We studied ZrO2 − La2/3Sr1/3MnO3 pillar–matrix thin films which were found to show anomalous magnetic and electron transport properties. With the application of an aberration-corrected transmission electron microscope, interfacial chemistry, and atomic-arrangement of the system, especially of the pillar–matrix interface were revealed at atomic resolution. Minor amounts of Zr were found to occupy Mn positions within the matrix. The Zr concentration reaches a minimum near the pillar–matrix interface accompanied by oxygen vacancies. La and Mn diffusion into the pillar was revealed at atomic resolution and a concomitant change of the Mn valence state was observed.

In functional oxides such as superconducting cuprates and ferromagnetic manganites, the investigation of electrically insulating columnar oxide defects embedded in the conducting matrix materials has experienced an increasing research activity in the past decade.1,2 From the fundamental point of view, this activity was initiated by the exploration of the crystalline structure and elemental distribution at the interface between the columnar defects and the matrix material with the subsequent strain- and defect-induced modification of their electronic and magnetic properties. Elemental and electronic reconstructions at the interface between complex oxides with different functionalities are challenging phenomena to be explored. They give rise to interference effects between the states across interfaces with the potential to generate new properties and functionalities.3 Research on self-assembled vertically aligned nanocomposite thin films with two immiscible components hetero-epitaxially grown on single crystal substrates represents another branch of these activities.1,4 These structures have the advantage of utilizing the functionalities of both components with the possibility to tune the material properties by tailoring the interface-to-volume ratio, hetero-epitaxial strain, or modifying the cation valence state. From the application point of view, columnar non-conducting BaZrO3 or SrZrO3 defects in superconducting Y Ba2Cu3O7 thin films have been proven to enhance the flux-line pinning properties drastically and are used in several 3rd generation commercial superconducting tape fabrication technologies.5 Additionally, the thermoelectric figure of merit ZT = (S2σ/κ) T (with S being the Seebeck-coefficient, σ and κ being the electrical and thermal conductivities, respectively) can be enlarged by introducing precipitates and columnar defects, thus reducing the phonon part of the thermal conductivity in order to achieve an “electron crystal/phonon glass”-type material.6 

In ultrathin La2/3Sr1/3MnO3 (LSMO) films (thickness ∼45 nm) with a low density (0–6 mol. %) of ZrO2 precipitates an unusual low temperature resistivity increase associated with quantum interference effects of the electron waves was observed in conjunction with anomalous magnetic anisotropies.7–9 Evidently, electronic transport of LSMO is sensitively influenced by the presence of ZrO2 precipitates. Whereas up to now, only macroscopic properties of ZrO2 − LSMO pillar-matrix systems (charge transport and magnetism) have been studied, microscopic properties at the atomic level were not investigated at all. Here, we report the structure and interfacial chemistry of these precipitates, which are found to exist as pillars in the LSMO matrix. We use aberration-corrected scanning transmission electron microscopy (STEM) and simultaneous electron energy-loss spectroscopy (EELS) to reveal the structure, composition, and valence state at atomic resolution10 especially for the pillar–matrix interface. The work is regarded as a case study for similar pillar-matrix systems based on functional ceramics with insulating precipitates.

Zirconium oxide (ZrO2) and lanthanum strontium manganese oxide (La2/3Sr1/3MnO3, LSMO) were co-deposited epitaxially on (001) single-crystalline lanthanum aluminum oxide (LaAlO3, LAO) substrate by pulsed laser deposition. Stoichiometric amounts of LSMO and ZrO2 according to (1 − x) LSMO + xZrO2, with x = 0, 0.06, 0.2, and 0.3, were used. Details of the material growth process can be found in Ref. 7.

In order to obtain 3-dimensional information, electron transparent specimens for transmission electron microscopy (TEM) studies were prepared perpendicular (cross-sectional view) and parallel (plan-view) to the substrate by grinding, dimpling, and low temperature (under liquid nitrogen) argon ion thinning with a precision ion polishing system (PIPS, Gatan, model 691).

High resolution transmission electron microscopy (HRTEM) images from cross-sectional specimens were acquired using an image-aberration-corrected JEOL ARM 200CF microscope operated at 200 keV. High-angle annular dark-field (HAADF) images and EELS spectrum images from plan-view specimens were obtained using a probe-aberration-corrected JEOL ARM200CF microscope operated at 200 keV, equipped with a Gatan GIF Quantum ERS imaging filter with dual-EELS acquisition capability. The experimental convergence angle was 30 mrad for HAADF and EELS imaging. The corresponding inner and outer collection semi-angles for HAADF were set to 54-220 mrad. The inner and outer collection semi-angles for annular dark field (ADF) images acquired simultaneously during EELS spectrum imaging with Gatan ADF detector were 72–172 mrad, and the collection angle for EELS spectrum imaging was 72 mrad. Multivariate statistical analysis (MSA) was performed to reduce the noise of the EEL spectra with weighted principle-component analysis (PCA). From HAADF images dislocations have been identified and strain distributions were calculated using geometric phase analysis (GPA) software from HREM Research, Inc.

The pillar structure is verified for both plan-view and side-view specimens. Fig. 1(a) shows the top-view HAADF image of the ZrO2 (30 mol. %)−LSMO thin film with clear circular precipitates. The aberration-corrected HRTEM image [Fig. 1(b)] of the side-view specimen reveals a columnar structure within the film and the film thickness is measured to be about 45 nm. The phase separation manifests itself in a Moiré contrast arising from the overlap of ZrO2 and LSMO crystal lattices having different lattice parameters. It shows that the pillars extend all the way through the LSMO film until the substrate. Even though the vast majority of the pillars extend from the bottom to the top of the film, some pillars extend straight with a relatively sharp edge, while other pillars are bent.

FIG. 1.

(a) Annular dark field (ADF) image of a plan-view 70 mol. %LSMO-30 mol. %ZrO2 sample; (b) HREM image of the side-view specimen with 30 mol. % ZrO2; (c) three-color EELS spectrum image of the area shown in (a) with Zr-L2,3 in orange, Mn-L2,3 in blue and La-M4,5 in green; (d) a schematic view of the sample.

FIG. 1.

(a) Annular dark field (ADF) image of a plan-view 70 mol. %LSMO-30 mol. %ZrO2 sample; (b) HREM image of the side-view specimen with 30 mol. % ZrO2; (c) three-color EELS spectrum image of the area shown in (a) with Zr-L2,3 in orange, Mn-L2,3 in blue and La-M4,5 in green; (d) a schematic view of the sample.

Close modal

The EELS spectrum image of Fig. 1(a) is shown in Fig. 1(c) where the intensities of Zr − L2,3, La − M4,5, and Mn − L2,3 edges are displayed in orange, green, and blue, respectively. Oxygen is distributed everywhere and is not shown here. It is clear that pillar-shaped structures of ZrO2 have formed. They show circular or elliptical, sometimes faceted circumferences. The light orange regions close to the pillars and in the matrix are attributed to bent pillars, small pillars, or pillars ending inside the film. These regions were excluded from the following interface studies. Details of these regions are shown in the supplementary information25 (Fig. S1 and Fig. S2). A schematic view of the whole specimen is presented in Fig. 1(d).

From a statistical analysis of the pillar areas, a diameter of (5.1 ± 0.8) nm was deduced for the 30 mol. % ZrO2 specimen and of (3.7 ± 0.8) nm for the 20 mol. % ZrO2 sample.

Fig. 2(a) displays a representative high magnification HAADF image of a pillar with sharp interfaces. In the matrix region, brighter columns correspond to heavy La/Sr (A-site) ions and weaker columns to the lighter mixed Mn-O (B-site) ions. In the pillar region, only Zr columns are visible because oxygen columns are invisible in HAADF due to the small scattering cross section. Towards the interface to the matrix, the column brightness is reduced. As will be explained later, this is most likely related to the substitution of Zr by Mn. We note that bent pillars do not exhibit this type of contrast (see supplementary information Fig. S1 and Fig. S2) and were excluded from strain and elemental analysis.

FIG. 2.

(a) and (d) HAADF images of a plan-view 70 mol. %LSMO-30 mol. %ZrO2 sample. Fourier filtered images using (b) and (e) {100} reflections and (c) and (f) {010} reflections.

FIG. 2.

(a) and (d) HAADF images of a plan-view 70 mol. %LSMO-30 mol. %ZrO2 sample. Fourier filtered images using (b) and (e) {100} reflections and (c) and (f) {010} reflections.

Close modal

Since a [001]-oriented LAO substrate was used, the orientation relationship between the thin film and the substrate can be determined from the HAADF and HRTEM images of the top-view and cross-section specimens. To facilitate the correlation of crystallographic orientation between LAO, LSMO, and ZrO2, we use here LAO11 and LSMO12 in the tetragonal notation (space group I4/mcm, No. 140) and ZrO2 in the tetragonal system (space group P42/NMCS, No. 137).13 The corresponding crystal structure can be found in Fig. S3. With our low sample-preparation temperature of 770 °C,7 pure ZrO2 is expected to crystallize in the monoclinic phase14 which however does not match with the square atomic arrangement seen in the HAADF image [Fig. 2(a)]. We believe that tetragonal or cubic ZrO2 is formed by partial substitution of Zr by Mn atoms as will be explained in detail later. Here, we describe it in the tetragonal system to facilitate our description. Thus, the orientation relationships of LSMO and ZrO2 with respect to the LAO substrate are as follows:

LSMO[100]//LAO[100], LSMO[001]//LAO[001],

ZrO2[110]// LSMO[100], ZrO2[001]//LSMO[001].

The presence of faceted interfaces between ZrO2 and LSMO [Fig. 2(a)] indicates a preference towards {110}ZrO2/{100}LSMO and {100}ZrO2/{110}LSMO interfaces.

As reported in a previous paper on this group of material,9 the a and b values of LSMO are 5.4472 Å, 5.4483 Å, 5.4624 Å, and 5.4471 Å for LSMO films with 0 mol. %, 3 mol. %, 6 mol. %, and 20 mol. % ZrO2, respectively. This means that the LSMO crystal lattice expands by the addition of ZrO2, but partially relaxes again at high ZrO2 content.

We use the a and b values of LSMO with 20 mol. % ZrO2. For ZrO2, depending on sample preparation, a values range from 3.562 Å13 to 3.646 Å.15 Thus, LSMO and ZrO2 are expected to have a mismatch of about (5.3%–7.5%). For the 30 mol. % ZrO2 sample, by measuring the plane distances of ZrO2 pillars and surrounding matrix, we found a lattice mismatch between LSMO and ZrO2 of 5.6%–9.2%. These values are large enough to initiate the nucleation of misfit dislocations at the interfaces, which are confirmed by TEM. The visibility of misfit dislocations in the ab-plane can be enhanced by Fourier filtering selecting {100} and {010} reflections in the GPA analysis [Fig. 2]. For the pillar shown in Fig. 2(a), the misfit dislocations appear in pairs along a and b planes, as shown in Figs. 2(b) and 2(c). Figs. 2(e) and 2(f) show the dislocations along a and b planes for the pillars depicted in Fig. 2(d). It is directly visible that the dislocations along b planes are not paired. Choosing the x-axis parallel to [100] and y-axis parallel to [010] direction and using the matrix as reference area, we get the symmetric strain-field image, see supplementary Fig. S4. It is obvious that the compressive strain extends into the matrix regions where non-paired misfit dislocations occur, whereas the strain is relaxed for the matrix regions with paired misfit dislocations.

These examples show that the overall structure of the pillars has not reached elastic equilibrium yet. This is because, according to the lattice misfit, the pillar size has reached a value which only just enables nucleation of a misfit dislocation. Probably because of insufficient thermal activation (low deposition temperature, 770 °C (Ref. 7)) or high deposition rate, not every pillar has managed to nucleate a sufficient number of misfit dislocations and remains in a “superstrained” state with tensile stresses on the ZrO2 side of the interface. The formation of misfit dislocations is one way to relax the strain in a two-component system. Alternatively, strain can be accommodated by interdiffusion. This reduces the abruptness of the interfaces in terms of lattice misfit and chemical potential.16 In the present system, we found several evidences for interdiffusion which will be discussed below; (i) Zr atoms are detected in the LSMO matrix; (ii) Mn atoms are found within the ZrO2 pillars; (iii) a third Mn-rich phase is formed connecting adjacent pillars; details of the latter will be presented in a separate paper.

The atomic-resolution EELS spectrum image (SI) of the matrix region in the 30 mol. %ZrO2 sample (processed with weighted PCA to improve the signal-to-noise ratio of spectra at each pixel) is shown in Fig. 3. La and Sr [Figs. 3(b) and 3(c)] occupy the same locations as expected from the LSMO structure. Mn and Zr [Figs. 3(d) and 3(e)] take the same location which confirms that Zr is present in the matrix and occupies Mn/O column positions. The O map shows minimum values at La/Sr positions while being present everywhere else, which is as expected from the LSMO structure.

FIG. 3.

(a) ADF image of the LSMO matrix area in a plan-view 70 mol. %LSMO-30 mol. %ZrO2 sample. EELS spectrum image of (b) La-M4,5 (integration window 821–868 eV), (c) Sr-L2,3 (1935–2066 eV), (d) Mn-L2,3 (627–678 eV), (e) Zr-L2,3 (2218–2404 eV), and (f) O-K (520–578 eV).

FIG. 3.

(a) ADF image of the LSMO matrix area in a plan-view 70 mol. %LSMO-30 mol. %ZrO2 sample. EELS spectrum image of (b) La-M4,5 (integration window 821–868 eV), (c) Sr-L2,3 (1935–2066 eV), (d) Mn-L2,3 (627–678 eV), (e) Zr-L2,3 (2218–2404 eV), and (f) O-K (520–578 eV).

Close modal

To obtain an overview of the elemental distribution, EELS line profiles across several ZrO2/LSMO interfaces are shown in Fig. 4 along with an ADF image (Fig. 4(a)). The EELS line profiles are normalized to the maximum areal density for each element. The ADF image contrast in the region left of dashed lines of Fig. 4 is similar to that found in orange regions of Fig. 1 (also shown in Fig. S2). We therefore suppose that this area does not represent the pure LSMO matrix but is influenced by the existence of a small or inclined pillar. Neglecting this region, we find that La and Mn concentrations quickly drop within the pillar region, however not reaching zero concentration. This shows that La and Mn atoms are present in the zirconia lattice. Sr shows a less steep gradient across the interface than the other cations and is absent inside the pillar.

FIG. 4.

(a) ADF image of a plan-view 70 mol. %LSMO-30 mol. %ZrO2 sample. EELS spectrum line profile of areal density of (b) La-M4,5 and Sr-L2,3, and (c) Mn-L2,3 and Zr-L2,3 of the line drawn with identical integration window to Fig. 3 in (a). The areal densities are normalized to the maximum value for each element.

FIG. 4.

(a) ADF image of a plan-view 70 mol. %LSMO-30 mol. %ZrO2 sample. EELS spectrum line profile of areal density of (b) La-M4,5 and Sr-L2,3, and (c) Mn-L2,3 and Zr-L2,3 of the line drawn with identical integration window to Fig. 3 in (a). The areal densities are normalized to the maximum value for each element.

Close modal

A detailed elemental distribution from an area free from small or inclined pillars is shown in the atomically resolved EELS maps as displayed in Fig. 5 along with the simultaneously acquired ADF image. In the bulk regions, the elements are distributed as expected from the structure models of LSMO and ZrO2: within LSMO, La [Fig. 5(b)] and Sr columns [Fig. 5(c)] take the same positions, Mn-O [Figs. 5(d) and 5(f)] columns are located in the center of four La/Sr columns, Zr columns are located at Mn-O columns, which is consistent with the results presented in Fig. 3, and pure oxygen [Fig. 5(f)] is located between La/Sr positions. Mn is octahedrally surrounded by oxygen atoms. Along the 〈001〉-projection, mixed Mn–O columns and pure O columns exist. In the Mn map, the Mn–O columns are clearly resolved. Because of the limited spatial resolution of the oxygen map [Fig. 5(f)], the columns containing oxygen are not visible separately but appear as horizontal and vertical stripes in LSMO. Within ZrO2, Mn [Fig. 5(d)], Zr [Fig. 5(e)], and O [Fig. 5(f)] columns are atomically resolved. La is found to exist within the pillar, while the concentration is too low to be atomically resolved by this analysis (Fig. 5(b)). Mn is found to be located in the Zr columns. However, the gradual increase of the Zr concentration profile (Fig. 5(k)) and the ADF image intensity (Fig. 5(g)) from the interface to the pillar center suggests that lighter Mn substitutes heavier Zr inside the pillar, which contributes to the darker contrast at the interface and the stabilization of ZrO2.

FIG. 5.

(a) ADF image of a plan-view 70 mol. %LSMO-30 mol. %ZrO2 sample at the pillar-matrix interface (matrix, left of the dotted line; pillar, right of the dotted line). Atomic-resolution EELS spectrum image of areal density of (b) La-M4,5, (c) Sr-L2,3, (d) Mn-L2,3, (e) Zr-L2,3, and (f) O-K with identical integration window to Fig. 3 of the pillar-matrix interface area shown in (a).The integrated line profiles of images (a) to (f) are shown in (g) to (l) respectively. The integrated areal density was normalized to the maximum value in each profile. The x-axis in (g) to (l) is the location axis as shown in (a), and the y-axis is the normalized intensity.

FIG. 5.

(a) ADF image of a plan-view 70 mol. %LSMO-30 mol. %ZrO2 sample at the pillar-matrix interface (matrix, left of the dotted line; pillar, right of the dotted line). Atomic-resolution EELS spectrum image of areal density of (b) La-M4,5, (c) Sr-L2,3, (d) Mn-L2,3, (e) Zr-L2,3, and (f) O-K with identical integration window to Fig. 3 of the pillar-matrix interface area shown in (a).The integrated line profiles of images (a) to (f) are shown in (g) to (l) respectively. The integrated areal density was normalized to the maximum value in each profile. The x-axis in (g) to (l) is the location axis as shown in (a), and the y-axis is the normalized intensity.

Close modal

From the atomically resolved EELS maps and the ADF image, line profiles were obtained by integrating intensities in the direction parallel to the interface. These are displayed in Figs. 5(g)-5(l). We mark the interface position by a dotted line, right of which no Sr signal was detected. For the ADF image, we find an increasing intensity followed by a damping on the LSMO side when approaching the interface. Such a damping is also visible on the ZrO2 side of the interface. This damping is obvious in every ADF or HAADF image as a dark circular region around each pillar. Since HAADF image intensity is proportional to the atomic number, the intensity damping here can be directly correlated with the elemental distribution at the interface. Oxygen vacancies were found on both LSMO and ZrO2 sides of the interface, as can be seen from Figs. 5(f) and 5(l).

The valence state of Mn was checked by doing nonlinear least square fitting of a Gaussian peak to the background subtracted Mn L3 peaks, as shown in Fig. 6. Spectra from the matrix region and the pillar center [Fig. 6(a)] show Mn L3 peak positions of 639.75 eV and 638.25 eV, respectively, i.e., a difference of 1.5 eV. Comparison with literature data17 shows that this can be interpreted as a reduced valence state of Mn in the pillar. Applying the same procedure to the Mn spectrum image in Fig. 5(d) from the area in Fig. 6(b), which is a replica of Fig. 5(a), we obtain the Mn L3 peak position map in Fig. 6(c), showing a decrease of Mn valence state from the matrix to the pillar.

FIG. 6.

(a) Nonlinear Gaussian least square fitting on background subtracted Mn-L3 edge of the two spectra from pillar and matrix region of the plan-view 70 mol. %LSMO-30 mol. %ZrO2 sample; (b) Same as Fig. 5(a), clockwise rotated by 90°; (c) Mn-L3 peak positions obtained from Gaussian fitting of the background subtracted Mn-L3 spectrum image recorded from the area in (b).

FIG. 6.

(a) Nonlinear Gaussian least square fitting on background subtracted Mn-L3 edge of the two spectra from pillar and matrix region of the plan-view 70 mol. %LSMO-30 mol. %ZrO2 sample; (b) Same as Fig. 5(a), clockwise rotated by 90°; (c) Mn-L3 peak positions obtained from Gaussian fitting of the background subtracted Mn-L3 spectrum image recorded from the area in (b).

Close modal

The concentration profiles shown in Fig. 5 show enhanced La and Mn concentrations at the pillar-matrix interface. We note here that this enhancement is hardly visible in between close pillars. Sr is slightly depleted in this interface region whereas Zr is almost absent there. It seems that La occupies Sr positions close to the interface and the overall interface composition resembles LaMnO3−x. As shown in Fig. 6, the Mn valence at the interface gradually decreases as compared to the LSMO matrix, probably to a value between 2.5 and 3 for about two atomic columns before the interface line and to a value between 2 and 2.5 in the pillar. Because the La valence is 3+, charge balance can be obtained by introducing oxygen vacancies into the interface region. A loss of oxygen in this region is indeed shown in Fig. 5(l). Assuming a Mn valence of 2.5+, x would be 0.25, corresponding to an oxygen vacancy concentration of 8%.

It is known that Mn can stabilize the tetragonal or cubic structure of ZrO2 at low temperatures by varying Mn concentrations.18,19 Our observation of tetragonal or cubic ZrO2 pillars and the presence of Mn within the ZrO2 pillars indicate that this stabilization is a driving force for Mn dissolution within the pillars. This has also been observed by other groups in the past, e.g., Refs. 19 and 20. For charge balance reasons, oxygen vacancies have to accompany the MnZr substitution. This fits well with our observation of a loss of oxygen in the interface region where the concentration of MnZr is particularly high. The incorporation of oxygen vacancies permits stabilized zirconia to conduct O2− ions, provided there is sufficient vacancy mobility, a property that increases with temperature.

As a consequence of the MnZr substitution, the replaced Zr ions diffuse into the LSMO matrix. The detection of Zr in LSMO is consistent with former results,21 where Zr-substituted La0.7Sr0.3Mn1−xZrx O3 with 0 < x < 0.20 was investigated by neutron diffraction revealing that substitutional Zr4+ occupies the Mn site. Zr4+ most likely replaces Mn4+ ions,21 because of charge balance. However, one has to notice that the ionic radius of Zr4+ is 0.72 Å whereas that of Mn4+ is only 0.53 Å. This introduces strain and increases the lattice parameter of LSMO.9 This imposes an upper limit to the soluble Zr amount. At concentrations above this solubility limit, ZrO2 precipitates nucleate, leading to a reduction of strain because of the smaller lattice parameter of the ZrO2 precipitates compared to the LSMO matrix. Strain is then mainly localized at the precipitate–matrix interface.

Fig. S5 in the supplementary information shows that ZrO2 precipitates form even in the specimen with 6 mol. % ZrO2. This shows that the solubility of Zr in LSMO must be less than 6 mol. %, which corresponds to 6.4% Mn substituted by Zr in LSMO. There are literature data claiming higher solubility, e.g., 10%21 and 20%,22 but this is probably due to the higher processing temperature in these studies. As noted above, the limited solubility may partially be due to the strain imposed by the large Zr ions in the LSMO matrix. Moreover, as mentioned by Kim et al.,21 oxygen vacancies and interdiffusion may also play an important role. Oxygen vacancies have been found and studied before in LSMO.23 The concentration of oxygen vacancies are correlated with the sample preparation temperature and oxygen pressure. The existence of oxygen vacancies would convert some Mn4+ to Mn3+, thus reduce the number of Mn4+ positions that can be replaced with Zr4+. This is clearly confirmed by our results shown in Figs. 4 and 5: Zr reaches a minimum in the LSMO accompanied by oxygen vacancies. Therefore, the combination of charge balance, strain, and oxygen vacancies is the possible cause for the low Zr solubility in the matrix. These results are relevant for the equilibrium constitution of the ZrO2/Zr/LSMO system.

A discussion of the implications of the observed microstructure for electron transport properties is beyond the scope of the present article. Probably, these implications are manifold due to the delicate interplay of charge-balance effects induced by the substitutional Mn-Zr exchange and the inhomogeneous oxygen vacancy distribution. It is well known that in LSMO, electron transport is determined by the double-exchange mechanism between Mn atoms of different valences (Mn3+-O-Mn4+). The substitution of Mn with Zr can therefore be expected to influence this mechanism and thus the electronic transport in this system. Most importantly, however, the pillars act as scattering centers modifying the phase of the scattered electron wave function which can give rise to effects such as weak localization. Here, the pillar size and its densities play an important role. However, there might also be more subtle influences by the atomic substitutions mentioned above.

In summary, we have presented atomic-scale studies of the structure and chemistry of the ZrO2–LSMO pillar matrix system. We showed that ZrO2 precipitates form at concentrations as low as 6 mol. %. Precipitates mainly form pillars extending the entire LSMO film. Substantial interdiffusion is found at the LSMO–ZrO2 interface with Mn replacing Zr in ZrO2 (thus stabilizing the cubic or tetragonal phase) and Zr replacing Mn atoms in LSMO. Charge balance requires the combination of change of the Mn valence state and oxygen vacancy formation which are observed to segregate at the interface. Strain analyses show that the system has not yet reached elastic equilibrium. It is clear that LSMO as well as pillar regions is strained because of the misfit, however, also modified by the interdiffusion. In the La(1−y)Sry MnO3 system, the magnetic properties are directly related to the y value.24 Therefore, we believe that our results are not only relevant for the understanding the mechanisms of electron transport and magnetism in this material system, like the observed anomalous transport properties and localization transition, but also pave the way for a deeper microscopic understanding of the electronic properties of complex oxide interfaces at an atomistic level.

This work has been initiated by Professor J. Zhang (Department of Physics, Shanghai University). Samples have been prepared by Dr. Y. Gao during his Ph.D. work at Max Planck Institute for Solid State Research (MPI-FKF). The research leading to these results has received funding from the European Union Seventh Framework Program [FP/2007/2013] under Grant Agreement No. 312483 (ESTEEM2).

1.
H. J.
Liu
,
L. Y.
Chen
,
Q.
He
,
C. W.
Liang
,
Y. Z.
Chen
,
Y. S.
Chien
,
Y. H.
Hsieh
,
S. J.
Lin
,
E.
Arenholz
,
C. W.
Luo
,
Y. L.
Chueh
,
Y. C.
Chen
, and
Y. H.
Chu
,
ACS Nano
6
(
8
),
6952
(
2012
);
[PubMed]
A.
Imai
,
X.
Cheng
,
H. L. L.
Xin
,
E. A.
Eliseev
,
A. N.
Morozovska
,
S. V.
Kalinin
,
R.
Takahashi
,
M.
Lippmaa
,
Y.
Matsumoto
, and
V.
Nagarajan
,
ACS Nano
7
(
12
),
11079
(
2013
);
[PubMed]
H.
Zheng
,
Q.
Zhan
,
F.
Zavaliche
,
M.
Sherburne
,
F.
Straub
,
M. P.
Cruz
,
L. Q.
Chen
,
U.
Dahmen
, and
R.
Ramesh
,
Nano Lett.
6
(
7
),
1401
(
2006
).
[PubMed]
2.
V.
Moshnyaga
,
B.
Damaschke
,
O.
Shapoval
,
A.
Belenchuk
,
J.
Faupel
,
O. I.
Lebedev
,
J.
Verbeeck
,
G.
Van Tendeloo
,
M.
Mucksch
,
V.
Tsurkan
,
R.
Tidecks
, and
K.
Samwer
,
Nat. Mater.
2
(
4
),
247
(
2003
).
3.
H. Y.
Hwang
,
Y.
Iwasa
,
M.
Kawasaki
,
B.
Keimer
,
N.
Nagaosa
, and
Y.
Tokura
,
Nat. Mater.
11
(
2
),
103
(
2012
);
[PubMed]
J.
Mannhart
and
D. G.
Schlom
,
Science
327
(
5973
),
1607
(
2010
).
[PubMed]
4.
A. P.
Chen
,
Z. X.
Bi
,
H.
Hazariwala
,
X. H.
Zhang
,
Q.
Su
,
L.
Chen
,
Q. X.
Jia
,
J. L.
Macmanus-Driscoll
, and
H. Y.
Wang
,
Nanotechnology
22
(
31
),
315712
(
2011
).
5.
A.
Goyal
,
S.
Kang
,
K. J.
Leonard
,
P. M.
Martin
,
A. A.
Gapud
,
M.
Varela
,
M.
Paranthaman
,
A. O.
Ijaduola
,
E. D.
Specht
,
J. R.
Thompson
,
D. K.
Christen
,
S. J.
Pennycook
, and
F. A.
List
,
Supercond. Sci. Technol.
18
(
11
),
1533
(
2005
);
J. W.
Sinclair
,
Y. L.
Zuev
,
C.
Cantoni
,
S. H.
Wee
,
C.
Varanasi
,
J. R.
Thompson
, and
D. K.
Christen
,
Supercond. Sci. Technol.
25
(
11
),
115003
(
2012
).
6.
R. G.
Yang
,
G.
Chen
, and
M. S.
Dresselhaus
,
Phys. Rev. B
72
(
12
),
125418
(
2005
);
R.
Robert
,
S.
Romer
,
A.
Reller
, and
A.
Weidenkaff
,
Adv. Eng. Mater.
7
(
5
),
303
(
2005
).
7.
Y. Z.
Gao
,
G. X.
Cao
,
J. C.
Zhang
, and
H. U.
Habermeier
,
Phys. Rev. B
85
(
19
),
195128
(
2012
).
8.
Y. Z.
Gao
,
J. C.
Zhang
,
G. X.
Cao
,
X. F.
Mi
, and
H. U.
Habermeier
,
Solid State Commun.
154
,
46
(
2013
).
9.
Y. Z.
Gao
,
J. C.
Zhang
,
X. W.
Fu
,
G. X.
Cao
, and
H. U.
Habermeier
,
Prog. Nat. Sci.-Mater.
23
(
2
),
127
(
2013
).
10.
D.
Häussler
,
L.
Houben
,
S.
Essig
,
M.
Kurttepeli
,
F.
Dimroth
,
R. E.
Dunin-Borkowski
, and
W.
Jäger
,
Ultramicroscopy
134
,
55
(
2013
);
D. A.
Muller
,
L. F.
Kourkoutis
,
M.
Murfitt
,
J. H.
Song
,
H. Y.
Hwang
,
J.
Silcox
,
N.
Dellby
, and
O. L.
Krivanek
,
Science
319
(
5866
),
1073
(
2008
);
[PubMed]
S.
Lazar
,
Y.
Shao
,
L.
Gunawan
,
R.
Nechache
,
A.
Pignolet
, and
G. A.
Botton
,
Microsc. Microanal.
16
(
4
),
416
(
2010
);
[PubMed]
J. C.
Idrobo
,
W.
Walkosz
,
R. F.
Klie
, and
S.
Ögüt
,
Ultramicroscopy
123
,
74
(
2012
).
[PubMed]
11.
A. J.
Hatt
and
N. A.
Spaldin
,
Phys. Rev. B
82
(
19
),
195402
(
2010
).
12.
S. V.
Trukhanov
,
I. O.
Troyanchuk
,
I. A.
Bobrikov
,
V. G.
Simkin
, and
A. M.
Balagurov
,
J. Surf. Invest.: X-Ray, Synchrotron Neutron Tech.
1
(
6
),
705
(
2007
);
N.
Kallel
,
N.
Ihzaz
,
S.
Kallel
,
A.
Hagaza
, and
M.
Oumezzine
,
J. Magn. Magn. Mater.
321
(
15
),
2285
(
2009
).
13.
H. P.
Ding
,
A. V.
Virkar
, and
F.
Liu
,
Solid State Ionics
215
,
16
(
2012
).
14.
G.
Bayer
,
J. Am. Ceram. Soc.
53
(
5
),
294
(
1970
);
R. E.
Smallman
and
R. J.
Bishop
,
Modern Physical Metallurgy and Materials Engineering
, 6th ed. (
Butterworth-Heinemann
,
Oxford
,
1999
).
15.
N.
Igawa
and
Y.
Ishii
,
J. Am. Ceram. Soc.
84
(
5
),
1169
(
2001
).
16.
S. W.
Kang
,
H. J.
Park
,
T.
Kim
,
T.
Dann
,
O.
Kryliouk
, and
T.
Anderson
,
Phys. Status Solidi C
2
(
7
),
2420
(
2005
).
17.
H. Y.
Tan
,
J.
Verbeeck
,
A.
Abakumov
, and
G.
Van Tendeloo
,
Ultramicroscopy
116
,
24
(
2012
);
L. A. J.
Garvie
,
A. J.
Craven
, and
R.
Brydson
,
Am. Mineral.
79
(
5-6
),
411
(
1994
).
18.
S.
Saravanakumar
,
J.
Kamalaveni
,
M. P.
Rani
, and
R.
Saravanan
,
J. Mater. Sci.: Mater. Electron.
25
(
2
),
837
(
2014
);
M.
Lajavardi
,
D. J.
Kenney
, and
S. H.
Lin
,
J. Chin. Chem. Soc.
(Taipei, Taiwan) 47(5), 1055 (2000)
19.
C. C.
Appel
,
Ionics
1
(
5–6
),
406
(
1995
).
20.
T.
Kawada
,
N.
Sakai
,
H.
Yokokawa
, and
M.
Dokiya
,
Solid State Ionics
53
,
418
(
1992
).
21.
M. S.
Kim
,
J. B.
Yang
,
Q.
Cai
,
W. J.
James
,
W. B.
Yelon
,
P. E.
Parris
, and
S. K.
Malik
,
J. Appl. Phys.
102
(
1
),
013531
(
2007
).
22.
S.
Roy
and
N.
Ali
,
J. Appl. Phys.
89
(
11
),
7425
(
2001
).
23.
A. A.
Alharbi
,
M.
Alkahtani
, and
O.
Al-Dossary
,
AIP Conf. Proc.
1370
,
116
(
2011
);
S.
Picozzi
,
C.
Ma
,
Z.
Yang
,
R.
Bertacco
,
M.
Cantoni
,
A.
Cattoni
,
D.
Petti
,
S.
Brivio
, and
F.
Ciccacci
,
Phys. Rev. B
75
(
9
),
094418
(
2007
).
24.
S.
Majumdar
and
S.
van Dijken
,
J. Phys. D: Appl. Phys.
47
(
3
),
034010
(
2014
).
25.
See supplementary material at http://dx.doi.org/10.1063/1.4904819 for structure model and more TEM results, Figures S1-S5.

Supplementary Material