Visualizing the interfacial evolution from charge compensation to metallic screening across the manganite metal-insulator transition

Abstract

Electronic changes at polar interfaces between transition metal oxides offer the tantalizing possibility to stabilize novel ground states yet can also cause unintended reconstructions in devices. The nature of these interfacial reconstructions should be qualitatively different for metallic and insulating films as the electrostatic boundary conditions and compensation mechanisms are distinct. Here we directly quantify with atomic-resolution the charge distribution for manganite-titanate interfaces traversing the metal-insulator transition. By measuring the concentration and valence of the cations, we find an intrinsic interfacial electronic reconstruction in the insulating films. The total charge observed for the insulating manganite films quantitatively agrees with that needed to cancel the polar catastrophe. As the manganite becomes metallic with increased hole doping, the total charge build-up and its spatial range drop substantially. Direct quantification of the intrinsic charge transfer and spatial width should lay the framework for devices harnessing these unique electronic phases.

ARTICLE

Received 2 Jul 2013 | Accepted 18 Feb 2014 | Published 17 Mar 2014

Julia A. Mundy1, Yasuyuki Hikita2, Takeaki Hidaka3, Takeaki Yajima2,4, Takuya Higuchi3, Harold Y. Hwang2,5, David A. Muller1,6 & Lena F. Kourkoutis1,6

Electronic changes at polar interfaces between transition metal oxides offer the tantalizing possibility to stabilize novel ground states yet can also cause unintended reconstructions in devices. The nature of these interfacial reconstructions should be qualitatively different for metallic and insulating lms as the electrostatic boundary conditions and compensation mechanisms are distinct. Here we directly quantify with atomic-resolution the charge distribution for manganitetitanate interfaces traversing the metalinsulator transition. By measuring the concentration and valence of the cations, we nd an intrinsic interfacial electronic reconstruction in the insulating lms. The total charge observed for the insulating manganite lms quantitatively agrees with that needed to cancel the polar catastrophe. As the manganite becomes metallic with increased hole doping, the total charge build-up and its spatial range drop substantially. Direct quantication of the intrinsic charge transfer and spatial width should lay the framework for devices harnessing these unique electronic phases.

DOI: 10.1038/ncomms4464

Visualizing the interfacial evolution from charge compensation to metallic screening across the manganite metalinsulator transition

1 School of Applied and Engineering Physics, Cornell University, Ithaca, New York 14853, USA. 2 Stanford Institute for Materials and Energy Sciences, SLAC National Accelerator Laboratory, Menlo Park, California 94025, USA. 3 Department of Advanced Materials Science, The University of Tokyo, Kashiwa, Chiba 277-8561, Japan. 4 Department of Materials Engineering, The University of Tokyo, Tokyo 113-8656, Japan. 5 Geballe Laboratory for Advanced Materials, Department of Applied Physics, Stanford University, Stanford, California 94305, USA. 6 Kavli Institute at Cornell for Nanoscale Science, Ithaca, New York 14853, USA. Correspondence and requests for materials should be addressed to L.F.K. (email: mailto:[email protected]

Web End [email protected] ).

NATURE COMMUNICATIONS | 5:3464 | DOI: 10.1038/ncomms4464 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 1

& 2014 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms4464

The interface between transition metal oxides is a playground for stabilizing novel phases not apparent in the parent compounds17. This is evidenced by the observation

of conductivity1 and later superconductivity4 and magnetism8 at the interface between two non-magnetic insulators LaAlO3 and

SrTiO3 as well as novel magnetic phases2 and correlated electron phases7. Manganites, such as La1 xSrxMnO3, exhibit many of the

exotic bulk phases seen in the complex oxides, with paramagnetic insulating (xr0.2) and ferromagnetic metallic ground states (x40.2) at room temperature and additional charge and orbital ordered phases accessible at low temperatures9,10. Next generation spin tunnel junctions could be formed by exploiting the room temperature ferromagnetism in ultrathin layers of La0.67Sr0.33MnO3 (ref. 11). Observed interfacial dead layers have,

however, limited the scaling of potential devices1214. While explicit attempts15,16 to offset the interface dipole17,18 and to eliminate extrinsic defects14 have reduced the observed dead layer thickness, direct mapping of these intrinsic electronic and magnetic reconstructions at the interface can be obscured by variations in growth techniques leading to competing defects.

The intrinsic reconstructions at the interface between an insulating polar lm and non-polar substrate, for example a LaMnO3 lm on a SrTiO3 {100} substrate, are typically thought of in terms of the polar catastrophe model19,20: the presence of alternating positively and negatively charged layers on a non-polar surface causes the build-up of a non-zero electric eld and an electric potential, which diverges with sample thickness. The

polar catastrophe can be alleviated by transferring charge to the interfacemanifest as either the presence of an extra half of an electron or hole at an n-type (LaO/TiO2) and p-type (MnO2/SrO)

interface, respectivelyyet also through cation or oxygen vacancies and/or atomic displacements21. Note that oxygen vacancies in SrTiO3, for example, can give rise to itinerant electrons22,23 yet can only cancel the polar catastrophe for p-type interfaces20. For metallic and insulating lms, the nature of the interfacial reconstructions should be qualitatively different as the electrostatic boundary conditions and compensation mechanisms are distinct. Some metal/semiconductor oxide interfaces, for example metallic non-polar SrRuO3 lms on SrTiO3 substrates24, can be well described as ideal Schottky barriers. Others, such as conducting polar La1-xSrxMnO3/SrTiO3 interfaces15,24, require additional dipoles to compensate for the interface charge. This interface dipole could take the form of an extra screening charge or atomic displacements of the constituent atoms.

Studying these electronic reconstructions at La1-xSrxMnO3/

SrTiO3/SrTiO3 interfaces provides an ideal platform to probe the interface across the manganite metalinsulator transition. The large changes in electronic properties are induced by continuous hole doping while maintaining the same crystal topology via the Mn-O backbone. The interfacial changes in the composition and bonding information can be exquisitely probed at the atomic-scale with scanning transmission electron microscopy (STEM) in combination with electron energy loss spectroscopy (EELS)25.

a

b c d

La

HAADF Mn Ti

Potential [MnO2]0.9

[MnO2]0.9

[MnO2]0.9

[MnO2]0.9

[SrO]0

[TiO2]0

[TiO2]0

e f

[La/Sr-O]+0.9

[La/Sr-O]+0.9

[La/Sr-O]+0.9

[La/Sr-O]+0.9

Figure 1 | EELS spectroscopic mapping of the La0.9Sr0.1MnO3/SrTiO3 interface. Simultaneously recorded high-angle annular dark-eld STEMimage (a) and false-colour La, Mn and Ti elemental concentration maps (bd) of the interface. (e) Combined elemental map with Ti in blue, Mn in green and La in red. (f) Cartoon of an atomically abrupt La0.9Sr0.1MnO3/SrTiO3 interface with the diverging electric potential due to the polar discontinuity at the interface. A corresponding area in the spectroscopic map is indicated by a white box in (e). Scale bar, 1 nm.

2 NATURE COMMUNICATIONS | 5:3464 | DOI: 10.1038/ncomms4464 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2014 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms4464 ARTICLE

In the single-particle picture, the energy loss due to inelastic scattering from core-level transitions provides a site-specic measurement of the local density of states above the Fermi level, partitioned by chemical species and angular momentum due to the dipole selection rules26. For transition metal L2,3 edges of relevance for probing the multivalent titanium and manganese atoms, however, strong core-hole effects produce a signicant deviation between the local density of states and the observed energy loss near edge ne structure26. Nevertheless, comparison with reference spectra has permitted atomic-resolution two-dimensional mapping of oxidation states2729.

Here, we provide a systematic study of a series of La1 xSrx MnO3/Nb:SrTiO3 (LSMO/STO) interfaces from x 0 to

x 0.5, prepared under identical growth conditions, to disen

tangle the extrinsic and intrinsic effects at the interface through a range of states on the manganite phase diagram. By measuring the concentration and valence of all cations in the system using EELS, we explicitly quantify the electronic charge transfer apparent as a cation valence change. We show a quantitative agreement between the polar catastrophe model and the total electron transfer to the manganese sites of the insulating LSMO lms, yet a considerable drop in the total charge transfer and spatial extent with the onset of screening in the metallic La0.7Sr0.3MnO3.

ResultsElemental mapping of the La0.9Sr0.1MnO3/SrTiO3 interface.

We investigate thin lms of La1 xSrxMnO3 with x 0, 0.1, 0.2,

0.3 and 0.5 grown on SrTiO3. To demonstrate the ability of EELS to chemically map valence changes, interdiffusion and vacancies at atomic-resolution, we rst investigate the La0.9Sr0.1MnO3/SrTiO3

interface. The high-angle annular dark-eld STEM image shown in Fig. 1a indicates that the interface is coherent and free of dislocations. From the EELS elemental maps in Fig. 1be, we note an asymmetry in cation intermixing as the B-site manganese/titanium sublattice only shows one monolayer of intermixing, while the A-site lanthanum shows considerable interdiffusion over approximately four monolayers (Supplementary Fig. 1). The polar

discontinuity shown in Fig. 1f and Supplementary Fig. 2 cannot be alleviated by A-site interdiffusion in and of itself, however, asymmetric diffusion could reduce (or enhance) the interfacial potential offset (Supplementary Fig. 3). Surprisingly though, preferential diffusion between the A-site cations for the La0.9Sr0.1O/

TiO2 interface without changes in the B-site cation valence serves to increase the potential build-up rather than decrease it. On the other hand, cation vacancies could provide negatively charged defects to reduce the potential build-up in the lm30,31. Despite the observed asymmetric interdiffusion, the total cation concentration across the interface does not deviate by more than 35% from that of a cation vacancy-free interface, even for the most diffuse interface, as shown in Supplementary Figs 4 and5. There are mechanisms whereby local deviations from stoichiometry can alter the local cation valence. For instance, Mn 2 could substitute on the La (Sr) site in La-decient

LaxMnO3-d (refs 32,33). From the measured deviations in (La Sr) composition proles, this could account for up to

35% of the A-sites, or up to 611% of the Mn 2 that we observe spectroscopically. Therefore, extrinsic defect-driven mechanisms, while possibly present, do not play a dominant role in our lms.

Charge mapping at the La0.9Sr0.1MnO3/SrTiO3 interface. As the polar discontinuity was not alleviated by cation vacancies, we now turn to charge modulation. The Mn-L2,3 edge binned parallel to the interface is shown in Fig. 2a; near the interface the major peaks shift to lower energies, indicative of a reduced manganese valence state. From this edge, two distinct components were extracted with multivariate curve resolutiona bulk Mn3.1 spectra and a lower valence stateas shown in Fig. 2b. A nonnegative non-linear least squares t to the full Mn-L2,3 edge

reveals the presence of a signicant fraction of a lower valence state at the interface as shown in the binned line prole in Fig. 2c and full 2-D t in Fig. 2d. The residual from the t is presented in Supplementary Figs 6 and 7. A similar analysis of titanium did not yield a change in the valence below the nominal Ti 4 of bulk

SrTiO3 as discussed in Supplementary Fig. 8. This indicates that

a

c d e

5

1

4

Position (nm)

0.5

3

0

2

Position (nm)

0.5

b

1

Interface

0

Bulk

Intensity (a.u.)

1

2

3

4

640 645 650 655 Energy loss (eV)

Concentration (a.u.)

Figure 2 | Reduced Mn valence at the La0.9Sr0.1MnO3/SrTiO3 interface. (a) The background-subtracted and gain-corrected Mn-L2,3 edge spectra averaged parallel to the interface. (b) Two distinct components are extracted by multivariate curve resolution. For the interface component, the major peaks of the Mn L-edge are shifted to lower energies, indicative of a reduced Mn valence. The Mn spectra across the spectroscopic image are t to the two components as shown binned in (c) and in the full 2-D map in (d). (e) Concentration map plotting Ti, Mn and La in blue, green and red, respectively. Scale bar, 1 nm.

NATURE COMMUNICATIONS | 5:3464 | DOI: 10.1038/ncomms4464 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 3

& 2014 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms4464

the majority of the electronic transfer to the interface resides on the manganese sites in our samples rather than titanium sites as was claimed for LaMnO3/SrTiO3 superlattices34. Note that diffusion of a manganese atom onto a titanium site should drive the manganese valence up from the nominal bulk valence of

3.1 towards the titanium valence of 4. Experimentally,

however, we show that the interfacial manganese valence instead decreases. Similarly, intermixing on the A-site, as observed in Supplementary Fig. 5, would place Sr atoms on the La sites in the LSMO lms. Without changes in the B-site cation valence, this once again would serve to increase the interfacial manganese valence, in contrast to the experimentally observed decrease. Thus, the changes in manganese valence observed cannot be interpreted by interdiffusion-driven valence changes.

Charge quantication across the metalinsulator transition. To probe how the charge transfer varies across the metalinsulator transition, this analysis was extended to the full series of LSMO lms from x 0 to x 0.5. The 2-D concentration maps shown in

Fig. 3a indicate that the considerable A-site intermixing at the LaMnO3/SrTiO3 interface did not persist for the strontium-doped lms. To characterize the manganese valence at the interface, we performed a non-negative non-linear least squares t of the manganese signal across the interface to reference spectra for Mn3, Mn 3.5 and Mn2, Fig. 3b, obtained from the bulk

LaMnO3, La0.5Sr0.5MnO3 lms and the literature35, respectively. We note that while the Mn3.5 spectrum represents a superposition of Mn3 and Mn 4 spectra, the mathematical decomposition of the full data set into this spectral basis is

equivalent to that which could be achieved using a basis including the Mn4 spectrum. The residuals from the ts are presented in

Supplementary Fig. 9, demonstrating the reliability of the ts across the series of samples with the basis spectra chosen. A statistically signicant concentration of Mn 2 is observed at the interface for xo0.3. Furthermore, the line proles, Fig. 3c, show a signicant change in manganese valence near the interface, even in the absence of the Mn2 component for the x 0.3 interface.

The total electronic charge transfer to the manganese ions near the interface can be calculated by subtracting the measured valence from the nominal bulk. Figure 4a shows the computed charge transfer across the interface, which is integrated to nd the total charge transfer in Fig. 4b. We demonstrate a quantitative agreement between the polar catastrophe electronic reconstruction model in Supplementary Fig. 2 and the measured charge on the interfacial manganese sites for xr0.2. For xZ0.3, signicantly less charge is measured with no statistically signicant electronic transfer for x 0.5 despite the prediction of 0.25 e .

We note that this sharp discontinuity as a function of doping in the measured charge build-up at the interface coincides with the room temperature metalinsulator transition at x 0.3. In

metallic lms, electronic charge transfer to the interface is expected to equilibrate the chemical potentials and to balance interface charges. The magnitude of the interface charge can be different from the polar catastrophe model because the simple ionic picture is no longer valid to describe the electrostatic potential. It is, therefore, not surprising that the magnitude of the compensating charge required for the metallic lms deviates from that expected for insulating lms.

a b

x = 0

x = 0.1 x = 0.2 x = 0.3 x = 0.5

Mn+3.5

Mn+3

Intensity (a.u.)

Mn+2

640 650 660

Energy loss (eV)

1 0.5 1 0.5 1

c

4

Position (nm)

3 2 1 0 1 2

0 0.5 1 0.5 1 0.5

Normalized concentration (a.u.)

x = 0

x = 0.1 x = 0.2 x = 0.3 x = 0.5

Figure 3 | Mn valence changes at the series of La1 xSrxMnO3/SrTiO3 interfaces. Spectroscopic images for x 0, 0.1, 0.2, 0.3 and 0.5 lms are

shown left to right in (a). Ti is plotted in blue, Mn in green and La in red. The x 0.5 image shows cation ordering not observed in the other lms. (b) Three

Mn reference spectra for Mn2, Mn 3 and Mn3.5, used to determine the Mn valence across the interface. (c) The results of the non-negative non-linear least squares t of the components in (b) for x 0, 0.1, 0.2, 0.3 and 0.5 ordered left to right. Error bars plot the s.e. of the mean generated

from ve binned regions from each spectroscopic image. Scale bar, 1 nm.

4 NATURE COMMUNICATIONS | 5:3464 | DOI: 10.1038/ncomms4464 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2014 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms4464 ARTICLE

a b c

x = 0 x = 0.1 x = 0.2 x = 0.3 x = 0.5

0.5

0.4

0.3

0.2

0.1

0.1 0.2 0.3

Insulator

10

Total excess charge Spatial distribution

Insulator

5

Position (unit cells)

5

0

Total excess charge (e)

Polar discont. model

4

FWHM (unit cells)

3

2

Metal

Metal

5 0 0.1 0.1

Excess charge (e)

0 0 0.1 0.2 0.3 0.4 0.5

0

0

0 0.1

0 0.1

0 0.1

0

Figure 4 | Excess charge accumulation at the manganitetitanate interfaces. The excess charge can be computed from the deviation between the nominal bulk and the measured Mn valence across the interface. (a) The total excess charge per unit cell is plotted across the interface for x 0, 0.1, 0.2,

0.3 and 0.5 ordered left to right. The 0 position denotes the interface, dened by the midpoint of the Mn concentration prole. A Gaussian t was added as a guide to the eye. (b) The total excess charge integrated across the interface and compared with the polar discontinuity model. For insulating manganite lms, the total charge observed quantitatively agrees with that needed to cancel the polar catastrophe. As the manganite becomes metallic for xZ0.3, the total charge build-up drops substantially. (c) The full width at half maximum of the Gaussian charge proles from a shows a discontinous drop at the insulator-metal transition. A dashed line is added through the mean width for the insulating samples at 4.4 unit cells. Error bars are the s.e. of the mean.

0.4 0.5

1

Sr doping (x)

Charge delocalization across the metalinsulator transition. Finally, we note that not only does the total electronic charge transfer to the interface change in the proximity of the metalinsulator transition, but also the width of the reconstructed region. As shown in Fig. 4c, despite the decrease in transferred charge with increasing x, the width of the electronic reconstruction region remains roughly constant for xr0.2 with the width of the charge proles constant at 4.4 unit cells. Surprisingly, the signicant intermixing observed for the LaMnO3/SrTiO3 interface in comparison to the other lms did not impact the width of charge transfer. In contrast, the small amount of charge transferred to the x 0.3 interface has a narrower width with mod

ulations observed only over two unit cells. This length scale in the metallic state is consistent with previous theoretical36 and experimental37 measurements of screening lengths of B1-2 unit cells for x 0.33 LSMO lms. Thus, both the magnitude

and the spatial width of the excess electrons at the LSMO/STO interface are tuned across this phase transition.

DiscussionIn summary, we probed a series of La1 xSrxMnO3 thin lms on

SrTiO3 ranging from the paramagnetic insulating to ferromagnetic metallic regions of the phase diagram. By mapping the valence changes of the manganese sites across the interface, we show a quantitative agreement between the total electron transfer to the interface for the insulating region of the phase diagram and a signicantly reduced charge transfer for the metallic lms. Moreover, we measured the spatial extent of the resultant excess electrons residing at the interface. Despite differences in chemical intermixing, the electronic reconstruction width remains constant at about four unit cells for the insulating lms, yet narrows to two unit cells as LSMO becomes metallic. The engineering of the spatial extent of the interface charge should be useful in the construction of novel electronic phases at oxide interfaces.

Methods

Thin lm fabrication. We investigate thin lms of La1 xSrxMnO3 with x 0, 0.1,

0.2, 0.3 and 0.5 grown by pulsed laser deposition on TiO2-terminated 0.01 wt% Nb-doped SrTiO3 {100} substrates. For each structure, the substrate was pre-annealed under 5 10 6 Torr of oxygen partial pressure (PO ) at 950 C for 30 min after

which the La1 xSrxMnO3 (LSMO) thin lm was deposited at a laser uence of

0.4 J cm 2 using a KrF excimer laser at a growth temperature of 750 C and PO of

1 10 3 Torr.

Electron microscopy and spectroscopy. Cross-sectional TEM specimens were investigated on the 5th-order aberration corrected 100 keV Nion UltraSTEM equipped with an Enna spectrometer. The microscope conditions were optimized for EELS spectroscopic imaging with a probe size of B1 , EELS energy resolution of B0.6 eV and a dispersion of 0.3 eV per channel. Elemental concentrations and bonding information were extracted from the simultaneous acquisition of the Ti-L2,3, Mn-L2,3, O-K and La-M4,5 edges. To measure the concentration of all the cations in the system, the Nion UltraSTEM, now equipped with a Quena Dual-EELS spectrometer, was used to measure the Sr-L2,3 edge

simultaneously with the Ti, Mn and La edges. For all samples, the Ti signal from the substrate and the La edges were used to accurately calibrate the energy shift and dispersion between the samples. Further details of the data acquisition and processing is given in Supplementary Discussion and the corresponding Supplementary Fig. 10.

References

1. Ohtomo, A. & Hwang, H. Y. A high-mobility electron gas at the LaAlO3/

SrTiO3 heterointerface. Nature 427, 423426 (2004).2. Chakhalian, J. et al. Magnetism at the interface between ferromagnetic and superconducting oxides. Nat. Phys. 2, 244248 (2006).

3. Ramesh, R. & Spaldin, N. A. Multiferroics: progress and prospects in thin lms. Nat. Mater. 6, 2129 (2007).

4. Reyren, N. et al. Superconducting interfaces between insulating oxides. Science 317, 11961199 (2007).

5. Garcia-Barriocanal, J. et al. Spin and orbital Ti magnetism at LaMnO3/SrTiO3 interfaces. Nat. Commun. 1, 82 (2010).

6. Chakhalian, J., Millis, A. J. & Rondinelli, J. Whither the oxide interface. Nat. Mater. 11, 9294 (2012).

7. Monkman, E. J. et al. Quantum many-body interactions in digital oxide superlattices. Nat. Mater. 11, 855859 (2012).

8. Bert, J. A. et al. Direct imaging of the coexistence of ferromagnetism and superconductivity at the LaAlO3/SrTiO3 interface. Nat. Phys. 7, 767771 (2011).

9. Tokura, Y. et al. Origins of colossal magnetoresistance in perovskite-type manganese oxides. J. Appl. Phys. 79, 52885291 (1996).

10. Moreo, A., Yunoki, S. & Dagotto, E. Phase separation scenario for manganese oxides and related materials. Science 283, 20342040 (1999).

11. Yamada, H. et al. Engineered interface of magnetic oxides. Science 305, 646648 (2004).

12. Gajek, M. et al. Tunnel junctions with multiferroic barriers. Nat. Mater. 6, 296302 (2007).

13. Huijben, M. et al. Critical thickness and orbital ordering in ultrathin La0.7Sr0.3MnO3 lms. Phys. Rev. B 78, 94413 (2008).

14. Kourkoutis, L. F., Song, J. H., Hwang, H. Y. & Muller, D. A. Microscopic origins for stabilizing room-temperature ferromagnetism in ultrathin manganite layers. Proc. Natl Acad. Sci. 107, 1168211685 (2010).

15. Hikita, Y., Nishikawa, M., Yajima, T. & Hwang, H. Y. Termination control of the interface dipole in La0.7Sr0.3MnO3/Nb:SrTiO3 (001) Schottky junctions.

Phys. Rev. B 79, 73101 (2009).

NATURE COMMUNICATIONS | 5:3464 | DOI: 10.1038/ncomms4464 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 5

& 2014 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms4464

16. Boschker, H. et al. Preventing the reconstruction of the polar discontinuity at oxide heterointerfaces. Adv. Funct. Mater. 22, 22352240 (2012).

17. Riedl, T., Gemming, T., Drr, K., Luysberg, M. & Wetzig, K. Mn Valency at La0.7Sr0.3MnO3/SrTiO3 (001) thin lm interfaces. Microsc. Microanal. 15, 213221 (2009).

18. Samet, L. et al. EELS study of interfaces in magnetoresistive LSMO/STO/LSMO tunnel junctions. Eur. Phys. J. B 34, 179192 (2003).

19. Harrison, W. A., Kraut, E. A., Waldrop, J. R. & Grant, R. W. Polar heterojunction interfaces. Phys. Rev. B 18, 44024410 (1978).

20. Nakagawa, N., Hwang, H. Y. & Muller, D. A. Why some interfaces cannot be sharp. Nat. Mater. 5, 204209 (2006).

21. Pentcheva, R. & Pickett, W. E. Avoiding the polarization catastrophe in LaAlO3 overlayers on SrTiO3 (001) through polar distortion. Phys. Rev. Lett. 102, 107602 (2009).

22. Tufte, O. N. & Chapman, P. W. Electron mobility in semiconducting strontium titanate. Phys. Rev. 155, 796802 (1967).

23. Frederikse, H. P. R. & Hosler, W. R. Hall Mobility in SrTiO3. Phys. Rev. 161, 822827 (1967).

24. Minohara, M., Ohkubo, I., Kumigashira, H. & Oshima, M. Band diagrams of spin tunneling junctions La1 xSrxMnO3/Nb:SrTiO3 and SrRuO3/Nb:SrTiO3

determined by in situ photoemission spectroscopy. Appl. Phys. Lett. 90, 132123 (2007).25. Muller, D. A. Structure and bonding at the atomic scale by scanning transmission electron microscopy. Nat. Mater. 8, 263270 (2009).

26. Rez, P. & Muller, D. A. The theory and interpretation of electron energy loss near-edge ne structure. Annu. Rev. Mater. Res. 38, 535558 (2008).

27. Muller, D. A. et al. Atomic-scale chemical imaging of composition and bonding by aberration-corrected microscopy. Science 319, 10731076 (2008).

28. Tan, H., Verbeeck, J., Abakumov, A. & Van Tendeloo, G. Oxidation state and chemical shift investigation in transition metal oxides by EELS. Ultramicroscopy 16, 2433 (2011).

29. Kourkoutis, L. F. From imaging individual atoms to atomic resolution 2D mapping of bonding. In Proceedings of the 17th International Microscopy Congress, Rio de Janeiro 354355 (2010).

30. Warusawithana, M. P. et al. LaAlO3 stoichiometry is key to electron liquid formation at LaAlO3/SrTiO3 interfaces. Nat. Commun. 4, 2351 (2013).

31. Sato, H. K., Bell, C., Hikita, Y. & Hwang, H. Y. Stoichiometry control of the electronic properties of the LaAlO3/SrTiO3 heterointerface. Appl. Phys. Lett.

102, 251602 (2013).32. Orgiani, P. et al. Multiple double-exchange mechanism by Mn2 doping in manganite compounds. Phys. Rev. B 82, 205122 (2010).

33. Galdi, A. et al. Magnetic properties and orbital anisotropy driven by Mn2 in nonstoichiometric LaxMnO3 d thin lms. Phys. Rev. B 83, 064418 (2011).

34. Garcia-Barriocanal, J. et al. Charge leakage at LaMnO3/SrTiO3 interfaces. Adv. Mater. 22, 627632 (2010).

35. Kurata, H., Lefevre, E., Colliex, C. & Brydson, R. Electron-energy-loss near-edge structures in the oxygen K-edge spectra of transition-metal oxides. Phys. Rev. B 47, 13763 (1993).

36. Satpathy, S., Popovi, Z. S. & Vukajlovi, F. R. Electronic Structure of the Perovskite Oxides: La1 xCaxMnO3. Phys. Rev. Lett. 76, 960963, 1996).

37. Hong, X., Posadas, A. & Ahn, C. H. Examining the screening limit of eld effect devices via the metal-insulator transition. Appl. Phys. Lett. 86, 142501 (2005).

Acknowledgements

We thank C. J. Fennie for useful discussions and E. J. Kirkland and M. Thomas for technical assistance. J.A.M., D.A.M. and L.F.K. acknowledge support by the U.S. Department of Energy, Ofce of Basic Energy Sciences, Division of Materials Sciences and Engineering under Award #DE-SC0002334. This work made use of the electron microscopy facility of the Cornell Center for Materials Research (CCMR) with support from the National Science Foundation Materials Research Science and Engineering Centers (MRSEC) programme (DMR 1120296) and NSF IMR-0417392. J.A.M. acknowledges nancial support from the Army Research Ofce in the form of a National Defense Science & Engineering Graduate Fellowship and from the National Science Foundation in the form of a NSF Graduate Research Fellowship. Y.H., T.Y. and H.Y.H acknowledge support by the Department of Energy, Ofce of Basic Energy Sciences, Materials Sciences and Engineering Division, under contract DE-AC02-76SF00515.

Author contributions

Electron microscopy and spectroscopy experiments were performed by J.A.M., L.F.K. and D.A.M. Film growth, transport measurements and structural characterization were performed by Y.H., T.H., T.Y., T.H. and H.Y.H. The manuscript was prepared by J.A.M.,L.F.K. and D.A.M. All authors discussed results and commented on the manuscript.

Additional information

Supplementary Information accompanies this paper at http://www.nature.com/naturecommunications

Web End =http://www.nature.com/ http://www.nature.com/naturecommunications

Web End =naturecommunications

Competing nancial interests: The authors declare no competing nancial interests.

Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions/

Web End =http://npg.nature.com/ http://npg.nature.com/reprintsandpermissions/

Web End =reprintsandpermissions/

How to cite this article: Mundy, J. A. et al. Visualizing the interfacial evolution from charge compensation to metallic screening across the manganite metalinsulator transition. Nat. Commun. 5:3464 doi: 10.1038/ncomms4464 (2014).

6 NATURE COMMUNICATIONS | 5:3464 | DOI: 10.1038/ncomms4464 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2014 Macmillan Publishers Limited. All rights reserved.