Kenji Watanabe](https://orcid.org/0000-0003-3701-8119), Xiaoyang Zhu, P. James Schuck, Cory R. Dean, Xavier Roy, Andrew J. Millis, Ting Cao, Angel Rubio, Abhay N. Pasupathy, D. N. Basov

## Rights

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

## Other metadata

[Engineering anisotropic electrodynamics at the graphene/CrSBr interface](https://mdr.nims.go.jp/datasets/8b6410a7-7267-41a9-897b-07535ff71e70)

## Fulltext

Engineering anisotropic electrodynamics at the graphene/CrSBr interfaceArticle https://doi.org/10.1038/s41467-025-56804-yEngineering anisotropic electrodynamics atthe graphene/CrSBr interfaceDaniel J. Rizzo 1 , Eric Seewald1, Fangzhou Zhao 2, Jordan Cox 3,Kaichen Xie 4, Rocco A. Vitalone1, Francesco L. Ruta 1,5, Daniel G. Chica3,Yinming Shao 1,6, Sara Shabani1, Evan J. Telford 1,3, MatthewC. Strasbourg 7,Thomas P. Darlington 1,7, Suheng Xu 1, SiyuanQiu1, Aravind Devarakonda 1,5,Takashi Taniguchi 8, Kenji Watanabe 9, Xiaoyang Zhu 3, P. James Schuck 7,Cory R. Dean 1, Xavier Roy 3 , Andrew J. Millis1, Ting Cao 4,Angel Rubio 2,10,11, Abhay N. Pasupathy 1 & D. N. Basov 1Graphene is a privileged 2D platform for hosting confined light-matter exci-tations known as surface plasmon polaritons (SPPs), as it possesses lowintrinsic losses and a high degree of optical confinement. However, the iso-tropic nature of graphene limits its ability to guide and focus SPPs, making itless suitable than anisotropic elliptical and hyperbolicmaterials for polaritoniclensing and canalization. Here, we present graphene/CrSBr as an engineered2D interface that hosts highly anisotropic SPP propagation acrossmid-infraredand terahertz energies. Using scanning tunneling microscopy, scattering-typescanning near-field optical microscopy, and first-principles calculations, wedemonstratemutual doping in excess of 1013cm–2 holes/electrons between theinterfacial layers of graphene/CrSBr. SPPs in graphene activated by chargetransfer interact with charge-induced electronic anisotropy in the interfacialdoped CrSBr, leading to preferential SPP propagation along the quasi-1Dchains that compose each CrSBr layer. Thismultifaceted proximity effect bothcreates SPPs and endows them with anisotropic propagation lengths thatdiffer by an order-of-magnitude between the in-plane crystallographic axesof CrSBr.Two-dimensional (2D) van der Waals (vdW)materials are ideal atomic-scale media for generating confined light spanning terahertz (THz)1–4,mid-5–22 /near-infrared23 (MIR/NIR) and visible24–26 energies. Thesematerials possess phononic9–16, electronic1–3,5–8,21–23, or excitonic24–26properties that cause the permittivity to become negative, providingthe necessary conditions for hosting confined light-matter excitationsknown as polaritons. In general, 2D materials can support enhancedoptical confinement, high electronic and dielectric tunability, and lowlosses, leading to the realization of polaritons displaying ballisticpropagation5, in-plane13–16,24 and out-of-plane hyperbolicity10,12,Received: 6 January 2025Accepted: 30 January 2025Check for updates1Department of Physics, Columbia University, New York, NY, USA. 2Theory Department, Max Planck Institute for Structure and Dynamics of Matter and Centerfor Free-Electron Laser Science, Hamburg, Germany. 3Department of Chemistry, Columbia University, New York, NY, USA. 4Department of Materials Scienceand Engineering, University ofWashington, Seattle, WA, USA. 5Department of Applied Physics and AppliedMathematics, Columbia University, New York, NY,USA. 6Department of Physics, Pennsylvania State University, University Park, PA, USA. 7Department of Mechanical Engineering, Columbia University, NewYork, NY, USA. 8Research Center for Materials Nanoarchitectonics, National Institute for Materials Science, 1-1 Namiki, Tsukuba, Japan. 9Research Center forElectronic and Optical Materials, National Institute for Materials Science, 1-1 Namiki, Tsukuba, Japan. 10Center for Computational Quantum Physics, FlatironInstitute, New York, New York, USA. 11Nano-Bio Spectroscopy Group, Universidad del País Vasco UPV/EHU, San Sebastián, Spain.e-mail: djr2181@columbia.edu; xr2114@columbia.edu; apn2108@columbia.edu; db3056@columbia.eduNature Communications |         (2025) 16:1853 11234567890():,;1234567890():,;http://orcid.org/0000-0003-4587-4863http://orcid.org/0000-0003-4587-4863http://orcid.org/0000-0003-4587-4863http://orcid.org/0000-0003-4587-4863http://orcid.org/0000-0003-4587-4863http://orcid.org/0000-0001-7355-7406http://orcid.org/0000-0001-7355-7406http://orcid.org/0000-0001-7355-7406http://orcid.org/0000-0001-7355-7406http://orcid.org/0000-0001-7355-7406http://orcid.org/0000-0003-4266-0990http://orcid.org/0000-0003-4266-0990http://orcid.org/0000-0003-4266-0990http://orcid.org/0000-0003-4266-0990http://orcid.org/0000-0003-4266-0990http://orcid.org/0000-0001-8479-8301http://orcid.org/0000-0001-8479-8301http://orcid.org/0000-0001-8479-8301http://orcid.org/0000-0001-8479-8301http://orcid.org/0000-0001-8479-8301http://orcid.org/0000-0002-8746-9420http://orcid.org/0000-0002-8746-9420http://orcid.org/0000-0002-8746-9420http://orcid.org/0000-0002-8746-9420http://orcid.org/0000-0002-8746-9420http://orcid.org/0000-0002-2891-0028http://orcid.org/0000-0002-2891-0028http://orcid.org/0000-0002-2891-0028http://orcid.org/0000-0002-2891-0028http://orcid.org/0000-0002-2891-0028http://orcid.org/0000-0002-9494-9166http://orcid.org/0000-0002-9494-9166http://orcid.org/0000-0002-9494-9166http://orcid.org/0000-0002-9494-9166http://orcid.org/0000-0002-9494-9166http://orcid.org/0009-0007-8062-1075http://orcid.org/0009-0007-8062-1075http://orcid.org/0009-0007-8062-1075http://orcid.org/0009-0007-8062-1075http://orcid.org/0009-0007-8062-1075http://orcid.org/0000-0002-3498-3936http://orcid.org/0000-0002-3498-3936http://orcid.org/0000-0002-3498-3936http://orcid.org/0000-0002-3498-3936http://orcid.org/0000-0002-3498-3936http://orcid.org/0000-0002-1456-5489http://orcid.org/0000-0002-1456-5489http://orcid.org/0000-0002-1456-5489http://orcid.org/0000-0002-1456-5489http://orcid.org/0000-0002-1456-5489http://orcid.org/0000-0002-6095-7854http://orcid.org/0000-0002-6095-7854http://orcid.org/0000-0002-6095-7854http://orcid.org/0000-0002-6095-7854http://orcid.org/0000-0002-6095-7854http://orcid.org/0000-0002-1467-3105http://orcid.org/0000-0002-1467-3105http://orcid.org/0000-0002-1467-3105http://orcid.org/0000-0002-1467-3105http://orcid.org/0000-0002-1467-3105http://orcid.org/0000-0003-3701-8119http://orcid.org/0000-0003-3701-8119http://orcid.org/0000-0003-3701-8119http://orcid.org/0000-0003-3701-8119http://orcid.org/0000-0003-3701-8119http://orcid.org/0000-0002-2090-8484http://orcid.org/0000-0002-2090-8484http://orcid.org/0000-0002-2090-8484http://orcid.org/0000-0002-2090-8484http://orcid.org/0000-0002-2090-8484http://orcid.org/0000-0001-9244-2671http://orcid.org/0000-0001-9244-2671http://orcid.org/0000-0001-9244-2671http://orcid.org/0000-0001-9244-2671http://orcid.org/0000-0001-9244-2671http://orcid.org/0000-0003-2967-5960http://orcid.org/0000-0003-2967-5960http://orcid.org/0000-0003-2967-5960http://orcid.org/0000-0003-2967-5960http://orcid.org/0000-0003-2967-5960http://orcid.org/0000-0002-8850-0725http://orcid.org/0000-0002-8850-0725http://orcid.org/0000-0002-8850-0725http://orcid.org/0000-0002-8850-0725http://orcid.org/0000-0002-8850-0725http://orcid.org/0000-0003-1300-6084http://orcid.org/0000-0003-1300-6084http://orcid.org/0000-0003-1300-6084http://orcid.org/0000-0003-1300-6084http://orcid.org/0000-0003-1300-6084http://orcid.org/0000-0003-2060-3151http://orcid.org/0000-0003-2060-3151http://orcid.org/0000-0003-2060-3151http://orcid.org/0000-0003-2060-3151http://orcid.org/0000-0003-2060-3151http://orcid.org/0000-0002-2744-0634http://orcid.org/0000-0002-2744-0634http://orcid.org/0000-0002-2744-0634http://orcid.org/0000-0002-2744-0634http://orcid.org/0000-0002-2744-0634http://orcid.org/0000-0001-9785-5387http://orcid.org/0000-0001-9785-5387http://orcid.org/0000-0001-9785-5387http://orcid.org/0000-0001-9785-5387http://orcid.org/0000-0001-9785-5387http://crossmark.crossref.org/dialog/?doi=10.1038/s41467-025-56804-y&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1038/s41467-025-56804-y&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1038/s41467-025-56804-y&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1038/s41467-025-56804-y&domain=pdfmailto:djr2181@columbia.edumailto:xr2114@columbia.edumailto:apn2108@columbia.edumailto:db3056@columbia.eduwww.nature.com/naturecommunicationselectronically-tunable topology9,17, and canalization9,13 Among thesematerials, only graphene can host high-quality photonic propagationin the monolayer limit due to the uniquely high mobility of Diracquasiparticles. Indeed, quality factors of 25 or greater have beenobserved for graphene surface plasmon polaritons (SPPs)5,6 at roomtemperature and 150or greater at low temperature5. Conversely, othervdW media typically require in excess of ~102 layers to achieve similarpropagation lengths10. On the other hand, the isotropic nature ofgraphene limits its potential utility for lateral confinement and chan-neling of photonicmodes –properties that have been observed in bulkslabs of intrinsically anisotropic materials such as V2O527, α-MoO313–17and CrSBr24. The ability to impose uniaxial photonic properties onpristine sheets of graphene – free of nano-structuring and post-processing – would enable in-plane manipulation and confinement oflight propagating in an atomically-thin material.Heterostructuring has recently been demonstrated as a viableroute toward tuning the behavior of polaritons in vdWmaterials11,17,20–22,28. Here, optically-active modes in adjoining layerscouple to 2D polaritons and create hybrid modes17–19,29. In addition,emergent phenomena arising from interfacial charge transfer cansignificantly influence photonic behavior in 2D heterostructures.Specifically, charge-transfer heterostructures (CTHs) enable non-volatile generation of SPPs in graphene17,20–22, and spatially-tunablelosses of phonon-polaritons (PhPs) in hexagonal boron nitride (hBN)11.CTHs also allow nanometer-scale control of local conductivity, andthus support plasmonic cavities20,30, edge plasmon polaritons22, andplasmonicpoint-scatters21,22. Therefore, interfacing graphenewith a 2Dmaterial possessing a different work function and optical anisotropywould leverage the combined effects of charge-transfer and opticalcoupling to control the directionality of graphene SPPs.In this study, we demonstrate the ability to create uniaxial SPPs ingraphene through heterostructuring with the air-stable vdW magnetCrSBr (Fig. 1A). CrSBr possesses significant in-plane structural aniso-tropy (Fig. 1B) that is reflected in its electronic and optical properties24.Each CrSBr layer is composed of 1D chains that result in a quasi-1Delectronic structure31–34, where the CrSBr a-axis hosts flat conductionbands (CBs) that suppress efficient electrical conductivity while the b-axis CBs are more dispersive and are thus electrically conductive.Using a combination of scattering-type scanning near-field opticalmicroscopy (s-SNOM) and scanning tunneling microscopy and spec-troscopy (STM/STS), we visualize emergent electronic and nano-optical behavior arising from interfacial charge transfer and electronicanisotropy at the graphene/CrSBr interface. Our STM/STS resultsprovide strong evidence of >0.5 eV shift in the Dirac-point energy(EDirac) of graphene, indicating significant charge transfer with CrSBr(n > 1013cm–2) and confirming a theoretical prediction35. STM data alsoreveal topographic and density-of-states (DOS) features associatedwith a second-order moiré pattern, exhibiting a long-range, atomicallyclean interface.In additional, we used s-SNOM to visualize the charge transfer-enabled SPPs ingraphene/CrSBr. Thedispersive behavior of these SPPsquantifies the magnitude of interfacial charge transfer consistent withthe STS data. Our s-SNOM data further reveal a roughly order-of-magnitude difference in the SPP propagation lengths at MIR fre-quencies between the two in-plane crystallographic axes of CrSBr,alongwith a systematic suppression of the b-axis SPP group velocity atTHz frequencies. We find that this highly anisotropic SPP behaviorarises due the exceptional enhancement of optical anisotropy withinthe electron-doped interfacial layer of CrSBr. Here, the doped CrSBrpossesses anisotropic electron-hole excitations that impart direction-dependent damping on SPPs with respect to the CrSBr in-plane crys-tallographic axes (Fig. 1A, B). The totality of our analysis indicates thatemergent intra- and interband transitions play a significant role in theobserved SPP anisotropy. This interpretation is supported by first-principles density functional theory (DFT) calculations, and provides anovel route for in-plane manipulation of confined light in atomically-thin media.ResultsAtomically-resolved topography and electronic structureFigure 2A shows a characteristic STM topographic image of a gra-phene/CrSBr heterostructure. Short- and long-range modulations inthe atomic-scale landscape are observed that resemble striped moirépatterns previously reported on heterostructures of graphene withanisotropic 2Dmaterials36–38. A series of Bragg peaks visible in the fastFourier transform (FFT) of STM topography (Fig. 2A; right panel)indicates the origin of these features. These include the graphene(orange circles) and CrSBr (cyan circles) atomic lattices, and a second-ordermoiré pattern (yellow circles) consistent with a 4° rotation of theCrSBr a-axis with respect to the graphene armchair axis. Our ability tosimultaneously image the graphene and CrSBr atomic lattices alongwith the second-order moiré pattern demonstrates that graphene/CrSBr forms high-quality, atomically clean interfaces with minimalstructural and twist disorder.An STS spectrum averaged over the field of view in Fig. 2A isshown in Fig. 2B. The point spectrum shows local density of states(LDOS) features derived from both graphene and CrSBr. A widespectral tail is observe spanning negative sample biases along withbroad shoulders centered at 1.0V and 2.5 V. We note that the spectradisplay intensitymodulations likely due tomoiré-induced variations inthe local vacuum potential (Fig. S1). In addition, we can resolve ani-sotropic patterns in the vicinity of point defects, revealing electronicanisotropy that aligns with the crystallographic axes of CrSBr (Fig. S1).To isolate for LDOS features representative of graphene, we focus onsample biases spanning –0.2 to 0.8 V (Fig. 2B, inset). We observe auniaxial plasmon polaritonCrSBr (quasi-1D semiconductor)graphene h+e–s-SNOMMIRablocalized electronsmobile electronsshort-range SPPslong-range SPPsCrSBrgrapheneABabFig. 1 | Schematic of charge transfer and uniaxial plasmon polariton propa-gation in graphene/CrSBr heterostructures. A Overview of s-SNOM measure-ments in graphene/CrSBr heterostructures. The difference in work functionsbetween these layers leads to hole-doped graphene and electron-doped CrSBr.Uniaxial surface plasmon polaritons (SPPs) are generated upon illumination of theAFM tip with mid-infrared (MIR) light. B CrSBr displays a highly anisotropic crystalstructure, forming 1D chains along the in-plane a-axis. The resulting electronicstructure is quasi-1D, with b-axis carriers possessing amuch highermobility than a-axis carriers. Proximate interactions between graphene SPPs and electron-holeexcitations in the underlying CrSBr leads to preferential SPP propagation along the1D a-axis chains.Article https://doi.org/10.1038/s41467-025-56804-yNature Communications |         (2025) 16:1853 2www.nature.com/naturecommunicationsseries of LDOSminima in the range0V to 0.54V. Assigning EDirac to theglobal minimum in the STS spectrum suggests significant interfacialcharge transfer corresponding to 0.54 eV hole-doping of graphene(the Dirac point is EDirac = 0.54 eV above the Fermi level), which gives atransferred charge density of n = 1πEDirac_vF� �2= ~2 × 1013cm–2 accordingto the Dirac cone model (here the graphene Fermi velocity isvF = 108cm/s). Indeed, measurement of the CrSBr work function usingKPFM yields a value of WCrsBr≈ 5 eV (Fig. S2), which is larger than thatof graphene (Wgraphene = 4.6 eV)39 and is consistentwith interfacial holedoping of graphene and electron doping of CrSBr in graphene/CrSBrheterostructures.Characterization of the SPP dispersionInterfacial charge transfer in graphene/CrSBr can be unambiguouslyquantified from s-SNOM measurements probing the frequency-dependent SPP behavior. Figure 2C shows a map of the near-fieldamplitude S4 at ω = 905 cm–1 collected on graphene draped over theedge of trilayer CrSBr (CrSBr edge indicated by black dashed line).While SPP fringes are typically observed near native graphene edges,the large difference in the graphene conductivity coinciding with theCrSBr edge generates a boundary capable of reflecting tip-launchedSPPs and in-coupling free-space light in analogy to a true grapheneedge. Thus, the characteristic fringe pattern in Fig. 2C is evidence ofstrong graphene-CrSBr interfacial charge transfer. We note that inFig. 2C, the observed SPPs are propagating along the CrSBr a-axis.To further quantify the behavior of graphene/CrSBr SPPs, wecollect a series of images for frequencies spanning 850 to 1020 cm–1.Taking the SPP line profile at each frequency (Fig. S3), we can extractthe complex-valued wavevector (q= q1 + iq2) using established fittingprocedures accounting for both tip-launched and edge-launchedSPPs6,22. In particular, the experimental SPP dispersion ωðq1Þ encodesthe graphene charge carrier density and is plotted in Fig. 2D (red cir-cles). We compare the experimental a-axis SPP dispersion to the cal-culated imaginary component of the p-polarized reflection coefficient,Im rp, whose maxima trace the expected SPP dispersion. To calculateIm rp, we input reported optical parameters40,41, our a-polarized far-field measurements of CrSBr (Fig. S4), and EDirac = 0.5 eV for graphene(as informed by our STS measurements) (Fig. 2D). The experimentaldispersion aligns well withmaxima in Im rp despite there being no freeparameters in our model. Thus, our near-field data validate theassignment of EDirac = 0.54 eV shown in Fig. 2B, confirming significantcharge transfer in graphene/CrSBr heterostructures.We speculate thatthe additional LDOS minima observed in the 0V to 0.54V range inFig. 2B arise due to one or a combination of (1) Van Hove singularitiesin the electron doped CrSBr layer, (2) inelastic tunneling42 and/or (3)superlattice Dirac points emerging frommoiré-induced Brillouin zonefolding43,44.Uniaxial SPPsWhile s-SNOMmeasurements of a-axis propagating SPPs yield severalobservable fringes, we find that b-axis SPPs have a more subtle–2 –1 0 1 2EDirac = 540 meV0.80.60.40.20–0.2dI/dV (10–2 arb. units)Sample Bias (V)graphene/CrSBrBdI/dV (arb. units)Sample Bias (V)0123453456784 5 632DFrequency (102  cm–1)q1 (105 cm–1)8.59.09.510.00 75A STM Topography (pm) Fast-Fourier Transform CS 4 Amplitude (rel.)0.71.4 graphene/CrSBr (ω = 905 cm–1)graphene/CrSBr200 nmExperimentEF = 0.5 eV  Im r p (arb. units)901 Å–15 nmFig. 2 |Multi-modal characterizationof interfacial charge transfer in graphene/CrSBr heterostructures. A Left panel: Atomically-resolved topographic STMimage of a graphene/CrSBr heterostructure (VS = 1.2 V, I = 50 pA, T = 5.7 K). Rightpanel: Fast Fourier transform (FFT) of topographic data shows Bragg peaks asso-ciated with the graphene atomic lattice (orange circles), the CrSBr atomic lattice(cyan circles) and the second-ordermoiré pattern (yellowcircles) corresponding toa twist angle of ~4° between the graphene armchair axis and CrSBr a-axis. B STScollectedon agraphene/CrSBrheterostructure (VS = 1.8 V, I = 50pA). Inset: LowbiasSTS (VS = 0.3 V, I = 100pA) shows a dI/dVminimumat 540mV corresponding to theDirac-point energy (EDirac) of graphene shifted due to interfacial charge transferwith the underlyingCrSBr. Additional nearbydI/dVminimaare also observedat0 V,150mV, and 330mV.CTypical s-SNOM image of a graphene/CrSBr heterostructureshowing oscillations in the near-field S4 amplitude that are characteristic of SPPs(ω = 905 cm–1; S4 normalized relative to the value in the graphene/CrSBr bulk). Thegraphene is draped over the CrSBr edge (dashed black line), creating a sharp gra-dient in the graphene charge density that acts as a hard boundary for plasmonicreflections. D The experimental SPP dispersion for graphene/CrSBr (red circles)extracted from the lineprofiles of SPP fringes collected at different frequencies (seeFig. S3). Calculated Im rp for the experimental stack using an input value ofEF = 0.5 eV for the graphene Fermi energy.Maxima in the Im rp correspondwell withthe experimental dispersion, indicating that the SPP behavior is consistent with an0.5 eV shift in EDirac of graphene due to charge transfer with the underlying CrSBr.Article https://doi.org/10.1038/s41467-025-56804-yNature Communications |         (2025) 16:1853 3www.nature.com/naturecommunicationsappearance. Figure 3A shows the corner of an exfoliated CrSBrmicrocrystal that is encapsulated with graphene. At this frequency(880 cm–1), at least five observable fringes emanate along the a-axisfrom the top edge, while SPP fringes are nearly unobservable along theb-axis. The inset of Fig. 3A shows the average SPP line profiles alongboth CrSBr crystallographic axes. One faint b-axis fringe is observable(blue curve), while the a-axis curve shows multiple significant oscilla-tions (red curve). This behavior is observed at all experimental MIRfrequencies (Fig. S3), indicating SPP damping with significant in-planeanisotropy. To quantify this anisotropy, we extract the SPP Q-factor(Q= q1q2) from the experimental SPP fringe profile (Fig. 3B).We find thatSPPs have significant additional losses along the b-axis compared tothe a-axis (i.e.,Qa >Qb) with up to an order-of-magnitude difference inthe associated Q-factors across the range ω = 850 – 1020 cm–1. NotethatQa is comparable though smaller than nominal values obtained forSPPs measured under similar conditions in high quality hBN-encapsulated structures (QhBN ≈ 20)6, while Qb is likewise significantlysuppressed. We remark that the conductivity due to Dirac electrons ofgraphene is isotropic. It is therefore evident that the CrSBr in ourstructures is acting not only as a reservoir for interfacial charge dopingbut also imposes anisotropic damping on graphene SPPs in the MIRthat depends on the underlying orientation of the CrSBr crystal-lographic axes. Notably, the observed difference in Qa and Qb is far inexcess of what can be attributed to the intrinsic optical anisotropy ofundoped CrSBr (Fig. S4), which shows almost no anisotropy (dashedcurve in Fig. S3D), even though pristine CrSBr has strong anisotropy inits lattice.SPP anisotropy at THz energiesNano-optical measurements at THz frequencies also reveal significantanisotropic SPP behavior. Figure 3C (left panel) shows a characteristicTHz space-time map of the near-field SPP electric field (ΔENF )45 run-ning parallel to the a-axis, with the CrSBr edges corresponding to theleft and right edges of the panel. Extrema in the space-time map tracethe SPP worldlines (red arrows in Fig. 3C, left panel) whose slopeprovide an explicit measure of the SPP group velocity, vg =2ΔxΔt , asQ151050ab300 nmABFrequency (102 cm–1)8.5 9.0 9.5 10.0S 4 (rel.)1.60.660–40Position (μm)S 4 (rel.)0.80.91.00 0.2 0.4 0.6 0.8 1.0b-axis SPPa-axis SPPQa (exp.)Qb (exp.) vga = 22 ± 3 μm/psvgb = 18 ± 3 μm/psNormalized Counts (10–1)01234Time Delay (102  fs)2.01.51.00.50DTHz SPP Group Velocity (μm/ps) 10 15 20 25Position (μm) 0 02 4 6 8 10 12 14 2 4 6 8reduction of vgC Terahertz TDS (ΔENF)graphene/CrSBr (ω = 880 cm–1)a-axis b-axisFig. 3 | Proximity-induced anisotropy and uniaxial SPPs in graphene/CrSBrheterostructures. A Map of the near-field S4 amplitude in a graphene/CrSBr het-erostructure showing multiple SPP fringes propagating along the CrSBr a-axis,while b-axis fringes are highly suppressed. Inset: The average line profiles of SPPfringes along the a-axis (red curve) versus the b-axis (blue curve) show a sig-nificantly diminished decay length for the latter (ω = 880 cm–1; S4 normalizedrelative to the value in the graphene/CrSBr bulk). B The experimentally-extractedfrequency-dependent Q-factor for SPPs propagating along the a-axis (red circles)and b-axis (blue circles). Error bars are extracted from the standard error of q1 andq2 when fitting the line profiles in Fig. S3. C Left panel: Space-timemap of the near-field plasmon electric field (ΔENF ) conducted along the a-axis of a graphene/CrSBrheterostructure (See Fig. S7 for background subtraction procedure). The edges ofthe images correspond to the edges of CrSBr. The red arrows indicate extrema thatcorrespond to the worldlines of propagating SPPs. The group velocity of a-axismodes vag is twice the slope of the worldline. Right panel: The same as the left panelbut for b-axis propagating modes with blue arrows indicating the SPP worldlineswith groupvelocity vbg .DHistogramsof the associated valuesof vag (redbars) and vbg(blue bars) extracted from N ≥ 18 worldlines for each direction showing a sys-tematic suppression of vbg compared to vag .Article https://doi.org/10.1038/s41467-025-56804-yNature Communications |         (2025) 16:1853 4www.nature.com/naturecommunicationsrecently discovered45. Extracting the a-axis group velocity vag from ourspace-time maps yields an average value of 22 ± 3 μm/ps. Space-timemaps collected along the b-axis also showSPPworldlines (Fig. 3C, rightpanel) whose associated group velocity vbg is 18 ± 3 μm/ps. Histogramsof the measured a- and b-axis group velocities can be found in Fig. 3D,revealing a systematic reduction of vbg compared to vag . Thus, ourspace-time maps demonstrate that proximity-induced suppression ofb-axis SPP propagation extends from the MIR down to THz energies.Mechanism for Uniaxial SPPsIn order to understand the microscopic origins of SPP anisotropy, weperform first-principles DFT calculations on model heterostructuresconsisting of graphene onmonolayer CrSBr (Fig. 4B) and graphene onbilayer CrSBr (Fig. S5A–C). The monolayer-on-monolayer band struc-ture is plotted in Fig. 4B (see Fig. S5 formonolayer-on-bilayer and spin-polarized band structures). Characteristic features of the isolatedgraphene and CrSBr layers can be clearly identified, such as the gra-phene Dirac cone and the quasi-1D structure of the CrSBr CB. Notably,the theoretical value of EDirac is ~0.50 eV (n = ~2 × 1013cm–2 within theDirac cone model) indicating significant hole-doping of the graphenelayer—in good agreement with both tunneling and nano-opticalexperiments (Fig. 2B, D), as well as previous theoretical predictions35.We note that most of the charge (>80% according to our calculations)transferred from graphene is found to be localized at the top-mostCrSBr layer due to the formation of a potential gradient within CrSBr.In addition, the calculated EDirac = 0.5 eV in the graphene/bilayer CrSBrcalculation (Fig. S5D) is mirrored in the graphene/monolayer CrSBranalysis (Fig. 4B), further indicating that the charge transfer is confinedto the interfacial CrSBr layer. Indeed, other charge-transfer hetero-structures are predicted to possess charge localization primarily in theinterfacial layers due to similar effects46.Our DFT calculations also show that the conduction band ofCrSBr is now significantly electron doped and can thus support newintra- and interband transitions. Notably, free-carriers associatedwiththe conduction band of doped CrSBr are highly anisotropic(ma =mΓ→Χ = 3.1me and mb =mΓ→Υ = 0.2me) indicating that dopedCrSBr ismuchmore conductive along the b-axis than along the a-axis.Hence, the associated anisotropic Drude response of electron-dopedCrSBr is likely to impart direction-dependence to both plasmonicgroup velocity and damping. In the case of interband transitions, theconduction band of doped CrSBr resides in close proximity toanother unoccupied band (Fig. 4A) whose energy separation is ani-sotropic and overlaps with the SPP energies probed in our experi-ment. Thus, emergent interband transitions accessed throughelectron doping CrSBr will likely also induce anisotropy in the SPPresponse of graphene/CrSBr.Γ Χ S Υ Γ ΓE - EF (eV)–0.500.51.51.0AΧ’ S’ Υ ΓE - EF (eV)–1.0–0.501.00.5B0 00.1 0.2 0.3 0.4JDOS (10–2 states eV–1Å–2)0123456JDOS (10–2 states eV–1Å–2)0123456C DE (eV)0 0.2 0.4 0.6 0.8 1.002468100.1 0.30.2 0.40 0.2 0.4 0.6 0.8 1.00246810E (eV)CrSBr Monolayer (Electron-Doped)Doped Fermi Energy by charge transferGraphene/CrSBr Heterostructureqa = 2.0 × 105 cm–1qa = 4.0 × 105 cm–1qa = 6.0 × 105 cm–1qa = 8.0 × 105 cm–1qb = 2.2 × 105 cm–1qb = 4.4 × 105 cm–1qb = 6.6 × 105 cm–1qb = 8.8 × 105 cm–1qa = 2.0 × 105 cm–1qa = 4.0 × 105 cm–1qa = 6.0 × 105 cm–1qa = 8.0 × 105 cm–1qb = 2.2 × 105 cm–1qb = 4.4 × 105 cm–1qb = 6.6 × 105 cm–1qb = 8.8 × 105 cm–1S’X’ΓYSXΓYFig. 4 | First-principles calculation of the JDOS for electron-doped CrSBrmonolayer and graphene/monolayer CrSBr heterostructure. A Band structureof free-standingmonolayer CrSBr calculated byDFTwith PBE functional. The solidgreen line indicates the Fermi level of the CrSBr due to charge transfer withgraphene (The charge transfer is ~0.03 electrons per CrSBr unit cell). The grayrectangle represents the first Brillouin zone of monolayer CrSBr with the highsymmetry points indicated. B Band structure of graphene/monolayer CrSBr het-erostructure calculated by DFT with PBE functional. The first Brillouin zones ofgraphene (gray hexagon), CrSBr (gray rectangle), and their heterostructure(black rectangle) are shown with the high symmetry points of the supercellindicated. The points X’ and S’ are the zone-folded analogs of the X and S points in(A). C The calculated joint density of state (JDOS) for an electron-doped CrSBrmonolayer with the Fermi level marked by the solid green line in panel (A). TheJDOS corresponds to the plasmon damping transition with momentum q andtransition energy E from the occupied and unoccupied manifolds, defined byJDOS E,qð Þ= Rδ E � Ec k+qð Þ � Ev kð Þ� �� �d3k. Red curves denote the JDOS corre-sponding to the plasmon damping transition in the a direction (jqj=qa), and bluecurves the b direction (jqj= qb). Curveswith different hues denote different valuesof qa and qb as indicated in the plot legend. D The calculated JDOS for the gra-phene/monolayer CrSBr heterostructure. We note that transitions between bandfolded states do not contribute to the plasmon damping process. Therefore, theplotted quantity excludes transitions between zone-folded states. The insets inboth (C) and (D) show the JDOS over a larger energy range. The anisotropy in JDOSis significant below ~0.4 eV, where the transitions are mostly from the flat bandsnear the Fermi level in the a direction, and from the very dispersive bands near theFermi level in the b direction, respectively. At higher energy the JDOS is quiteisotropic since the transitions involve the more isotropic band complexes above1.0 eV and below −0.5 eV in panel (A).Article https://doi.org/10.1038/s41467-025-56804-yNature Communications |         (2025) 16:1853 5www.nature.com/naturecommunicationsTo explore the respective roles of intra- and interband transitionson the creation of uniaxial SPPs, we next investigate the origin ofdirection-dependent surface plasmon damping from first-principlescalculations. We evaluate the overall plasmon scattering rate γ usingFermi’s golden rule:γ E,qð Þ= 2π_ZδðE � Ec k+qð Þ � Ev kð Þ� �ψc,k+qjHintjψv,kD E������2Þd3k ð1ÞWhere Ec and Ev are the energies of unoccupied and occupied states,respectively, k are the crystal momenta, q the SPP momenta, and Hintrepresents the electron-photon interaction Hamiltonian (e.g., anelectric dipole interaction Hint =emcA � p, with A representing thevector potential of the incident field, and p= i_∇ representing themomentum operator). The form of γ includes the joint density ofstates (JDOS =RδðE � ½Ecðk+qÞ � EvðkÞ�Þd3kÞ where the integrand ismodulated by the matrix elements of Hint. We have checked that thelatter matrix element varies by less than 10% for different kx . This isbecause the wavefunction character along the Γ to X direction doesnot change significantly. Along the Γ to Y direction, only states close tothe minima of the parabolic band are relevant to transitions thatinfluence plasmon damping, and thus the matrix element staysapproximately constant. Therefore, we use the JDOS rather than thefull expression in Eq. 1 to quantify the anisotropic plasmon scatteringrate by sampling electronic transitions whose crystal momenta matchthe momenta of the SPPs in the MIR (� 2� 8× 105 cm–1). Here weconsider twomodel structures: (1) free-standingmonolayer CrSBr thathas been electron-doped to match the charge transfer with graphene(Fig. 4A), and (2) a graphene/monolayer CrSBr heterostructure(Fig. 4B). By comparing the results of freestanding CrSBr and itsheterostructure with graphene, we aim to parse the separate effects ofcharge doping and the presence of the graphene layer on thecalculated scattering rate.Along the a-axis, the calculated JDOS for the free-standing dopedCrSBr shows two peaks at ~0.0 eV and ~0.14 eV (red curves, Fig. 4C)that mirror JDOS peaks in graphene/monolayer CrSBr at ~0.03 eV and~0.20 eV (red curves, Fig. 4D). For both structures, the lower energypeaks mainly come from intraband transitions in the two closelypacked flat bands near the Fermi level along Γ � X , while the higherenergy peaks mainly come from interband transitions between thesebands. We note that for graphene/monolayer CrSBr, there are alsopacked bands due to zone folding (Fig. 4B). Transitions between stateswhere momentum conservation is only made possible due to zonefolding will be improbable since the wavefunction overlap betweeninitial and final states is small, and thus the matrix elementψc,k +qjHintjψv,kD Ewill be negligible. Therefore, transitions betweenzone-folded states are excluded from our calculation. The plasmondamping contribution from the graphene layer ismodest compared tothat of the CrSBr layer due to the large dispersion of the graphenelayer, which is confirmed by the similarity between the two resultsshown in Fig. 4C, D. Notably, both model structures show a JDOSminimum (i.e., suppressed plasmon damping) along the a-axis around0.1 eV, corresponding to the energy range covered in our MIRexperiments.In contrast, the JDOS along the b-axis (blue curves, Fig. 4C, D)reveals a series of peaks that are roughly 10 timeshigher than thea-axisJDOS around 0.1 eV. Here, the b-axis peaks are composed of bothintraband and interband transitions in roughly equal proportion. Sincethe quality factor scales with the ratio between the plasma frequencyand the scattering rate (Q � ωpγ )22 and the a- and b-axis plasma fre-quencies are similar (Fig. S3), our calculated damping ratio explainsthe measured order-of-magnitude reduction of the Q factor along theb-axis compared to the a-axis for SPPs in the MIR. The notion of anenhanced b-axis scattering rate is also consistent with our nano-THzdata, since the group velocity of low-momentum THz plasmons is infact impacted by the scattering rate as described in ref. 45. Thereduction of the b-axis THz group velocity by 30% in Fig. 3C, D istherefore in accord with the scenario of an increased b-axis scatteringrate. Thus, the totality of our analyses of SPPs in the MIR and THzregimes reveal that anisotropicSPPpropagation in graphene/CrSBr is acooperative effect of strong electronic anisotropy in CrSBr and prox-imate charge transfer.DiscussionWe have performed amulti-modal STM and s-SNOMexperimental andtheoretical study of graphene/CrSBr heterostructures, revealing sig-nificant proximity-induced reciprocal charge transfer andelectronically-mediated plasmonic anisotropy. By leveraging sensitiv-ity to both the local electronic and nano-optical behavior of graphene/CrSBr, we unravel the subtle interplay of intrinsic optical properties,electronic anisotropy, and emergent plasmonic damping. Combinedwith theoretical insights provided by first-principles calculations, ourobservations yield a complete mechanistic picture for engineeringuniaxial SPPs in graphene, and enable 2Dmanipulation of polaritons inan atomically-thin material.Our results have significant implications for the manipulation ofgraphene SPPs and 2D polaritons in general, and outline a novelapproach for inducing anisotropy in intrinsically isotropic media.Indeed, previously proposed routes for achieving uniaxial or aniso-tropic SPPs rely on complex device fabrication17 or intrinsic anisotropyin the SPP host-medium47— significantly limiting the potential materialplatforms for engineering 1D light. Using our proximity-basedapproach, it is now possible to use uniaxial graphene SPPs to directplasmonically-mediated energy-transfer in 2D on a non-volatile plat-formwith nanoscale precision—enabling 2D lensing, waveguiding, andcanalization in themonolayer limit. Our heterostructure could thus beintegrated into optical circuits to provide space-efficient directional-coupling of photonic elements (e.g., waveguides, resonators, andemitters), acting as both a directional conduit and momentum filter.The behavior observed in this study is generic to graphene-basedcharge transfer heterostructures composed of anisotropic semi-conducting building blocks (e.g., black phosphorous, ReSe2), intro-ducing a new class of quasi-1D plasmonic heterostructures. We alsoanticipate that this behavior can be further manipulated throughmodulation of the graphene/CrSBr chemical potential and twist-engineeringmultiple graphene-CrSBr interfaces. In addition, it is likelythat other interlayer effects can be exploited in graphene/CrSBr totune directional SPP transport, including anisotropic shake-offbands48, moiré-induced Brillouin zone folding43,44, and quasi-1Dcharge density wave formation. Finally, we foresee opportunities forelectronic and plasmonic manipulation of 2D magnetically-orderedphases in CrSBr using our charge-transfer platform and vice versa.MethodsDevice fabricationSingle crystals of CrSBr were synthesized using a chemical vaportransport reaction with source and sink zone temperatures of 930 °Cand 850 °C, respectively. Additional details of the CrSBr synthesis canbe found in ref. 49.For s-SNOMdevices, flakes of few-layer CrSBr, graphene, and few-layer hBN (<8 nm) were obtained via mechanical exfoliations. We notethat CrSBr naturally cleaves along the principal a- and b-axes, with ahigh a- to b-axis aspect ratio in resulting microcrystals. As such, therelative orientation of graphene plasmon propagation with respect tothe underlying CrSBr can be readily identified from the macroscopicfeatures of the stack; this assignment was confirmed with polarizedphotoluminescence experiments (Fig. S6). Flake thicknesses wereidentified optically and confirmed with atomic force microscopy(AFM). The hBN/graphene/CrSBr heterostructure was prepared withArticle https://doi.org/10.1038/s41467-025-56804-yNature Communications |         (2025) 16:1853 6www.nature.com/naturecommunicationsthe dry stamp method50. A transparent polydimethylsiloxane (PDMS)cube (~ 1 × 1 × 1 mm3) was covered by a thin polycarbonate (PC) poly-mer film and used to pick up the top hBN layer. After picking up thegraphene layers with the hBN, the heterostructure was transferredonto the trilayer CrSBr at elevated temperatures ( ~ 120–180 °C). Thesample surface was then washed with chloroform, acetone, and iso-propyl alcohol to remove the melted PC polymer. To further removeresidual transfer polymer and interfacial bubbles, the surfaces of hBN/graphene/CrSBr heterostructures were imaged with approximately 1nN of contact force using contact-mode AFM with line spacingof < 100nm.For STM experiments, exfoliated flakes were instead picked up inthe following sequence: hBN, CrSBr, then graphene. Polymer-supported stacks were then flipped and deposited onto a Si/SiO2substratewith no furtherprocessing. Electrical contactwasestablishedto the graphene layer by microsoldering using Field’s metals51.Scanning tunneling microscopy and spectroscopySTM/STS measurements were performed in a home-built, ultra-highvacuum system at 5.7 K. Atomically sharp tips were electrochemicallyetched from tungsten wire and spectroscopically calibrated usingShockley surface states on single crystal Au(111)52. Multiple indepen-dently prepared tips were used to verify the accuracy and reproduci-bility of the measurements.Scanning Near-field Optical MicroscopyThe MIR s-SNOM and KPFM measurements were performed on acommercial Neaspec system under ambient conditions using com-mercial ArrowTM AFM probes with nominal resonant frequencies off = 75 kHz or 256 kHz. Tunable continuous wave quantum cascadelasers produced by Daylight Solutions were used spanning wave-lengths from 9 to 11.7 μm. The detected signal was demodulated at thefourth harmonic of the tip tapping frequency in order tominimize far-field contributions to the scattered light. The fourth harmonic of thenear-field scattering amplitude (S4) and phase (Φ4) were collectedsimultaneously using a pseudoheterodyne interferometry technique.The THz s-SNOM measurements were performed on a commer-cial Neaspec system under ambient conditions using AFM tips pro-duced by Rocky Mountain Nanotechnology, LLC with a nominalresonant frequency of 30– 80kHz. The THz broadband pulse is gen-erated and detected using a pair of photo-conductive antennas (PCAs,Menlo Systems GmbH). The THz radiation from the PCA emitter iscollimated by a TPX lens and focused onto the tip and sample by aparabolic mirror. The scattered field is detected by an unbiased PCA inthe time domain using a 50 fs, 780 nm gate beam, and the photo-current signal is demodulatedby a lock-in amplifier at harmonics of thetip tapping frequency.Ab-initio calculations of graphene/CrSBr heterostructuresFirst principles calculations were performed utilizing DFT imple-mented in the Quantum ESPRESSO package53. Norm-conservingpseudopotentials were employed alongside a plane-wave energy cut-off of 90 Ry54. For structural relaxation, the spin-polarized Perdew-Burke-Ernzerhof exchange-correlation functional was employed withvan derWaals corrections (PBE-D3)55. The structures were fully relaxeduntil the forceon eachatomwas<0.005 eV/Å. Inmonolayer CrSBr, thelattice constants along the a and b axes were determined to be 3.54Åand 4.72 Å, respectively. In monolayer graphene, the lattice constant awas relaxed to 2.465 Å. The graphene/monolayer CrSBr hetero-structure was constructed with a supercell of 5 × 2 graphene rectan-gular conventional cell containing 4 carbon atoms stacked atop a 6 × 1CrSBr supercell, aligning the a and b axes of CrSBr along the armchairand zigzag directions of the graphene monolayer, respectively. Thegraphene monolayer experiences compressive strain of ~0.5% alongthe armchair direction, and compressive strain of ~4% along the zigzagdirection. The vdW spatial gap between the CrSBr layer (from top Bratoms) and the graphenemonolayer is 3.46 Å. A vacuum region of 15 Åwas added in the out-of-plane direction to avoid interaction betweenperiodic images. Brillouin zone sampling in the graphene/monolayerCrSBr heterostructure was performed using an 8 × 30 × 1 k-grid. Dipolecorrection was applied in all calculations of the graphene/monolayerCrSBr heterostructure56. A Gaussian smearing of 1meV was adoptedfor electron occupation. The first-principles calculation of JDOS for theestimation of plasmondamping is performedon a 24 × 96 × 1 k-grid forthe graphene/monolayer CrSBr heterostructure with supercell, and ona 144 × 96 × 1 k-grid for the free-standing monolayer CrSBr calculationwith doping, so the smallest sampled wavevector for plasmon damp-ing is 2 × 105 cm–1 alongbotha and b axes. TheBrillouin zone unfoldingin the JDOS calculation of the graphene/monolayer CrSBr supercellutilizes the BandUp code57,58. The RPA dielectric function calculationsare performed using the BerkeleyGW code59. The inverse of thedielectric function is calculated on a 12 × 48 × 1 k-grid, with a cutoffenergy of 8.0Ry, and a slab truncationof theCoulomb interaction. Thesmallest sampled plasmon wavevector in the dielectric function cal-culation is 4 × 105 cm–1 along both a and b axes, which is limited by thelarge computational cost of large supercells. The frequency depen-dence of the inverse dielectric function is calculated by the Adler-Wiser60,61 formalism implemented in the BerkeleyGW code, with anenergy broadening of 5meV. The inverse dielectric function shows theplasmon peaks with linewidths on the order of 20meV, with a largerlinewidth for the plasmon mode along the b-axis.Data availabilityAll raw data presented in the manuscript are available through theFigshare public repository linked to this manuscript.References1. Bandurin, D. A. et al. Resonant terahertz detection using grapheneplasmons. Nat. Commun. 9, 5392 (2018).2. Alonso-González, P. et al. Acoustic terahertz graphene plasmonsrevealed by photocurrent nanoscopy. Nat. Nanotechnol. 12,31–35 (2017).3. Low, T. & Avouris, P. Graphene plasmonics for terahertz to mid-infrared applications. ACS Nano 8, 1086–1101 (2014).4. Gonçalves, P. A. D. et al. Quantum surface-response of metalsrevealed by acoustic graphene plasmons. Nat. Commun. 12,3271 (2021).5. Ni, G. X. et al. Fundamental limits to graphene plasmonics. Nature557, 530–533 (2018).6. Woessner, A. et al. Highly confined low-loss plasmons ingraphene–boron nitride heterostructures. Nat. Mater. 14,421–425 (2015).7. Chen, J. et al. Optical nano-imaging of gate-tunable grapheneplasmons. Nature 487, 77–81 (2012).8. Fei, Z. et al. Gate-tuning of graphene plasmons revealed by infrarednano-imaging. Nature 487, 82–85 (2012).9. Li, P. et al. Collective near-field coupling and nonlocal phenomenain infrared-phononic metasurfaces for nano-light canalization. Nat.Commun. 11, 3663 (2020).10. Ni, G. et al. Long-lived phonon polaritons in hyperbolic materials.Nano Lett. 21, 5767–5773 (2021).11. Rizzo, D. J. et al. Polaritonic probe of an emergent 2D dipole inter-face. Nano Lett. 23, 8426–8435 (2023).12. Dai, S. et al. Tunable phonon polaritons in atomically thin van derWaals crystals of boron nitride. Science 343, 1125–1129 (2014).13. Duan, J. et al. Multiple and spectrally robust photonicmagic anglesin reconfigurable α-MoO3 trilayers. Nat. Mater. 22, 867–872(2023).14. Zheng, Z. et al. A mid-infrared biaxial hyperbolic van der Waalscrystal. Sci. Adv. 5, eaav8690 (2019).Article https://doi.org/10.1038/s41467-025-56804-yNature Communications |         (2025) 16:1853 7www.nature.com/naturecommunications15. Ma, W. et al. In-plane anisotropic and ultra-low-loss polaritons in anatural van der Waals crystal. Nature 562, 557–562 (2018).16. Zheng, Z. et al. Highly confined and tunable hyperbolic phononpolaritons in Van Der Waals semiconducting transition metal oxi-des. Adv. Mater. 30, 1705318 (2018).17. Ruta, F. L. et al. Surface plasmons induce topological transition ingraphene/α-MoO3 heterostructures. Nat. Commun. 13, 3719(2022).18. Dai, S. et al. Graphene on hexagonal boron nitride as a tunablehyperbolic metamaterial. Nat. Nanotechnol. 10, 682–686 (2015).19. Brar, V. W. et al. Hybrid surface-phonon-plasmon polariton modesin graphene/monolayer h-bn heterostructures. Nano Lett. 14,3876–3880 (2014).20. Kim, B. S. Y. et al. Ambipolar charge-transfer graphene plasmoniccavities. Nat. Mater. 22, 838–843 (2023).21. Rizzo, D. J. et al. Nanometer-scale lateral p–n junctions in graphene/α-RuCl3 heterostructures. Nano Lett. 22, 1946–1953 (2022).22. Rizzo, D. J. et al. Charge-transfer plasmonpolaritons at graphene/α-RuCl3 interfaces. Nano Lett. 20, 8438–8445 (2020).23. Shao, Y. et al. Infrared plasmons propagate through a hyperbolicnodal metal. Sci. Adv. 8, eadd6169 (2022).24. Ruta, F. L. et al. Hyperbolic exciton polaritons in a van der Waalsmagnet. Nat. Commun. 14, 8261 (2023).25. Sternbach, A. J. et al. Femtosecond exciton dynamics in WSe2optical waveguides. Nat. Commun. 11, 3567 (2020).26. Hu, F. et al. Imaging exciton–polariton transport in MoSe2 wave-guides. Nat. Photonics 11, 356–360 (2017).27. Taboada-Gutiérrez, J. et al. Broad spectral tuning of ultra-low-losspolaritons in a van derWaals crystal by intercalation.Nat. Mater. 19,964–968 (2020).28. Reserbat-Plantey, A. et al. Quantum nanophotonics in two-dimensional materials. ACS Photonics 8, 85–101 (2021).29. Zhu, Y. et al. Electroluminescence as a probe of strongexciton–plasmon coupling in few-layer WSe2. Nano Lett. 24,525–532 (2024).30. Vitalone, R. A. et al. Charge transfer plasmonics in bespoke gra-phene/α-RuCl3 cavities. ACS Nano 18, 29648–29657(2024).31. Klein, J. et al. The bulk van der Waals layered magnet CrSBr is aquasi-1D material. ACS Nano 17, 5316–5328 (2023).32. Wu, F. et al. Quasi-1D electronic transport in a 2D magnetic semi-conductor. Adv. Mater. 34, 2109759 (2022).33. Ziebel, M. E. et al. CrSBr: an air-stable, two-dimensional magneticsemiconductor. Nano Lett. 24, 4319–4329 (2024).34. Wilson, N. P. et al. Interlayer electronic coupling on demand in a 2Dmagnetic semiconductor. Nat. Mater. 20, 1657–1662 (2021).35. Xie, K., Zhang, X.-W., Xiao, D. & Cao, T. Engineering magneticphases of layered antiferromagnets by interfacial charge transfer.ACS Nano 17, 22684–22690 (2023).36. Plumadore, R., Al Ezzi, M. M., Adam, S. & Luican-Mayer, A. Moirépatterns in graphene–rheniumdisulfide vertical heterostructures. J.Appl. Phys. 128, 044303 (2020).37. Qiu, Z. et al. Giant gate-tunable bandgap renormalization andexcitonic effects in a 2D semiconductor. Sci. Adv. 5,eaaw2347 (2019).38. Liu, Y. et al. Tailoring sample-wide pseudo-magnetic fields on agraphene–black phosphorus heterostructure.Nat. Nanotechnol. 13,828–834 (2018).39. Yu, Y.-J. et al. Tuning the graphene work function by electric fieldeffect. Nano Lett. 9, 3430–3434 (2009).40. Caldwell, J. D. et al. Sub-diffractional volume-confined polaritons inthe natural hyperbolic material hexagonal boron nitride. Nat.Commun. 5, 5221 (2014).41. Kučírková, A. & Navrátil, K. Interpretation of infrared transmittancespectra of SiO2 thin films. Appl. Spectrosc. 48, 113–120 (1994).42. Zhang, Y. et al. Giant phonon-induced conductance in scanningtunnelling spectroscopy of gate-tunable graphene. Nat. Phys. 4,627–630 (2008).43. Yankowitz, M. et al. Emergence of superlattice Dirac points in gra-phene on hexagonal boron nitride. Nat. Phys. 8, 382–386 (2012).44. Li, Y. et al. Anisotropic band flattening in graphene with one-dimensional superlattices. Nat. Nanotechnol. 16, 525–530 (2021).45. Xu, S. et al. Electronic interactions inDiracfluids visualizedby nano-terahertz spacetime interference of electron-photonquasiparticles.Sci. Adv. 10, eado5553 (2024).46. Biswas, S., Li, Y., Winter, S. M., Knolle, J. & Valentí, R. ElectronicProperties of α-RuCl3 in Proximity to Graphene. Phys. Rev. Lett. 123,237201 (2019).47. Low, T. et al. Plasmons and screening in monolayer and multilayerblack phosphorus. Phys. Rev. Lett. 113, 106802 (2014).48. Ulstrup, S. et al. Observation of interlayer plasmon polaron in gra-phene/WS2 heterostructures. Nat. Commun. 15, 3845 (2024).49. Bae, Y. J. et al. Exciton-coupled coherent magnons in a 2D semi-conductor. Nature 609, 282–286 (2022).50. Wang, L. et al. One-dimensional electrical contact to a two-dimensional material. Science 342, 614–617 (2013).51. Girit, Ç. Ö. & Zettl, A. Soldering to a single atomic layer. Appl. Phys.Lett. 91, 193512 (2007).52. Chen, W., Madhavan, V., Jamneala, T. & Crommie, M. F. Scanningtunneling microscopy observation of an electronic superlattice atthe surface of clean gold. Phys. Rev. Lett. 80, 1469–1472 (1998).53. Giannozzi, P. et al. QUANTUM ESPRESSO: a modular and open-source software project for quantum simulations of materials. J.Phys.: Condens. Matter 21, 395502 (2009).54. Hamann, D. R. Optimized norm-conserving Vanderbilt pseudopo-tentials. Phys. Rev. B 88, 085117 (2013).55. Grimme, S. Semiempirical GGA-type density functional con-structed with a long-range dispersion correction. J. ComputationalChem. 27, 1787–1799 (2006).56. Bengtsson, L. Dipole correction for surface supercell calculations.Phys. Rev. B 59, 12301–12304 (1999).57. Medeiros, P. V. C., Stafström, S. & Björk, J. Effects of extrinsic andintrinsic perturbations on the electronic structure of graphene:Retaining an effective primitive cell band structure by bandunfolding. Phys. Rev. B 89, 041407 (2014).58. Medeiros, P. V. C., Tsirkin, S. S., Stafström, S. & Björk, J. Unfoldingspinor wave functions and expectation values of general operators:Introducing the unfolding-density operator. Phys. Rev. B 91,041116 (2015).59. Deslippe, J. et al. BerkeleyGW: a massively parallel computerpackage for the calculation of the quasiparticle and optical prop-erties of materials and nanostructures. Computer Phys. Commun.183, 1269–1289 (2012).60. Wiser, N. Dielectric constant with local field effects included. Phys.Rev. 129, 62–69 (1963).61. Adler, S. L. Quantum theory of the dielectric constant in real solids.Phys. Rev. 126, 413–420 (1962).AcknowledgementsResearch on charge transfer plasmons is solely supported as part of theEnergy Frontier Research Center on Programmable Quantum Materialsfunded by the U.S. Department of Energy (DOE), Office of Science, BasicEnergy Sciences (BES), under Award No DE-SC0019443. K.W. and T.T.acknowledge support from the JSPS KAKENHI (Grant Numbers21H05233 and 23H02052), the CREST (JPMJCR24A5), JST and WorldPremier International Research Center Initiative (WPI), MEXT, Japan. Thefirst-principles calculations in Fig. S5 are based uponwork supported bythe National Science Foundation under Award No. DMR-2339995. F. Z.acknowledges the support of the Alexander von Humboldt-Stiftung forArticle https://doi.org/10.1038/s41467-025-56804-yNature Communications |         (2025) 16:1853 8www.nature.com/naturecommunicationsthefinancial support from theHumboldt Research Fellowship. Synthesisof the CrSBr crystals was in part supported by the National ScienceFoundation (NSF) through the Columbia Materials Science and Engi-neering Research Center on Precision-Assembled Quantum Materials(DMR-2011738). T.C. utilized advanced computational, storage, andnetworking infrastructure provided by the Hyak supercomputer systemand funded by the University of Washington Molecular EngineeringMaterials Center at the University of Washington (NSF MRSEC DMR-2308979). A.D. acknowledges support from the Simons FoundationSociety of Fellows program (Grant No. 855186).Author contributionsD.J.R. performed all s-SNOM measurements and analysis. E.S. and S.S.performed the STM experiments and analysis. F.Z. and A.R. performedDFT calculations of the plasmonic properties and analysis of plasmonlifetime. K.X. and T.C. performed DFT calculations and analysis. J.C.,E.J.T., and A.D. fabricated heterostructures for s-SNOM and STMexperiments. R.A.V. and S.X. aided THz imaging and analysis. F.L.R.modeled the near-field data. Y.S. and S.Q. performed KPFM and far-fieldoptical characterization of CrSBr. D.G.C. performed growth and char-acterization of CrSBr single crystals. M.C.S. and T.P.D. performed PLmeasurements. K.W. and T.T. performed growth and characterization ofhBN single crystals. A.J.M., A.R., P.J.S., and X.Z. advised and participatedin experimental design and interpretation. C.R.D. and X.R. advisedsynthesis and device fabrication efforts. A.N.P. advised STM experi-ments. D.N.B. advised s-SNOMexperiments. Themanuscriptwaswrittenwith input from all authors.Competing interestsThe authors declare no competing interests.Additional informationSupplementary information The online version containssupplementary material available athttps://doi.org/10.1038/s41467-025-56804-y.Correspondence and requests for materials should be addressed toDaniel J. Rizzo, Xavier Roy, Abhay N. Pasupathy or D. N. Basov.Peer review information Nature Communications thanks the anon-ymous reviewer(s) for their contribution to thepeer reviewof thiswork. Apeer 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) 2025Article https://doi.org/10.1038/s41467-025-56804-yNature Communications |         (2025) 16:1853 9https://doi.org/10.1038/s41467-025-56804-yhttp://www.nature.com/reprintshttp://creativecommons.org/licenses/by/4.0/http://creativecommons.org/licenses/by/4.0/www.nature.com/naturecommunications Engineering anisotropic electrodynamics at the graphene/CrSBr interface Results Atomically-resolved topography and electronic structure Characterization of the SPP dispersion Uniaxial SPPs SPP anisotropy at THz energies Mechanism for Uniaxial SPPs Discussion Methods Device fabrication Scanning tunneling microscopy and spectroscopy Scanning Near-field Optical Microscopy Ab-initio calculations of graphene/CrSBr heterostructures Data availability References Acknowledgements Author contributions Competing interests Additional information