Interfacial tuning of chiral magnetic interactions for large topological Hall effects in LaMnO3/SrIrO3 heterostructures

See allHide authors and affiliations

Science Advances  03 Jul 2020:
Vol. 6, no. 27, eaaz3902
DOI: 10.1126/sciadv.aaz3902


Chiral interactions in magnetic systems can give rise to rich physics manifested, for example, as nontrivial spin textures. The foremost interaction responsible for chiral magnetism is the Dzyaloshinskii-Moriya interaction (DMI), resulting from inversion symmetry breaking in the presence of strong spin-orbit coupling. However, the atomistic origin of DMIs and their relationship to emergent electrodynamic phenomena, such as topological Hall effect (THE), remain unclear. Here, we investigate the role of interfacial DMIs in 3d–5d transition metal-oxide-based LaMnO3/SrIrO3 superlattices on THE from a chiral spin texture. By additively engineering the interfacial inversion symmetry with atomic-scale precision, we directly link the competition between interfacial collinear ferromagnetic interactions and DMIs to an enhanced THE. The ability to control the DMI and resulting THE points to a pathway for harnessing interfacial structures to maximize the density of chiral spin textures useful for developing high-density information storage and quantum magnets for quantum information science.


Magnetic chirality is fundamentally linked to broken inversion symmetry, but it is rarely found in nature because only a few bulk materials have crystal structures that violate this symmetry (1, 2). By contrast, inversion symmetry can be readily broken at interfaces within epitaxial heterostructures and superlattices. Moreover, inversion symmetry breaking combined with strong spin-orbit coupling (SOC) can lead to a large chiral Dzyaloshinskii-Moriya interaction (DMI), HDMI = Dij · (Si × Sj), which imparts a handedness to the magnetic exchange interaction between adjacent spins Si and Sj with the strength and direction expressed by the DMI vector Dij (3, 4). At a thin-film interface, Dij lies in the plane and drives the formation of chiral Néel-type domain walls (5) or spin textures like the magnetic skyrmion, as depicted in Fig. 1 (A and B) (6). Furthermore, the Berry phase associated with the interaction between charge carriers and a chiral spin texture i.e., with scalar spin chirality χijk = Si · Sj × Sk gives rise to an emergent magnetic field and topological Hall effect (THE) (7, 8) that links the local magnetic spin texture to an electrical response. Altogether, these effects are promising phenomena that must be understood for future devices based on the electrical detection and manipulation of magnetic information (7, 919).

Fig. 1 Interfacial engineering of the DMI to control chiral magnetism in oxide superlattices.

(A) A schematic illustrating an effective DMI stabilized by chiral spin textures from multiple interfaces (Deff). (B) A schematic of the THE (ρTHE) that results from the Berry phase accumulated by a charge carrier traversing a chiral spin texture. The latter acts as an emergent effective magnetic field (Beff) in real space. (C and D) Control of the inversion symmetry in LaMnO3/SrIrO3 superlattices by artificially modifying the interfacial termination. Conventional growth of a superlattice with a BO2-type layer termination creates different top and bottom interface structures (D12), as illustrated in (D). We simply differentiate the interface structure according to the number of connecting SrO atomic layers [i.e., D1 indicates the stacking sequence (moving upward) LaO-MnO2-LaO]. (E) DMI can be controlled within an ABO3/A′B′O3 perovskite heterostructure by engineering the interfacial layer structures by inserting SrO layers during the film growth, thus controlling inversion symmetry. Interfaces can be inversion symmetric, where the top interface is D2 and the bottom interface is the mirror −D2 as in superlattice D22, or inversion symmetry can be broken when the interface structures are different, as shown for superlattice D13 (combining D1 and D3).

The discovery that DMI can be tuned using different material combinations, structures, and film architectures has led to an explosion of interest in the study of thin films with 3d ferromagnet and heavy (i.e., 4d and 5d) element layers. In all-metal multilayers, skyrmion phases were observed within a range of temperatures and applied fields that extended well beyond the narrow phase stability regions that were first found in bulk noncentrosymmetric systems (2, 5, 14, 15, 18, 2022). Most recently, magnetic superlattices and multilayers with cooperative DMI from multiple interfaces enabled the study of current-induced skyrmion motion and stability up to room temperature (14, 15, 23, 24). Despite remarkable recent advancements, it also remains a challenge to control chiral magnetic interactions to meet the size and stability requirements of spin textures for practical applications in high-density memory devices, and there is a lack of understanding of the connection between electronic transport phenomena and chiral magnetism (7, 11, 19, 25).

Recently, it was found that the addition of oxide layers into metal multilayers could strongly enhance the strength of DMIs through interfacial electrostatics and charge transfer (26), while strong electronic correlations could enable exceptionally large THEs in oxide thin films (27). In particular, owing to the large SOC of the 5d element Ir and the layer-by-layer stacking and atomic control of growth achievable in epitaxial oxides, SrIrO3-based heterostructures provide an excellent avenue to study the origin of interfacial DMI and the associated Hall response (2831).

Here, we show the emergence of a highly robust chiral magnetic phase exhibiting a large THE in LaMnO3/SrIrO3 superlattices. By controlling the film architecture and interface structures, we determine that the THE originates from DMI created at the 3d-5d interface. A key observation is our ability to modulate the THE by nearly an order of magnitude through engineering of the interface symmetry with the nonmagnetic A-site layers. We discuss the relationship between THE and the role of atomic interface structure to drive the competition between interfacial collinear ferromagnetic (FM) interactions and chiral DMI as a strategy to miniaturize chiral spin textures emerging from interfaces.


Epitaxial [(LaMnO3)n/(SrIrO3)n]m superlattices were grown on (001) SrTiO3 substrates using pulsed laser deposition (PLD). Superlattices with n = 1 to 12 unit cells (u.c.) were created with the layer repetition (m) chosen to create a total thickness of ∼50 nm. X-ray diffraction scans (fig. S1) showed that all samples were phase pure and coherently strained. The natural AO-BO2-AO-BO2 layered structure of the perovskite ABO3 shown in Fig. 1C (i.e., A = La or Sr, B = Mn or Ir, respectively, for LaMnO3 and SrIrO3) can enable systematic control over inversion symmetry of the structure by carefully tuning the layer stacking at individual interfaces. This control of the interface structure is illustrated in Fig. 1 (D and E) for three examples of [(LaMnO3)n/(SrIrO3)n]m superlattices with n and m fixed, which are either inversion symmetric or asymmetric depending on the arrangement of the interfacial A-site layers. The atomic-scale control of PLD enables each of these structures to be systematically constructed, as discussed in Materials and Methods, and verified using high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) imaging as shown in fig. S2. This interfacial control is of central importance to identifying the competing effects between interfacial magnetic coupling and DMI.

The impact of interfacial coupling between the 3d and 5d oxide layers is reflected in the variation of field- and temperature-dependent magnetic properties with the layer thickness n (figs. S3 and S4). Compared to a single LaMnO3 layer that is FM when grown using PLD (32), LaMnO3/SrIrO3 superlattices with n = 1 to 3 u.c. had a lower Curie temperature (TC) and saturation magnetization (MS) and a larger magnetic coercivity (HC). We found a nonmonotonic relationship between MS and n, with the largest MS observed when n = 5. This result is consistent with the overall magnetization depending on bulk and interface magnetizations with contributions from Ir and Mn, as observed in other manganite/iridate systems (30, 33, 34) (see further consideration of the interface magnetism in Discussion). Furthermore, interfacial octahedral rotations influence the B-O-B bond angle, and interfacial charge transfer and redistribution alter the orbital occupancy, both of which modify the interfacial exchange interactions and magnetization in perovskite heterostructures (3537). These phenomena, which act typically on short (0.4 to 2 nm) length scales (35), may account for the nonmonotonic relationship between layer thickness and MS in LaMnO3/SrIrO3, as has been observed for other manganite superlattices (38). An increase in TC with increasing n up to ∼150 K and a decrease in HC were observed for n = 5 to 12 u.c. All films had an easy-plane magnetic anisotropy. However, the difference in HC measured for H applied in-plane and out-of-plane was sizeable only for n ≤ 3 u.c. (fig. S4C). The large HC enhancement for n ≤ 3 u.c. resembles that of SrMnO3/SrIrO3 and LaxSr1−xMnO3/SrIrO3 superlattices with magnetism dominated by charge transfer–driven interfacial collinear FM interactions (30, 33). By contrast, the magnetism of superlattices with large LaMnO3 thicknesses (n = 12 u.c.) was similar to that of single LaMnO3 layers. Thus, the overall dependence on layer thickness suggests a competition between the interfacial interactions and intrinsic ferromagnetism of LaMnO3.

To understand the influence of the interface-induced emergent magnetism on electronic transport, we measured the Hall resistivity (ρxy) as a function of H at 10 K, which is well below TC for all superlattices. The Hall resistivity of a typical collinear ferromagnet is described by the relationship ρxy = ρOHE + ρAHE = R0H + RSM, where the ordinary (ρOHE) and anomalous (ρAHE) Hall resistivities are linear responses of the applied field (H) and magnetization (M), with respective scale factors R0 and RS (8). As shown in Fig. 2A, for short- or long-period superlattices (i.e., n = 3 and 12 u.c., respectively), the Hall resistivity is satisfactorily described by the combination of an ordinary Hall effect (OHE) and anomalous Hall effect (AHE), indicating a collinear FM state. For example, the Hall effect of n = 3 is similar to that of SrMnO3/SrIrO3 superlattices that are dominated by collinear interfacial FM interactions (30). This result is also consistent with the known minimum critical thickness rule for the heavy layer, below which any DMI may be too weak to establish chiral domain walls (39). Furthermore, the FM state becomes energetically favorable even in the presence of DMI when the magnetic anisotropy is large (40), consistent with FM coinciding with an HC enhancement for n ≤ 3 LaMnO3/SrIrO3 superlattices. For long-period superlattices (i.e., n = 12 u.c.), the AHE due to ferromagnetism can be understood as a dominating effect of the intrinsic ferromagnetism of the thick layer over a diminishing (i.e., volume averaged) effect of interfacial DMI (41).

Fig. 2 The Hall effect in LaMnO3/SrIrO3 superlattices.

(A) The Hall effect ρxy measured at 10 K and shown for superlattices n = 3 to 12 u.c. D12-type superlattices. (B) The magnetization measured for the n = 5 u.c. superlattice, which is similar for all n on the scale shown. (C) Temperature dependence of the Hall resistivity (ρxy) for the n = 5 u.c. superlattice measured in fields up to 30 T. (D) Determination of the topological Hall resistivity ρTHE of n = 5 u.c. at several temperatures obtained for the decreasing (+H to −H) applied magnetic field measurements. (E) The topological Hall resistivity (ρTHE) shown as functions of temperature and applied field. (F) Temperature dependence of the maximum value of ρTHE taken from (D) and (E). Data in (A) to (D) are shown with a vertical offset for clarity.

For intermediate n = 4 to 6 u.c. superlattices, however, a strongly enhanced Hall response with a dramatically different behavior emerges. In particular, a large nonmonotonic dependence on the magnetic field and a clear hysteretic behavior were observed in magnetic fields as large as 9 T, the maximum magnetic field used in Fig. 2A. Whereas the magnetization saturates for H > 2 T (Fig. 2B and fig. S3), the hysteresis observed in the Hall resistivity does not follow the net M or H. We noted that while multiple decoupled FM layers with opposite signs of AHE may induce field-dependent anomalies in the Hall resistivity (42, 43), we observed no features in the magnetization that indicate the distinct formation of two FM layers in LaMnO3/SrIrO3. This additional Hall component for n = 4 to 6 u.c. superlattices can be attributed to a chiral magnetic phase induced by interfacial DMI leading to a THE (discussed in further detail below) (1, 7). This is described by an additional term, ρTHE, in the Hall effect (i.e., ρxy = ρOHE + ρAHE + ρTHE). To understand the origin of the chiral magnetism and its relationship to the observed THE, we further investigated the temperature-dependent Hall effect with the n = 5 u.c. superlattice in magnetic fields up to 30 T. At T = 300 K, well above TC, a clear linear response was observed, as shown in Fig. 2C, which is indicative of the OHE with n-type behavior. For temperatures below TC ~ 150 K, an AHE due to collinear FM develops with RS similar in magnitude to the short-period n = 3 u.c. superlattice. Below 50 K, the strong enhancement in ρxy was observed as a pronounced curvature, followed by the onset of hysteresis in large H.

The THE was isolated from ρOHE and ρAHE using the standard approach. The OHE was taken to be the slope of ρxy at 30 T, and the AHE was determined by extrapolating the measured magnetization to 30 T to intersect the linear H region of ρxy, as shown in Fig. 2D. The resulting THE is shown as a function of T and H in Fig. 2E, which was measured on sweeping H from +30 to −30 T. The circles and dotted line provide an estimate of the phase stability region with |ρxy| < 1 μΩ⋅cm, which is wide compared to previous reports for systems with interfacial DMI. The maximum THE at each temperature shown in Fig. 2F is quite large compared to layered oxides or bulk materials with strong DMI that are typically on the order of a few nΩ⋅cm to 1 μΩ⋅cm. Recent theoretical investigations have also identified the possibility of strong DMI and THE originating at the iridate/manganite interface, consistent with our experimental observations [see fig. S5, section S2, and (44)]. These results point to a highly robust chiral phase stabilized in the intermediate-period LaMnO3/SrIrO3 superlattices with n = 4 to 6 u.c.

The relationships among inversion symmetry, DMI, and the THE were further tested by examining n = 5 u.c. superlattices grown with the different interface layer structures D12, D22, and D13 that are shown schematically in Fig. 1 (D and E). The hypothesis is that the extent of inversion symmetry breaking with respect to the LaMnO3 layer should be reflected in the strength of the total DMI that is created additively from multiple interfaces in the superlattice. Because of the THE results from the interaction between charge carriers and a spatially varying magnetic spin texture, the magnitude of the THE depends on the winding number and density of the chiral spin features that establish the total topological charge (nt) of the system (7, 8). For Néel-type skyrmions, in the adiabatic limit of the Berry phase description, ρTHE is proportional to the skyrmion density, nsk = nt, and inversely proportional to the skyrmion size, Dsk ~ nsk−1/2, for a close packed arrangement of the spin texture (2, 20, 45). Note that the skyrmion density and thus THE is determined by the competition between DMI, FM exchange, and magnetic anisotropy energy terms governing the magnetic system. For our LaMnO3/SrIrO3 superlattices with different A-site layer structures, the Hall effect is shown in Fig. 3 (A and B). We observe that the amplitude of ρTHE measured at T = 10 K for D13 is ~75 μΩ·cm, which is near an order of magnitude larger than D22 and a factor of 3 larger than D12. The enhanced THE is independent of the conventional magnetic properties (e.g., TC, HC, or MS), which is consistent with the THE being directly linked to the interface structure and DMI. We note that the global inversion symmetry breaking due to the surface and substrate are unavoidable and that other magnetic interactions existing in the superlattice (e.g., dipolar interactions) may help to stabilize chiral structures even with weakened DMI that may contribute to a persistent THE in even the most symmetric superlattice structure. Relative to the naturally occurring superlattice D12, we attribute the suppressed ρTHE of D22 to the symmetric interface structure creating a weakened total DMI, and we attribute the large enhancement of ρTHE in D13 to the largest total DMI from fully different top and bottom interface layer structures.

Fig. 3 Tunable THE in LaMnO3/SrIrO3 superlattices.

(A) Hall resistivity after subtracting the OHE measured at 10 K for n = 5 u.c. superlattices with interface types D22, D12, and D13 shown in Fig. 1D. (B) The maximum value of (left axis) ρTHE and (right axis) relative spin feature size (Δnt−1/2) measured at 10 K. (C) An illustration of the relative change in density of topological charges (nt) obtained from the observed variation in ρTHE that demonstrates a miniaturization of the spin feature size (Δnt−1/2) by tuning the interfacial magnetic interactions using highly inversion-asymmetric interfacial structures.

Because the enhanced Hall effect created by the controlled film architecture and interface structures of our 3d-5d system is consistent with effects of interfacial DMI, we consider the relationship between DMI, THE, and real-space spin texture. In thin films, the interface geometry necessarily sets the spin rotation direction induced by DMI to favor chiral Néel-type domain walls (5). For this reason, although we do not have a direct image of a chiral spin texture in LaMnO3/SrIrO3, we propose that the LaMnO3/SrIrO3 system may host Néel-type skyrmions or chiral domain walls, while we strongly believe that our observation of the THE indicates the potential formation of skyrmions (see the Supplementary Materials for more details on skyrmion confirmation). Although establishing a quantitative description of skyrmion textures directly from Hall effects remains a challenge for complex systems with multiple magnetic layers, conduction pathways, or possible nonadiabatic effects (7, 8, 18, 25, 27, 46, 47), a qualitative proportionality between the magnitude of the THE and nt appears robust (7, 20, 47). Consequently, an increase in ρTHE signifies a denser array of chiral features and hence a smaller feature size, even when the absolute feature size is difficult to establish. Thus, we illustrate the impact of different interfacial A-site layer structures of the n = 5 u.c. LaMnO3/SrIrO3 superlattice to a possible relative change in chiral spin feature density Δnt = ρTHE/[ρTHE(D22)] and feature size (Δnt−1/2) in Fig. 3 (B and C) (7) (section S2). This indicates that engineering the nonmagnetic interfacial atomic layers is an effective strategy for tuning the chiral interactions and inversion symmetry to ultimately shrink the size of chiral spin textures originating from interfaces. From the topological Hall resistivity, we estimate that the size of potential skyrmions can be as small as 2 nm (see section S2 for further details).

To gain deeper insight into the magnetic interaction energy terms that govern magnetic domain formation in LaMnO3/SrIrO3 superlattices, we examined the temperature dependence of the spontaneous (i.e., zero field) domain structure using photoemission electron microscopy (PEEM) combined with Mn L3-edge x-ray magnetic circular dichroism (XMCD). Because of the grazing incidence geometry of the incoming x-rays, XMCD-PEEM provides a spatial map of the dominant in-plane projection of the Mn magnetization. Images of an n = 5 u.c. superlattice measured while cooling from 100 to 32 K are shown in Fig. 4A. These images reveal a dense domain structure formed at temperatures near the onset T ~ 90 K of the THE. The LaMnO3/SrIrO3 superlattice showed essentially no fluctuations in shape or domain coalescence upon decreasing temperature; this contrasts with the typical behavior of collinear FM phases (see section S4 and fig. S6 for further analysis and interpretation of PEEM images). This result suggests a highly energetically favorable domain structure formed in LaMnO3/SrIrO3 that is strongly affected by an energy term that is not typical of FM systems but that is consistent with strong DMI.

Fig. 4 Domain structure and element-specific magnetism of a LaMnO3/SrIrO3 superlattice.

(A) XMCD-PEEM images at various temperatures for an n = 5 u.c. D12 superlattice and (B) XAS and (C) XMCD measured at 10 K and 5 T for the Mn and Ir L-edges for n = 5 u.c.

Because XMCD-PEEM allows the direct imaging of nanoscale magnetic features, it is well suited to identify chiral domain wall and skyrmion textures (16). To examine the impact of DMI on the domain structure in LaMnO3/SrIrO3 superlattices, we performed micromagnetic simulations of the spontaneous domain structure of a simple manganite layer with interfacial DMI using Object Oriented Micromagnetic Framework (48). We found that the structures observed with XMCD-PEEM closely resemble the mixed stripe and bubble features that can evolve into a skyrmion phase in a sizable applied magnetic field and which are only observed when a DMI term is included in the magnetic simulation parameters (see section S3 and figs. S5 to S7 for detailed comparison of XMCD-PEEM images and micromagnetic simulations).

To examine the element-specific electronic and magnetic details to understand the origin of the magnetism of the LaMnO3/SrIrO3 system, we conducted x-ray absorption spectroscopy (XAS) and XMCD measurements of Mn and Ir L-edges of n = 5 u.c. LaMnO3/SrIrO3 superlattice, as shown in Fig. 4 (B and C). The magnetism originates predominantly from the LaMnO3 layer, as revealed by a large Mn XMCD signal, consistent with FM LaMnO3 films grown by PLD (32). We observed null L3-edge and weak L2-edge Ir XMCD signal that indicates a small Ir magnetization in the n = 5 u.c. LaMnO3/SrIrO3 superlattice. By comparison, the Ir and Mn XMCD of short-period SrMnO3/SrIrO3 and La2/3Sr1/3MnO3/SrIrO3 superlattices with strong interfacial FM interactions showed clear XMCD at both the L3- and L2-edges (30, 33), indicating a much larger Ir magnetization in those systems. Our lack of observed L3-edge XMCD indicates change in the element-specific magnetism, with a possible canting of the Ir moment away from the applied field direction for the n = 5 LaMnO3/SrIrO3 superlattice, compared to previously studied FM iridate/manganite systems. These observations are consistent with the Ir-Mn interfacial coupling modified by weakened collinear interfacial interactions and strong DMI in the n = 5 u.c. LaMnO3/SrIrO3 superlattice versus a predominance of strong collinear interfacial exchange interactions in short-period (n = 1 to 3 u.c.) superlattices that qualitatively agrees with the layer thickness–dependent onset of THE in LaMnO3/SrIrO3 superlattices.


Because the DMI is determined by magnetic interactions within a single atomic layer of the interface, local interface structures are expected to have a critical role in the emergence of THEs. To understand the mechanism responsible for the THE of LaMnO3/SrIrO3 superlattices, we consider the importance of interfacial chemistry and symmetry that drive the Mn valence and exchange interactions that govern the magnetism of artificially engineered manganite superlattices (36, 37). Specifically, because of charge leakage that is effective over 1 to 2 u.c., short-period (n ≤ 3 u.c.) [(LaMnO3)2n/(SrMnO3)n] superlattices display double exchange–driven ferromagnetism and conductivity that are similar to those of the La0.66Sr0.33MnO3 alloy instead of insulating antiferromagnetism intrinsic to the constituent materials. We propose that the presence of both interfacial LaO and SrO layers can promote a mixed Mn valence and a similar charge leakage in short-period LaMnO3/SrIrO3 superlattices. This phenomenon can lead to a strengthened interfacial FM double exchange that directly competes with DMI and thus result in the appearance of an AHE due to an FM ground state in our n ≤ 3 u.c. LaMnO3/SrIrO3 superlattices. This behavior is in agreement with conventional understanding that the stabilization of chiral phases depends critically on the competition between DMI, collinear FM interactions, and magnetic anisotropy of a system (19, 39, 40). It is also likely that the distinct local Mn environment and magnetism at the polar-nonpolar interface bridged by LaO and SrO layers differ from the uniform LaO-MnO2-LaO environment of the bulk. This difference in interface and bulk environments likely contributes to the strong influence of interface versus intrinsic properties in the intermediate n systems, such as nonmonotonic n and THE dependence. Furthermore, our ability to modulate the magnitude of the THE for the n = 5 u.c. superlattice reveals that control of the interface structure driven by interfacial A-site layers can be a highly effective method for tuning DMI.

In summary, the interfacial inversion symmetry of 3d-5d LaMnO3/SrIrO3 superlattices is deliberately controlled to modify the competition between chiral DMI and intrinsic collinear ferromagnetism. This interfacial symmetry control led to a large THE, which originates from a highly robust chiral magnetic phase, potentially hosting skyrmions. Our findings shed light on development of novel pathways to stabilize ultrahigh-density chiral spin textures important for developing topologically protected quantum magnets.


Sample synthesis and structural characterization

All samples were synthesized by pulsed laser epitaxy with a substrate temperature, oxygen partial pressure, and laser fluence of 700°C, 100 mTorr, and 1.0 J/cm2, respectively, on atomically flat TiO2-terminated (001) SrTiO3 substrates with a KrF eximer laser (λ = 248 nm). The crystal structure, phase purity, and orientation of these films and superlattices were confirmed by x-ray reflectivity and diffraction measurements using a four-circle x-ray diffractometer. Superlattices with n = 1 to 12 u.c. thick layers were prepared using a typical two-target growth alternating LaMnO3 and SrIrO3 to produce the structure type D12. For superlattices with n = 5 u.c., two additional superlattice structures were created by controlling the interface growth by replacing either the bottom or top 1 u.c. of LaMnO3 with SrMnO3; this resulted in the structures D22 or D13, respectively.

HAADF-STEM imaging was carried out in Nion UltraSTEM200 operated at 200 kV. The microscope is equipped with a cold field-emission gun and an aberration corrector for subangstrom resolution. An inner angle of 78 mrad was used for HAADF imaging. The convergence semiangle for the electron probe was set to 30 mrad.

Physical properties characterization

The magnetization was determined with a 7 T Quantum Design MPMS3, measured using conventional techniques from hysteresis loop measurements at 10 K and up to 7 T, after subtracting a linear background to correct for the diamagnetic response of the SrTiO3 substrate. The electronic transport measurements were performed with a 14 T Quantum Design Physical Property Measurement System with custom electronics. The electrical contacts were made by ultrasonic soldering Au wires with In solder in a van der Pauw configuration. Magnetotransport properties of thin films under high magnetic field up to 30 T were measured at the National High Magnetic Field Laboratory (NHMFL) in Tallahassee.

X-ray absorption experiments

X-ray absorption, magnetic circular dichroism, and linear dichroism were measured at the Advanced Photon Source at Argonne National Laboratory at beamlines 4-ID-C and 4-ID-D. Measurements of Mn L2,3-edge were done in total electron yield and fluorescence yield modes at a grazing angle to the sample surface. The XMCD at 10 K was obtained from the average of ±5 T magnetic fields applied in the film plane to correct for possible nonmagnetic artifacts. All spectra were normalized to a maximum XAS intensity at the L3-edge to facilitate direct comparison between the Ir and Mn L-edges. Variable-temperature XMCD-PEEM images were measured at the Advanced Light Source at Lawrence Berkeley National Laboratory at sector 11.0.1 using the PEEM-3 endstation. Images were collected on films grown on a conducting Nb-doped (0.05%) SrTiO3 substrate and were measured using the Mn L3-edge XMCD in zero applied magnetic field at a grazing angle of 30% to the film surface, which provides a predominant (86.6%) contrast for the in-plane magnetization versus 50% for the out-of-plane magnetization.


Supplementary material for this article is available at

This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial license, which permits use, distribution, and reproduction in any medium, so long as the resultant use is not for commercial advantage and provided the original work is properly cited.


Acknowledgments: We thank M. R. Fitzsimmons, T. Z. Ward, and G. Eres for feedback on an early version of this manuscript. We also thank H. Liu, L. Yin, and J. Shen from Fudan University for assistance with magnetic force microscopy experiments and L. Debeer-Schmitt of the High Flux Isotope Reactor at Oak Ridge National Laboratory for assistance with small-angle neutron scattering experiments. Funding: This work was supported by the U.S. Department of Energy (DOE), Basic Energy Sciences, Materials Sciences and Engineering Division. Use of the Advanced Photon Source and the Advanced Light Source, which are DOE Office of Science User Facilities, was supported by DOE’s Office of Science under contracts DE-AC02-06CH11357 and DE-AC02-05CH11231, respectively. The Hall effect measurements in high magnetic fields were performed at the NHMFL, which is supported by NSF cooperative agreement no. DMR-1644779 and the state of Florida. Author contributions: E.S., J.N., and H.N.L. designed the experiment and wrote the manuscript with input from all authors. E.S. and J.N. performed sample growth, Hall effect, and magnetometry measurements. J.M.O., E.S., and M.B. performed high magnetic field measurements at the NHFML with the support of E.S.C. R.V.C. performed XMCD-PEEM measurements and analysis at the ALS. A.R. and C.S. provided assistance with sample growth and electronic transport measurements. E.S., J.N., C.S., and R.D.D. performed hard and soft x-ray absorption experiments and analysis with the support of Y.C., D.H., and J.W.F. at the APS. X.G. and S.Y. were responsible for STEM measurements and analysis. T.F. and J.N. carried out micromagnetic simulations, and S.O. carried out theoretical simulations. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. Additional data related to this paper may be requested from the authors.

Stay Connected to Science Advances

Navigate This Article