Proximity induced high-temperature magnetic order in topological insulator--ferrimagnetic insulator heterostructure

Letter pubs.acs.org/NanoLett Proximity Induced High-Temperature Magnetic Order in Topological Insulator - Ferrimagnetic Insulator Heterostructure Murong Lang,*,†,ll Mohammad Montazeri,†,ll Mehmet C. Onbasli,‡ Xufeng Kou,† Yabin Fan,† Pramey Upadhyaya,† Kaiyuan Yao,† Frank Liu,‡ Ying Jiang,§ Wanjun Jiang,† Kin L. Wong,† Guoqiang Yu,† Jianshi Tang,† Tianxiao Nie,† Liang He,*,† Robert N. Schwartz,† Yong Wang,§ Caroline A. Ross,‡ and Kang L. Wang*,† † Department of Electrical Engineering, University of California, Los Angeles, California 90095, United States Department of Materials Science and Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, United States § Center of Electron Microscopy, State Key Laboratory of Silicon Materials, Department of Materials Science and Engineering, Zhejiang University, Hangzhou, 310027, China ‡ S Supporting Information * ABSTRACT: Introducing magnetic order in a topological insulator (TI) breaks time-reversal symmetry of the surface states and can thus yield a variety of interesting physics and promises for novel spintronic devices. To date, however, magnetic effects in TIs have been demonstrated only at temperatures far below those needed for practical applications. In this work, we study the magnetic properties of Bi2Se3 surface states (SS) in the proximity of a high Tc ferrimagnetic insulator (FMI), yttrium iron garnet (YIG or Y3Fe5O12). Proximity-induced butterfly and square-shaped magnetoresistance loops are observed by magneto-transport measurements with out-of-plane and in-plane fields, respectively, and can be correlated with the magnetization of the YIG substrate. More importantly, a magnetic signal from the Bi2Se3 up to 130 K is clearly observed by magneto-optical Kerr effect measurements. Our results demonstrate the proximity-induced TI magnetism at higher temperatures, an important step toward roomtemperature application of TI-based spintronic devices. KEYWORDS: Magnetic topological insulator, proximity effect, high temperature, ferrimagnetic insulator YIG, magnetoresistance, magneto-optical Kerr effect T effective way to break the TRS in TIs, substantial efforts have been made to introduce ferromagnetic order in a TI by doping 3d transition metal elements such as Cr, Fe, and Mn.14−19 However, to date the Curie temperature (Tc) of magnetically doped TIs has only reached ∼35 K, which is much lower than required for practical applications.17,19−21 Hence, realization of magnetic order in TIs at higher temperature remains one of the major challenges in the field. One promising approach to break the TRS in TI is to utilize a magnetic proximity effect at the interface of the TI and a topologically trivial magnetic material. While theoretical work suggests the legitimacy of this approach,22,23 limited experimental work has been reported in the literature.15,24−26 Recent efforts on EuS/TI heterostructure strongly suggest the presence of magnetic proximity effects at the interface. However, the effect is limited to low temperatures (<20 K) due to the low he recently discovered time-reversal-invariant topological insulator (TI), a novel state of quantum matter, has led to the flourishing of unique physics along with promises for innovative electronic and spintronic applications.1−3 For example, consequent to the time-reversal symmetry (TRS) the spin of the Dirac-like surface states (SS) is tightly locked to the momentum, resulting in a spin-polarized current at the surface of the TI which is immune to direct backscattering.1,3−7 Furthermore, these topological properties are robust against nonmagnetic external perturbations as long as TRS is preserved. Such properties make TIs perfect candidates for low dissipation spintronic devices. Breaking TRS introduces axion electrodynamics physics manifested by a gapped Dirac spectrum as well as the topological magnetoelectric effect, which gives rise to the direct coupling of electric and magnetic fields. Various exotic phenomena, such as the quantum anomalous Hall effect,8−10 giant magneto-optical Kerr effect,11 chiral mode conduction channels,6,12 and magnetic monopole effect13 have been experimentally discovered or theoretically predicted. As an © XXXX American Chemical Society Received: March 14, 2014 Revised: May 19, 2014 A dx.doi.org/10.1021/nl500973k | Nano Lett. XXXX, XXX, XXX−XXX Nano Letters Letter Figure 1. Atomic force microscopy (AFM), X-ray diffraction (XRD), and VSM characterizations of YIG/GGG and STEM of Bi2Se3/YIG/GGG. (a) A typical AFM image of 50 nm YIG thin film grown on GGG (111) substrate. The YIG layers have a surface rms roughness less than 0.3 nm over the scanned areas of 1 μm × 1 μm. (b) XRD spectrum of YIG grown on GGG in ω-2θ scans between 2θ = 40° to 130°. The peaks of the film and substrate were aligned in most cases because of epitaxial growth. Inset: a spectrum with an expanded angle scale, which indicates that the YIG film has a (111) orientation. (c) Magnetic hysteresis loop for 50 nm in-plane magnetized YIG measured by VSM at 300 K. The in-plane saturation moment is 142 emu/cm3 and the coercivity is less than 5 Oe. (d) A typical STEM-HAADF image of Bi2Se3/YIG/GGG heterostructure, showing atomically sharp interfaces of YIG/GGG (yellow dashed lines, left panel) and Bi2Se3/YIG (red dashed lines, right panel). Curie temperature of EuS layer.25,26 In this work, for the first time a high-temperature (∼130 K) proximity-induced magnetic order is demonstrated at the interface between Bi2Se3 and a high Tc ferrimagnetic insulator (FMI), yttrium iron garnet (YIG, Y3Fe5O12). This is a significant step toward realizing TIbased spintronics at room temperature. By a combination of temperature-dependent magneto-transport measurements and magneto-optical Kerr effect (MOKE) magnetometry, we provide direct evidence of the correlation between the magnetization of the YIG layer and the transport properties of Bi2Se3. A butterfly shape or a square shape hysteretic magnetoresistance (MR) is observed when the external field is perpendicular or parallel to the sample plane, respectively, which is correlated with the magnetization reversal of the YIG. Furthermore, the MOKE data suggest that a magnetic order develops in the TI at the interface with its spin presumed to be antiparallel to the magnetization of the YIG up to at least 130 K. Consistent with numerical simulations, while the YIG substrate shows in-plane anisotropy at room temperature, a canting of magnetization toward out-of-plane direction is clearly observed at lower temperatures, most likely due to magnetocrystalline anisotropy. The magnetic configuration of YIG is expected to be useful to control the TRS of Bi2Se3 surface states, due to the penetration of magnetic order from the YIG into the Bi2Se3 at the interface. The YIG was a 50 nm thick single crystal film grown epitaxially by pulsed-laser deposition on a paramagnetic gallium gadolinium garnet (GGG) (111) substrate at 650 °C. YIG is a well-known FMI with Tc well above the room temperature (Tc ∼ 550 K) and has been extensively studied as a prototype magnetic insulator for spin waves and magnonic physics including spin-Seebeck and spin-Hall effects.27−30 A low deposition rate (<1 nm/min) and a small lattice mismatch between YIG and GGG (<0.2%) allowed high quality epitaxial growth of YIG on GGG with low surface roughness.31 Atomic force microscopy showed that the YIG layer had a surface rootmean-square roughness less than 0.3 nm over the scanned area of 1 μm × 1 μm as shown in Figure 1a. X-ray diffraction, Figure 1b, showed YIG with a (111) orientation,32 and no other phases were observed. Figure 1c displays the room-temperature magnetic hysteresis loops measured by a vibrating sample magnetometer (VSM), showing the YIG films are magnetically soft and isotropic in the film plane. An in-plane saturation moment (MS) of 142 emu/cm3 and coercivity less than 5 Oe were measured at room temperature. To form a TI/FMI heterostructure, 8 quintuple layers (QLs) of single crystal Bi2Se3 were grown on the YIG/GGG substrate in a PerkinElmer molecular beam epitaxy (MBE) system under an ultrahigh vacuum environment at 200 °C (see Supporting Information S1 for growth method).33−35 For comparison with the Bi2Se3/YIG sample, 8 QLs of Bi2Se3 films were also deposited on highly resistive Si(111) using the same technique. High-resolution scanning transmission electron microscopy (STEM) and energy dispersive X-ray spectroscopy (EDX) (see Supporting Information S2) were employed to verify the high quality of the sample with no signature of intermixing at the interface.36 Interdiffusion of materials at the interface is not expected due to the high stability of YIG and the large difference in the growth temperatures. Figure 1d shows the high-angle annular dark-field (HAADF) images of YIG/GGG and Bi2Se3/YIG interfaces, acquired by a Fischione HAADF detector. The atomically sharp interface indicated by the dashed line in the left panel of Figure 1d demonstrates the high-quality epitaxial growth of YIG on the GGG substrate. The right panel B dx.doi.org/10.1021/nl500973k | Nano Lett. XXXX, XXX, XXX−XXX Nano Letters Letter Figure 2. MR of Bi2Se3/YIG with out-of-plane and in-plane magnetic field applied. (a) The nonhysteretic WAL behavior is obtained with out-ofplane field applied in Bi2Se3/Si. The parabolic MR curve is observed with in-plane field applied. (b) Temperature-dependent MR with an out-ofplane field applied. The hysteretic feature associated with the magnetic property of TI bottom surface disappears above 25 K, as the Bi2Se3 bulk component starts to dominate at higher temperature. Inset: An optical image of a Hall bar device structure with a 10 μm scale bar. (c) Temperaturedependent MR with in-plane field applied. Similarly, due to the increasing bulk conduction contribution, the square-shaped feature cannot be observed above 25 K. (d) Anisotropic mangnetoresistance in Bi2Se3/YIG sample at H = 4 kOe with AMR ratio ∼0.2%. contribution of the topological surface states. On the other hand, the Hmin of each hysteresis loop showed a weak temperature dependence, to be discussed in detail later, implying that the magnetic property may survive at higher temperatures, due to the high Tc of the YIG. With an in-plane magnetic field, however, the MR hysteresis loops were square shaped with two sharp steps within each sweep (see Figure 2c). For instance, sweeping from negative to positive field, a sharp increase in MR appeared at Hr = −1.45 kOe following by a sharp drop at Hf = 1.84 kOe, as indicated by the arrows. Hr and Hf did not vary much as the temperature increased, which can be attributed to the modest temperature dependence of YIG magnetization within this temperature range (see Supporting Information S5). The comparison between the temperature dependence of the two MR geometries (Figures 2b-c) suggests that while their hysteretic MR shared a common origin in the proximity effect, the switching process of the YIG differed for different field directions. Similar hysteretic behaviors of magnetoresistance were observed in other TI/YIG devices (see Supporting Information S6). Moreover, we have performed in-plane rotation magnetic field measurement at H = 4 kOe, in which the MR results of Bi2Se3/YIG exhibit an anisotropic MR (AMR) behavior as shown in Figure 2d. The AMR ratio of Bi2Se3/YIG is measured at ∼0.2% while the control sample (Bi2Se3/Si) does not exhibit any AMR as expected. It is important to note that the ∼1% amplitude of the hysteretic MR is much larger than AMR ratio of 0.2%, which indicates that the AMR is not the dominant mechanism in the hysteretic behavior. Nevertheless, both MR and AMR strongly suggest the presence of magnetic order in the TI induced by the magnetic substrate. To further investigate the magnetic properties of the TI/FMI bilayer and verify the role of broken/preserved TRS in the spectrum of TI, MOKE measurements were performed at various temperatures to study both the out-of-plane (polar mode) and in-plane components (longitudinal mode) of the magnetic moment of the YIG layer and the Bi2Se3/YIG bilayer. shows magnified quintuple layers of Bi2Se3 grown on YIG in which the sharp interface between the Bi2Se3 and YIG is indicated by the red dashed line. To investigate the magnetic response of Bi2Se3/YIG heterostructure, the samples were patterned into standard Hall bar devices with 10 μm length and 10 μm width. Fourprobe magneto-transport measurements were conducted in a Quantum Design Physical Properties Measurement System (PPMS) (see Supporting Information S3 for device fabrication and electrical characterization methods). The carrier density of the Bi2Se3/YIG sample was ∼8.2 × 1012 cm−2 at 2 K, comparable to the density of ∼7 × 1012 cm−2 in the Bi2Se3/Si sample (see Supporting Information S4). Figure 2a presents the longitudinal MR measurements of a nonmagnetic control sample, Bi2Se3/Si. With an out-of-plane magnetic field, the MR exhibited weak antilocalization (WAL) behavior with a sharp cusp at the low-field region. With H (magnetic field) applied in the plane, however, the MR cusp feature disappeared completely. Instead, the MR showed a parabolic H2 dependence, which results from the Lorentz deflection of carriers.37 These are characteristic transport features expected in a nonmagnetic TI, which have been reported previously for high quality TI thin films.35,37−40 In sharp contrast to the nonhysteretic MR of Bi2Se3/Si, the longitudinal MR of Bi2Se3/YIG showed hysteresis loops in the low field region distorting the WAL and parabolic MR backgrounds, with the application of perpendicular and inplane field, respectively (Figures 2b-c). In Figure 2b, the hysteresis loop shows a butterfly shape with two separate minima of MR, at Hmin = ± 90 Oe most prominent at 2 K. The magnetic feature in the MR of the TI indicates the presence of the proximity effect at the interface. As the temperature increased, the loop was gradually obscured by the background MR signal and could not be clearly resolved above 25 K (see Supporting Information S5). The reduction of the hysteretic signal is attributed to the increasing transport contribution of the bulk Bi2Se3 channel at higher temperature. This indicates the interfacial origin of the hysteretic loop presumably with C dx.doi.org/10.1021/nl500973k | Nano Lett. XXXX, XXX, XXX−XXX Nano Letters Letter Figure 3. Temperature-dependent magnetization by MOKE measurements. (a,b) Polar and longitudinal MOKE of YIG substrate. (c,d) Polar and longitudinal MOKE of Bi2Se3/YIG heterostructure. All curves are vertically shifted for clarification. The hysteresis loops of panels a and c show different signs, suggesting that the exchange coupling induced spin polarization of Bi2Se3 is opposite to YIG magnetization. The out-of-plane magnetization of Bi2Se3/YIG can be clearly seen up to 130 K (c), indicating the high-temperature magnetic order of TI induced by proximity effect. The longitudinal MOKE of Bi2Se3/YIG can be measured up to 77 K (d), presenting a coupled but distinct switching behavior from that of YIG (b). (e) Schematic of the Bi2Se3/YIG heterostructure in the presence of YIG domain pattern. The red and blue arrows indicate up and down magnetization, respectively. (f) Cross-sectional view of Bi2Se3/YIG interface. Bi2Se3 bottom surface state is gapped through exchange interaction by YIG out-of-plane magnetization at domain regions, while SS remains gapless at the center of domain wall. the basis of reported parameters for YIG at low temperature, as well as micromagnetic simulations, the shape anisotropy is expected to be the dominant term but the magnetocrystalline anisotropy leads to a small out-of-plane tilt of the magnetization and hysteresis in the out-of-plane loops at lower temperatures (see Supporting Information S9 for simulation results). The polar and longitudinal MOKE signals of Bi2Se3/YIG differed from those of YIG in the sign of the MOKE signal as well as the shape of the longitudinal loops. However, the switching fields were the same as those of the YIG films, and hysteresis occurred up to at least 130 K for the polar signal and 77 K for the longitudinal signal (Figure 3c,d). Because of a decreased signal-to-noise ratio at higher temperatures, the polar MOKE of Bi2Se3/YIG does not provide conclusive evidence of survival of the proximity effect above 130 K. The data indicate a contribution to the MOKE signal from the Bi2Se3 layer, suggestive of magnetic order induced by the proximity effect at the Bi2Se3/YIG interface. This could be explained by exchange interaction, that is, hybridization between the p-orbital of Bi2Se3 and d-orbital of Fe in YIG at the interface, which induces a spin polarized state on the Bi2Se3 side. The direction of spin polarization of this state is presumed to be opposite to the YIG magnetization, as suggested by a recent theoretical work on proximity effect between a TI and a ferromagnetic insulator.23 However, the possibility of an optical origin for the reversal of the sign of the polar MOKE cannot be completely ruled out. In order to clarify microscopic origin of the proximity effect, comprehensive theoretical investigation is required. In addition, The sample was mounted in a cryostat and cooled to 4.4 K at zero magnetic field. A laser beam with wavelength of 750 nm was focused either on the YIG or the Bi2Se3/YIG portion of the device and MOKE measurements were performed at increasing temperatures (see Supporting Information S7 for MOKE measurement setup). Figure 3a,b shows both polar and longitudinal MOKE measurements of the YIG. The polar measurements indicated low remanence, and a saturation field that increased with temperature from ∼0.4 kOe at 4.4 K to ∼0.6 kOe at 130 K. However, hysteretic steps were seen in the loops near saturation with a magnitude that increased at low temperature. The longitudinal loops showed a substantial remanence and abrupt switching at ∼2 kOe in the range of 4.4 to 35 K, decreasing at higher temperature. The opposite temperature dependence of saturation field for the polar and longitudinal modes is mainly because the out-of-plane anisotropy is weakened while the in-plane anisotropy is enhanced as temperature increases (see Supporting Information S8). The existence of out-of-plane magnetization in YIG at low temperature is explained below. The net magnetic anisotropy is determined by contributions from shape anisotropy, magnetocrystalline anisotropy, and magnetoelastic anisotropy. The shape anisotropy favors in-plane magnetization whereas the magnetocrystalline anisotropy favors magnetization along the ⟨111⟩ directions at 90 and 35° to the film plane. Magnetoelastic effects are weak but thermal mismatch between the YIG and GGG gives a small out-of-plane magnetoelastic anisotropy. On D dx.doi.org/10.1021/nl500973k | Nano Lett. XXXX, XXX, XXX−XXX Nano Letters Letter Figure 4. The comparison between MOKE and magneto-transport of Bi2Se3/YIG. (a,b) In panel a, regions I and III, the YIG substrate is single domain with magnetic moment aligned with the perpendicular magnetic field. When magnetic field is swept from region I to ∼−90 Oe, as indicated by the dashed line, the domain nucleation is initiated in the YIG substrate, which is illustrated by a sharp increase in the polar MOKE of YIG (a) and correspondingly the sharp drop in the Bi2Se3/YIG (b). (c) WAL background subtracted ΔMR shows a shape increase of resistance at ∼90 Oe, which is influenced by the multidomains in the YIG substrate. (d,e) In-plane magnetization of YIG (d) and Bi2Se3/YIG (e) measured by a longitudinalmode MOKE setup. The peaks in the hysteresis loop may come from the domain nucleation. (f) Parabolic background subtracted MR with in-plane field applied, clearly displaying two resistance states. The switching field is consistent with the MOKE data (d,e). increases with increasing of the positive field, the number of domains decreases and hence the ΔMR of TI is gradually reduced as shown in Figure 4c. Finally, when the field enters region III (∼0.5 kOe), the YIG substrate is saturated in the opposite direction and the MR of the TI rejoins the WAL background. The right panel of Figure 4 illustrates a similar comparison of MOKE and MR measurements of both YIG and TI/YIG with an in-plane magnetic field. Contrary to the out-ofplane case, with in-plane field the TRS is restored at larger magnetic fields (regions I and III) causing a gapless spectrum of the SS. In region II, again, the domain nucleation leads to a rapid increase in the MR owing to the multidomain formation of YIG. Figure 4f presents the resistance after subtraction of the parabolic background. There is no appreciable change in the ΔMR at region II with an in-plane field, which is consistent with the constant magnetization of YIG (Figure 4d) and TI/ YIG (Figure 4e). The presence of domain walls during the reversal process of the YIG can result in the formation of additional chiral mode conducting channel in the TI bottom surface states, as theories predicted,1,3 which will lower the resistance. However, our experimental results show an increase of resistance in region II, suggesting that the chiral modes conduction has a minor contribution here, which could be because the device dimension is much larger than the domain size of YIG estimated in the order of 100 nm. As mentioned before, the smaller AMR ratio (∼0.2%) compared with the ΔMR ratio (∼1%) in Figure 4c,f suggests that the hysteretic ΔMR is not dominated by AMR. In addition to AMR and increased scattering due to spatially nonuniform SS spectrum, another possible effect that could contribute to the increase of MR is the domain wall resistance due to the mistracking of carriers spin and the background magnetization texture.42 The spin mistracking could be especially important for the surface states given their spin-momentum locking characteristic. In order to clarify the possible different mechanisms in the hysteretic MR, further theoretical and experimental work is required. to exclude any contribution from GGG substrate, we performed MOKE measurements for Bi2Se3/GGG control sample. Linear paramagnetic signals were observed for both GGG and Bi2Se3/ GGG (see Supporting Information S10). Breaking the TRS requires an out-of-plane component of the magnetization of the YIG that is readily available in our case owing to the canting of the magnetization. This can introduce a gap in the bottom SS due to proximity-induced magnetic ordering from the exchange coupling between the Bi2Se3 bottom surface and the top surface of YIG. Figure 3e schematically shows the coupling between Bi2Se3 and the perpendicular component of the magnetization of the YIG in which up and down magnetization is indicated by the red and blue arrows, respectively. At fields below saturation, domain walls in the YIG can be formed, as illustrated in Figure 3e. In the middle of the wall, the out-of-plane magnetization vanishes leading to a gapless region of the SS, shown schematically in Figure 3f. To explore the influence of the YIG on the transport of the surface states of the TI, we compare the low-temperature MOKE measurements of both Bi2Se3/YIG and YIG with the corresponding magneto-transport measurements in Figure 4. For sufficiently large negative perpendicular magnetic field (region I in Figure 4a), the YIG substrate is a single domain with perpendicular magnetic moment, creating a gap in the SS at the bottom interface of the TI.41 As the field increases, the perpendicular magnetization component of the YIG and the Bi2Se3/YIG drops abruptly at −90 Oe (Figure 4a,b). At the same field, instead of continuing the WAL behavior, a sharp rise of the longitudinal MR is observed, as shown by subtracting the WAL background in Figure 4c. The multidomain configuration during the reversal process of the YIG may lead to gapped surface states in the domain regions and gapless SS in the domain walls, resulting in a spatially nonuniform spectrum for the surface states.1,3 This may introduce additional scattering for SS carriers that results in an overall increase of resistance in the device. As the perpendicular component of magnetization E dx.doi.org/10.1021/nl500973k | Nano Lett. XXXX, XXX, XXX−XXX Nano Letters Letter (6) Qi, X.-L.; Hughes, T. L.; Zhang, S.-C. Phys. Rev. B 2008, 78 (19), 195424. (7) Zhang, H.; Liu, C.-X.; Qi, X.-L.; Dai, X.; Fang, Z.; Zhang, S.-C. Nat. Phys. 2009, 5 (6), 438−442. (8) Chang, C.-Z.; Zhang, J.; Feng, X.; Shen, J.; Zhang, Z.; Guo, M.; Li, K.; Ou, Y.; Wei, P.; Wang, L.-L.; Ji, Z.-Q.; Feng, Y.; Ji, S.; Chen, X.; Jia, J.; Dai, X.; Fang, Z.; Zhang, S.-C.; He, K.; Wang, Y.; Lu, L.; Ma, X.C.; Xue, Q.-K. Science 2013, 340 (6129), 167−170. (9) Liu, C.-X.; Qi, X.-L.; Dai, X.; Fang, Z.; Zhang, S.-C. Phys. Rev. Lett. 2008, 101 (14), 146802. (10) Yu, R.; Zhang, W.; Zhang, H.-J.; Zhang, S.-C.; Dai, X.; Fang, Z. Science 2010, 329 (5987), 61−64. (11) Tse, W.-K.; MacDonald, A. H. Phys. Rev. Lett. 2010, 105 (5), 057401. (12) Tserkovnyak, Y.; Loss, D. Phys. Rev. Lett. 2012, 108 (18), 187201. (13) Qi, X.-L.; Li, R.; Zang, J.; Zhang, S.-C. Science 2009, 323 (5918), 1184−1187. (14) Chen, Y. L.; Chu, J.-H.; Analytis, J. G.; Liu, Z. K.; Igarashi, K.; Kuo, H.-H.; Qi, X. L.; Mo, S. K.; Moore, R. G.; Lu, D. H.; Hashimoto, M.; Sasagawa, T.; Zhang, S. C.; Fisher, I. R.; Hussain, Z.; Shen, Z. X. Science 2010, 329 (5992), 659−662. (15) Kou, X.; He, L.; Lang, M.; Fan, Y.; Wong, K.; Jiang, Y.; Nie, T.; Jiang, W.; Upadhyaya, P.; Xing, Z.; Wang, Y.; Xiu, F.; Schwartz, R. N.; Wang, K. L. Nano Lett. 2013, 13 (10), 4587−4593. (16) Kou, X.; Lang, M.; Fan, Y.; Jiang, Y.; Nie, T.; Zhang, J.; Jiang, W.; Wang, Y.; Yao, Y.; He, L.; Wang, K. L. ACS Nano 2013, 7 (10), 9205−9212. (17) Hor, Y. S.; Roushan, P.; Beidenkopf, H.; Seo, J.; Qu, D.; Checkelsky, J. G.; Wray, L. A.; Hsieh, D.; Xia, Y.; Xu, S. Y.; Qian, D.; Hasan, M. Z.; Ong, N. P.; Yazdani, A.; Cava, R. J. Phys. Rev. B 2010, 81 (19), 195203. (18) Liu, M.; Zhang, J.; Chang, C.-Z.; Zhang, Z.; Feng, X.; Li, K.; He, K.; Wang, L.-l.; Chen, X.; Dai, X.; Fang, Z.; Xue, Q.-K.; Ma, X.; Wang, Y. Phys. Rev. Lett. 2012, 108 (3), 036805. (19) Chang, C.-Z.; Zhang, J.; Liu, M.; Zhang, Z.; Feng, X.; Li, K.; Wang, L.-L.; Chen, X.; Dai, X.; Fang, Z.; Qi, X.-L.; Zhang, S.-C.; Wang, Y.; He, K.; Ma, X.-C.; Xue, Q.-K. Adv. Mater. 2013, 25 (7), 1065− 1070. (20) Kou, X. F.; Jiang, W. J.; Lang, M. R.; Xiu, F. X.; He, L.; Wang, Y.; Yu, X. X.; Fedorov, A. V.; Zhang, P.; Wang, K. L. J. Appl. Phys. 2012, 112 (6), 063912−6. (21) Zhang, D.; Richardella, A.; Rench, D. W.; Xu, S.-Y.; Kandala, A.; Flanagan, T. C.; Beidenkopf, H.; Yeats, A. L.; Buckley, B. B.; Klimov, P. V.; Awschalom, D. D.; Yazdani, A.; Schiffer, P.; Hasan, M. Z.; Samarth, N. Phys. Rev. B 2012, 86 (20), 205127. (22) Luo, W.; Qi, X.-L. Phys. Rev. B 2013, 87 (8), 085431. (23) Eremeev, S. V.; Men’shov, V. N.; Tugushev, V. V.; Echenique, P. M.; Chulkov, E. V. Phys. Rev. B 2013, 88 (14), 144430. (24) Vobornik, I.; Manju, U.; Fujii, J.; Borgatti, F.; Torelli, P.; Krizmancic, D.; Hor, Y. S.; Cava, R. J.; Panaccione, G. Nano Lett. 2011, 11 (10), 4079−4082. (25) Wei, P.; Katmis, F.; Assaf, B. A.; Steinberg, H.; Jarillo-Herrero, P.; Heiman, D.; Moodera, J. S. Phys. Rev. Lett. 2013, 110 (18), 186807. (26) Yang, Q. I.; Dolev, M.; Zhang, L.; Zhao, J.; Fried, A. D.; Schemm, E.; Liu, M.; Palevski, A.; Marshall, A. F.; Risbud, S. H.; Kapitulnik, A. Phys. Rev. B 2013, 88 (8), 081407. (27) Uchida, K.-i.; Adachi, H.; Ota, T.; Nakayama, H.; Maekawa, S.; Saitoh, E. Appl. Phys. Lett. 2010, 97 (17), 172505−3. (28) Jiang, W.; Upadhyaya, P.; Fan, Y.; Zhao, J.; Wang, M.; Chang, L.-T.; Lang, M.; Wong, K. L.; Lewis, M.; Lin, Y.-T.; Tang, J.; Cherepov, S.; Zhou, X.; Tserkovnyak, Y.; Schwartz, R. N.; Wang, K. L. Phys. Rev. Lett. 2013, 110 (17), 177202. (29) Serga, A. A.; Chumak, A. V.; Hillebrands, B. J. Phys. D: Appl. Phys. 2010, 43 (26), 264002. (30) Kelly, O. d. A.; Anane, A.; Bernard, R.; Youssef, J. B.; Hahn, C.; Molpeceres, A. H.; Carretero, C.; Jacquet, E.; Deranlot, C.; Bortolotti, P.; Lebourgeois, R.; Mage, J. C.; de Loubens, G.; Klein, O.; Cros, V.; Fert, A. Appl. Phys. Lett. 2013, 103 (8), 082408−4. In summary, a Bi2Se3/YIG heterostructure has been used to cleanly probe the surface magnetic behavior of Bi2Se3 due to the magnetic proximity effect. Our measured butterfly and square-shaped magnetoresistance loops provided direct evidence of magnetic coupling between the insulating YIG and Bi2Se3. More significantly, for the first time magnetic ordering at the Bi2Se3/YIG interface has been demonstrated at temperatures up to at least 130 K by MOKE measurements. The engineering of a TI and FMI heterostructure will open up numerous opportunities to study high temperature TI-based spintronic devices, in which the TI is controlled by breaking the TRS using a FMI with perpendicular magnetization component. A YIG film with out-of-plane anisotropy at >300 K could potentially manipulate the magnetic properties of a TI may even above room temperature. ■ ASSOCIATED CONTENT S Supporting Information * Sample growth and TEM characterization methods, device fabrication and characterization, EDX mapping of Bi2Se3/YIG/ GGG heterostructure, polar and longitudinal MOKE measurement setup, comprehensive temperature dependence of polar and longitudinal MOKE, MOKE measurements of GGG and Bi2Se3/GGG, magnetic anisotropy of YIG/GGG, temperature dependence of MR ratio and Hs in magnetoresistance, temperature dependence of longitudinal resistance and carrier density, and magneto-transport data of other TI/YIG samples. This material is available free of charge via the Internet at http://pubs.acs.org. ■ AUTHOR INFORMATION Corresponding Authors *E-mail: langmr@ucla.edu. *E-mail: wang@seas.ucla.edu. *E-mail: liang.heliang@gmail.com. Author Contributions ll M.L. and M.M. contributed equally to this work. Notes The authors declare no competing financial interest. ACKNOWLEDGMENTS This work was supported in part by the FAME Center, one of six centers of STARnet, a Semiconductor Research Corporation program sponsored by MARCO and DARPA. The authors also acknowledge the support from the Western Institute for Nanoelectronics (WIN) and the DARPA Meso program under Contract Nos. N66001-12-1-4034 and N66001-11-1-4105. Y.W. acknowledges the support of National Science Foundation of China (11174244, 51390474), the National 973 Program of China (2013CB934600), Zhejiang Provincial Natural Science Foundation of China (LR12A04002), National Young 1000 Talents Program of China, and the Ministry of Education (20120101110087). M.L. and M.M. acknowledge technical support from L. Chang and K. Murata and helpful discussions with I. Ovchinnikov and P. Khalili. ■ ■ REFERENCES (1) Qi, X.-L.; Zhang, S.-C. Rev. Mod. Phys. 2011, 83 (4), 1057−1110. (2) Moore, J. E. Nature 2010, 464 (7286), 194−198. (3) Hasan, M. Z.; Kane, C. L. Rev. Mod. Phys. 2010, 82 (4), 3045. (4) Kane, C. L.; Mele, E. J. Phys. Rev. Lett. 2005, 95 (14), 146802. (5) Moore, J. E.; Balents, L. Phys. Rev. B 2007, 75 (12), 121306. F dx.doi.org/10.1021/nl500973k | Nano Lett. XXXX, XXX, XXX−XXX Nano Letters Letter (31) Goto, T.; Onbasli, M. C.; Ross, C. A. Opt. Express 2012, 20 (27), 28507−28517. (32) Sun, Y.; Song, Y.-Y.; Chang, H.; Kabatek, M.; Jantz, M.; Schneider, W.; Wu, M.; Schultheiss, H.; Hoffmann, A. Appl. Phys. Lett. 2012, 101 (15), 152405−5. (33) Lang, M.; He, L.; Kou, X.; Upadhyaya, P.; Fan, Y.; Chu, H.; Jiang, Y.; Bardarson, J. H.; Jiang, W.; Choi, E. S.; Wang, Y.; Yeh, N.-C.; Moore, J.; Wang, K. L. Nano Lett. 2012, 13 (1), 48−53. (34) Lang, M.; He, L.; Xiu, F.; Yu, X.; Tang, J.; Wang, Y.; Kou, X.; Jiang, W.; Fedorov, A. V.; Wang, K. L. ACS Nano 2011, 6 (1), 295− 302. (35) Kim, Y. S.; Brahlek, M.; Bansal, N.; Edrey, E.; Kapilevich, G. A.; Iida, K.; Tanimura, M.; Horibe, Y.; Cheong, S.-W.; Oh, S. Phys. Rev. B 2011, 84 (7), 073109. (36) Jiang, Y.; Wang, Y.; Sagendorf, J.; West, D.; Kou, X.; Wei, X.; He, L.; Wang, K. L.; Zhang, S.; Zhang, Z. Nano Lett. 2013, 13 (6), 2851−2856. (37) He, H.-T.; Wang, G.; Zhang, T.; Sou, I.-K.; Wong, G. K. L.; Wang, J.-N.; Lu, H.-Z.; Shen, S.-Q.; Zhang, F.-C. Phys. Rev. Lett. 2011, 106 (16), 166805. (38) Chen, J.; Qin, H. J.; Yang, F.; Liu, J.; Guan, T.; Qu, F. M.; Zhang, G. H.; Shi, J. R.; Xie, X. C.; Yang, C. L.; Wu, K. H.; Li, Y. Q.; Lu, L. Phys. Rev. Lett. 2010, 105 (17), 176602. (39) Kong, D.; Chen, Y.; Cha, J. J.; Zhang, Q.; Analytis, J. G.; Lai, K.; Liu, Z.; Hong, S. S.; Koski, K. J.; Mo, S.-K.; Hussain, Z.; Fisher, I. R.; Shen, Z.-X.; Cui, Y. Nat. Nanotechnology 2011, 6 (11), 705−709. (40) Wang, J.; DaSilva, A. M.; Chang, C.-Z.; He, K.; Jain, J. K.; Samarth, N.; Ma, X.-C.; Xue, Q.-K.; Chan, M. H. W. Phys. Rev. B 2011, 83 (24), 245438. (41) Wray, L. A.; Xu, S.-Y.; Xia, Y.; Hsieh, D.; Fedorov, A. V.; Hor, Y. S.; Cava, R. J.; Bansil, A.; Lin, H.; Hasan, M. Z. Nat. Phys. 2011, 7 (1), 32−37. (42) Levy, P. M.; Zhang, S. Phys. Rev. Lett. 1997, 79 (25), 5110− 5113. G dx.doi.org/10.1021/nl500973k | Nano Lett. XXXX, XXX, XXX−XXX