Interface‐Coupled BiFeO3/BiMnO3 Superlattices with Magnetic Transition Temperature up to 410 K

This research was funded by the Engineering and Physical Sciences Research Council, (EP/P50385X/1), the European Research Council (ERC-2009-AdG 247276 NOVOX). The work at Texas A&M was funded by the U.S. National Science Foundation (DMR-1401266). The work at Los Alamos was supported by the U.S. Department of Energy through the LANL/LDRD program and was performed, in part, at the Center for Integrated Nanotechnologies, a U.S. Department of Energy, Office of Basic Energy Sciences user facility. Use of the National Synchrotron Light Source, Brookhaven National Laboratory, was supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences, under Contract No. DE-AC02-98CH10886.

polarization is very small (0.1 µ C cm −2 at ≈90 K) or it is even paraelectric. [ 4 ] In BFO, strong FE behavior (Ferroelectric transition temperature, T C,FE ≈ 1103 K) coexists together with antiferromagnetism (AFM). BFO has an incommensurate cycloidally modulated G-type AFM structure with a large period of 62 nm (Néel temperature, T N ≈ 650 K). [ 5,6 ] Despite having ferroelectric and magnetic ordering at high temperatures, BFO has drawbacks for practical applications. Most notably the weak FM-like magnetic behavior results from destruction of its incommensurate magnetic structure leading to AFM. [ 6 ] With strong ferromagnetic behavior in BMO and strong ferroelectric behavior in BFO, various studies have suggested combining BMO and BFO in order to achieve room-temperature (RT) multiferroicity. [7][8][9][10] One approach is to modify the local spin confi guration of Fe by chemical doping at the B -site. [ 8,9 ] By 50% doping, this simple concept could result in ordered double perovskite ferrites of Bi 3+ 2 Fe (3+ n )+ TM (3− n )+ O 2− 6 (TM = transition metal: Cr, Mn, Co, Ni, and Cu) giving antiferromagnetically coupled high spin state of the TM and Fe. However, this concept has not been widely realized due to the lack of cation ordering as a result of negligible differences in ionic valences and ionic radii. [11][12][13][14][15] On the other hand, disordered double perovskites (notably BiFe 0.5 Mn 0.5 O 3 (BFMO) and Bi 2 CoMnO 6 (BCMO)) show a strongly enhanced T C albeit with lower magnetic moment than expected for the ordered structure. [ 16,17 ] Pálová et al. extended the idea of ordered double perovskites by theoretically studying BFMO with an ordered nanoscale checkerboard structure (NCB-BFO/BMO, see Figure 1 a), i.e., columnar B -site ordering. [ 10 ] Their calculations predict that an ordered NCB-BFO/BMO would be ferroelectric and ferrimagnetic (FiM) with M S of 3.8 µ B /Fe-Mn pair and T C of 406 K, as a result of magnetic ordering arising from the superexchange coupling between the neighboring AFM-Fe and FM-Mn. [ 10 ] Unfortunately, spontaneous columnar B -site ordering is not expected and high quality growth of artifi cial 1:1 BFO/BMO superlattices along the 〈110〉 direction is very hard.
Therefore, it is worth considering (001)-oriented BMO/BFO superlattices (SL-BFO/BMO) instead (Figure 1 b). Similar to NCB-BFO/BMO, interfacial Fe-Mn ferromagnetic ordering and a concomitant increase in T C can be expected in SL-BFO/ BMO. In this structure, small local variations in magnetic behavior will also be expected since the number of Fe-Mn nearest neighbors is halved for a 1:1 SL-BFO/BMO compared to NCB-BFO/BMO. [ 10 ] Various studies on short period perovskite superlattices have shown that the magnetic behavior can be enhanced by the interaction between the B -site cations. For example, Ueda et al. reported ferromagnetism at 375 and 230 K in (111)-oriented In the fi eld of multiferroics, the coexistence of magnetism and ferroelectricity in perovskite oxides (ABO 3 ) has drawn increasing interest due to their potential for multifunctional devices, for example, magnetoelectric random access memory, tunable multi functional spintronic devices including four-state memory devices, and spin fi lters. [1][2][3] Room temperature operating multiferroics are currently of intense scientifi c interest for these devices. [1][2][3] An important requirement for multiferroics is a strong coupling between the ferroic properties. Two of the most widely studied multiferroics are BiMnO 3 (BMO) and BiFeO 3 (BFO). BMO is one of the rare multiferroic materials that can possess both ferromagnetism (FM) and ferroelectricity (FE). It has a paramagnetic-ferromagnetic transition temperature ( T C ) of 105 K and a magnetic saturation moment ( M S ) of 3.6 Bohr magneton/formula unit ( µ B /f.u.). Unfortunately, its electrical 1:1 LaCrO 3 / LaFeO 3 and LaMnO 3 /LaFeO 3 superlattices, respectively, while the parent bulk materials are AFM. [ 18,19 ] Moreover, interfacial FM has been observed at 95 and 75 K in (001)-oriented CaMnO 3 /CaRuO 3 and LaNiO 3 /CaMnO 3 superlattices, respectively even though CaMnO 3 is AFM and both CaRuO 3 and LaNiO 3 are paramagnetic. [20][21][22] Additional changes in the B -B′ magnetic interactions may result from internal charge transfer from B to B′. [20][21][22][23][24] Internal charge transfer has been observed in, e.g., (001)-oriented LaNiO 3 /LaMnO 3 superlattices from Mn 3+ to Ni 3+ , similar to rock salt ordered La 2 NiMnO 6 . [ 23,24 ] As a result, interface-induced ferromagnetism was found in the paramagnetic LaNiO 3 layer. Moreover, the lattice strain from the substrate as well as the superlattice interfaces can affect the magnetic behavior as a result of changes in orbital ordering or bonding distortion of the B O 6 octahedra. [25][26][27][28][29] Here, in SL-BFO m /BMO m ( m = 1, 2, 4, and 8 unit cells (u.c.)), tuning Fe-Mn magnetic interactions through interfacial strain and charge engineering is demonstrated. Detailed analysis of both structural and magnetic features of superlattices using X-ray diffraction (XRD) and X-ray absorption spectroscopy (XAS) supports our conclusions. We show the presence of three magnetic transitions in SL-BFO/BMO. For all SL fi lms, T C1 is observed at around 145 ± 15 K, which is signifi cantly higher than the T C of bulk BMO (≈100 K). [ 4 ] m = 1, 2 superlattices even showed ferromagnetic behavior at room temperature with a T C2 of ≈410 K for m = 1 and ≈320 K for m = 2, whereas m = 4, 8 showed a T C,BMO of 96 K which is close to bulk BMO. Recently, a spin glass was observed for BFO/BMO multilayers with m = 1. [ 30 ] These samples were suggested to be relaxed and had a single out-of-plane fi lm lattice parameter of ≈3.93 Å which is considerably smaller than c pc = 3.970 ± 0.005 Å (pc is pseudocubic), as measured for our m = 1 fi lms, as shown later. This difference is likely the result of the use of different growth conditions since the creation of highly epitaxial in-plane strained BFO and BMO is diffi cult. We have found that it is necessary to carefully optimize the growth conditions to create the desired highly strained, heteroepitaxial fi lms with strong ferrimagnetic coupling between BFO and BMO.
To verify both the epitaxy as well as the phase purity of the superlattices of this work, we measured out-of-plane XRD scans (see Figure 2 a). The scans show that all superlattices grew epitaxially on SrTiO 3 (STO), showing pseudocubic (001) pc refl ections. [ 31 ] No superlattice peaks from the multilayer were observed; however, this does not indicate the lack of a high quality SL-BFO/BMO structure since any minor interface roughening reduces the intensity of these peaks dramatically due to the small difference between scattering factors of BMO and BFO. A minor contribution of monoclinic BMO (<1%) was observed in all fi lms. The monoclinic phase could be expected to form as a result of strain relaxation. [ 31 ] For the superlattices with m = 4, 8, a minor contribution of Bi 2 O 3 was present. Note that the impurity phases were not observed by transmission electron microscopy (TEM). TEM images across the SL layers showed a single uniform structure (see inset of Figure 2 a for a representative TEM image of a 1/1 SL-BFO/BMO taken around the substrate-fi lm interface).
To investigate the out-of-plane lattice parameters in more detail, symmetric reciprocal space maps (RSMs) around STO (004) were conducted (see Figure 2 b). For the short period superlattice ( m = 1), a single fi lm peak ( c pc = 3.970 ± 0.005 Å) corresponding to the SL-BFO/BMO was observed. The single peak indicates that the two lattices are strain coupled to give a single out-of-plane lattice parameter. The large c parameter is consistent with in-plane compression of the fi lm lattices by the smaller STO lattice ( a pc-bulk BMO = 3.935 Å, a pc -bulk BFO = 3.962 Å and a bulk STO = 3.905 Å), and consequent out-of-plane elastic lattice extension of both the BMO and BFO lattices.
As indicated by the dashed lines, with increasing m a clear separation of the out-of-plane fi lm peaks for BMO and BFO occurs. A distinct increase of the BFO out-of-plane lattice parameter was found (to c pc = 3.992 ± 0.005 Å) when m increased to 8. On the other hand, for BMO only a small decrease of the out-of-plane lattice parameter occurred with increasing m (from c pc = 3.970 ± 0.005 Å for m = 1 to c pc = 3.950 ± 0.005 Å for m = 8). The small decrease in c pc for BMO with increasing m is consistent with the BMO bulk lattice being closer to STO and hence being subject to less in-plane compression. This was explored further from asymmetric RSM scans around the STO (103) refl ection which reveals both the in-plane and out-ofplane strain states in the fi lms ( superlattices and full in-plane strain of both the BMO and BFO fi lms in the superlattice structure. Interestingly, the BMO ( H0L ) peak hardly relaxed with m , while some peak broadening of the BFO ( H0L ) peak was observed indicating some in-plane relaxation. This means that although small changes in the BMO structure were observed, the change in tetragonal distortion ( c / a ) of the BMO layer was small ( c / a = 1.018 for m = 1 to c / a = 1.013 for m = 8). For BFO a shift to lower reciprocal L value was also observed with increasing m , consistent with the downward shift of STO (004) refl ection of Figure 2 b. The results show that minor in-plane relaxation of the BFO occurs for m = 4, 8 and the out-of-plane lattice parameter increases toward bulk values, i.e., c pc of BFO increases with m . The appearance of in-plane relaxation with increasing m indicates that the BMO/BFO interface supports BFO to maintain the strain state. [ 32 ] To compare the magnetic properties of SL-BFO/BMO with bulk BMO, we measured the fi eld-cooled (FC, 500 Oe) temperature dependence of the magnetic moment ( M-T ) for all superlattices ( Figure 3 a). To determine the transition temperature, we undertook fi rst derivatives of the M-T curves ( dM/dT ). For m = 4 and 8, a magnetic transition is observed at 96 K (the value for bulk BMO, T C,BMO ). For all the fi lms, a clear magnetic transition is observed at around 145 ± 15 K ( T C1 ), which is signifi cantly higher than the T C of bulk BMO. Interestingly, for m = 1, 2, a second magnetic transition is observed above room temperature ( T C2 ), at 410 K for m = 1, and at 320 K for m = 2. To determine the origin of T C1 and T C2 , we deconvoluted the M-T curves into two different transitions by using the saturation magnetic moment from the M-H loops measured at various temperatures (see Figure S1 in the Supporting Information).
From the deconvoluted M-T curves, we can deduce that the high T C2 for m = 1, 2 originates from the change of the Mn environment in BMO. The environment of Mn in BMO for m = 1, 2 is different to m = 4, 8 since BMO is directly coupled to BFO, giving a single, strained BFO-BMO lattice ( Figure 2 ) similar to the NCB-BFO/ BMO lattice (Figure 1 a). T C1 is observed for all the fi lms. A plot of T C1 , T C2 , and T C,BMO versus m for the various superlattices is given in the inset of Figure 3 a. Figure 3 b shows the relationship between T C1 and the out-of-plane lattice parameter of BFO.
To get a better understanding of the magnetic behavior of these superlattices, we also measured the magnetic hysteresis loops   [ 14,19,33 ] The lower M S value of our superlattices is likely to be caused by complex effects resulting from a mixture of AFM and FM ordering which can arise from internal charge transfer leading to mixed ionic valences for Mn and Fe. [ 19 ] Also note that the in-plane BMO/BFO interface is present along the step edges. However, the effect of in-plane Fe-Mn interactions on the M S of our superlattices can be expected to be within the error of the measured values as the fraction is very small compared to the fraction of out-of-plane Fe-Mn interactions («1 %).
The superlattices show an increase in coercivity ( H C ) (Figure 3 c) compared to bare BMO fi lms grown on STO ( H C = 470 Oe). [ 33 ] While for the m = 8 superlattice the enhancement of H C was doubled ( H C = 900 Oe), for the m = 1 superlattice a much more dramatic enhancement was found ( H C = 5.6 kOe).
To clarify the origin of the increase of H C , we deconvoluted the experimental M-H loops into two different transitions (loop1 and loop2, see Figure S2 in the Supporting information). A squarer and narrower loop (loop1) is simulated by subtracting a trial function of the form M = M S tanh{ α ( H ± H C )} from the experimental data. The residual M -H loop (loop2) is of a reasonable form and has a much higher coercivity. For the m = 8 sample, the H C s are 400 Oe (loop1) and 4.5 kOe (loop2), and the H C of loop1 is close to that of plain BMO fi lms. [ 33 ] For the m = 1 sample, the M-H loop is decomposed into two M-H curves with H C s of 1.2 (loop1) and 9.2 (loop2) kOe. Therefore, it seems reasonable to assume that in both cases the loop1 relates to BMO. The enhanced H C for m = 1 can be understood by the general thickness dependence of H C which tends to decrease with increase of magnetic layer thickness. [ 34,35 ] The broader loop2 presumably originates from a ferromagnetic BFO phase with T C1 of ≈145 K for which the magnetic hardness can be understood in terms of an exchange interaction between FM-and a residual AFM-BFO matrix. [ 36,37 ] Adv. Mater. Interfaces 2016, 3, 1500597 www.advmatinterfaces.de www.MaterialsViews.com XAS near the Mn L 2,3 and Fe L 2,3 absorption edges was carried out to determine the Mn and Fe oxidation states (see Figure 4 a,b). A clear change in the fi ngerprint of the Mn spectra versus m is present, indicating a variation of the oxidation state of Mn. Comparing our Mn spectra with those of simple oxides, the line shape of the m = 8 superlattice spectrum is close to that of Mn 2 O 3 , indicating a predominance of Mn 3+ . [ 38,39 ] By decreasing m , the peak around 641 eV becomes stronger and well defi ned. Comparing the Mn-L edge peak shape of our m = 1 superlattice with those of simple oxides, the peak shape of Mn for the m = 1 superlattice is similar to the peak shape of Mn 3 O 4 , suggesting that approximately one-third of all Mn ions are fi lled by Mn 2+ . [38][39][40] Comparing the Fe spectra, a small change in line shape versus m is present as well. The spectrum of the m = 8 superlattice shows a sharp peak at ~710 eV with a clear shoulder at lower energy. This strongly suggests a predominance Fe 3+ valence with a small fraction of Fe 4+ . [ 40,41 ] By reducing m , the observed shoulder at lower energy is partially merged into the strong peak at 710 eV. Comparing our spectra with the literature, this change in line shape indicates an increase of Fe 4+ concentration when reducing m . [ 41,42 ] The observed trends for Fe and Mn valence states versus m are in agreement with each other.
Summarizing the XAS results, for the m = 1 superlattices, we fi nd clear mixed valence states of the B -site cations, Fe 3+/4+ and Mn 2+/3+ , while for m = 8 superlattices we have a predominance of Fe 3+ and Mn 3+ , but also some Fe 4+ and Mn 2+ are present. This suggests that charge transfer occurs from Fe to Mn near the interface. Considering the band gaps ( Δ ) of both BFO and BMO, charge transfer from BFO to BMO is remarkable since Δ BFO is larger than Δ BMO , 2.8 and 1.6 eV respectively. [ 33,43 ] An alternative explanation for the presence of Fe 3+/4+ and Mn 2+/3+ would be leakage of electrons from the Fe-site into the BMO layer. Such leakage is predicted to be on the order of 0.07 e − /Mn in order to stabilize interfacial FM in CaMnO 3 /CaRuO 3 , and is suggested to occur experimentally in SL of CaRuO 3 /CaMnO 3 and LaNiO 3 /CaMnO 3 . [20][21][22] Note that such a leakage has only been observed when CaRuO 3 and LaNiO 3 are metallic. BFO is an insulator, but BFO fi lms are often not highly resistive owing to the presence of defects in the structure, allowing small charge leakage. In addition, charge transfer has been suggested to occur at the interface of BFO and half-metallic La 0.7 Sr 0.3 MnO 3 to explain interface magnetism. [ 44 ] Adv. Mater. Interfaces 2016, 3, 1500597 www.advmatinterfaces.de www.MaterialsViews.com For all our superlattices, we found two magnetic transitions and clear magnetization at temperatures above the T C of BMO. For the very short period superlattices, m = 1, 2, magnetization was even found above room temperature. Since the change in tetragonal distortion of BMO was minimal and remained above 1, the variation in magnetic behavior is not expected to result from a change in Jahn-Teller distortion. [ 45 ] Therefore, to understand the origin of the magnetic states and the enhancement in T C , we consider the various superexchange interactions between Fe 4+/3+ and Mn 2+/3+ that may be present in our m = 1 superlattices, taking the Goodenough-Kanamori (GK) rule into account. [ 46 ] First of all, according to these rules, most of the possible interactions are AFM (Fe 4+ -O-Mn 3+ , Fe 3+ -O-Mn 2+ , Fe 3+ -O-Fe 3+ , Mn 2+ -O-Mn 2+ , Fe 4+ -O-Mn 2+ ). [ 14,46 ] Also the Fe 4+ -O-Fe 3+ interaction can be expected to be AFM as found in La 1− x Sr x FeO 3 . [ 47,48 ] The presence of these AFM interactions may only decrease the total M S of the superlattices. Therefore, to understand the magnetic behavior of the BMO/BFO superlattices we focus on the remaining interactions, ferromagnetic First of all, the Mn 3+ -O-Mn 3+ interactions present in BMO are likely to be present in the superlattices too. Normally, Mn 3+ -O-Mn 3+ interactions give rise to an AFM interaction. However, due to the presence of Bi 6s lone pairs, BMO has 3 dimensional (3D) magnetic structure, leading to FM behavior. [ 49 ] For the m = 4, 8 superlattices, we can expect bulklike BMO to be present. For these superlattices, a clear magnetic transition is present around 100 K, which is close to the T C of bulk BMO. Therefore, this transition likely originates from the Mn 3+ -O-Mn 3+ interactions and is named T C,BMO . For the m = 1,2 superlattices, we would expect the Mn 3+ -O-Mn 3+ interaction to be mainly in-plane. This also confi rms why no clear transition is found around 100 K.
Three dimensional ferromagnetic Fe 4+ -O-Fe 4+ interactions are observed in BaFeO 3 , leading to a T C of 115 K and a M S of 3.5 µ B /Fe. [ 50 ] Interestingly, the T C of BaFeO 3 , is fairly close to the T C1 of our SL-BMO/BFO (140-160 K). Moreover, BaFeO 3 shows an enhancement of T C with increasing unit-cell volume. Interestingly, this trend is similar to the behavior of T C1 in our superlattices where the unit cell volume of BFO increases with m as a result of the increase in out-of-plane lattice (Figures 2  and 3 c). These similarities strongly suggest that T C1 may originate from the FM Fe 4+ -O-Fe 4+ interactions. Therefore, let us take a closer look the found magnetic moment for T C1 . First of all, the deconvoluted M-T curves with T C1 show that magnetic moment increases when decreasing m (see Figure S1c in the Supporting information). This is in agreement with the increase of Fe 4+ content for lower m . The higher magnetic moment found for m = 2 compared to m = 1 can be explained by the greater 3D character of the Fe 4+ -O-Fe 4+ interactions for m = 2. Secondly, M S is 0.78 µ B /Fe at 10 K and 5 T from the deconvoluted M-H loop2 (as shown in Figure S2 in the Supporting Information) was found for m = 1, when the magnetic moment would come solely from Fe-Fe interactions. If we assume ~ 1 / 3 of the Fe ions to be Fe 4+ in the single layers of BFO, then we may expect a maximum magnetic moment of approximately 1.2 µ B /Fe for m = 1. However, a magnetic moment of 1.2 µ B /Fe would mean that Fe 3+ and Fe 4+ are fully separated within the layer. As Fe 3+ and Fe 4+ in-plane intermixing is expected, the presence of Fe 3+ -O-Fe 4+ AFM interactions would reduce the total magnetic moment, explaining the observed M S of 0.78 µ B /Fe for m = 1. The resemblance between BaFeO 3 and our superlattices, and the observation of Fe 4+ in the XAS spectra suggest that T C1 originates from ferromagnetic Fe 4+ -O-Fe 4+ interactions in BFO. [ 50 ] The origin of T C1 is schematically pictured in Figure 4 c.
As mentioned in the introduction, ferromagnetic interactions between AFM BFO and FM BMO are predicted to result in a T C of ≈406 K for NCB-BFO/BMO. [ 10 ] The high T C2 of up to 410 K in our very short period SL-BFO/BMO is very close to this predicted T C . SL-BFO/BMO fi lms have the same lowest energy magnetic ordering as NCB-BFO/BMO: ferrimagnetic ordering of AFM-Fe and FM-Mn (Figure 4 c). [ 10 ] The absence of T C2 for m = 4, 8 superlattices is consistent with the increasing out-of-plane lattice parameter in BFO structure with increasing m , and the separation of BFO and BMO into two distinct lattices (Figures 2 and 3 c) rather than a single strain BFO-BMO lattice. Here, the distortions of the B O 6 octahedra at the BMO/ BFO interfaces may be not suitable for optimum magnetic coupling. [25][26][27][28] Similar effects of strongly sensitive magnetic properties to structural distortion have been observed previously in superlattices of [(La 2/3 Ca 1/3 MnO 3 ) 4 /(BaTiO 3 ) 4 ] 20 (SL-LCMO/BTO) and [(La 2/3 Sr 1/3 MnO 3 ) 3 /(BaTiO 3 ) 3 ] 25 (SL-LSMO/BTO). Strong enhancements of T C were found in these strained superlattices, 1000 K for SL-LCMO/BTO, and 650 K for SL-LSMO/ BTO, whereas in relaxed bulk and plain thin fi lms of LCMO and LSMO lower T C s of only 250 and 370 K, respectively, were achieved. [ 25 ] Another possibility that we cannot discount for the origin of the high T C2 phase in our fi lms is double exchange FM coupling of Mn 2+ -O-Mn 3+ in BMO as a clear mixed valence of Mn 2+ and Mn 3+ was observed in the m = 1, 2 superlattices (Figure 4 a). According to several reports on electron doped LaMnO 3 , e.g., La 1− x Ce x MnO 3 , FM ordering with T C of 250 K has been reported. [ 46,47 ] Also, in La 1− x Ce x MnO 3 for x up to 0.3, T C increases as the amount of Mn 2+ increases. However, the T C of La 1− x Ce x MnO 3 is still much lower than T C2 of SL-BFO/BMO. But this could be due to a difference in B O 6 octahedral rotations which has been shown to affect the T C of LSMO significantly. [ 26 ] In addition, this double exchange FM coupling of Mn 2+ -O-Mn 3+ should result in metallic behavior below T C2 caused by electron hopping, just as for La 1− x Ce x MnO 3 with x = 0.3. [ 51,52 ] However, independent of m , all our superlattices are highly resistive, which indicates that the Mn 2+ -O-Mn 3+ is not the dominant interaction. Therefore, we strongly expect the ferrimagnetic interaction between AFM-BFO and FM-BMO to be the origin of T C2 .
In summary, in BFO/BMO superlattices of very high crystalline quality we observed above room temperature magnetic transition temperatures for very short period superlattices. A T C of 410 K was observed for m = 1, and a T C of 320 K was observed for m = 2. This T C phase, T C2 , agrees remarkably well to a similarly structured nanoscale checkerboard BFO/BMO lattice, which was shown by theory to have a T C of 406 K as a result of FM BMO-AFM BFO interactions. A lower T C , T C1 , was observed in all the superlattices ( m = 1-8 u.c.) and can be explained by ferromagnetic Fe 4+ -O-Fe 4+ interactions in BFO. Artifi cial superlattices of ferroelectric BFO and BMO with interface-induced high temperature magnetism are an important prerequisite for creating practical magnetoelectric systems.

Experimental Section
Film Fabrication : BFO m /BMO m superlattices ( m = 1, 2, 4, 8 u.c.) of ≈50 nm total thickness were grown on (001)-oriented STO substrates by pulsed laser deposition using a KrF laser ( λ = 248 nm). Commercial ceramic BFO and BMO targets were ablated at a fl uence of 1.6 J/cm 2 and a repetition rate of 2-5 Hz. The substrate to target distance was fi xed at 12 cm to ensure a slow growth rate which was needed to maintain coherent growth. [ 31 ] During growth the substrate temperature was ≈640 °C and the oxygen pressure was fi xed at 100 mTorr. The fi lms were cooled down in 300 Torr O 2 by switching off the heater. To determine the growth rates of BMO and BFO, the thicknesses of single phase BFO and BMO fi lms were calibrated by X-ray refl ectivity (XRR). The growth rates, 0.01 Å/pulse for BFO and 0.08 Å/pulse for BMO, were subsequently set to control the fabrication of the superlattices. It was found that the BMO did not grow layer-by-layer which could be expected since the growth is complicated by the highly volatile Bi as well as the large lattice mismatch between pseudocubic BMO and STO. [ 31 ] It is noted that the interface roughening of our superlattices increased with fi lm thickness as confi rmed by XRR.
Magnetic Analysis : Magnetization measurements versus temperature and fi eld ( H ) were performed using a vibrating sample magnetometer (VSM, Princeton) and a superconducting quantum interference device (SQUID) magnetometer (Quantum Design, MPMS). To investigate the valence state of Fe and Mn in the fi lms, XAS was performed on beamline U4B at the National Synchrotron Light Source (NSLS), Brookhaven National Laboratory. The spectra were measured by total electron yield (TEY) mode at 300 K after pulsing at 9.5 kOe. To prevent charging, silver paste was used at the corner of the sample to create a conducting path.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.