Letter | Published:

# Ligand-field helical luminescence in a 2D ferromagnetic insulator

## Abstract

Bulk chromium tri-iodide (CrI3) has long been known as a layered van der Waals ferromagnet1. However, its monolayer form was only recently isolated and confirmed to be a truly two-dimensional (2D) ferromagnet2, providing a new platform for investigating light–matter interactions and magneto-optical phenomena in the atomically thin limit. Here, we report spontaneous circularly polarized photoluminescence in monolayer CrI3 under linearly polarized excitation, with helicity determined by the monolayer magnetization direction. In contrast, the bilayer CrI3 photoluminescence exhibits vanishing circular polarization, supporting the recently uncovered anomalous antiferromagnetic interlayer coupling in CrI3 bilayers2. Distinct from the Wannier–Mott excitons that dominate the optical response in well-known 2D van der Waals semiconductors3, our absorption and layer-dependent photoluminescence measurements reveal the importance of ligand-field and charge-transfer transitions to the optoelectronic response of atomically thin CrI3. We attribute the photoluminescence to a parity-forbidden d–d transition characteristic of Cr3+ complexes, which displays broad linewidth due to strong vibronic coupling and thickness-independent peak energy due to its localized molecular orbital nature.

Van der Waals layered materials offer fascinating opportunities for studying light–matter interactions in the 2D limit. For instance, monolayer semiconducting transition metal dichalcogenides (for example, WSe2) enable coupling between the helicity of light and the valley degree of freedom4. In all non-metallic 2D materials to date, it has been established that tightly bound Wannier–Mott excitons dominate the intrinsic optical response3, and there has been rapid progress in studying 2D excitonic interactions, dynamics and spin/valley physics3,5. However, none of these 2D materials possesses long-range magnetic order. A monolayer semiconductor or insulator with intrinsic magnetism would enable the study of novel photo-physical phenomena and the interplay with underlying magnetic order, possibly involving physics incompatible with the Wannier–Mott excitonic picture.

On the other hand, the exploration of ferromagnetism in non-metallic bulk materials has a long history. Early studies examined the intrinsic ferromagnetic ordering of a variety of insulating and semiconducting materials, including, for example, the ferrites and ferrospinels6, Cr trihalides7, Eu chalcogenides8 and Cr spinels9. Later, with the introduction of magnetic dopants into non-magnetic II–VI and III–V semiconductors, diluted magnetic semiconductors captured widespread attention10, boosted by the discovery of ferromagnetism in Mn-doped InAs (ref. 11) and GaAs (ref. 1) in the 1990s12. Central to progress in these fields, optical experiments have led to a deep understanding of electronic structure, magnetization dynamics and interactions between magnetism and light8,13,14,15. While the fascinating physics in the quantum structures of diluted magnetic semiconductors has propelled spintronics research over the last few decades16, there has been a comparative lack of non-metallic nanoscale materials hosting intrinsic magnetism.

Recently, several van der Waals magnetic insulators have emerged as a promising platform for exploring light–matter interactions in the monolayer limit2,17,18. In particular, a magneto-optical Kerr effect study2 revealed 2D ferromagnetism in monolayer CrI3 (Fig. 1a). Historically, the bulk chromium trihalides were found to behave as Mott insulators with an optical response governed by ligand-field and charge-transfer transitions13,19,20,21,22, which are highly localized photo-excitations between molecular orbitals. Therefore, atomically thin chromium trihalides may provide a new platform from which to study 2D optical physics under the influence of intrinsic magnetic ordering beyond the excitonic picture. In this work, we reveal magnetization-determined ligand-field transitions in 2D ferromagnetic CrI3 by magneto-photoluminescence (PL) and reflection spectroscopy.

We prepared monolayer CrI3 on sapphire substrates by mechanical exfoliation of bulk crystals. For the PL experiments, we excited with a linearly polarized laser at 1.96 eV and analysed the circularly polarized PL components (see Methods). All measurements were done under normal incidence excitation and collection. We first measured the monolayer CrI3 PL at 75 K, above the Curie temperature (T C) of 45 K (ref. 2), without applying magnetic field. As shown in Fig. 1b, there is a single broad emission feature centred at ~1.1 eV with a full-width at half-maximum of about 180 meV. We observed no other PL features at higher energy. The red and blue spectra correspond to $${{\rm{\sigma }}}^{+}$$and $${{\rm{\sigma }}}^{-}$$ circularly polarized PL, respectively. The two components are indistinguishable, which is consistent with the absence of magnetic order above T C.

Remarkably, on cooling to 15 K (well below T C) in the absence of a magnetic field, we observe that the $${{\rm{\sigma }}}^{+}$$ and $${{\rm{\sigma }}}^{-}$$ PL diverge in intensity, with $${{\rm{\sigma }}}^{+}$$ much brighter than $${{\rm{\sigma }}}^{-}$$ (Fig. 1c). Defining the PL circular polarization, $$\rho =\frac{{I}_{+}-{I}_{-}}{{I}_{+}+{I}_{-}}$$, where $${I}_{\pm }$$ is the peak intensity of $${{\rm{\sigma }}}^{\pm }$$ PL, we find $$\rho \approx$$ 0.45. Since the linearly polarized excitation does not break the time-reversal symmetry of the system, the observation of net circular emission demonstrates the appearance of spontaneous out-of-plane magnetic ordering in monolayer CrI3. While the peak narrows slightly (~10%) on cooling from 75 K to 15 K, its position does not change. Any energy splitting between the two polarizations is smaller than ~1 meV and unresolvable. The temperature dependence of ρ is shown in Fig. 1e. The onset of finite ρ at 45 K is consistent with T C for a monolayer2.

We next explore the effect of an applied out-of-plane magnetic field. Figure 2a shows the circularly polarized PL spectra at +0.5 T. Here the polarization is reversed relative to that shown in Fig. 1c (which was taken at zero field), due to a flip in the magnetization by the applied field. If the field is then lowered to 0 T, the spectra remain unchanged (Fig. 2b). When the opposite field, −0.5 T, is then applied, the magnetization is reversed again and the PL returns to its original state with a stronger $${{\rm{\sigma }}}^{+}$$ component than $${{\rm{\sigma }}}^{-}$$. This state is in turn preserved when the field is returned to zero (Fig. 2c,d). Additional data demonstrating this behaviour for a monolayer on a SiO2/Si substrate are presented in Supplementary Section 1. We show ρ over a cycle of the magnetic field in Fig. 2e, where the observed hysteresis loop is clearly the hallmark of ferromagnetic behaviour. The saturation polarization is ±0.5, and the coercive field is ~55 mT. Both circular polarization components show a linear power dependence (Supplementary Section 2), leading to a power-independent ρ.

The field-dependent measurements unambiguously demonstrate that the helicity of the CrI3 PL is determined by its magnetic ordering, and thus that measurement of PL helicity can be used as a simple probe of its magnetic phases. Figure 3a–c presents the circular polarization-resolved PL from a bilayer of CrI3 at 15 K, at magnetic fields −1, 0 and +1 T respectively. We see strong circularly polarized PL at ±1 T (Fig. 3a,c), consistent with full spin polarization in bilayer CrI3 at such high fields. In contrast with the monolayer, no net circular polarization is seen at zero field (Fig. 3b), which implies vanishing net out-of-plane magnetization. As shown in Fig. 3d, the polarization ρ is negligible between ± 0.65 T and jumps abruptly to a value of −0.5 above 0.65 T and +0.5 below −0.65 T. This magnetic-field-dependent PL polarization forms a close parallel to the magneto-optical Kerr effect response in ref. 2, which also vanishes in the same field range. The consistency of the results for these two distinct experimental probes reinforces the interpretation that at low fields, bilayer CrI3 consists of two ferromagnetic monolayers that are antiferromagnetically coupled.

What is the origin of luminescence in atomically thin CrI3? In well-studied 2D semiconductors, such as WSe2, the photo-response can be described by band-to-band transitions with strong excitonic effects. Tightly bound Wannier–Mott excitons dominate the optical spectrum3. In CrI3, however, the 3d electrons are much more spatially localized, and thus to understand the optical response, a molecular orbital approach is more suitable. In fact, ligand field theory, which predicts intra-atomic dd transitions and higher-energy charge-transfer transitions within metal–ligand complexes23, has previously been used to interpret the optical spectra of bulk transition metal halides (refs 3,4,5,6,7). While prior studies have focused on absorption/reflection in CrCl3 and CrBr3, there are few reports on the optical properties of CrI3 (refs 1,3,19), and none examining the PL. Furthermore, all previous work was limited to bulk crystals containing thousands of layers. We now present evidence that these intrinsic localized photo-excitations dominate in 2D CrI3.

We first reiterate that the monolayer PL intensity is linear in excitation power, as illustrated in Supplementary Fig. 2. Taken together with the tight link between the PL helicity and layer-dependent magnetic phases, this rules out the possibility that the dominant PL contribution is from defect-bound excitons, which tend towards saturation at higher excitation intensity and are often seen in the low-energy PL spectra of other 2D semiconductors3,24. To investigate the electronic response more broadly, we measured the differential reflectance of monolayer CrI3 on sapphire (Fig. 4a), which is proportional to its absorbance25. There is a weak peak near 1.5 eV along with stronger features around 2 eV and 2.7 eV. Using the ligand-field framework, we can attribute the 2.7 eV peak as well as the strongest two peaks near 2 eV to dipole-allowed ligand-to-metal charge-transfer (LMCT) transitions between the iodine 5p orbitals and the Cr3+ 3d orbitals (Supplementary Sections 3 and 4). The 1.5 eV transition has not been discussed much before in the literature. As a result of the approximate octahedral symmetry of the iodine ligands around each Cr3+ site (Fig. 1a), the d 3 configuration (4F term) in isolated Cr3+ splits into an 4A2 ground state and 4T2 and 4T1 excited states of the t 2g and e g orbitals in CrI3. We assign the 1.5 eV peak to the lowest energy transition between these levels, from 4A2 to 4T2. Despite being electric-dipole-forbidden by the Laporte parity selection rule, dd transitions can become weakly allowed by mixing with odd-parity states, such as produced by phonons26. In addition, the trigonal field of the nearest-neighbour Cr atoms (Fig. 1a) eliminates the local inversion symmetry of each Cr site, providing a static odd-parity field to allow the dd transitions. We deduce that the absorbance of the 4T2 transition is about 0.7% for monolayer CrI3 (see Methods), which is an order of magnitude lower than the absorbance for the A and B excitonic resonances in monolayer semiconducting transition metal dichalcogenides27, such as MoS2. The weak absorption underscores the weakly allowed parity-forbidden nature of the dd transition.

Our assignment of the reflection features is consistent with the results of prior experiments and recent calculations28,29 on bulk CrCl3 and CrBr3. To connect these with CrI3, in Fig. 4b, we plot the absorption peak energies of bulk CrI3 that we measured (Supplementary Section 4) against those of CrCl3 and CrBr3 from previous studies13,19,20,21,22. The relationship between the optical spectra of the Cr trihalides then becomes clear. The large energetic shift of the two high-energy peaks between different ligand species confirms their charge-transfer origin. On the other hand, the dd transitions exhibit weaker dependence on the ligand. Interestingly, in view of the decreased ligand-field strength of I in CrI3, we expect the lowest LMCT transition to overlap the highest dd transition (4A2 to 4T1). This suggests that the origin of the low-energy shoulder at 1.8 eV is absorption to the 4T1 level, enhanced by its energetic proximity to the strong charge-transfer transitions near 2 eV. Thus, it is likely that both LMCT and dd transitions play a role in the recently observed magneto-optical Kerr response2 near 2 eV in few-layer CrI3.

We can now begin to understand the origin of PL in atomically thin CrI3. The clear correlation between the three Cr trihalide compounds in the energies of the lowest absorption peak and PL (Fig. 4b) implies that the monolayer CrI3 PL arises from the 4T2 to 4A2 dd transition. The ~400 meV Stokes shift between the PL and 1.5 eV reflection peak is a consequence of the Franck–Condon principle and strong electron–lattice coupling26 (Fig. 4c, Huang–Rhys factor ~ 10, see Supplementary Section 5). The extracted ~1.3 eV 4T2 zero-phonon energy is also consistent with the estimated ligand-field splitting of 1.2 eV determined from the angular overlap between the Cr–I σ and π bonds (Supplementary Section 5). The wide PL linewidth is a signature of dd luminescence broadened by vibronic modes, characteristic of transition metal ions in solids26. This strong vibronic mixing precludes the formation of well-resolved phonon sidebands in the spectra26. In addition, our layer-dependent study demonstrates that the PL peak energy is independent of the thickness of CrI3 (Fig. 4d,e), highlighting the localized nature of the ligand-field excitation. As the dd transition is parity-forbidden, the observed PL intensity will sensitively depend on the origins of local symmetry breaking, such as from odd-parity phonons and trigonal distortion, and on the interactions with excited states and spin–orbit coupling. Determining the strengths of these interactions will be essential for understanding the relative intensity of the PL helicities observed in ferromagnetic CrI3. We also note that PL intensity per layer (that is, normalized by the number of layers) increases with increasing thickness (Fig. 4e), suggesting that the dd relaxation processes are affected by interlayer or substrate interactions.

The above measurements establish the prominence of highly localized optical excitations in atomically thin CrI3. While transport measurements will be necessary to confirm its electrical properties, mono- and few-layer CrI3 are likely to be good insulators, judging from the localized optical response and the large bulk resistivity30. The ligand-field luminescence helicity displays clear signatures of the underlying magnetic order, suggesting that the optical spectra will serve as an important probe of magnetic order in atomically thin CrI3. Our results enlarge the landscape of light–matter interactions in 2D materials, and unlock new opportunities to study and control ligand-field spectra in the 2D limit in the presence of magnetic ordering. Further magneto-optical studies on the thickness and polarization dependence of the charge-transfer and ligand-field transitions may shed light on the nature of the intralayer and interlayer exchange interactions. Beyond this, our work previews exciting possibilities to use CrI3 as an atomically thin magnetic insulator in van der Waals heterostructures. Complementary to the popular non-magnetic layered insulator hexagonal boron nitride, 2D CrI3 will serve as a substrate, interfacial layer and tunnel barrier for engineering magnetic proximity effects, exploring spin-dependent tunnelling phenomena, and designing novel magneto-optoelectronic devices with spontaneous helical light emission.

## Methods

### Sample fabrication

Bulk CrI3 crystals were grown by chemical vapour transport, as described in detail in refs 2,30 . Monolayer and bilayer CrI3 samples were then obtained by mechanical exfoliation from bulk CrI3 onto a 0.5-mm-thick c-plane sapphire substrate in an Ar-filled glovebox. We also fabricated and measured samples on 285 nm SiO2/Si (Supplementary Figs 1 and 3d). We confirmed the optical contrast of bilayer CrI3 on sapphire by transferring a bilayer from 285 nm SiO2/Si (on which the optical contrast has been established2) to sapphire. Thus, we determined the optical contrast on sapphire to be ~0.035 and ~0.07 at 631 nm for monolayer and bilayer CrI3, respectively. Samples were kept under an inert atmosphere or vacuum during the entire fabrication and measurement process.

### Optical measurements

Low-temperature optical measurements were performed in a closed-cycle cryostat with a superconducting magnet with the axis directed out of the sample plane. For the PL measurements, the sample was excited by a HeNe laser (632.8 nm) focused to a ~1 µm spot diameter. A low power of 10 µW was used to avoid sample heating and degradation. A dichroic beamsplitter reflected the collected PL, which was then spatially filtered through a confocal pinhole (to avoid collecting nearby bulk CrI3 PL), dispersed by a 1.2 μm blaze grating, and detected by a liquid-nitrogen-cooled InGaAs linear photodiode array (Princeton Instruments). The InGaAs detector was spectrally calibrated using Hg emission lines. The excitation and detection polarization were controlled using linear polarizers and achromatic near-infrared half- and quarter-wave plates. Peak intensities were calculated by averaging 100 points (~30 meV) about the peak centre. For the white-light differential reflection measurements, we spatially filtered a tungsten halogen lamp through a pinhole and focused the beam to a ~3 µm spot size on the CrI3. The reflected light was deflected with a beamsplitter and detected by a spectrometer and Si CCD (charge-coupled device) or InGaAs array, which enabled measurements from 1 to 3 eV. To obtain the differential reflectance, we subtracted and normalized the CrI3 reflectance by the reflectance of the bare sapphire substrate. The 4T2 absorbance was determined as $$\frac{1}{4}\left({n}^{2}-1\right)\frac{{\rm{\Delta }}R}{R}$$ (ref. 25), where $$n\approx$$ 1.76 is the ordinary refractive index of sapphire at 1.5 eV (ref. 31).

### Data availability

The data that support the plots within this paper and other findings of this study are available from the corresponding author upon reasonable request.

Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

## References

1. 1.

Dillon, J. F. & Olson, C. E. Magnetization, resonance, and optical properties of the ferromagnet CrI3. J. Appl. Phys. 36, 1259–1260 (1965).

2. 2.

Huang, B. et al. Layer-dependent ferromagnetism in a van der Waals crystal down to the monolayer limit. Nature 546, 270–273 (2017).

3. 3.

Mak, K. F. & Shan, J. Photonics and optoelectronics of 2D semiconductor transition metal dichalcogenides. Nat. Photon. 10, 216–226 (2016).

4. 4.

Xu, X., Yao, W., Xiao, D. & Heinz, T. F. Spin and pseudospins in layered transition metal dichalcogenides. Nat. Phys. 10, 343–350 (2014).

5. 5.

Schaibley, J. R. et al. Valleytronics in 2D materials. Nat. Rev. Mater. 1, 16055 (2016).

6. 6.

Snoek, J. L. New Developments in Ferromagnetic Materials (Elsevier, New York, 1947).

7. 7.

Tsubokawa, I. On the magnetic properties of a CrBr3 single crystal. J. Phys. Soc. Jpn 15, 1664–1668 (1960).

8. 8.

Wachter, P. The optical electrical and magnetic properties of the europium chalcogenides and the rare earth pnictides. CRC Crit. Rev. Solid State Sci. 3, 189–241 (1972).

9. 9.

Baltzer, P. K., Lehmann, H. W. & Robbins, M. Insulating ferromagnetic spinels. Phys. Rev. Lett. 15, 493–495 (1965).

10. 10.

Furdyna, J. K. Diluted magnetic semiconductors. J. Appl. Phys. 64, R29–R64 (1988).

11. 11.

Ohno, H., Munekata, H., Penney, T., von Molnár, S. & Chang, L. L. Magnetotransport properties of p-type (In,Mn)As diluted magnetic III-V semiconductors. Phys. Rev. Lett. 68, 2664–2667 (1992).

12. 12.

Ohno, H. et al. Ga,Mn)As: A new diluted magnetic semiconductor based on GaAs. Appl. Phys. Lett. 69, 363–365 (1996).

13. 13.

Dillon, J. F., Kamimura, H. & Remeika, J. P. Magneto-optical properties of ferromagnetic chromium trihalides. J. Phys. Chem. Solids 27, 1531–1549 (1966).

14. 14.

Burch, K. S., Awschalom, D. D. & Basov, D. N. Optical properties of III-Mn-V ferromagnetic semiconductors. J. Magn. Magn. Mater. 320, 3207–3228 (2008).

15. 15.

Miles, P. A., Westphal, W. B. & Von Hippel, A. Dielectric spectroscopy of ferromagnetic semiconductors. Rev. Mod. Phys. 29, 279–307 (1957).

16. 16.

Dietl, T. & Ohno, H. Dilute ferromagnetic semiconductors: Physics and spintronic structures. Rev. Mod. Phys. 86, 187–251 (2014).

17. 17.

Gong, C. et al. Discovery of intrinsic ferromagnetism in 2D van der Waals crystals. Nature 546, 265–269 (2017).

18. 18.

Lee, J.-U. et al. Ising-type magnetic ordering in atomically thin FePS3. Nano Lett. 16, 7433–7438 (2016).

19. 19.

Grant, P. M. & Street, G. B. Optical properties of the chromium trihalides in the region 1-11 eV. Bull. Am. Phys. Soc. II 13, (1968).

20. 20.

Pollini, I. & Spinolo, G. Intrinsic optical properties of CrCl3. Phys. Status Solidi 41, 691–701 (1970).

21. 21.

Bermudez, V. M. & McClure, D. S. Spectroscopic studies of the two-dimensional magnetic insulators chromium trichloride and chromium tribromide—I. J. Phys. Chem. Solids 40, 129–147 (1979).

22. 22.

Nosenzo, L., Samoggia, G. & Pollini, I. Effect of magnetic ordering on the optical properties of transition-metal halides: NiCl2, NiBr2, CrCl3, and CrBr3. Phys. Rev. B 29, 3607–3616 (1984).

23. 23.

Figgis, B. N. & Hitchman, M. A. Ligand Field Theory and its Applications (Wiley, New York, 2000).

24. 24.

Tongay, S. et al. Defects activated photoluminescence in two-dimensional semiconductors: interplay between bound, charged, and free excitons. Sci. Rep. 3, 2657 (2013).

25. 25.

McIntyre, J. D. E. & Aspnes, D. E. Differential reflection spectroscopy of very thin surface films. Surf. Sci. 24, 417–434 (1971).

26. 26.

Henderson, B. & Imbusch, G. F. Optical Spectroscopy of Inorganic Solids (Clarendon, Oxford, 1989).

27. 27.

Li, Y. et al. Measurement of the optical dielectric function of monolayer transition-metal dichalcogenides: MoS2, MoSe2, WS2, and WSe2. Phys. Rev. B 90, 205422 (2014).

28. 28.

Shinagawa, K., Sato, H., Ross, H. J., McAven, L. F. & Butler, P. H. Charge-transfer transitions in chromium trihalides. J. Phys. Condens. Matter 8, 8457 (1996).

29. 29.

McAven, L. F., Ross, H. J., Shinagawa, K. & Butler, P. H. The Kerr magneto-optic effect in ferromagnetic CrBr3. J. Phys. B At. Mol. Opt. Phys. 32, 563 (1999).

30. 30.

McGuire, M. A., Dixit, H., Cooper, V. R. & Sales, B. C. Coupling of crystal structure and magnetism in the layered, ferromagnetic insulator CrI3. Chem. Mater. 27, 612–620 (2015).

31. 31.

Malitson, I. H. Refraction and dispersion of synthetic sapphire. J. Opt. Soc. Am. 52, 1377–1379 (1962).

## Acknowledgements

The authors thank D. Gamelin for insightful discussions on the optical response of CrI3, and A. Majumdar for testing the measurement system. Work at the University of Washington was mainly supported by the Department of Energy, Basic Energy Sciences, Materials Sciences and Engineering Division (DE-SC0018171), and University of Washington Innovation Award. Work at MIT has been supported by the Center for Integrated Quantum Materials under NSF grant DMR-1231319 as well as the Gordon and Betty Moore Foundation’s EPiQS Initiative through Grant GBMF4541 to P.J.-H. Device fabrication has been partly supported by the Center for Excitonics, an Energy Frontier Research Center funded by the US Department of Energy (DOE), Office of Science, Office of Basic Energy Sciences under Award Number DESC0001088. The contribution of D.H.C. is supported by DE-SC0002197. Work at CMU is supported by DOE BES DE-SC0012509. W.Y. is supported by the Croucher Foundation (Croucher Innovation Award), the RGC of Hong Kong (HKU17305914P), and the HKU ORA. Work at ORNL (M.A.M.) was supported by the US Department of Energy, Office of Science, Basic Energy Sciences, Materials Sciences and Engineering Division. X.X. and D.X. acknowledge the support of a Cottrell Scholar Award. S.G. and L.Y. are supported by NSF grant no. DMR-1455346 and EFRI-2DARE-1542815. X.X. acknowledges the support from the State of Washington funded Clean Energy Institute and from the Boeing Distinguished Professorship in Physics.

## Author information

### Affiliations

1. #### Department of Physics, University of Washington, Seattle, WA, USA

• Kyle L. Seyler
• , Ding Zhong
• , Bevin Huang
• , David H. Cobden
•  & Xiaodong Xu
2. #### Department of Physics, Massachusetts Institute of Technology, Cambridge, MA, USA

• Dahlia R. Klein
• , Efrén Navarro-Moratalla
•  & Pablo Jarillo-Herrero
3. #### Department of Physics, Washington University, St Louis, MO, USA

• Shiyuan Gao
•  & Li Yang
4. #### Department of Physics, Carnegie Mellon University, Pittsburgh, PA, USA

• Xiaoou Zhang
•  & Di Xiao
5. #### Materials Science and Technology Division, Oak Ridge National Laboratory, Oak Ridge, TN, USA

• Michael A. McGuire

• Wang Yao
7. #### Department of Materials Science and Engineering, University of Washington, Seattle, WA, USA

• Xiaodong Xu

### Contributions

X.X., K.L.S. and P.J.-H. conceived the experiment. K.L.S. built the experimental set-up and carried out the measurements, assisted by D.Z. and B.H., supervised by X.X. Crystal growth, characterization and device fabrication at MIT were carried out by D.R.K. and E.N.-M., supervised by P.J.-H. Device fabrication at UW was carried out by K.L.S., D.Z. and B.H., with crystal grown and characterized by M.A.M. at ONRL. K.L.S. and X.X. analysed and interpreted the data with theoretical support from X.Z., D.X., W.Y., S.G. and L.Y. K.L.S., X.X., D.H.C. and P.J.-H. wrote the manuscript with input from all authors. All authors discussed the results.

### Competing financial interests

The authors declare no competing financial interests.

### Corresponding authors

Correspondence to Pablo Jarillo-Herrero or Xiaodong Xu.

## Electronic supplementary material

1. ### Supplementary Information

Supplementary notes, Supplementary Figures 1–5, and Supplementary References 1–12

### DOI

https://doi.org/10.1038/s41567-017-0006-7

• ### Routing the emission of a near-surface light source by a magnetic field

• F. Spitzer
• , A. N. Poddubny
• , I. A. Akimov
• , V. F. Sapega
• , L. Klompmaker
• , L. E. Kreilkamp
• , L. V. Litvin
• , R. Jede
• , G. Karczewski
• , M. Wiater
• , T. Wojtowicz
• , D. R. Yakovlev
•  & M. Bayer

Nature Physics (2018)

• ### Two-dimensional itinerant ferromagnetism in atomically thin Fe3GeTe2

• Zaiyao Fei
• , Bevin Huang
• , Paul Malinowski
• , Wenbo Wang
• , Tiancheng Song
• , Joshua Sanchez
• , Wang Yao
• , Di Xiao
• , Xiaoyang Zhu
• , Andrew F. May
• , Weida Wu
• , David H. Cobden
• , Jiun-Haw Chu
•  & Xiaodong Xu

Nature Materials (2018)

• ### Electrical control of 2D magnetism in bilayer CrI3

• Bevin Huang
• , Genevieve Clark
• , Dahlia R. Klein
• , David MacNeill
• , Efrén Navarro-Moratalla
• , Kyle L. Seyler
• , Nathan Wilson
• , Michael A. McGuire
• , David H. Cobden
• , Di Xiao
• , Wang Yao
• , Pablo Jarillo-Herrero
•  & Xiaodong Xu

Nature Nanotechnology (2018)

• ### 2D magnetism gets hot

Nature Nanotechnology (2018)