Letter pubs.acs.org/NanoLett
Assessment of Strain-Generated Oxygen Vacancies Using SrTiO3 Bicrystals Si-Young Choi,*,† Sung-Dae Kim,† Minseok Choi,*,† Hak-Sung Lee,† Jungho Ryu,‡ Naoya Shibata,§ Teruyasu Mizoguchi,∥ Eita Tochigi,§ Takahisa Yamamoto,⊥ Suk-Joong L. Kang,# and Yuichi Ikuhara§ †
Materials Modeling and Characterization Department and ‡Functional Ceramics Group, Korea Institute of Materials Science, Changwon 642-831, Korea § Institute of Engineering Innovation, School of Engineering, The University of Tokyo, Tokyo 113-8656, Japan ∥ Institute of Industrial Science, The University of Tokyo, Tokyo 153-8505, Japan ⊥ Department of Quantum Engineering, Nagoya University, Nagoya 464-8603, Japan # Department of Materials Science and Engineering, Korea Advanced Institute of Science and Technology, Daejeon 305-701, Korea S Supporting Information *
ABSTRACT: Atomic-scale defects strongly influence the electrical and optical properties of materials, and their impact can be more pronounced in localized dimensions. Here, we directly demonstrate that strain triggers the formation of oxygen vacancies in complex oxides by examining the tilt boundary of SrTiO3 bicrystals. Through transmission electron microscopy and electron energy loss spectroscopy, we identify strains along the tilt boundary and oxygen vacancies in the strain-imposed regions between dislocation cores. First-principles calculations support that strains, irrespective of their type or sign, lower the formation energy of oxygen vacancies, thereby enhancing vacancy formation. Finally, current−voltage measurements confirm that such oxygen vacancies at the strained boundary result in a decrease of the nonlinearity of the I−V curve as well as the resistivity. Our results strongly indicate that oxygen vacancies are preferentially formed and are segregated at the regions where strains accumulate, such as heterogeneous interfaces and grain boundaries. KEYWORDS: Perovskites, strain engineering, oxygen vacancies, bicrystals
T
or ion bombardment,12,13 that strain can stimulate the formation of oxygen vacancies in oxides. Here, we performed a comprehensive study to understand the correlation between strain and the formation of oxygen vacancies in oxides. The major finding is the first direct demonstration that oxygen vacancies are confined near the strain-imposed boundaries of STO and that the confinement can be enhanced with greater strain. In this work, we prepared bicrystals with [001] symmetric tilt grain boundaries (GBs) and then investigated the local areas with a different strain accumulation amounts between dislocation cores, rather than at the cores, in contrast to earlier studies using STO bicrystals.14−22 The spatial distributions of strain and oxygen vacancies in the strained region were characterized by spherical aberration-corrected scanning transmission electron microscopy (STEM), including high-angle annular dark-field (HAADF) and low-angle annular dark-field (LAADF) techniques. The LAADF technique is mainly taken into
he strain-mediated control of materials properties has been extensively studied, as mechanical strain is a simple and predictive parameter for controlling certain properties. In complex oxides, drastic changes in the chemical and physical properties of the oxides are observed with an introduction of strain by means of lattice distortion or through lattice mismatches. In SrTiO3 (STO), for example, biaxial strain leads to a phase transition from paraelectric to ferroelectric at room temperature, showing a correlation between the strain and the transition temperature.1 It has also been reported that strain can induce coupling between ferroelectricity and ferromagnetism, that is, multiferroicity.2−4 Changes in materials properties are attributed not merely to strain but also to strain-generated oxygen vacancies, which are typical defects in oxides. Previous studies5−9 suggested that unprecedented properties at STO/oxide interfaces are likely associated with the interface quality, such as oxygen deficiencies and local strain. It has been reported that oxygen vacancies do not diffuse out of the interface but are rather localized and distributed along the interface,7,9 even after an oxidizing treatment.10 These studies strongly indicate, along with hightemperature annealing under low oxygen partial pressures10,11 © 2015 American Chemical Society
Received: March 31, 2015 Revised: May 20, 2015 Published: May 22, 2015 4129
DOI: 10.1021/acs.nanolett.5b01245 Nano Lett. 2015, 15, 4129−4134
Letter
Nano Letters
Figure 1. HAADF STEM images (a,c) and LAADF-STEM images (b,d) of the 6°-tilt and the 10°-tilt boundaries, respectively, where the red and blue circles indicate Sr−Sr and Ti−O atomic columns and the white lines indicated by the red arrow are the grain boundary planes. The images of the LAADF-STEM show detailed contrast in regions between dislocation cores, indicating that strain is accumulated along the 10° tilted GB.
account because, in contrast to HAADF, this provides details of the strained regions from the collection of low-order diffractions of electrons when the LAADF detector is close to an on-axis position.10,20,23−25 High-resolution transmission electron microscopy (HRTEM) and electron energy loss spectroscopy (EELS) were employed, and first-principles defect calculations were also performed to support the experimental observations. The electrical properties, being related to the vacancies, were examined using I−V characteristic measurements. Figure 1 shows HAADF and LAADF STEM images of a symmetrically tilt boundary in STO. Two distinct features are observed. One is a dislocation arrangement. In general, dislocations in oxides are stabilized through the creation of two partial dislocations with Burgers vectors of b1 and b2 from a perfect dislocation with a Burgers vector of b (|b|2 > |b1|2 + |b2|2). When dislocations dissociate into partial dislocations, the subsequent alignment of the partial dislocations readily relaxes the local strain near the dislocation cores. In alumina, for instance, dislocations dissociate into two partial dislocations via a climbing process and then align parallel to the GB plane.21,26 Likewise, it has been reported that a dislocation at a low-angle tilt STO boundary dissociates according to the following process:17,19 a[010] → a/2[010] + a/2[010] in [001]. On the other hand, the dislocation array in the present study does not follow the above process. The low-angle tilt STO boundary consists of pure edge dislocations with a Burgers vector of a[010]. The measured distance between dislocation cores is ∼3.7 nm at the 6°-tilt boundary and ∼2.3 nm at the 10°-tilt boundary (Figure 1a,c), which are in good agreement with the distance (3.8 and 2.2 nm, respectively) calculated by Frank’s formula.27 This behavior is attributed to the structural complexity of the perovskite oxide or to the strong covalent bonding between titanium and oxygen atoms, implying that the excess strain can be relaxed in another way on the GB plane. The other feature is the presence of different type of dislocation cores and residual strains. The 6°-tilt boundary has a Sr-excess core (indicated by the label A) and a Ti-excess core (indicated by the label B), and they alternately appear along the grain boundary, whereas the 10°-tilt boundary contains only Ti-excess cores with a zigzag arrangement. Such a difference
Figure 2. Measured strains along the symmetrical tilt GBs using the geometrical phase analysis method: (a) HRTEM image of the 6°-tilt boundary, where (b−d) are the measured in-plane strain distributions εxx, εyy, and εxy around the dislocation cores of the 6°-tilt boundary, respectively. Red denotes compressive stress, blue denotes tensile stress, and green indicates no detectable strain. The strain distributions around each dislocation core are nearly identical. (e) Line profile of the strain distribution from the 6°-tilt boundary denoted by the dotted rectangles in (a−d). The profile represents εxx (red), εyy (green), and εxy (blue) along the GBs, as indicated by the red rectangular boxes in the HRTEM images. The residual strains in the regions between dislocation cores are mostly compensated for, as indicated by the yellow box in (e). 4130
DOI: 10.1021/acs.nanolett.5b01245 Nano Lett. 2015, 15, 4129−4134
Letter
Nano Letters
dislocation cores are ideally aligned in a line, the stress and strain become negligible by compensating for the compressive and tensile stresses between the dislocation cores. On the other hand, the 10°-tilt boundary contains a strain distribution which is anisotropic and distorted, and εxx and εxy are particularly prominent (Figure 3b,d). Such a nonsymmetric distribution must be related to the zigzagged array of Ti-excess dislocation cores to endure the larger tilt angle (Figure 1c). Using first-principles calculations based on density functional theory, the formation energy of an oxygen vacancy in our STO systems is evaluated. Note that we considered a neutral oxygen vacancy, as the issue of a stable charge state is debatable. Several previous studies29−31 claimed that the oxygen vacancy would be a single donor-like vacancy, releasing one electron to the conduction band, whereas others reported that the vacancy was doubly charged32 or neutral.33 However, consideration of the neutral charge state is natural in the present case because the Fermi level is positioned near the conduction band edge of the Nb-doped STO system. In our calculations, the effects of two types of strain (biaxial and shear strains) on the formation of an oxygen vacancy are investigated. Regarding the biaxial strain, as shown in Figure 4a, the strain (both compressive and tensile)
Figure 3. (a) HRTEM image of the 10°-tilt boundary, where (b−d) are the measured in-plane strain distributions εxx, εyy, and εxy around the dislocation cores of the boundary, respectively. Strain distributions around each dislocation core are anisotropic and distorted. (e) Line profile of the strain distribution from the 10°-tilt boundary denoted by the dotted rectangles in (a−d). Residual strains remain in the regions between dislocation cores, as indicated by the red rectangles in (e).
can affect strain relaxation around the dislocation cores. To examine the details of the strain distribution around cores, the LAADF-STEM technique was utilized. We note that extra contrast appears in between cores of the 10°-tilt boundary, indicating that residual strain exists in the region, in contrast to the image of the 6°-tilt boundary (Figure 1b,d). Presumably, owing to the zigzag pattern of the cores, the compressive and tensile strains along the dislocation core array cannot be fully relaxed, even between the cores. As shown in Figures 2 and 3, we further visualized the strain distribution along the GBs by plotting two-dimensional strain maps for GBs. All of the information was extracted from HRTEM images (Figures 2a and 3a) of the GBs using a geometric phase analysis method (GPA, HREM Research Inc.).28 Each dislocation at the GB is surrounded by tensile (blue) and compressive (red) strains, as shown in Figures 2b−d and 3b−d. Our results show that the strains are mostly diluted in the region between the two dislocation cores for the 6°-tilt boundary, whereas they remain for the 10°-tilted GB, which is consistent with the results of the LAADF-STEM image analysis. The 6°-tilt boundary shows that the strain distribution (εxx, εyy, and εxy) around each dislocation core is nearly identical and periodically arrayed, resulting in negligible strains between the cores. The stresses accompanying the strains can simply be calculated using elastic theory with two assumptions that the dislocation core effects are not taken into consideration and that the core radius is within 5b (Supporting Information Figure S1). The obtained results also show that, when
Figure 4. Calculated relative formation energy of oxygen vacancies under (a) biaxial and (b) shear strain. The value of the formation energy in the unstrained bulk STO was set as a reference. Regardless of the strain type, strain effectively decreases the formation energy of oxygen vacancies.
lowers the formation energy of the oxygen vacancy in STO, making it easier for vacancies to form. When the calculated formation energy in the unstrained STO was set as a reference, the relative formation energy was −0.14 eV at 2% compressive strain and −0.24 eV at 2% tensile strain. A stronger strain of 4% yields a more significant reduction in the formation energy, −1.05 and −1.14 eV under compressive and tensile strains, respectively, indicative of enhanced vacancy formation in the strained region. In the case of shear strain, its effect is brought 4131
DOI: 10.1021/acs.nanolett.5b01245 Nano Lett. 2015, 15, 4129−4134
Letter
Nano Letters
Figure 5. Electron energy loss spectra (EELS) for Ti and O simultaneously recorded across the symmetric tilt GBs: (a) Ti-L2,3 and (b) O-K edge EELS spectra from the 6°-tilt boundary. (d) Ti-L2,3 and (e) O-K edge EELS spectra from the 10°-tilt boundary. (c,f) ADF images of the 6°-tilt and 10°-tilt boundaries, respectively. The electron beam positions for acquiring the EELS spectra are denoted by the yellow boxes in the ADF images.
into the calculation by imposing a tilt angle θ (Figure 4b). The percent of shear strain γ is quantified as follows: tan−1(γ %) = tan−1(2γ/100) = θ. We find that shear strain also significantly reduces the formation energy of the oxygen vacancy from the undertrained values of −0.12 eV at 2% and −0.47 eV at 4% shear strain. We attempt to understand the effect of strain on the formation of oxygen vacancies. As described above, the vacancy formation energy decreases under both compressive and tensile strains. This finding is different from previous reports of other prototypical perovskite oxides, such as BaTiO334 and PbTiO3,35 while similar behavior was found in La1−xSrxCoO3.36 Because the formation energy of an oxygen vacancy decreases regardless of changes in the cell volume, under either negative (compressive strain) or positive (tensile strain), a change in the cell volume or bond length does not provide a robust explanation of the reduced formation energy under strain. We posit that the decrease in the formation energy is related to a decrease in the band gap of strained STO with an increase of any type of strain, shear, or biaxial. According to our calculations, the band gap decreases with any type of strain, such as biaxial (in agreement with ref 37), shear, and even hydrostatic strain. Consequently, the vacancy formation energy is directly related to changes in the strain-induced band gap, especially in the case of a band gap reduction (Supporting Information Figure S2). Next, the EELS spectra were examined. A change in the valence charge from the nominal value is often considered as a fingerprint of defect formation, as the defect can transfer electrons or holes to its neighboring atoms. In STO, the
nominal valence-charge of Ti atoms is +4 and the amount of Ti charge can be perturbed by atomic defects, such as oxygen vacancies. To quantify the Ti valence-charge in the strained region, the EELS spectra were obtained from the bulk region (“B”) to a GB region (“A”) for the 6°-tilt and 10°-tilt boundaries. For the 6°-tilt boundary, there is no significant difference in the Ti-L edge when probing from “B” to “A” (Figure 5a). Here, t2g peaks (blue arrow) are clearly observed throughout the spectra. For the 10°-tilt boundary, a conspicuous difference in the spectra is noted. The red arrow in Figure 5d indicates a collapse of the t2g peaks of the Ti-L edges at “A.” The atomic structure in region “A” is different from that in the dislocation core region but similar to that in the bulk region (“B”) (Figure 5f). An EELS simulation (Supporting Information Figure S3) also supports the contention that the decrease of the peak intensity of Ti t2g originates from a weaker covalence state between the Ti d-shell and the O p-shell,38 that is, a change in the Ti valence state from +4 to +3. We now focus on the shape of the O-K edge in the EELS spectra, which is known to be generated by the unoccupied O-p density of states by a core hole.10 For the 10°-tilted GB, the first peak marked with the red arrow in Figure 6e is suppressed in “A” (red lines), which comes from a decrease in the Ti valencestate and/or electrons released from oxygen vacancies.39 The conduction band near the O p−Ti t2g orbital is partially occupied by electrons and thus the excitation of electrons from the core orbital is relatively limited, thereby weakening the first peak in the O-K edge (Supporting Information Figure S4, the 4132
DOI: 10.1021/acs.nanolett.5b01245 Nano Lett. 2015, 15, 4129−4134
Letter
Nano Letters
In conclusion, our electron microcopy results, theoretical calculations, EELS spectra, and I−V measurement results clearly indicate that strain fields can form oxygen vacancies in oxide materials. Two types of model STO bicrystals were fabricated by changing the tilt angle. At the 6°-tilt boundary, dislocation cores are arrayed in a line; therefore, Sr-excess and Ti-excess core structures alternately appear. The 10°-tilt boundary, however, only has Ti-excess dislocation cores, which are arrayed in a zigzag manner. Such a zigzagged arrangement gives rise to residual strain, especially shear strain, between the dislocation cores. Through electron microscopy and spectroscopy techniques, we identified the presence of excess oxygen vacancies in the strained regions of the low-angle tilt boundary. First-principles defect-formation energy calculations show that the formation of oxygen vacancies could be triggered by strain irrespective of its type or sign. Electrically, the 10°-tilt boundary exhibited less nonlinearity of the I−V curve and less higher boundary resistivity than those of the 6°tilt boundary, which supports the higher content of oxygen vacancies in the highly strained region, as the diluted DSB is attributed to the compensation of negatively charges near the 10°-tilt boundary.
■
Figure 6. Current−voltage properties of (a) the Σ1 boundary without misorientation, (b) the 6°, and (c)10°-tilt boundaries: the Σ1 boundary shows ohmic behavior, while the 6° and 10°-tilt boundaries exhibit nonlinearity due to the DSB. The dotted lines indicate the reference slopes of an I−V curve without nonlinearity.
ASSOCIATED CONTENT
S Supporting Information *
Details of the experimental and computational methods are included. The stress field along an imaginary symmetrical tilt GB (tilt angle of 10°) calculated using numerical equations, the relative formation energy of oxygen vacancies calculated as a function of the ratio of the calculated band gap under strains and hydrostatic pressure, and the calculated Ti-L and O-K edge EELS spectra are shown. The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.nanolett.5b01245.
calculated dependency of an oxygen vacancy on the O-K edge). Using the Hartree-Slater method, the O/Ti ratio in strained region “A” at the 10°-tilt boundary is calculated to be ∼18% (2.46 ± 0.1), which is lower than that in the bulk region of “B” and far from the stoichiometric value of O/Ti = 3 (Figure 5f). Meanwhile, the ratio for “A” in the 6°-tilt boundary is ∼2.92 ± 0.09, close to the stoichiometry (Figure 5c). These EELS analysis results strongly indicate the preferential presence of oxygen vacancies between dislocation cores at the 10°-tilt boundary. Finally, the electrical properties of STO bicrystals are measured in terms of the I−V electrical characteristic. Figure 6 shows a plot of the I−V curves obtained from 6° and 10°-tilt boundaries together with that of the Σ1 boundary for comparison. The I−V curve of the Σ1 boundary shows an ohmic contact, indicative of no experimental drawback in the electrical measurement as well as bicrystal fabrication. In contrast, the I−V curve of the 6°-tilt boundary is not linear with a nonlinearity coefficient α of 2.4. These measurements are in good agreement with our previous report,40 which showed that the nonlinearity of the I−V curve and the boundary resistivity increase with an increase in the tilt angle. However, the I−V behavior measured at the 10°-tilt boundary deviates from the expected results. The 10°-tilt boundary exhibits decreased nonlinearity (α = 2.1) and boundary resistivity compared to the 6°-tilt boundary. This can be explained in terms of the double Schottky barrier (DSB)39,40 weakened by oxygen vacancies. It was reported that the DSB originates from negatively charged cationic vacancies (e.g., Sr vacancies)32,41 that accumulate in the vicinity of the boundary, which correlates with the nonlinearity of the I−V curve and the boundary resistivity.40 The DSB height becomes lower as the concentration of negative charges decreases at the boundary. On the basis of the above results, it is natural that oxygen vacancies exist near the GB and then compensate for the negatively charged vacancies, thus accordingly decreasing the DSB at the GB.
■
AUTHOR INFORMATION
Corresponding Authors
*(S.Y.C.) E-mail:
[email protected]. *(M.C.) E-mail:
[email protected]. Author Contributions
S.Y.C. and S.D.K. contributed equally to this study. S.Y.C. performed all of the experiments. M.C carried out the firstprinciples calculations for the oxygen vacancies. S.Y.C. and S.D.K. analyzed the electron microscopy data. Theoretical calculations of the EELS spectra were performed by T.M., and H.S.L., J.R., E.T., N.S., and T.Y. held deep and fruitful discussions of the results. S.J.L.K. and Y.I. are the principal investigators of the laboratory in which the research was performed. S.D.K., M.C., and S.Y.C. wrote the manuscript. All of the authors read and approved the final manuscript. The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS This research was supported by the Global Frontier Program through the Global Frontier Hybrid Interface Materials (GFHIM) program of the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT & Future Planning (2013M3A6B1078872). 4133
DOI: 10.1021/acs.nanolett.5b01245 Nano Lett. 2015, 15, 4129−4134
Letter
Nano Letters
■
(33) Mitra, C.; Lin, C.; Robertson, J.; Demkov, A. A. Phys. Rev. B 2012, 86 (15), 155105. (34) Yang, Q.; Cao, J. X.; Ma, Y.; Zhou, Y. C.; Jiang, L. M.; Zhong, X. L. J. Appl. Phys. 2013, 184110. (35) Yang, Q.; Cao, J. X.; Zhou, Y. C.; Ma, Y.; Zhong, X. L. Appl. Phys. Lett. 2013, 142911. (36) Donner, W.; Chen, C.; Liu, M.; Jacobson, A. J.; Lee, Y. L.; Gadre, M.; Morgan, D. Chem. Mater. 2011, 23, 984−988. (37) Berger, R. F.; Fennie, C. J.; Neaton, J. B. Phys. Rev. Lett. 2011, 107 (14), 146804. (38) The EELS simulations indicate that the change of Ti4+ to Ti3+ results in a slight shift in Ti-L and the disappearance of its t2g (Supporting Information Figure S3), which is consistent with the literature reports [refs 17, 20−22]. The appearance of t2g in the Ti-L spectrum is attributed to the strong covalence state between the Ti dshell and O p-shell. (39) Cantoni, C.; Gazquez, J.; Granozio, F. M.; Oxley, M. P.; Varela, M.; Lupini, A. R.; Pennycook, S. J.; Aruta, C.; di Uccio, U. S.; Perna, P.; Maccariello, D. Adv. Mater. 2012, 24, 3952−3957. (40) Yamamoto, T.; Mizoguchi, T.; Choi, S. Y.; Sato, Y.; Shibata, N.; Ikuhara, Y. Mater. Sci. Forum 2007, 558 (2007), 851−856. (41) Mizoguchi, T.; Sato, Y.; Buban, J. P.; Matsunaga, K.; Yamamoto, T.; Ikuhara, Y. Appl. Phys. Lett. 2005, 87 (24), 241920.
REFERENCES
(1) Haeni, J. H.; Irvin, P.; Chang, W.; Uecker, R.; Reiche, P.; Li, Y. L.; Choudhury, S.; Tian, W.; Hawley, M. E.; Craigo, B.; Tagantsev, A. K.; Pan, X. Q.; Streiffer, S. K.; Chen, L. Q.; Kirchoefer, S. W.; Levy, J.; Schlom, D. G. Nature 2004, 430 (7001), 758−61. (2) Fennie, C. J.; Rabe, K. M. Phys. Rev. Lett. 2006, 97 (26), 267602. (3) Lee, J. H.; Rabe, K. M. Phys. Rev. Lett. 2010, 104 (20), 207204. (4) Lee, J. H.; Fang, L.; Vlahos, E.; Ke, X.; Jung, Y. W.; Kourkoutis, L. F.; Kim, J. W.; Ryan, P. J.; Heeg, T.; Roeckerath, M.; Goian, V.; Bernhagen, M.; Uecker, R.; Hammel, P. C.; Rabe, K. M.; Kamba, S.; Schubert, J.; Freeland, J. W.; Muller, D. A.; Fennie, C. J.; Schiffer, P.; Gopalan, V.; Johnston-Halperin, E.; Schlom, D. G. Nature 2010, 466 (7309), 954−958. (5) Cen, C.; Thiel, S.; Hammerl, G.; Schneider, C. W.; Andersen, K. E.; Hellberg, C. S.; Mannhart, J.; Levy, J. Nat. Mater. 2008, 7 (4), 298−302. (6) Garcia-Barriocanal, J.; Rivera-Calzada, A.; Varela, M.; Sefrioui, Z.; Iborra, E.; Leon, C.; Pennycook, S. J.; Santamaria, J. Science 2008, 321 (5889), 676−80. (7) Herranz, G.; Basletic, M.; Bibes, M.; Carretero, C.; Tafra, E.; Jacquet, E.; Bouzehouane, K.; Deranlot, C.; Hamzic, A.; Broto, J. M.; Barthelemy, A.; Fert, A. Phys. Rev. Lett. 2007, 98 (21), 216803. (8) Ohtomo, A.; Hwang, H. Y. Nature 2004, 427 (6973), 423−6. (9) Siemons, W.; Koster, G.; Yamamoto, H.; Harrison, W. A.; Lucovsky, G.; Geballe, T. H.; Blank, D. H.; Beasley, M. R. Phys. Rev. Lett. 2007, 98 (19), 196802. (10) Muller, D. A.; Nakagawa, N.; Ohtomo, A.; Grazul, J. L.; Hwang, H. Y. Nature 2004, 430 (7000), 657−61. (11) Yamada, H.; Miller, G. R. J. Solid State Chem. 1973, 6, 169−177. (12) Henrich, V. E.; Dresselhaus, G.; Zeiger, H. J. Phys. Rev. B 1978, 17, 4908. (13) Kan, D.; Terashima, T.; Kanda, R.; Masuno, A.; Tanaka, K.; S, C.; Kan, G.; Ishizumi, A.; Kanemitsu, Y.; Shimakawa, Y.; Takano, M. Nat. Mater. 2005, 4, 816−819. (14) Zhang, Z.; Sigle, W.; De Souza, R. A.; Kurtz, W.; Maier, J.; Rühle, M. Acta Mater. 2005, 53 (19), 5007−5015. (15) De Souza, R. A. Phys. Chem. Chem. Phys. 2009, 11 (43), 9939− 9969. (16) Metlenko, V.; Ramadan, A. H. H.; Gunkel, F.; Du, H.; Schraknepper, H.; Hoffmann-Eifert, S.; Dittmann, R.; Waser, R.; De Souza, R. A. Nanoscale 2014, 6 (21), 12864−12876. (17) Klie, R. F.; Beleggia, M.; Zhu, Y.; Buban, J. P.; Browning, N. D. Phys. Rev. B 2003, 68, 214101. (18) Du, H.; Jia, C.-L.; Houben, L.; Metlenko, V.; De Souza, R. A.; Waser, R.; Mayer, J. Acta Mater. 2015, 89, 344. (19) Zhang, Z.; Sigle, W.; Kurtz, W.; Rühle, M. Phys. Rev. B 2002, 66, 214112. (20) Fitting, L.; Thiel, S.; Schmehl, A.; Mannhart, J.; Muller, D. A. Ultramicroscopy 2006, 106 (11−12), 1053−61. (21) Zhang, Z.; Sigle, W.; Kurtz, W. Phys. Rev. B 2004, 144103. (22) Klie, R. F.; Browning, N. D. Appl. Phys. Lett. 2000, 77, 3737. (23) E, H. S.; Silcox, J. Ultramicroscopy 1995, 58, 6−17. (24) Perovic, D. D.; Rossow, C. J.; Howie, A. Ultramicroscopy 1993, 52, 353−359. (25) Yu, Z.; Muller, D. A.; Silcox, J. J. Appl. Phys. 2004, 95, 3362. (26) Tochigi, E.; Shibata, N.; Nakamura, A.; Yamamoto, T.; Ikuhara, Y. Acta Mater. 2008, 56, 2015−2021. (27) Frank, F. C. Philos. Mag. VII/Ser. 1951, 42, 809−819. (28) Johnson, C. L.; Hÿtch, M. J.; Buseck, P. R. Proc. Natl. Acad.. Sci. U.S.A. 2004, 101, 17936−17939. (29) Moos, R.; Hardtl, K. H. J. Am. Ceram. Soc. 1997, 80 (10), 2549− 2562. (30) Choi, M.; Oba, F.; Kumagai, Y.; Tanaka, I. Adv. Mater. 2013, 25, 86. (31) Janotti, A.; Varley, J. B.; Choi, M.; Van de Walle, C. G. Phys. Rev. B 2014, 90 (8), 085202. (32) Tanaka, T.; Matsunaga, K.; Ikuhara, Y.; Yamamoto, T. Phys. Rev. B 2003, 68 (20), 205213. 4134
DOI: 10.1021/acs.nanolett.5b01245 Nano Lett. 2015, 15, 4129−4134
