Introduction

Epitaxial synthesis of complex oxides has stimulated considerable interest in creating novel functionalities and physical properties, where various means are used to control the close interactions among the order parameters, including lattice, spin, charge, and orbital1,2,3,4. Heterostructures of oxide materials have also played an important role in discovering novel phenomena as they can produce well-defined interfaces to couple electronic and magnetic ground states, structure, lattice, crystallographic symmetry, etc. Most studies on the epitaxial growth of complex oxides have focused on isostructural materials, e.g. perovskites on perovskites. While many binary oxides, such as TiO2 and VO2, also offer intriguing physical properties5,6,7,8,9,10,11, only a few substrates are available with similar structures (lattice parameters and crystal symmetry) for epitaxy thereon. Thus, gaining fundamental insight into the epitaxial growth of binary oxide thin films on lattice and symmetry mismatched substrates is of vital importance for exploring their unprecedented potential12,13,14.

Recently, high quality VO2 polymorphs were successfully stabilized as epitaxial thin films using pulsed laser epitaxy (PLE) on perovskite substrates, such as SrTiO315,16,17. Among VO2 polymorphs17, bronze-phase VO2(B) has a monoclinic structure (with C2/m symmetry) whose lattice constants are a = 12.03, b = 3.69, c = 6.42 Å, and β = 106.6°18, whereas SrTiO3 (with Pm3m symmetry) has a cubic structure with the lattice constant of 3.905 Å. Note that while many previous studies focused on rhombohedral (R) and monoclinic (M1) phase VO2, recent studies in developing advanced energy storage found VO2(B) to be a promising cathode material for Li ion batteries19,20,21.

In oxide epitaxy, the important role of octahedral connectivity has been well established in perovskite-based oxide heterostructures22,23,24. Previous studies for heterostructures with a large symmetry mismatch observed the formation of a thin interlayer between the film and substrate induced by either phase transition25,26,27,28,29,30 or phase separation31,32,33. Moreover, incorporation of such an interfacial buffer layer seems to be critical, enabling the epitaxial growth between large-symmetry-mismatch materials34,35,36,37,38. Interestingly, a previous study based on a STEM observation reported the formation of an imperfect structure at a VO2(B)/SrTiO3 interface16. However, details about the interface structure have not been explored.

In this work, we report how two very dissimilar materials can form an epitaxial heterostructure by aberration-corrected scanning transmission electron microscopy (STEM) imaging and electron energy-loss spectroscopy (EELS). We have found an interfacial bi-layer at the VO2(B)/STO interface that enables epitaxial growth of a structurally complex, low symmetry film on a high symmetry substrate. It is rather surprising that VO2(B) films with corner- and edge-sharing oxygen octahedra (see Fig. 1a and b) can be epitaxially grown on STO with corner-sharing octahedra, despite the different oxygen networks and the large biaxial lattice mismatch.

Figure 1
figure 1

Atomic structure of VO2(B).

(a,b) Schematics and (c,d) corresponding cross-sectional HAADF images of VO2(B) seen along (a,c) the [100] and (b,d) [010] directions. The hexagon in (d) indicates the large open channel in VO2(B) useful for ionic conduction.

Results and Discussion

High quality VO2(B) epitaxial films were grown on (001)-oriented STO by PLE under well-optimized growth conditions. The details on the epitaxial growth and crystal quality as well as associated physical properties can be found elsewhere17. Figure 1 shows atomic structure projections and corresponding cross-sectional high-angle annular dark-field (HAADF) images taken along the [100]VO2(B) and [010]VO2(B) directions of VO2(B). In the Z-contrast HAADF images, the cation columns containing Ti (Z = 22), V (Z = 23), and Sr (Z = 38) are seen with intensities strongly dependent on their atomic number, while columns containing only light O (Z = 8) atoms are hardly visible. The image shown in Figure 1d provides the reason why VO2(B) is of particular interest for energy storage as the atomic structure seen from the [010]VO2(B) direction features an open framework that offers a good ionic diffusion pathway. The structural projection along [010]VO2(B) also reveals the clear symmetry mismatch between the film and substrate. Thus, we chose this orientation for the majority of the STEM investigations.

Figures 2a and b show cross-sectional HAADF images of an epitaxial VO2(B) film grown on a STO substrate. The images were taken along the [100]STO direction. As shown in Figures 2a and b, the film is found to contain at least two domains aligned parallel to the [100]VO2(B) and [010]VO2(B) directions, i.e. orthogonally positioned with respect to the [100] direction of the STO substrate. In fact, the film contains two additional domains that are rotated 180 degrees about the surface normal from those imaged in Fig. 2. A thin (typically ~2 nm thick) interfacial layer (IL) can be seen between the VO2(B) film and the STO substrate. Based on fast-Fourier transformation (FFT) analysis, an array of misfit dislocations has formed between the IL and the structurally relaxed VO2(B) film (as indicated in Fig. 2a and b, and in the corresponding FFT images in Figure S1 in Supporting Information). The interface between the STO substrate and the IL appears to be fully coherent. As shown in Figure 2a and b, the average spacing between dislocations observed along the [100]VO2(B) direction is 3.6 ± 0.9 nm, while it is 7.9 ± 1.1 nm when seen along the [010]VO2(B) direction. These spacings are in good agreement with calculated values of ~3.5 nm and ~7.2 nm obtained using the lattice mismatch of +5.5% for the [100]VO2(B) || [100]STO projection and −2.7% for the orthogonal [010]VO2(B)||[100]STO projection. This result reveals unambiguously that the large bi-axial lattice mismatch between the film and substrate is accommodated by the creation of dislocations at the VO2(B)/TL interface, i.e. strain-free VO2(B) epitaxial films are obtained.

Figure 2
figure 2

Microstructure of the VO2(B)/STO interface.

HAADF images show two growth twins orthogonally oriented along (a) the [100]VO2(B) and (b) [010]VO2(B) directions with respect to the [100]STO direction. (c) LAADF image taken from the image in (b), showing an extra intensity from the interfacial layer (IL) associated with increased electron beam dechanneling and, thus, scattering of electrons due to increased atomic disorder.

Figure 2c shows a low-angle annular dark-field (LAADF) image taken from the sample seen along the [010]VO2(B)/[100]STO direction. The LAADF image highlights the interlayer, which is substantially brighter than the film or the substrate. This brighter contrast in a LAADF image indicates that the IL has a higher level of structural disorder, which leads to the electron dechannelling of the incident beam39,40,41. Based on a geometric phase analysis (GPA), it is also seen that the VO2 in the IL undergoes a significant lattice expansion along the film surface normal as compared to the VO2(B) film (see Figure S2 in Supporting Information). This result is consistent with the rather large in-plane compression (−5.5%) of the VO2(B) film in the [100]VO2(B)/[100]STO projection that will cause the observed out-of-plane expansion.

Figure 3a shows a HAADF image of the IL taken along the [010]VO2(B) direction. While there is a region of the IL (Fig. 3b) that clearly duplicates the projected structure of the relaxed VO2(B) film above it (except that it is rotated 180 degrees about the surface normal), most of this layer (and its FFT, Fig. 3d,e) looks to be a superposition of [100], [−100], [010], and [0–10] projections of epitaxially strained VO2(B). The other important feature of the IL is the extra atomic layer between the STO and VO2(B) indicated with a black arrow in Fig. 3b and c. The atomic layer shows a periodic, but different arrangement of B-site atoms than that of the TiO2-terminated STO surface. The HAADF images show the out-of-plane lattice spacing between the topmost TiO2 layer of STO and the extra (Ti,V)-O layer to be 2.4 ± 0.1 Å, which is significantly larger than the 1.9 Å (001) plane spacing in STO. The intensity variations indicate that, in this [100]STO projection, the extra layer contains additional (Ti,V) columns with roughly 1/2 the B-site density of its neighboring atom columns.

Figure 3
figure 3

High-resolution observation of interfacial layer (IL).

(a) HAADF image showing that the IL consists of nanodomains, e.g. I and I′. (b,c) Magnified HAADF images taken from nanodomains I and I′ marked by the dashed rectangles in a) showing in greater detail the local atom arrangements. (d,e) FFT electron diffraction patterns obtained from the VO2(B) film and the IL (nanodomain-I), respectively. The red and black arrows between (b,c) indicate extra atom planes formed at the upper and lower sides of IL, respectively.

Spatially resolved STEM-EELS data from the interfacial region is presented in Fig. 4. Figure 4a and b respectively show element maps using the Ti-L2,3 and V-L2,3 signals taken from the same interfacial region shown in Fig. 4c. The V-L2,3 signal in Fig. 4a shows a chemically abrupt interface between the film and STO substrate. On the other hand, the Ti-L2,3 signal is seen to extend into the IL. Figure 4d shows background-subtracted Ti-L, V-L and O-K EELS profiles obtained layer-by-layer across the IL. Standard spectra obtained from single-crystalline VO2(B) (V4+) and V2O3 (V3+) thin films are also included for comparison. The peak position of V-L2,3 edges are seen to remain fixed indicating little to no change in the valence state of V in the IL and the VO2(B) film.

Figure 4
figure 4

Layer-by-layer EELS analysis.

Elemental maps for (a) V-L2,3 and (b) Ti-L2,3 signals obtained from the interface region shown in (c). (d) Back-ground subtracted Ti-L2,3, V-L2,3, and O-K spectra obtained across the interface. The EELS spectra numbered 1 through 9 are obtained from the local atomic planes indicated in (c). EELS profile intensity is normalized using the O-K edge beyond the ionization transitions to discrete states.

The Ti-L2,3 EELS fine structure obtained from the extra (Ti,V)-O layer on the STO substrate surface shows broadened L3 and L2 edges, as well as a shift of the eg peaks toward lower energy-loss (see Fig. 4d). The observed electronic state and atomic structure of this extra layer are in good agreement with previous theoretical simulations42 and STEM observations43 of a c (4 × 2) reconstructed STO (001) surface composed of a double-layer TiO2. The topmost layer of this reconstructed surface was predicted to contain clustered quartets of edge-sharing square-pyramidal TiO542,43. The extra (Ti,V)-O layer on TiO2-terminated STO can introduce edge sharing oxygen containing units, which are more consistent with the VO2(B) structure. To our knowledge, the formation of such an interface bi-layer can not be explained by any existing growth model that involves either phase transition25,26,27,28,29,30 or phase separation31,32,33 at film/substrate interfaces to accommodate inter-phase structural discontinuities.

The observed results reveal unambiguously, at the initial growth stage, the formation of an interfacial layer composed of VO2(B) nanodomains that enable the epitaxy of VO2(B) on STO. This epitaxy with a large symmetry mismatch involves a structural reconstruction process at the substrate surface to facilitate the symmetry transition between the two distinct component structures. The VO2(B) nanodomains form a fully coherent interface with the STO substrate and are subject to considerable lattice strain. Once the strain energy in the VO2(B) nanodomains exceeds a critical level, misfit dislocations are introduced, followed by the growth of fully relaxed VO2(B). The much larger domain size in the relatively strain-free film is an expected result of increased adatom mobility on the relaxed surface. Formation of the interfacial VO2(B) nanodomains indicates a nanoscale self-templating process that enables the epitaxy of strain-free VO2(B) films on STO substrates. Therefore, the results not only enable novel insights into atomic mechanism of complex heterostructure interface at an atomic scale, but also shed light on the epitaxial design of two materials with large symmetry and lattice mismatch.

Methods

Epitaxial synthesis

VO2(B) epitaxial films were deposited on (001) SrTiO3 substrates by pulsed laser epitaxy. A sintered ceramic VO2 target was ablated with a KrF excimer laser (λ = 248 nm) at a repetition rate of 5 Hz and laser fluence of 1 J·cm−2. The optimized substrate temperature and oxygen pressure to grow high quality thin films were 500 °C and 20 mTorr, respectively, and the samples were in-situ post-annealed in 1 atm of O2 for 1 hour at the growth temperature to ensure the oxygen stoichiometry. Detailed information on the synthesis of single-crystalline VO2(B) (V4+) and V2O3 (V3+) thin films utilized for EELS analysis can be found elsewhere14.

Scanning Transmission Electron Microscopy (STEM)

Cross-sectional specimens oriented along the [100] STO direction for STEM analysis were prepared using ion milling after mechanical thinning and precision polishing (using water-free abrasive). High-angle annular dark-field (HAADF) and low-angle annular dark-field (LAADF) imaging and electron-energy loss spectroscopy (EELS) analysis were carried out in Nion UltraSTEM200 operated at 200 keV. The microscope is equipped with a cold field-emission gun and a corrector of third- and fifth-order aberrations for sub-angstrom resolution. Inner/outer detector angles of 78/240 mrad and 30/63 mrad were used for HAADF and LAADF imaging, respectively. The convergence semi-angle for the electron probe was set to 30 mrad.

Additional Information

How to cite this article: Gao, X. et al. Nanoscale self-templating for oxide epitaxy with large symmetry mismatch. Sci. Rep. 6, 38168; doi: 10.1038/srep38168 (2016).

Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.