The superlattices (LCO_M/SRO_N)₁₂ (denoted by L_MS_N, where M and N denote the number of unit cells of LCO and SRO, respectively, in each bilayer period, and the subscript 12 indicates 12 repetitions of the LCO/SRO bilayer) were grown on TiO₂-terminated SrTiO₃ (STO) substrates (from MTI Corporation Inc., China) by pulsed laser deposition (PLD, PASCAL Co., Ltd. Japan) assisted with reflection high energy electron diffraction (RHEED). We have successfully designed and prepared various superlattices, among which L₉S₆, L₇S₆, and L₅S₅ were particularly investigated. During the growth of the superlattices, the substrate temperature, oxygen partial pressure, and repetition rate of the laser pulse were 750°C, 100 mTorr, and 2 Hz, respectively. The laser fluence set at ~ 2 J cm^-2 for the SRO layer and 1.2 J cm^-2 for the LCO layer.

X-ray characterizations, including X-ray diffraction (XRD), X-ray reflectivity (XRR), and reciprocal space mapping (RSM), of the superlattices were performed using a four-circle diffractometer (Cu Kα, SmartLab, Rigaku, Japan). Scanning transmission electron microscopy (STEM) was performed using an FEI Titan Themis G2 (FEI, USA), equipped with a double spherical-aberration corrector and a high brightness field-emission gun. The cross-section of the specimens was prepared using a FEI Helios 600i dual-beam FIB/SEM machine. Pt electron-beam deposition and C ion-beam deposition were conducted in order to protect the sample surface from beam damage. Both X-ray characterization and STEM experiments were conducted at room temperature.

MFM measurements were performed on the top surface of the superlattice films in a 12 T superconducting magnet placed in ultrahigh vacuum at the High Magnetic Field Laboratory in Hefei, China. The spatial resolution of the microscope is 10 nm. A rotation sample plate is mounted on the axis, allowing for an in-situ field angle ranging from 0° to 180°. For signal detection, we utilized custom piezoresistive cantilevers (SCL-Sensor Tech., PRSA-L300-F50-STD) characterized by a hollow beam structure and a resonance frequency of ~ 37 kHz. The tips were magnetically coated with a Ti (5 nm)/Co (30 nm)/Au (5 nm) film using electron beam evaporation and pre-magnetized perpendicular to the cantilever beam. The readout electronics consisted of a home-made active Wheatstone bridge preamplifier (gain 1,000×) interfaced with a commercial RHK R9 controller. Topographic images were acquired in tapping mode, while MFM images were acquired in constant-height mode approximately 100 nm above the sample surface. The resulting MFM contrast maps the resonant frequency shift caused by the out-of-plane stray field gradient, where dark and light regions represent attractive and repulsive magnetic forces, respectively. Post-processing of the SPM data was conducted using WSxM. All MFM images presented in this work were measured on the top surface of the as-grown superlattice films. Cross-sectional MFM was not performed in this study. The superlattice terminates with an SRO layer (topmost surface). The MFM probe detects magnetic stray fields emanating from this top SRO layer. The signal primarily reflects the magnetization state of the SRO layer because: SRO is metallic and ferromagnetic with strong stray-field contrast, while LCO is insulating with weak magnetic contrast at the measurement temperatures (5-300 K), and the 6uc (~ 2.4 nm) top SRO thickness is within the typical MFM detection depth.

All transport measurements were performed using a physical property measurement system (PPMS, Quantum Design, USA) equipped with a sample rotator. The magnetic-field-angle-dependent resistivity (MAR) is defined as MAR = [ρ(θ) - ρ(0˚)]/ρ(0˚), where θ is the angle of the magnetic field with respect to the film normal (e.g. 0˚ for field pointing out-of-plane). For the transport measurements, Hall electrodes were fabricated using the evaporation method. The current was applied in the plane of the film along the [100] direction of the STO substrate, and the magnetic field was rotated in the plane perpendicular to the current direction (the ac plane), with θ = 0° corresponding to the field applied perpendicular to the film plane.

An all-Python-based micromagnetic framework were employed that computes the magnetostatic field via 3D fast Fourier transform (FFT) convolution and the magnetization is relaxed toward a steady state through an energy-minimization alignment scheme. At each iteration the local moment is rotated toward the direction of the effective field with a damping-like coefficient serving purely as a convergence parameter until the system reaches a local energy minimum. The micromagnetic simulations were performed under two regimes: (A) a DMI-dominated regime and (B) an interfacial-exchange-dominated regime. The total energy functional includes exchange, uniaxial anisotropy (PMA for SRO, IMA for LCO), demagnetizing (computed via FFT-based convolution), interfacial DMI, Zeeman, and an effective interfacial exchange interaction term at the LCO/SRO boundary. The 2D model simulation and its comparison with the standard GPU-based solver mumax3 are provided in Supplementary Table 1. The parameters used for SRO and LCO layers are summarized in Supplementary Table 2, while the parameters used for Case A and Case B are listed in Supplementary Table 3.

The Hall resistivity was analyzed using a two-channel AHE model, which is expressed as ρ_xy = R₁M₁ + R₂M₂ + ρ_xy^OHE, where R₁ and R₂ are the anomalous Hall coefficients of two distinct magnetic channels, M₁ and M₂ are their respective magnetizations, and ρ_xy^OHE is the ordinary Hall resistivity. The two channels are assigned to represent the bulk-like SRO ferromagnetic component (weakly coercive S-loop) and the interfacial ferromagnetic coupling component (highly coercive L-loop), respectively. The model parameters were determined by least-squares fitting to the experimental Hall data. We note that this phenomenological model assumes isotropic magnetic properties and does not account for possible contributions from noncollinear spin textures or inhomogeneous magnetization distributions.

LCO/SRO superlattices were prepared by alternating growth of LCO and SRO on top of TiO₂-terminated STO substrates. Figure 1A shows the conceptual stacking of the LCO and the SRO layers on STO. The STEM image identifies each atomic layer with no ambiguity. RHEED images during sample preparation were captured for the three sets of samples [Figure 1B-D], which present the layer-by-layer nature of the growth. XRD measurements [Figure 1E] show rigid epitaxy of the superlattice to the substrate according to the result of the (002) substrate peak and the superlattice satellite peaks (SL0, ±1, ±2), including beating patterns due to the whole film thickness and the satellite peaks (SLx) due to each LCO/SRO layer pair. The XRR patterns were in good agreement with the theoretical fit for L₉S₆, L₇S₆ and L₅S₅ structures, respectively [Figure 1F]. The combination of RHEED, STEM, XRD and XRR shows the high atomic layer-by-layer quality for all of our superlattice samples.

The lattice constants of bulk STO, LCO, and SRO are approximately 3.905 Å, 3.837 Å, and 3.923 Å, respectively. Reciprocal space mapping (RSM) around the (103) reflection was performed to determine the in-plane lattice parameters of the superlattices [Supplementary Figure 1]. The RSM results confirm that the superlattices are coherently strained to the STO substrate. The SRO and LCO suffer from compressive and tensile strain, respectively. This corresponds to a tensile strain of +1.8% for LCO and a compressive strain of -0.5% for SRO. The out-of-plane lattice parameters were determined from the XRD data, showing elongation consistent with the Poisson effect under biaxial compressive strain. The strain state is an important factor influencing the magnetic anisotropy of both LCO and SRO layers, as compressive strain in SRO is known to favor perpendicular magnetic anisotropy, while tensile strain in LCO can stabilize ferromagnetic order with in-plane magnetic anisotropy.

We performed the cryogenic MFM measurements to directly image the magnetic domain evolution for our superlattice samples. Here, we focus on the L₉S₆ samples. Figure 2A shows a series of the MFM images at 75 K with the field applied perpendicular to the film plane and swept in a complete loop. Stripe-like domains emerge at ~ 1 T, become saturated at higher field strengths, and reappear as the field strength decreases back to 1 T in both positive and negative field-scanning directions. Figure 2B shows corresponding MFM images at 25 K. A notable difference is that at lower temperature, stripe-like domains appear at smaller field strength (denoted from cyan brackets) and persist to higher fields, indicating enhanced magnetic anisotropy pinning. Complete sets of MFM loop measurements with out-of-plane fields can be found in Supplementary Figures 2-5. The stripe-like domains are stabilized at intermediate fields (~ 1 T), but disperse at both low fields (< 0.5 T, near demagnetized state) and high fields (> 7 T, field-polarized state). By contrast, when the field is applied in-plane, no magnetic stripe is observed even at 10 T [Supplementary Figure 6 and 7], demonstrating that the stripe formation is highly anisotropic and favored by out-of-plane magnetic fields. While magnetic anisotropy has been reported in similar materials^[38-40], the particular stripe morphology observed here appears to be characteristic of this superlattice architecture.

To further investigate the magnetic anisotropy, we conducted the MAR measurements as a function of temperature and magnetic field. MAR plots for different experimental conditions are presented in Figure 3 for the L₉S₆ samples. Forward and backward field-angle scans are shown in Supplementary Figure 8. Hysteretic behavior of magneto-resistivity, indicative of coercive ferromagnetism, is observed but diminishes as higher temperature elevates the thermal fluctuations and higher field enhances the magnetization alignment. More importantly, a purely two-fold symmetric MAR signal would be expected if the SRO magnetization maintains a single, fixed orientation. Therefore, the emergence of any higher-order symmetry component may imply intrinsic spin reorientations within the SRO layers.

To quantitatively characterize the evolution of the MAR symmetry, we developed a harmonic decomposition model to assess the magneto-transport anisotropy. The MAR signal was fitted with three components, one two-fold and one four-fold and one eight-fold, capturing the dominant symmetries: MAR(θ) = A₀ + A₂cos(2θ + φ₂) + A₄cos(4θ + φ₄) + A₈cos(8θ + φ₈), where A₂, A₄, and A₈ are the amplitudes, and φ₂, φ₄, φ₈ are the remnant phases of the two-fold, four-fold and eight-fold components, respectively. The results were plotted in Figure 3. It is noted that a two-component decomposition with one two-fold and one four-fold underfits the MAR signal though A₄/A₂ still appears to be a good order parameter for the symmetry and its evolution with field and temperature (see Supplementary Figures 9 and 10), assuring the necessity of the eight-fold component. The two-fold component originates from the uniaxial anisotropy of the SRO layer with its spin canted by interfacial exchange coupling. The four-fold component arises from the superposition of contributions from multiple SRO sublayers with different spin orientations. The eight-fold component captures higher-order angular anisotropy from crystal symmetry breaking and spin-orbit coupling, consistent with eight-fold symmetry reported in other SRO superlattice systems^[41].

The fitting results reveal several important features. The systematic evolution of the three-component (2-fold, 4-fold and 8-fold) amplitudes with field and temperature is summarized in Supplementary Figure 11. The phase diagrams provide crucial quantitative support for the presence of spin reorientation. First, the amplitude A₂ of the dominant component increases monotonically with field strength at all temperatures, reflecting the progressive alignment of the bulk-like SRO spins toward the field direction. The secondary amplitudes A₄ and A₈ exhibit more complex field dependence, with non-monotonic evolution that reflects the competing effects of field-induced alignment, interfacial exchange pinning, and higher-order crystal symmetry. At 25 K, the two-fold amplitude A₂ show systematic field-dependent evolution, where the interfacial coupling is strongest. At 75 K, the 3 T and 5 T data show anomalous four-fold dominance due to proximity to the LCO Curie temperature, while the 7 T and 9 T data exhibit significantly reduced amplitudes as spins are highly aligned by strong fields.

To provide additional support of the competition between interfacial exchange coupling and DMI in the LCO/SRO superlattice, we performed 2D magnetostatic energy minimization simulations. Figure 4 shows the M_z spatial distribution maps and depth-dependent magnetization profiles. The simulation of the SRO magnetization at different magnetic fields is summarized in Supplementary Table 4. The magnetic stripe domains are saturated by field ≲ 7 T for the DMI-dominated regime (Case A) and field ≲ 9 T for the interfacial-exchange-dominated regime (Case B), indicating that the interfacial exchange interaction competes with the Zeeman energy. The depth-resolved magnetization profiles reveal a field-driven spin reorientation: at zero field, the average spin tilt angle θ ≈ 5.5° for Case A and θ ≈ 3.8° for Case B (with respect to out-of-plane direction), consistent with stripe-domain formation; with increasing field, θ increases to maximum at 5 T and decreases back to 0° at 9 T, reflecting the competition between the applied Zeeman energy and the interfacial exchange coupling. The average tilt angle of SRO spins at 5 T is ~ 15° for Case A and ~ 8° for Case B, consistent with the effect of DMI in stabilizing chiral or stripe-spiral textures. Since the thickness of each SRO layer (2.4 nm) is comparable to the exchange length, the entire layer responds coherently to interfacial coupling rather than maintaining a bulk-like out-of-plane orientation. This systematic canting favors the formation of anisotropic stripe domains.

With the MAR analysis and the simulation results, we therefore propose a vertically evolving spin texture within the SRO layers, driven by interfacial exchange coupling between the PMA-SRO and the IMA-LCO [Figure 5A]. The coupling may cant SRO spins adjacent to the interface toward the in-plane direction (inheriting the IMA of LCO), while SRO spins in the layer interior retain a stronger tendency toward the out-of-plane orientation (reflecting the PMA of bulk SRO). The phase evolution of the harmonic components directly demonstrates the proposed field-driven spin reorientation: as the magnetic field increases, the spin orientations in different SRO sublayers rotate at different rates, changing the relative phases of their transport contributions (See Supplementary Figure 12). The phase difference φ₄ - φ₂ ≈ -48° at 25 K, 3 T and -10° at 25 K, 5 T, directly demonstrating the field-driven spin reorientation.

This proposed noncollinear spin-reorientation model also suggests a correlation of the magneto-transport properties to the magnetic anisotropic stripe pinning. Figure 5B shows the MAR vs. θ plot for various fields at 25 K. The minimal resistivity occurs at θ = 90° (in-plane field), where no magnetic stripe is formed. The stripe-like domain formation introduces localized exchange interaction and enhanced scattering, thereby suppression of conductivity. Nevertheless, direct verification of the proposed vertically varying spin configuration would benefit from techniques such as polarized neutron reflectometry.

Our MFM results found no signature of the skyrmion structure, which is nanoscale chiral magnetic spot and is often regarded as topological Hall features. The absence of skyrmion-like features in our samples might be the result of the competition between interface exchange interactions and DMI^[40-44]. Figure 6 shows field-dependent Hall resistivity (after subtracting the ordinary Hall contribution) at 25 K with vertical magnetic fields, overlaid with MFM images at corresponding fields. Green symbols represent experimental data, and the red curve shows a fit using the conventional two-channel AHE model also conducted in earlier works^[31,32,45]. The fitting quality is reasonable except near the peak and valley regions (~ ±1.2 T), where significant deviations occur (See Supplementary Figure 10).

The MFM images reveal an anticorrelation between stripe-domain visibility and the Hall resistivity anomaly: stripe-like domains are pronounced at low fields (-0.05 T, -1.0 T) and high fields (> -5 T), but become significantly suppressed near -1.2 T - precisely where the Hall resistivity exhibits its peak feature. This anticorrelation suggests that the Hall anomaly may include contribution from spin-reorientation-induced transport anisotropy, possibly reflected by the peak structure: near -1.2 T, the competing interfacial and bulk anisotropies drive a transition from the noncollinear, stripe-domain state toward a more uniformly canted spin configuration, producing enhanced spin-chirality-dependent scattering that is not captured by the two-channel AHE model.

The two-channel AHE model assumes isotropic magnetic properties and thus provides no information about the detailed spin configuration in our samples. The systematic underfitting of peak and valley regions may indicate residual contributions from the proposed noncollinear spin texture, which could reflect complex interplay between spin-orbit coupling and vertically varying spin reorientation in our superlattices. The observed anticorrelation between stripe-domain evolution and Hall resistivity anomaly is suggestive of a topological contribution, though we cannot fully exclude inhomogeneous magnetization dynamics or domain-reconfiguration effects as alternative origins. Our combined MFM, MAR measurements and micromagnetic simulations thus provide critical constraints on the spin configuration, revealing how interfacial engineering can stabilize novel magnetic textures that suppress topological skyrmions while generating distinctive magneto-transport signatures through noncollinear spin arrangements. Although our observational results cannot quantitatively determine the relative energy scales of interfacial exchange interaction and DMI, the idea that interfacial exchange competes with or inhibits DMI is still a physically reasonable assumption that is consistent with the observed magnetic microstructure. Future work that combines element-specific spectroscopic analysis with first-principles calculations might be able to obtain the microscopic parameters needed to directly verify this assumption.

To assess the generality of our observations, we systematically compared the transport properties across the three superlattice structures: L₉S₆, L₇S₆, and L₅S₅. As shown in Supplementary Figures 13-16, all three samples exhibit qualitatively similar Hall resistivity anomalies and temperature-dependent resistivity and magnetization behavior, indicating that the observed magnetic phenomena are robust features of the LCO/SRO superlattice architecture. Notably, the two-channel AHE model fits reveal that the interfacial L-loop component exhibits prominent dependence on LCO layer thickness [Supplementary Figure 13]. The temperature-dependent resistivity shows a slope change at approximately 75 ~ 85 K for all samples [Supplementary Figure 15A and B], which we attribute to the Curie temperature of the LCO layers. The negative magneto-resistivity increases with decreasing temperature for all samples except peaked at respective critical temperatures as labeled by arrows with different colors [Supplementary Figure 15C]. The field-dependence of the magneto-resistivity is more significant for L₅S₅ [Supplementary Figure 15D], suggesting that the thickness of LCO layers has a complex correlation to the spin-dependent scattering. These thickness-dependent trends support our interpretation that the observed anisotropic stripe pinning and multi-fold MAR symmetry originate from interfacial exchange coupling. While these thickness-dependent trends are consistent with the proposed interfacial coupling picture, a more comprehensive study with additional samples - particularly structures with sub-unit-cell SRO thicknesses and variable LCO/SRO periodicities - would be needed to fully establish the quantitative relationship between layer thickness, strain state, and magnetic properties, and to rule out alternative explanations such as disorder-induced localization effects.