Interfacial Electronic and Magnetic Reconstructions in Manganite/Titanate Superlattices

Complex oxide heterointerfaces provide a spacious arena for creating emergent phenomena that are unattainable in the constituent bulk counterparts. Herein, The BaTiO3/La1‐xSrxMnO3 [BTO/LSMO (x)] superlattice (SL) as a model system to investigate the intimately coupled interfacial effects and their resultant phenomena is focused on. The experimental and Density Functional Theory (DFT) calculations reveal that the induced magnetism on Ti originates from both the charge transfer process between Mn and Ti at the titanate/manganite heterointerfaces and the cation intermixing, both contributing comparably to the overall magnetic effect. Upon changing x, the orbital reconstruction of the Mn 3d electrons, tailored by the strain state of the LSMO (x) layers, efficiently modifies the magnetic exchange coupling between Mn–Ti and Mn–Mn at the interfaces, which are proved to account for the modulations of the macroscopic magnetism of the SLs. This work demonstrates the significance of interfacial reconstructions for magnetism, paving a new pathway to manipulate the magnetic properties of artificial oxide heterostructures.

The manganite/titanate heterostructures have attracted intense interest [14] due to their potential applications such as magnetic tunnel junctions (MTJ). [15]For instance, an in-plane orbital polarization of the interfacial e g Mn electrons was observed in La 0.7 Sr 0.3 MnO 3 /SrTiO 3 (STO) interfaces.14c] The metallic/insulating characters were corroborated to play a crucial role in determining the electronic reconstruction process. [17]14d,e,18] This system hosts an exciting perspective of novel applications in next-generation electronics, sensors, energy harvesting devices and so on. [19]iven the strongly entangled degrees of freedom at the manganite/titanate heterointerfaces and their potential applications for future computing and memory devices, the understanding of the complicated couplings between manganite and titanate is warranted for creating or improving the desirable properties and applications.In this work, BTO/La 1-x Sr x MnO 3 [BTO/LSMO (x)] superlattices (SLs) with x fixed at 0.10, 0.20, 0.33, 0.50, and 0.70 were fabricated on (001)-oriented STO substrates to investigate these strongly coupled effects.These SLs are designed to provide the following benefits: First, the unique SL is endowed with the ability to intensify the impact of the interfacial effects, and thus unprecedented properties can potentially be achieved.Second, the affluent ground states of LSMO (x), including the FM-insulating state, FM-metallic state, and AFMinsulating state, [20] make the BTO/LSMO (x) heterointerface an ideal system to investigate the strongly entangled interactions at titanate/manganite heterointerfaces and the resultant physical properties.Third, the relatively large lattice constants and c/a ratio of the BTO layers ensure the steady out-of-plane orbital occupancy of the Ti 3d electrons, which strengthens the interfacial interactions between Mn and Ti.Last but not least, the identical orbital occupancy of Ti 3d electrons in the SLs with various x can also simplify the investigation of the titanate/manganite interfaces to some extent.
Our experimental and Density Functional Theory (DFT) calculations reveal that the induced magnetism on Ti originates from both the charge transfer process between Mn and Ti at the titanate/manganite heterointerfaces and the cation intermixing, both contributing comparably to the overall magnetic effect.The charge transfer can be suppressed with a decreasing electronegativity of LSMO layers as the amount of Sr doping increases (i.e., higher x in BTO/LSMO (x)), which correlates well with the observed magnitude of Ti magnetism.Intriguingly, the magnetic exchange coupling between Ti-Mn (Mn-Mn) at the interfaces is found to change from AFM (FM) to FM (AFM) types as the amount of Sr doping increases, which can be attributed to the strain-modulated orbital occupancy of the Mn 3d e g electrons.The modulations of the interfacial Mn-Mn exchange coupling should be responsible for the modulated macroscope magnetism of the SLs depending on the level of Sr doping.

The Epitaxial Growth of the BTO/LSMO SLs
The repeat number of the BTO/LSMO (x) bilayers in all [BTO/LSMO (x)] 10 SLs is fixed at 10, and the thicknesses of the BTO and LSMO (x) layers are kept at 5 and 20 uc, respectively.The high-quality epitaxial growth with a layer-by-layer mode of both BTO and LSMO (x) layers is realized in all the SLs (see Figure S1, Supporting Information).Figure S2 (Supporting Information), displays the representative high-angle annular darkfield (HAADF) (STEM) images of a BTO/LSMO (x = 0.2) SL.It is shown that the LSMO layer (with higher image contrast) with a thickness of 20 uc is separate from the BTO layer (with lower image contrast) that has a thickness of 5 uc.The interfacial layers are well crystallized without dislocations and other defects, which is in line with the alternate growth in LSMO and BTO films, demonstrating the epitaxial growth coherent interfaces of the SLs. [21]The corresponding spatially resolved electron energyloss spectroscopy (EELS) elemental maps for the Ba-M 4,5 , La-M 4,5 , Mn-L 2,3 , O-K, and Ti-L 2,3 edges further indicate the chemically uniform growth of the individual layers (Figure 1a).
To visualize the strain states of the LSMO and BTO layers in the BTO/LSMO SL, the lattice parameters are extracted from the HAADF by averaging the lattice parameters on the whole width of the image (top panel of Figure 1b), and the in-plane and outof-plane lattice spacing profiles are summarized in the middle and bottom panels of Figure 1b, respectively.Both the LSMO and BTO layers are fully constrained by the STO substrate in in-plane directions, as the two layers share the identical in-plane parameter (a ≈3.905 Å) as the STO substrate and remain uniform over the whole region.Since the films are less controlled by the substrates along the out-of-plane direction, the lattices of LSMO and BTO layers are close to their bulk states.A peak value of 4.07 Å is observed in the BTO region.Similar results are also obtained from another area in the same SL, signifying the fully stained states of all the sublayers throughout the whole SL (see Figure S3, Supporting Information).Reciprocal space mappings (RSMs) were performed on the (103) asymmetric plane to further elucidate the strain states for all the SLs with various x (Figure 1c).Apart from the main reflection of the SLs, additional satellite peaks with up to eight orders of Kiessig fringes are also visible for all the SLs, further confirming the highly ordered epitaxial growth of the SLs.As x increases, the main reflections from the SLs shift to high L values, demonstrating the reduction of the averaged out-of-plane lattice parameter of BTO/LSMO bilayers.The reduction of averaged lattice constants of SLs should be attributed to the decreased lattice constant of the LSMO layers with increasing x. [22] More importantly, the diffractions from all the SL samples have identical in-plane reciprocal vector (Q x ) in comparison to the STO substrates, validating that all the SLs are coherently strained by the STO substrates, which is consistent with the STEM results.
To investigate the cation intermixing of LSMO-BTO interface, the layer-by-layered EELS spectra along the growth direction of SLs were plotted (Figure S4, Supporting Information).Specifically, we focused on the Ba-M 4,5 , La-M 4,5 , Mn-L 2,3 , O-K and Ti-L 2,3 edges.Notably, in the BTO region, a weak La signal was detected.On further investigation of the La cation intermixing at the LSMO-BTO interface, the quantitative results indicate a La concentration within the BTO layer of ≈9.7%, suggesting the presence of cation intermixing at the interface.To address the potential influence of the dechanneling effect during EELS mapping for thick samples, which could lead to unclear interface, [23] we conducted EELS simulations.To ensure accurate simulation, the sample thickness was set to be 30 nm, consistent with the absolute thickness calculated through the Log-ratio method using the zero-loss peak from experimental data. [24]The outcomes, presented in Figure S5a (Supporting Information), showcased a distinct LSMO-BTO interface even at a 30 nm sample thickness, indicating that the dechanneling effect might not be the primary factor influencing the observed result.Subsequently, we performed an EELS simulation wherein 25% of the Ba in the BTO layer was randomly replaced with La, as shown in Figure S5b,c (Supporting Information), also indicating the existence of cation intermixing at the LSMO-BTO interface.

Interfacial Magnetic Reconstruction at BTO/LSMO Heterointerfaces
The electrical transport properties of all the samples were measured using the van der Pauw method. [25]The electrical behaviors of the LSMO (x) single films and the BTO/LSMO (x) SLs display remarkable differences (for details, see Figure S6, Supporting Information).For example, the single film with x = 0.10 (0.50) is metallic (insulating) at low temperatures, while the corresponding SL turns to be insulating (metallic).Given the fact that the BTO layers are insulating, and the electrical conduction of the SL samples mainly takes place through the LSMO (x) layers, the notable distinction between the magnetotransport properties of the SLs and the films implies the profound influence of the interfacial effects at Mn/Ti heterointerfaces.Concurrently, striking differences are also observed in the magnetic properties of the single films and SLs (for details, see Figure S7, Supporting Information).With increasing x from 0.10 to 0.50, the M S differ-ence (ΔM S ) between the single film and the SL with identical x monotonously increases (Figure 2a).For x = 0.70, ΔM S decreases to 0, due to the AFM nature of both the single film and the SL.The suppressed M S of the SLs by contrast with the single films should also originate from the strong couplings between Mn and Ti at BTO/LSMO (x) heterointerfaces.
As aforementioned, the spin alignments between Ti and Mn are strongly coupled at the heterointerfaces, which markedly modify the magnetic properties of the epitaxial system.Therefore, understanding the delicate spin interactions at Mn/Ti heterointerfaces is of tremendous significance to investigate the magnetic evolution of BTO/LSMO (x) SLs depending on x.We applied element-specific techniques to the BTO/LSMO (x) interfaces, namely synchrotron-radiation-based X-ray absorption spectroscopy (XAS) and X-ray magnetic circular dichroism (XMCD).All data were obtained by recording the total electron yield (TEY).Since the maximum probing depth of TEY is about 5-10 nm, these characterization methods are highly sensitive to the interfacial Ti magnetism via choosing the proper thickness of BTO. [26]The experimental setup is schematically shown in the inset of Figure 2b, where the angle between the X-ray propagation vector and sample surface was set at 30°to optimize for detecting the in-plane magnetic moment.Figure 2b demonstrates the representative XAS as well as the XMCD spectra at the Ti-and  Robust Mn ferromagnetism is detected for all the SLs except for x = 0.70 reflected by the strong XMCD signals.With increasing x, the Mn XMCD signal peaks at x = 0.20, displays the same trend as the results obtained from magnetic hysteresis loops (M-H curves).For the Ti-L 2,3 edge, a significant difference among the XMCD spectra can be observed.The curves corresponding to x = 0.10 and 0.20 possess positive integrals, whereas the curves corresponding to x = 0.33 and 0.50 have negative integrals (see Figure S9, Supporting Information), suggesting the distinct magnetic exchange coupling between Mn and Ti at the interfaces.Although it is not feasible to apply the sum rules to quantify the value of the spin moments, the XMCD integrals do provide information about the relative magnitude and orientation of both the spin (s z ) and orbital (l z ) moments of the Mn and the Ti ions.The results are summarized in Figure 2c (for details, see Figure S9, Supporting Information).Strong spin moments are observed at the Mn edges in all the SLs with x = 0.10, 0.20, 0.33, and 0.50, without detectable orbital moments, which are essentially quenched by the crystal field effect.The calculated Mn s z of the BTO/LSMO (x = 0.2) SL with the strongest ferromagnetism is 1.05 μ B /Mn, which is substantially smaller than that (2.66 μ B /Mn) obtained from the M-H curve at 80 K (see Figure S8, Supporting Information).14b] As x varies, the Mn spin moment reduces, and be-comes to be zero when x = 0.70, in accordance with the results obtained from M-H curves (see Figures S7 and S8, Supporting Information).
Visible Ti magnetism is also detected by the XMCD spectra in the BTO/LSMO (x) SLs (Figure 2b), despite the nonmagnetic nature of the bulk BTO.The integrated intensity under the Ti XMCD signal changes from a positive value to a negative value as x increases from 0.10 to 0.50, unambiguously suggesting the alteration of the magnetic exchange coupling between Ti and Mn at the interfaces (see Figure S9, Supporting Information).Moreover, the magnitude of the spin moments of Ti demonstrates rough monotonicity with x (Figure 2c), signifying the critical role of the LSMO (x) layers on the emerged Ti magnetism.14b,27] Therefore, the spin moments are substantially larger than the orbital moments, and the Ti magnetism should be dominated by the spin moments.Accordingly, the magnetic exchange coupling between Ti and Mn magnetization is AFM for the cases with x = 0.10 and 0.20, and then turns to be FM type as x increases to 0.33 and 0.50.It is noteworthy that bulk BTO does not exhibit magnetism due to the empty e g 3d 0 orbitals of Ti 4+ .The presence of the weak Ti magnetism here should be highly related to an interfacial charge redistribution process between Mn and Ti.

Charge Redistribution at Mn/Ti Heterointerfaces
An upward self-polarization (pointing toward the surface) is revealed to be maintained in all BTO layers in the BTO/LSMO SLs with various x by the piezoresponse force microscopy (PFM) and ABF-STEM measurements (for details, see Figure S10, Supporting Information).The density of hole carriers in the interfacial LSMO (x) may be electrostatically modulated by the selfpolarization of BTO layers, giving rise to two possible states, i.e., depletion or accumulation of charge carriers. [28]However, this contradicts the results obtained from the atomic-resolution EELS characterization, where the Mn valence at the two kinds of interfaces simultaneously increases (for details, see Figure S11, Supporting Information).This scenario signifies the negligible effect of the FE polarization on the charge redistribution.Considering the ultrathin BTO layers in the present SLs, the ferroelectricity of the BTO layers is expected to be weak, and so is the limited FE-polarization-driven charge screening effect.Therefore, other physical mechanisms should exist to account for the modulations of the valence state of Mn.
For oxide heterostructures, charge transfer from one perovskite compound to the other can be triggered by band alignment at the heterointerface. [29]The electronegativity difference between LSMO (x) and BTO layers may prompt the occurrence of the charge transfer process from LSMO (x) to BTO layers, thus giving rise to the higher valence of interfacial Mn.In specific, the electron transfers from the Mn 3+ ion to the adjacent Ti 4+ ion, and as a result, a Mn 4+ ion and a Ti 3+ ion are generated.In this regard, the valence of Ti ions is expected to be lowered due to the transferred electrons to the empty Ti 3d bands, which is corroborated by the Ti L-edge EELS result that the valence of Ti deviates from +4 to around +3.85 (see Figure S12, Supporting Information).This further confirms the occurrence of the charge transfer process.
The existence of the interfacial charge transfer in all BTO/LSMO (x) SLs is certified by the isotropic XAS at Ti L 2,3 edges (Figure 3).We found that the energy difference between the L 3 and L 2 edges becomes smaller with a decreasing amount of Sr dopants (x) (Figure 3a).This result signals the presence of valence variation of Ti ions in the BTO layers, and an increasing concentration of Ti 3+ with a decreasing x. [14b,30] Increasing Sr doping concentration (x) is expected to consume the Mn 3d band and reduce the electronegativity difference at BTO/LSMO (x) heterointerfaces, thus suppressing the charge transfer from LSMO (x) to BTO.Consequently, a reduced concentration of Ti 3+ is observed with increasing x (Figure 3b).14b,d] This is the reason why the detected Ti magnetism monotonously decays with increasing x.
Both the interfacial charge transfer and cation intermixing can contribute to the valance change of the Ti ions and the resultant Ti magnetism.Since it is experimentally difficult to separate them, detailedDFT investigations were carried out, which will be discussed in Section 2.6.

Orbital Reconstruction in BTO/LSMO SLs
So far, we have ascribed the emerging Ti magnetism to the interfacial charge transfer from Mn to Ti and interfacial cation intermixing.Nevertheless, the origin of the modulation of the magnetic exchange coupling at Mn/Ti interfaces has not been fully addressed.To disclose this issue, element-specific X-ray linear dichroism (XLD) spectroscopy at the Ti and Mn L 2,3 edges was carried out using linearly polarized X-ray beams in TEY mode to examine the Mn and Ti 3d orbital occupancy in the SLs (Figure 4).Upon rotating the X-ray incident angle, the polarization directions of the linearly polarized X-rays were set to be 90°and 30°c orresponding to in-plane (E//a, I ab ) and out-of-plane (E//c, I c ) polarization components, respectively, as shown in Figure 4a.Based on the XLD sum rule, the preferential orbital occupancy of the d electrons can be obtained from the difference in the energy position and intensity, and thus XLD (I ab -I c ) is one of the most widely used methods to explore the orbital occupancy of the d electrons. [31]epresentative XAS and XLD spectra of Ti and Mn L 2,3 edges measured from the SLs with various x are shown in Figure 4b.The XAS spectra are normalized by the Ti and Mn L 3 edge jump, respectively.31a] The SL with x = 0 shows a positive XLD integral of 1.40%, then it decays to 0.70% as x increases to 0.20, suggesting the suppressed out-of-plane d 3z 2 −r 2 orbital occupancy of the Mn 3d e g electrons.With further increasing x to 0.33, the XLD integral turns to be a negative value of −1.82%, correlating to the in-plane d x 2 −y 2 orbital occupancy.Upon further increasing x to 0.50 and 0.70, the XLD integral becomes more negative, i.e., −2.32% at x = 0.50 and −2.87% at x = 0.70.
For complex oxide heterostructures, the orbital occupancy of d electrons can be tuned by external stimulations, especially the epitaxial strain.31a] In the present case, the coherent epitaxial growth without strain relaxation has been validated for both the BTO and LSMO (x) layers via the STEM and RSM measurements (Figure 1).Namely, the strain states of the BTO and LSMO (x) layers can be estimated by comparing their bulk lattice parameters with that of the STO substrate, i.e.,  = a f −a s a s × 100%, where ɛ denotes the in-plane strain imposed on the SLs, and a s , a f are the bulk lattice parameters of the STO substrate and the film sublayers (BTO or LSMO (x) layers) in the film plane.Bulk BTO layer has a tetragonal structure with lattice parameters of a = 3.993 Å and c = 4.031 Å. [14d,32] When grown on the STO substrate (cubic, a = 3.905 Å), the BTO layers receive a large in-plane compressive strain (−2.25%), which further enlarges the intrinsically sizable c/a ratio of the BTO layers.This should be the primary origin of the persistent out-of-plane orbital occupancy of the Ti 3d t 2g electrons in all the BTO/LSMO (x) SLs.For the LSMO (x) layers, with increasing x, the lattice parameters of their bulk counterparts monotonously decrease from 3.927 to 3.800 Å, yielding an alternation of the strain states of the LSMO (x) layers from compressive to tensile states (Figure 4c).The similar tendency between strain states and orbital occupancy of the LSMO (x) layers substantiates the decisive role of the epitaxial strain on the orbital occupancy of Mn 3d electrons.

Physical Picture of the Coupled Interfacial Effects
Combining the results described above, we now try to construct a complete picture of the strong entangled interactions at the BTO/LSMO (x) interfaces regarding the charge redistribution, orbital occupancy, and interfacial exchange coupling.The band alignment across BTO/LSMO (x) interfaces facilitates a charge transfer process from Mn 3d to Ti 3d bands.Consequently, Ti magnetism is induced due to the nonempty 3d 1 band of Ti.Upon increasing x, the magnetic exchange coupling between Mn and Ti magnetization shows a sign of crossover from AFM type to FM type.Meanwhile, the Mn 3d orbital occupancy is also observed to be altered from out-of-plane d 3z 2 −r 2 orbital to in-plane d x 2 −y 2 orbital, which is mediated by the modified strain states of the LSMO (x) layers.The simultaneous alternation of the magnetic exchange coupling between Mn and Ti moments and the orbital occupancy of Mn 3d e g electrons demonstrates the intimate correlation between these two physical phenomena.
For ABO 3 perovskites with connecting corner-sharing oxygen octahedra, the B site cation is normally coordinated with the six oxygen anions.Due to Coulomb repulsion between the d orbital electrons and the charges from those oxygen anions, the energy of the d orbitals increases with the oxygen anions approaching the B site cations.Hence, this unique crystal environment breaks the five-fold degeneracies of the d orbitals into two higher energy e g orbitals (d x 2 −y 2 and d 3z 2 −r 2 ) and three lower energy t 2g orbitals (d xy , d yz and d xz ).The degeneracies between the e g and t 2g orbitals can be further lifted by the deformation of the oxygen octahedral structures due to internal (Jahn-teller distortion) or external (e.g., epitaxial strain) perturbations, and thus new electronic configurations are evoked. [33]In the present BTO/LSMO (x) SLs, three Mn 3d electrons half-fill the low-energy Mn t 2g orbital, and one 3d electron occupies the high-energy Mn e g orbital for the Mn 3+ ions, while the single electron occupies the low-energy t 2g orbital for the Ti 3+ ions.Therefore, the interfacial exchange coupling between Mn and Ti at the interfaces is essentially controlled by the hybridization between the occupied Mn e g orbitals and Ti t 2g orbitals.The alternation of the magnetic exchange coupling between Mn and Ti can be well explained by the Goodenough-Kanamori-Anderson (GKA) rules. [22,34]lthough hybridization between the occupied Mn d x 2 −y 2 and Ti t 2g orbitals across the interfaces are negligible, finite overlap between Ti t 2g and Mn d 3z 2 −r 2 orbitals may result from the octahedral distortions at the interfaces.To clarify the octahedral rotation behaviors across the BTO/LSMO (x) heterointerfaces, the annular bright-field STEM (ABF-STEM) was performed on the SL with x = 0.20 (see Figure 5; Figure S14, Supporting Information).Dramatic changes of the OOR angles are observed crossing the BTO/LSMO interfaces (Figure 5b,c).The LSMO layer possesses a large OOR (6.37°), which is close to its bulk state. [35]For the BTO layer, due to the absence of the OOR in its bulk state, the OOR angle is nearly 0°inside BTO layer.6a,36] The OOR in the interfacial LSMO region is substantially suppressed by the adjacent BTO layer, while a moderate OOR is also induced at the interfacial BTO region.The existence of the GdFeO 3 -type distortion at the BTO/LSMO interfaces ensures the hybridization between the Ti t 2g and Mn d 3z 2 −r 2 orbitals.
In the SLs with x ≤ 0.20, the LSMO layers suffer compressive strain (Figure 6a), and the electrons dominantly occupy the Mn d 3z 2 −r 2 and Ti d xz/yz orbitals (Figure 6c).The interfacial Ti-Mn exchange coupling is mainly governed by the superexchange interaction between the above occupied Mn 3d and Ti 3d orbitals, resulting in an AFM alignment between the Ti and Mn moments (Figure 6c).In the SLs with x ≥ 0.33, the LSMO layers suffer tensile strain (Figure 6b), and the Mn 3d e g electrons preferentially occupied the in-plane d x 2 −y 2 orbital (Figure 6d).The octahedral distortions yielding finite overlap between Ti t 2g and Mn d 3z 2 −r 2 orbitals also allow the virtual hopping of electron from the occupied Ti d xz/yz orbital to the empty Mn d 3z 2 −r 2 orbital.According to the GKA rules, FM coupling is established at Mn/Ti interfaces, thanks to the superexchange interaction through the virtual excitation of electrons (Figure 6d).A noticeable exchange bias (see Figure S15, Supporting Information) may be due to the presence of an AFM layer at the interface.Hence, we propose that in the SLs with x ≥ 0.33, the exchange coupling between Mn atoms at the interface is AFM, whereas it is ferromagnetic between Mn and Ti, as displayed in Figure 6f.14d] The tunable hybridization between Mn and Ti through O 2p orbitals at the heterointerfaces is also corroborated by the XAS spectra at O K edges measured from the BTO/LSMO SLs and LSMO single films with various x (see Figures S16 and S17, Supporting Information).Based on the above analysis, the magnetic configurations at BTO/LSMO (x) heterointerfaces are outlined.For the x = 0.10 and 0.20 cases, Mn-Ti (Mn-Mn) magnetic moments are antiferromagnetically (ferromagnetically) coupled, while for x = 0.33 and 0.50, the opposite is true (Figure 6e,f).
The delicate modulations of the magnetic exchange coupling at the Mn/Ti interfaces pave a pathway toward analyzing the evolution of the macroscopic magnetic properties of the BTO/LSMO (x) SLs depending on x.Considering the limited magnetism of Ti, the ferromagnetism in the SLs is mainly contributed by the LSMO (x) layers.While a substantial difference between the M S of the BTO/LSMO (x) SLs and LSMO (x) single films (ΔM S ) is observed for x = 0.33 and 0.50, the ΔM S for the cases with x = 0.10 and 0.20 is comparatively small, as summarized in Figure 2a.As elucidated by the physical pictures in Figure 6e,f, in BTO/LSMO (x = 0.33 and 0.50) SLs where the LSMO layers suffer a tensile strain, AFM layers are built in the interfacial LSMO regions.Due to the compensation between the adjacent Mn magnetizations in the AFM layers, the suppressed macroscopic magnetization of such SLs is anticipated.The relatively small ΔM S for the SLs with x = 0.10 and 0.20 is consistent with the FM nature of the interfacial LSMO (x) layers.

Density Functional Theory Studies
To understand experimental observation more in-depth, we have performed DFT calculations on BTO/LSMO SLs.We have considered two different approaches to model the observed magnetic moments on Ti-sites.a) First is through charge transfer between LSMO and BTO region due to band alignment and b) a second approach that involves interface in the SL with cation intermixing, which is probable while growing these SLs.Both approaches are studied on an SL with 8 uc of LSMO and 4 uc of BTO along (001) orientation with oxygen continuity as shown in Figure 7a.29a] But here we have two different compounds, i.e., BTO and LSMO, with different oxygen p energy levels ( F ) when considering their bulk phase, which results in mismatch of the Fermi energy (E F ).However, a system in equilibrium must have constant E F throughout the heterostructure, hence charge transfer occurs to balance the mismatch. [10,37]o explore the charge transfer and band alignment at the heterointerface, we first studied the bulk properties of BTO and LSMO. Figure 7b shows the layer-dependent DOS of TiO 2 in bulk BTO and BTO/LSMO SL and MnO 2 in bulk LSMO.As can be seen, the oxygen states in the bulk BTO and LSMO are at different energy levels.29a,37a] Usually, in semiconducting p-n junctions when two distinct semiconductors are interfaced, there the E F mismatch between p and n-doped sites is balanced by charge transfer.Similarly, in our case, the E F mismatch which is evident from layer-dependent DOS when compared to bulk and interface leads to charge transfer between LSMO and BTO region.
In semiconducting heterostructures, one might employ Anderson's or Schottky-Mott rules [35] to estimate the amount of charge transfer as a result of band mismatch, using work functions.But in oxide heterostructures, the concept of a single work function is invalid, as work function changes with respect to orientation and terminations. [38]For example, in STO, the work function of SrO-termination is 2.5 eV, while that of TiO 2termination is 4.2 eV, [39] which is almost double that of SrOtermination.Hence, previous reports have already proven that SLs.With increasing x, the strain state of the LSMO layers changes from compressive strain a) to tensile strain b).c,d) Sketches of the orbital occupancy of Mn and Ti 3d electrons for the BTO/LSMO (x) SLs with compressive c) and tensile strain d), respectively.Note that the color-filled orbitals indicate the preferentially occupied Mn and Ti orbitals.e,f) Proposed interfacial spin configuration and magnetic coupling mechanism at the BTO/LSMO (x) heterointerfaces with compressive e) and tensile strain f), respectively.
Anderson's method or Schottky-Mott methods are not applicable for transition metal oxide heterostructures. [38]But to have a rough estimate of charge transfer and direction of charge transfer, we considered Badder and Mulliken charge analysis.Figure S18 (Supporting Information), shows the charge density difference plots (CDD) of BTO/LMO SL.The charge density difference can be determined using the following formula: where Δ (BTO + LMO) represents the charge density of BTO/LMO SL.Δ (BTO) and Δ (LMO) are the total charge densities of the BTO and LMO regions alone.Both Δ (BTO) and Δ (LMO) are calculated with each component of the respective system at the same positions as in the combined SL. Figure S18 (Supporting Information), shows most of the charges is accumulated on the heterojunctions.And the focused Figure S18b (Supporting Information), shows charge accumulated regions around Ti and charge-depleted regions around the Mn via the oxygen bonds.Which clearly indicates the charge transfer of LMO to the BTO region via the octahedral oxygens.Further, Our Badder charge analysis results show that ≈0.27e [29a] charge has been transferred from the LMO region to the BTO region via oxygen continuity.Now the charge transfer is evident from the DOS and Bader charge analysis, we then explore the amount of charge transfer from LSMO to BTO region as a function of Sr doping.Figure S19 (Supporting Information), shows the layer projected DOS of TiO 2 in BTO and MnO 2 in LMO and LSMO bulks, the respective charge transfer obtained from Badder charge analysis is also mentioned in the respective panels.We can observe the shift in the O-p orbitals with an increase in Sr concentration in the LSMO.The noteworthy point is that both DOS and Badder charge analysis shows, with an increase in the concentration of Sr in LSMO, the band mismatch between the hetero layers reduces which in turn reduces the charge transfer between them.This explains the experimental observation of reduced magnetic moment with an increase in the concentration of Sr doping.
The induced magnetic moments on the Ti atoms which is a result of the above-mentioned charge transfer mechanism from the Mn 3d to Ti 3d orbitals via O 2p. Figure S20a (Supporting Information), shows the 2D magnetic density mapping of BTO/LSMO SL.The induced magnetic moments on the Ti are evident from the magnetic density plot.The integrated magnetic moments inside the Ti sphere range from 0.02 to 0.05 B/Ti depending on the integrating radii of the atomic sphere.As these magnetic moments are sensitive to the chosen integral radii, [40] we double-checked the calculation by varying the integral radii as well.The observed magnetic moments are smaller when compared to the experimental observations.Also, the change in the rotation angles with respect to layers is not exactly in agreement with the experimental observations.There is a sudden jump from ≈8°to ≈0°degrees when we move from the LSMO region to the BTO region as obtained in the optimized structure (shown in Figure S20c, Supporting Information).But From Figure 5, we observe a gradual change in the rotation angles from ≈8°t o ≈0°.This discrepancy can be accounted for if we consider the cation intermixing in our model, which will be discussed below.
Since cation intermixing was observed across the heterointerfaces, we built the model to investigate the contribution of such intermixing to the induced magnetic moments on Ti. Figure 7c shows the SL of the BTO/LSMO interface, where 25% of Ba of the interface was replaced by La and vice versa.Now the discrepancy on the rotation angle is solved, the rotation angle Vs the layer plot is shown in Figure 7d.Now the rotation from the LSMO region which is ≈8°is gradually reduced to ≈0°in the BTO region, the gradual change in rotation angle of Figures 5 and 7d further confirms the intermixing of La to the BTO region.The induced magnetic moment on Ti was found to be ≈0.05 to 0.08  B /Ti, when considering both interface charge transfer and cation intermixing.
In order to observe the contribution of La diffusion alone for the Ti moment, we considered a unit cell that replicates the interface of BTO and LSMO alone, as shown in Figure S20b (Supporting Information).In the bulk phase of BTO, Ba 2+ forces Ti to take +4 oxidation state to maintain the charge neutrality.Ti 4+ has 3d 0 valence configuration leading to zero magnetic moment.But when one La 3+ diffuses to the BTO region, to maintain charge neutrality one Ti 4+ becomes Ti 3+ .Now, Ti 3+ with 3d 1 electronic configuration gives rise to 1.0  B magnetic moment.To investigate further, we have considered bulk BTO and replaced 25% of Ba by La as shown in Figure S20b (Supporting Information).As expected, the Ti adjacent to the La dopants shows sizeable magnetic moments ≈0.1  B /Ti.This large magnetic moment on Ti is purely due to the La diffusion the BTO region.Which also explains the reduction of Ti 3+ population with the increasing concentration of Sr. Because, both Ba and Sr are in +2 oxidation which restricts the formation of Ti 3+ sub-lattices.It is important to highlight that, across all examined scenarios -including i) interfacial charge transfer alone, ii) cation intermixing alone, or both phenomena in conjunction -the levels of induced magnetism on Ti are at similar levels.It should also be noted that the actual degree of cation intermixing observed experimentally is 9.7%, which is significantly lower than the 25% estimated by our DFT calculations.This discrepancy is attributed to the constraints imposed by the limited size of the supercell used in our computational model.It is, therefore, reasonable to conclude that both interfacial charge transfer and cation intermixing influence the magnetic moments on Ti, contributing to the same order of magnitude.Furthermore, this impact diminishes as the concentration of Sr increases.

Conclusion
In summary, the strongly entangled interfacial effects and the resultant magnetic properties of the SLs constituted of BTO and LSMO (x) with various ground states were systematically investigated.The electron-doped 3d band of Ti ensures the emergence of weak magnetism at the interfacial Ti sites.Our experimental and DFT calculations reveal that the induced magnetism on Ti originates from the charge transfer process between Mn and Ti at the titanate/manganite heterointerfaces, as previously reported.However, it should be noted that a small amount of cation intermixing also plays a contributory role, which might have been previously overlooked.In addition, depending on the strain state of the SL, modulated by the Sr doping concentration of the LSMO (x) layers, the magnetic exchange coupling between Mn-Ti spins (Mn-Mn spins) at the interfaces can be switched, emphasizing the decisive potential of the orbital occupancy at the interface in tuning the exchange interactions.Our findings will pave a pathway to engineer the spin-related devices, such as MTJ, of which the performance is highly relied on the spin configurations at the interfaces.

Figure 1 .
Figure 1.Coherent epitaxial growth of the BTO/LSMO SLs.a) Cross-sectional high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) image and electron energy-loss spectroscopy (EELS) spectrum of the BaTiO 3 /La 1-x Sr x MnO 3 [BTO/LSMO (x)] superlattice (SL) with x = 0.20.The colored panels show the integrated intensities of Ba-M 4,5 , La-M 4,5 , Mn-L 2,3 , O-K, and Ti-L 2,3 edges, which indicate the distribution of elements across the interfaces.b) In-plane and out-of-plane lattice spacing profiles are obtained by averaging the lattice parameters on the whole width of the HAADF image.c) Reciprocal space mappings (RSMs) of BTO/LSMO SLs with x ranging from 0.10 to 0.70 around the substrate (103) reflections.The white dash-dot lines illustrate the L positions of the main reflections of the SLs.

Figure 2 .
Figure 2. Magnetic properties of BTO/LSMO SLs.a) The saturation magnetization (M S ) of the BTO/LSMO (x) SLs and LSMO (x) single films as a function of the Sr doping concentration (x).The corresponding M S differences (ΔM S ) between the SLs and the single films with identical x are also included for comparison.b) Typical X-ray absorption spectroscopy (XAS) and X-ray magnetic circular dichroism (XMCD) spectra at the Ti-(left panel) and Mn-L 2,3 (right panel) edges obtained from the SLs with various x at 80 K.The inset shows the experimental setup of the XMCD measurement.The circular polarized X-ray is incident grazingly (30°) onto the sample surface under ±1 T in-plane magnetic field.The XMCD signals were calculated from the difference between the I + and I − , where I + and I − denote XAS obtained from +1 T and −1 T, respectively.The integrations of XAS (light blue dot-dash line) and XMCD (red dash line) signals for x = 0.10 case are also included for reference.c) The corresponding Mn spin and orbital moments as well as Ti spin and orbital moments obtained from the application of the sum rules on the basis of the XAS and XMCD spectra in (b).
Mn-L 2,3 edges of BTO/LSMO SLs with various x.XMCD signals are obtained by calculating the difference between the I + and I − , where I + and I − denote the XAS intensity obtained under an inplane applied field of +1 T and −1 T, respectively.

Figure 3 .
Figure 3. Interfacial charge transfer across BTO/LSMO heterointerfaces.a) Ti L 2,3 X-ray isotropic absorption spectra of the SLs with various x.b) Energy difference between the e g and t 2g peaks of the Ti L 3 (green circles) and Ti L 2 (orange circles) absorption edges.The data have been plotted as a function of x.

Figure 4 .
Figure 4. Ti-and Mn-L 2,3 -edge XAS and XLD of BTO/LSMO SLs.a) The sketch of the XAS measurements using linear polarized lights with grazingincidence (GI) and normal-incidence (NI) configurations, respectively.The XAS intensity obtained from GI and NI is indexed as I c and I ab .b) Typical XAS and X-ray linear dichroism (XLD) spectra at the Ti-(left panel) and Mn-L 2,3 (right panel) edges measured from the SLs with various x.The XLD signal is obtained by subtracting I c from I ab , that is, XLD = I ab -I c .c) The integrated linear dichroism obtained from the XLD spectra in (b) normalized by the I ab + I c .The strain states of the BTO and LSMO layers are also included for comparison.

Figure 5 .
Figure 5. Oxygen octahedral rotation in BTO/LSMO SLs.a,b) Local oxygen octahedral coupling crossing the BTO/LSMO (x = 0.2) interfaces, including the HAADF a) and Intensity-inversed annular bright-field (ABF) images b) viewed from the [110] direction.Enlarged images show the MnO 6 octahedral rotation in LSMO region and Ti-O polar displacements in the BTO region, with the atomic model superimposed.c) Oxygen octahedral angles across the BTO/LSMO interfaces calculated from the ABF image.

Figure 6 .
Figure 6.Interfacial exchange coupling in BTO/LSMO SLs.a,b) Schematically illustrations of the strain states of the LSMO layers in the BTO/LSMO (x)SLs.With increasing x, the strain state of the LSMO layers changes from compressive strain a) to tensile strain b).c,d) Sketches of the orbital occupancy of Mn and Ti 3d electrons for the BTO/LSMO (x) SLs with compressive c) and tensile strain d), respectively.Note that the color-filled orbitals indicate the preferentially occupied Mn and Ti orbitals.e,f) Proposed interfacial spin configuration and magnetic coupling mechanism at the BTO/LSMO (x) heterointerfaces with compressive e) and tensile strain f), respectively.

Figure 7 .
Figure 7. BTO/LSMO Super lattice.a) 4 uc of BTO and 8 uc of LSMO.Here green, black, orange, blue, and pink atoms indicate, Ba, La, Sr, Ti, and Mn sites respectively.b) Layer projected DOS of (top panel) TiO 2 from BTO bulk, (middle panel) TiO 2 from BTO/LSMO SL, and (bottom panel) MnO 2 from LSMO bulk.Here the relative shift in the O-p states leads to charge transfer from LSMO to BTO region.Here grey shaded and, blue-shaded area indicates Ti and Mn projected states and red lines indicate O-p states of respective layers.c) 2 × 2 × 1 supercell of BTO/LMO with limited interfacial cation intermixing.d) Oxygen octahedral angles across the BTO/LMO interfaces with limited interfacial cation intermixing.