Strain Effects in Epitaxial VO2 Thin Films on Columnar Buffer-Layer

Dec 20, 2016 - Strain Effects in Epitaxial VO2 Thin Films on Columnar Buffer-Layer TiO2/Al2O3 Virtual Substrates .... AIP Advances 2017 7 (10), 105116...
0 downloads 0 Views 3MB Size
Download PDF
Research Article www.acsami.org

Strain Effects in Epitaxial VO2 Thin Films on Columnar Buffer-Layer TiO2/Al2O3 Virtual Substrates Eric Breckenfeld,† Heungsoo Kim,*,† Katherine Burgess,† Nicholas Charipar,† Shu-Fan Cheng,‡ Rhonda Stroud,† and Alberto Piqué† †

Naval Research Laboratory, 4555 Overlook Avenue, Washington, D.C. 20375, United States Nova Research, Inc., 1900 Elkin Street, Suite 230, Alexandria, Virginia 22308, United States

‡

S Supporting Information *

ABSTRACT: Epitaxial VO2/TiO2 thin film heterostructures were grown on (100) (m-cut) Al2O3 substrates via pulsed laser deposition. We have demonstrated the ability to reduce the semiconductor−metal transition (SMT) temperature of VO2 to ∼44 °C while retaining a 4 order of magnitude SMT using the TiO2 buffer layer. A combination of electrical transport and Xray diffraction reciprocal space mapping studies help examine the specific strain states of VO2/TiO2/Al2O3 heterostructures as a function of TiO2 film growth temperatures. Atomic force microscopy and transmission electron microscopy analyses show that the columnar microstructure present in TiO2 buffer films is responsible for the partially strained VO2 film behavior and subsequently favorable transport characteristics with a lower SMT temperature. Such findings are of crucial importance for both the technological implementation of the VO2 system, where reduction of its SMT temperature is widely sought, as well as the broader complex oxide community, where greater understanding of the evolution of microstructure, strain, and functional properties is a high priority. KEYWORDS: epitaxial VO2, TiO2 buffer layers, semiconductor-to-metal transition, strain effect, electrical switching

1. INTRODUCTION Epitaxial strain has been a powerful tool for controlling and inducing new properties in complex oxide thin films. Understanding the evolution of strain in these systems is essential to the development of strain-induced responses. By controlling lattice strain in various heteroepitaxial oxide thin films, researchers have gained access to a range of novel functionalities across a multitude of material systems. Consequently, the control of strain has emerged as an important consideration when engineering structural, electronic, and magnetic properties in many functional systems.1 Despite such great interest in the relationship between epitaxial strain and the evolution of properties in complex oxide thin films, a comprehensive understanding of strain evolution and relaxation remains underdeveloped. This contrasts with traditional semiconductor systems, where numerous studies have revealed that large strains are accommodated by either threedimensional growth and island formation2 or by breaking of the coherence of the epitaxial interface via misfit or threading dislocations.3,4 Theoretical approaches to predict the critical thickness for breakdown of epitaxial coherence have been famously conducted by researchers such as Frank and Van der Merwe5 as well as Matthews and Blakeslee.6 While using different approaches (energy minimization vs force balance), both models predict similar critical thicknesses for many This article not subject to U.S. Copyright. Published XXXX by the American Chemical Society

systems such as Si, Ge, and III−V semiconductor compounds.7,8 The same depth of study and understanding of strain relaxation, unfortunately, does not exist for many complex oxide films. Although more classical strain relaxation mechanisms (i.e., island formation and the formation of misfit dislocations arrays) are observed in some oxide systems,9−12 other effects can also occur to drive strain relaxation away from the classical predictions. A number of mechanisms for strain relaxation in oxide systems which can preclude the expected relaxation behavior have been experimentally observed in oxide systems, including bond angle adjustments,13 ferroelastic domain formation,14 and strain-induced phase transitions to alternate structural polymorphs.15 At the same time, considerable interest has been generated in the vanadium dioxide (VO2) system due to its sharp, temperature-driven semiconductor-to-metal transition (SMT),16 which is responsible for a very distinctive set of optical and electrical responses.17,18 The primary obstacle for the implementation of VO2 as a technologically relevant material is the relatively high phase transition temperature of ∼68 °C in bulk. Many efforts to reduce this temperature have Received: October 14, 2016 Accepted: December 20, 2016 Published: December 20, 2016 A

DOI: 10.1021/acsami.6b13112 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

ACS Applied Materials & Interfaces

Research Article

Figure 1. (a) Schematic (not to scale) of the VO2/TiO2/Al2O3 heterostructure as well as the epitaxial alignment of each layer. (b) XRD φ-scans of the VO2 (101), TiO2 (101), and Al2O3 (104) peaks of the heterostructure. (c−e) Schematic illustrations of (c) (100) m-Al2O3, (d) (001) rutile TiO2, and (e) (001) tetragonal VO2. Note in (c) that, for visual clarity, only the oxygen sublattice is shown. chamber was evacuated to a background pressure of ∼2 × 10−6 Torr. The TiO2 target was prepared from TiO2 powder (purity 99.99%, Aldrich) by pressing into 2.54 cm diameter at 10 000 kg and then sintering at 1200 °C for 24 h in O2 flowing atmosphere. The VO2 target (5 cm diameter; ACI Alloys) was used for VO2 film growth. The TiO2 buffer layers were grown at temperatures between 500 and 800 °C and an oxygen pressure of 5 mTorr to thickness of ∼150 nm. The VO2 layers were grown at an oxygen pressure of 10 mTorr and a temperature of 550 °C to thicknesses between 8 and 240 nm. Film thickness was determined with a stylus profilometer (KLA tencor P15). Following growth, the films were subjected to extensive structural and property studies. The electrical switching properties of the VO2 films were characterized at temperatures between 20 and 110 °C by a linear four-probe method using a current of 20 μA. X-ray diffraction (XRD) θ−2θ and reciprocal space mapping (RSM) studies [Rigaku rotating anode X-ray generator with Cu Kα radiation] were used to characterize the crystal structure of the films. Atomic force microscopy (AFM) [Digital Instrument, Dimension 3100 series] was used to evaluate the surface morphology of the films. Cross sections for scanning transmission electron microscopy (STEM) and electron energy loss spectroscopy (EELS) analysis were prepared via focused ion beam (FIB) lift out with an Ascend micromanipulator in a FEI Nova 600 FIB-SEM. Bright field and high-angle annular dark field (HAADF) images were collected on a NION Ultra STEM-X operated at 200 kV and ∼40 pA, with a 0.1−0.2 nm probe.

met with some success including cation doping, 19−21 introduction of oxygen vacancies,22−24 and hydrogenation.25 Among these various efforts, epitaxial strain26−28 has emerged as an effective way to control the transition temperature of VO2 thin films without diminishing the magnitude or sharpness of the SMT. Numerous substrates have been explored for epitaxial growth of VO2 thin films, but TiO2 in particular has emerged as an ideal substrate for reducing the transition temperature by applying tensile epitaxial strain. This is due to the close match in lattice constant between the two materials, with tetragonal VO2 (a = b = 0.4556 nm, JCPDS #76-0675) experiencing roughly 0.81% tensile epitaxial strain on rutile TiO2 (a = b = 0.4593 nm, JCPDS #21-1276) substrates. Such values of strain have been shown to lower the semiconductor-to-metal transition temperature to 25 °C or lower,29 making VO2 much more attractive for applications closer to room temperature. Unfortunately, it has been shown to be quite difficult to sustain this epitaxial strain for film thicknesses much higher than 10−15 nm before strain relaxation via misfit dislocations begins to occur.29 In this work, we demonstrate a route to synthesizing thick, partially strained VO2 films on columnar TiO2/Al2O3 virtual substrates. By growing (001) oriented TiO2 on m-cut sapphire substrates at various temperatures, we are able to control the crystallite size of the TiO2 layer to achieve an ideal virtual substrate microstructure for reducing strain relaxation in thick (60−100 nm) VO2 films displaying sharp, 4 order of magnitude semiconductor-to-metal transitions at relatively low temperatures (∼44 °C).

■

■

RESULTS AND DISCUSSION Figure 1a shows a schematic illustration of the VO2/TiO2/ Al2O3 heterostructure deposited for this work as well as the epitaxial alignment of each layer. To confirm the epitaxial growth and crystallographic alignment in this heterostructure, XRD φ-scans were performed on the VO2 (101), the TiO2 (101), and the Al2O3 (104) (Figure 1b). As expected, the Al2O3 exhibits only a single azimuthal peak due to the trigonal symmetry of alumina. The φ-scan of the TiO2 (101) shows four 90° spaced peaks indicating fourfold symmetry about the c-axis of TiO2 and one of these peaks appears at the same azimuthal angle as the Al2O3, verifying the epitaxial relationships between

EXPERIMENTAL SECTION

Epitaxial VO2/TiO2 thin film heterostructures were grown on (100) (m-cut) Al2O3 substrates via pulsed laser deposition (PLD) using a KrF excimer laser (LPX 300).30,31 The growth conditions for the VO2 films were optimized for the purpose of maximizing the magnitude and sharpness of the semiconductor-to-metal transition.30 In short, the B

DOI: 10.1021/acsami.6b13112 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

ACS Applied Materials & Interfaces

Research Article

Figure 2. (a) Resistivity (ρ) vs temperature (T) for 60 nm VO2 films grown on TiO2/Al2O3 buffer layers in which the TiO2 layer was grown at 600 °C (purple), 700 °C (blue), and 800 °C (red). (b) The first derivative with respect to temperature of the log of the resistivity shows the temperature of the semiconductor-to-metal phase transition. (c) ρ vs T for VO2/TiO2/Al2O3 heterostructures for different thicknesses of VO2. Note that the TiO2 layers for these films were grown at 700 °C. (d) Shows the corresponding d(log ρ)/dT. (e) ρ vs T for 8 nm VO2 on a single-crystal TiO2 substrate (black), a 60 nm VO2 film on a TiO2/Al2O3 (700 °C) buffer layer (blue), a 60 nm VO2 film on a single-crystal TiO2 substrate (red), and a 60 nm VO2 film on a single-crystal M-Al2O3 substrate (purple). (f) Shows the corresponding d(log ρ)/dT. All VO2 films were grown at 550 °C.

the TiO2 layer and Al2O3 substrate. The φ-scan of the VO2 (101) reveals that the four (101) peaks (tetragonal VO2) appear at the same azimuthal angles as the TiO2 (101) peaks, verifying the epitaxial growth on the TiO2 buffer layer. On the basis of these φ-scan results, the epitaxial relationships in VO2/ TiO 2 /Al 2 O 3 heterostructure can be established as (001)VO2∥(001)TiO2∥(100)Al2O3 and [100]VO2∥[100]TiO2∥[010]Al2O3. For clarity, the 2D projections of the atomic lattice configurations from a top-down perspective for the M-Al2O3, rutile TiO2 (r-TiO2), and t-VO2 layers are depicted in Figure 1c−e. On the basis of the observed alignment, the expected epitaxial lattice misfit strain values for as-grown TiO2 on the Al2O3 substrate are −3.5% along the [100]TiO2 direction and 6.1% along the [010]TiO2 direction, and the expected epitaxial strain values for the as-grown VO2 on bulk-value TiO2 are 0.8% along both the (100)[100] and (010) [010] directions. These misfit strain values are in good agreement with the observed epitaxial growth. Transport behavior as a function of temperature for a variety of VO2/TiO2/Al2O3 heterostructures is provided (Figure 2). We begin with heterostructures for 60 nm thick VO2 in which the TiO2 buffer layer was grown at 600, 700, and 800 °C (Figure 2a,b: purple, blue, and red, respectively). Beginning with the 600 °C film (purple), a ∼3.5 order of magnitude SMT occurring at roughly 48 °C is observed. For the 700 °C film (blue), a 4 order of magnitude transition occurring at 44 °C is observed. Finally, the 800 °C film (red) undergoes a 3 order of magnitude transition at 50 °C. Optical transmittance data for these VO2 films as a function of wavelength are well matched to the electrical transport behavior across the SMT (Figure S1 of the Supporting Information). The differences in electrical transport behavior between these films can be interpreted as a

difference in epitaxial strain. Generally, the VO 2 film experiencing the greatest tensile epitaxial strain would be expected to possess the lowest transition temperature. Thus, on the basis of the transport data, 700 °C can be identified as the ideal growth temperature for the TiO2 layer to most effectively strain the VO2 film. The VO2 films grown upon the 600 and 800 °C TiO2 layers have slightly higher SMT temperatures, indicating that they could be relaxed somewhat compared to the VO2 film grown upon 700 °C TiO2. This hypothesis can be tested by observing the transport character of thicker VO2 films grown upon 700 °C TiO2 (Figure 2c,d). Moving from 60 nm (blue) to 120 nm (purple) to 240 nm (red), the strain is expected to relax as the increasing strain energy nucleates misfit dislocations at the film interface.29,32 Resulting from this gradual relaxation of strain, the transition temperature is observed to increase and the SMT magnitude is observed to decrease. Note that both increasing transition temperatures and diminished SMT magnitude are observed in the VO2 films grown upon 600 and 800 °C TiO2, consistent with relaxation behavior. These strain effects are much more pronounced when comparing 60 nm VO2 on 700 °C buffer TiO2/Al2O3 to the same film grown directly upon Al2O3 substrates and (001)oriented TiO2 substrates (Figure 2e,f: blue, purple, and red, respectively). Note that it is well-established that thick (>20 nm) VO2 grown directly upon Al2O3 or (001) TiO2 substrates are fully relaxed.29 Consistent with prior literature, the transition temperature has increased to 66 °C for VO2/Al2O3 and 60 °C for VO2/TiO2 (Figure 2e,f), which is very close to the bulk transition temperature. For 8 nm VO2 upon (001) TiO2 substrates (Figure 2e,f: black), however, the observed transition temperature is quite low (23 °C), consistent with C

DOI: 10.1021/acsami.6b13112 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

ACS Applied Materials & Interfaces

Research Article

fully strained ultrathin VO2/TiO2.29 The transition temperature for our 60 nm VO2/TiO2/Al2O3 heterostructure is roughly halfway between the fully relaxed 60 nm VO2/Al2O3 and fully strained 8 nm VO2/TiO2. Thus, purely on the basis of transport properties, it appears that the 60 nm VO2 film is partially strained when grown epitaxially upon the 700 °C TiO2/MAl2O3 virtual substrate. The full set of transport properties for the aforementioned films as well VO2 films grown on 650 and 750 °C TiO2/M-Al2O3 virtual substrate can be found in Table 1. In addition to the SMT temperature on sample cooling and

thermore, recent studies on epitaxial perovskite oxides demonstrate that strain relaxation via misfit dislocations only tends to occur when alternate routes for strain relaxation are unavailable.37 In prior studies on VO2/TiO2/Al2O3 heterostructures, it is apparent that the VO2 films are not fully relaxing in the same way that VO2 films on single-crystal TiO2 substrates would be expected to relax (i.e., misfit dislocations). What is not apparent in these studies, however, is the nature of the alternate mechanism for partial strain relaxation enabled by these heterostructures, which is not available for VO2 films on single-crystal TiO2 substrates. The work of Bayati et al. posits that the small lattice misfit strain across the VO2/TiO2 interface may limit the ability of the interface to nucleate misfit dislocations. In short, the small misfit strain results in a large critical thickness for relaxation of ∼10 nm. Thus, Bayati et al. explain that misfit dislocations must glide large distances to the interface, which is difficult due to the high lattice friction found in ionic materials. On the basis of the difficulty of dislocation glide, the authors conclude that the films are observed to only partially relax.32 In our view, this explanation is somewhat unsatisfactory as it again does not account for the difference between the observed relaxation behavior in thick VO2 films on single-crystal TiO2 substrates (fully relaxed) and VO2/TiO2/ Al2O3 heterostructures (partially relaxed). Therefore, we go on to perform several structural characterizations (XRD RSM, AFM, and STEM) to ascertain the strain relaxation mechanisms which allow thick VO2/TiO2/Al2O3 heterostructures to persist in an intermediate strain state. To begin with, RSM studies were conducted to determine the strain state of various VO2/TiO2/Al2O3 heterostructures in which the TiO2 layer has been grown at different temperatures (Figure 3a−c). The RSM data for a 60 nm VO2 film on a TiO2 substrate is also provided (Figure 3d). RSMs were performed about the (112) diffraction peak for rutile TiO2. The dashed lines in Figure 3 correspond to the Qx value of bulk TiO2. The expected Qx and Qz values for TiO2 are marked in each panel with a triangle symbol and the expected Qx and Qz values for tetragonal VO2 (T-VO2) are marked with a square symbol. Therefore, VO2 peaks which fall on this dashed line possess the same in-plane lattice parameter as TiO2 and can be considered coherently strained to the TiO2 buffer layer. Furthermore, VO2 peaks which coincide with the square symbol possess the same lattice parameters as T-VO2 and can be considered fully relaxed. Peaks which fall between the dashed line and square symbol can be considered partially relaxed. We begin by discussing the VO2 films grown on 600 and 700 °C TiO2/Al2O3 buffer layers (Figure 3a,b, respectively). For both sets of films, the Qx value for the VO2 (112) peak falls between the dashed line and the expected Qx value of T-VO2 (square symbol). Furthermore, the Qz value of the VO2 (112) peak is larger than the expected Qz value of T-VO2 (square symbol). Both of these observations indicate that a film is partially strained to the TiO2 buffer layer with in-plane tensile strain and out-of-plane compressive strain. Moving on to the VO2 film grown on the 800 °C TiO2/Al2O3 buffer layer (Figure 3c), some differences emerge. First, the VO2 (112) peak is broader than the peaks observed for the 600 and 700 °C samples. Additionally, the Qx and Qz values of the VO2 (112) peak are much closer to the expected bulk values (square symbol). The peak broadening and the shift of the Qx and Qz values are consistent with a film that is relaxed via misf it dislocations but still in a partially strained state. It is also important to note the similarities between the 800 °C sample and the fully relaxed, 60 nm VO2 film grown on a single-crystal

Table 1. Comparison of Semiconductor-to-Metal Transition Behavior of VO2 Films Grown on TiO2/Al2O3 Buffer Layers Prepared at Different Temperatures sample 60 nm VO2 on 600 °C TiO2/ M-Al2O3 60 nm VO2 on 650 °C TiO2/ M-Al2O3 60 nm VO2 on 700 °C TiO2/ M-Al2O3 60 nm VO2 on 750 °C TiO2/ M-Al2O3 60 nm VO2 on 800 °C TiO2/ M-Al2O3 120 nm VO2 on 700 °C TiO2/ M-Al2O3 240 nm VO2 on 700 °C TiO2/ M-Al2O3 8 nm VO2/(001) TiO2 60 nm VO2/(001) TiO2 60 nm VO2/M-Al2O3

ΔH (°C)

Tc cooling (°C)

Tc heating (°C)

48.5

53.8

3180

5.3

47.9

53.6

4680

5.7

44.5

50.6

8850

6.1

50.1

55.9

3200

5.8

50.3

57.0

1490

6.7

53.9

58.7

5770

4.8

57.5

62.4

3790

4.9

22.9 60.0 65.9

30.4 62.7 71.4

1680 1220 5960

7.5 2.7 5.5

ρs ρM

heating, we also provide the width of the temperature hysteresis ρ (ΔH) and the magnitude of the electrical transition ( s ). ρM

There have been numerous studies examining the exact mechanisms by which lattice strain influences the temperature and character of the VO2 semiconductor-to-metal transition. The general expectation is that the SMT temperature decreases with the decrease of the lattice parameter along the c-axis of the rutile VO2. Given that there is a large change in the c-axis across the SMT, any externally applied strain applied to the tetragonal [001]-direction, such as epitaxial strain, has a significant effect on the characteristics of the transition.32 Mathematically, the relationship between the stress σ and the transition temperature Tc can be expressed as dTc/dσ = (ε0T0c )/ΔH, where ε0 is a material coefficient, T0c is the bulk transition temperature, and ΔH is the latent heat of the SMT. According to this, increasing the strain along the a-axis of tetragonal VO2 is expected to decrease the Tc.33 Our RSM results in Figure 3 clearly indicate that the strained films increase the a-axis lattice parameters and decrease the c-axis lattice parameters compared to the bulk tetragonal VO2, thus decreasing the Tc. The relationship between strain and SMT temperature has been observed by other groups experimentally23,32 and modeled via DFT calculations.34 Although the strain effect in VO2 thin films has been generally studied, a similar strain effect in VO2 particles has also been reported.35,36 The observation of an intermediate transition temperature (∼45 °C) for VO2/TiO2/M-Al2O3 heterostructures is not without precedent. Recent efforts by Bayati et al. on such heterostructures have demonstrated similar behavior.32 FurD

DOI: 10.1021/acsami.6b13112 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

ACS Applied Materials & Interfaces

Research Article

Figure 3. RSMs of 60 nm thick films of VO2 grown on TiO2/Al2O3 buffer layers in which the TiO2 layer has been grown at (a) 600 °C, (b) 700 °C, and (c) 800 °C. (d) An RSM of a 60 nm VO2 film grown on a single-crystal TiO2 substrate. All scans are about the (112) diffraction condition of the TiO2. The dashed lines represent the Qx value of bulk TiO2.

TiO2 substrate (Figure 3d). In the fully relaxed sample, we observe a VO2 (112) peak centered at the expected Qx and Qz values (square symbol) without significant distortion or elongation. These observations support the results of the electrical transport measurements. That is to say, they indicate that the 60 nm VO2 films grown on 600 and 700 °C TiO2/ Al2O3 buffer layers are partially strained to the TiO2 layer. The VO2 film grown on the 800 °C TiO2/Al2O3 buffer layer is also partially strained, although less than the lower temperature samples. Furthermore, based on the shape of the (112) T-VO2 RSM peak for the 800 °C sample, the mechanism for relaxation is different. The RSM studies do not, however, explain the nature of the different mechanisms for strain relaxation between these various structures. It is clear that 600 and 700 °C heterostructures exhibit different relaxation behaviors from that of 800 °C heterostructures, while all three heterostructures exhibit different relaxation behaviors from VO2 films grown directly on TiO2 substrates. To resolve these questions, a series of AFM and STEM studies were performed. Figure 4 shows AFM scans for TiO2/ Al2O3 buffer layers grown at various temperatures without the VO2 layer. The AFM data exhibit a clear trend of surface coarsening with increasing temperatures. Beginning with the 600 °C TiO2/Al2O3 films (Figure 4a), AFM scan reveals a finegrained surface with an average feature size of ∼45 nm. Moving on to the 700 °C TiO2/Al2O3 films (Figure 4b), there is a marked coarsening of the surface features, with an average feature size of ∼85 nm. Finally, the 800 °C TiO2/Al2O3 films (Figure 4c) exhibit the highest degree of coarsening, with an

Figure 4. AFM surface scans of TiO2 buffer layers grown on Al2O3 substrates at 600 °C (a), 700 °C (b), and 800 °C (c). Low-resolution STEM shows the cross section of a 45 nm VO2 film grown on a TiO2/ Al2O3 (700 °C) buffer layer (d).

average feature size of ∼160 nm. Low-resolution STEM images of VO2/TiO2(700 °C)/Al2O3 heterostructures reveal that the surface features observed in the AFM scans correspond to columnar TiO2 crystallites (Figure 4d). On the basis of the AFM and STEM data, it is clear that increasing the growth temperature of the TiO2 layer subsequently increases the lateral size of the columnar TiO2 crystallites. Higher resolution STEM images provide further insight into the structure of the VO2/ TiO2 interface (Figure 5). Bright-field scans exhibiting multiple E

DOI: 10.1021/acsami.6b13112 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

ACS Applied Materials & Interfaces

Research Article

performed on samples that are ∼80 nm thick. Thus, the STEM images are projections through roughly one TiO2 crystallite (∼80 nm diameter). When looking at the cross section of this interface, and considering its 3D nature, three regions can be identified (Figure 6c): the purely TiO2 region (bottom), the purely VO2 region (top), and the mixed TiO2/VO2 3D interface region (middle). STEM EELS studies showing the extent of apparent intermixing at the VO2/TiO2 interface confirm that this mixed region is roughly 8−10 nm thick, indicating a significant presence of both Ti and V atoms at this interface region (see Figure S2). However, as depicted in Figure 6b, this is approximately the thickness of the 3D interface, and thus the EELS data would be expected to detect a mixture of V and Ti atoms within that volume. It is important to note that there is no evidence of additional ion migration or intermixing beyond the 10 nm thick interfacial region. Even if there is some small amount of additional cation intermixing, it does not appear to be more significant than 1−3 nm, which is common for epitaxial oxide films. Regardless, this would not be expected to dominate the electrical response. By performing an inverse fast Fourier transform of the FFT pattern in Figure 5b, masked to show only the vertical lattice fringes, the location of interfacial misfit dislocations can be easily determined (Figure 6d). When the locations of those dislocations are translated to the original STEM image (Figure 6e), it is apparent that they cluster toward the middle and bottom of the 3D interface region. By mapping the vertical positions of the dislocations in the STEM cross section to their corresponding places on the curved 3D interface, a region can be identified along the perimeter of the TiO2 crystallites which readily allows for dislocation nucleation (Figure 6f). The results of the AFM and STEM studies help resolve the discrepancies observed previously. The differences in strain relaxation behavior between VO2 films on TiO2/Al2O3 buffer layers and those on single-crystal TiO2 substrates can be explained by considering the differences in microstructure. The boundary between the TiO2 crystallites has been shown to be a favorable region for dislocation nucleation. Additionally, crystal grain or antiphase boundaries have been repeatedly shown in the literature to help relax tensile strain in epitaxial oxide systems.38−40 Thus, the high density of grain/antiphase boundaries resulting from the columnar growth of the TiO2 buffer layer serves two functions. First, it provides a template which limits where dislocations are allowed to form, and second, it also provides an alternate pathway (through the antiphase boundary) to relax some of the strain at the VO2/ TiO2 interface. This explains why VO2 films persist in an intermediate strain state over a wider range of film thicknesses. However, as the growth temperature of the TiO2 layer is increased and the lateral crystallite size also increases, the density of grain boundaries rapidly decreases. When this alternate pathway for strain relaxation is gradually removed, the films begin to exhibit signs of additional misfit and threading dislocations, as is the case for the 800 °C sample. Single-crystal TiO2 substrates do not possess crystal grain boundaries, and thus such a route for strain relaxation is completely ruled out. This explains why epitaxial VO2 films on single-crystal TiO2 substrates relax almost immediately, generally above thicknesses of tens of nanometers. On the basis of the results of this study, it is observed that TiO2/Al2O3 films grown at 700 °C and possessing an average crystallite size of ∼85 nm provide the ideal virtual substrate for reducing the semiconductor-to-metal transition temperature of

Figure 5. (a) Bright-field STEM cross section of a VO2 film grown on a TiO2/Al2O3 buffer layer (700 °C). Note the columnar structure of the TiO2 crystallites. (b) HAADF cross section of the VO2/TiO2 interface of the same film. The location of the interface is marked with a dashed line. FFT diffraction pattern is shown in the inset of (b). The green diffraction pattern corresponds to the VO2 layer and the red pattern corresponds to the TiO2 layer.

TiO2 crystallites do not exhibit clear evidence of misfit or threading dislocations at the VO2/TiO2 interface (Figure 5a). HAADF images of the interface (Figure 5b) show clear alignment of the VO2 and TiO2 atomic columns and the overlap of fast Fourier transform (FFT) patterns (Figure 5b, inset) indicates a coherent epitaxial match. However, the VO2/ TiO2 interface observed in Figure 5b does not appear to be well-defined. Additional processing of the STEM images sheds light on this issue and clarifies the mechanism of partial relaxation in this system. Figure 6a shows a schematic illustration of the VO2/TiO2/ M-Al2O3 heterostructure. As highlighted in Figure 5a, the interface between the VO2 film and the TiO2 layer cannot be considered strictly planar. By depicting a single TiO2 crystallite, we can see that the interface between the two films possesses a distinct curvature (Figure 6b), which can be referred to as a “3D interface”. The cross-sectional STEM measurements are

Figure 6. (a) Cartoon depiction of the VO 2 /TiO 2 /M-Al 2 O 3 heterostructure. (b) Schematic of 3D interface between TiO2 columnar crystallite and VO2 film. (c) Cartoon cross section of 3D interface. (d) Processed image by performing an inverse fast Fourier transform of the FFT pattern of Figure 5b. The location of the misfit dislocations is highlighted with green and red lines aligned to the atomic columns. (e) Location of misfit dislocations arranged on unprocessed STEM image. (f) Region where dislocations are allowed to form as determined by STEM. F

DOI: 10.1021/acsami.6b13112 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

ACS Applied Materials & Interfaces thick VO2 films. This is likely due to the trade-off between the two strain relaxation mechanisms. Ideally, the density of grain boundaries is low enough to maintain a high degree of epitaxial tensile strain, but not so low that misfit dislocations begin to form (as found in 800 °C samples). The ideal conditions observed here are consistent with those observed in previous studies.32 Finally, the results from this work are consistent with previous models established to account for microstructure effects such as grain size, grain boundaries, and defects.41

■

REFERENCES

(1) Schlom, D. G.; Chen, L.-Q.; Pan, X.; Schmehl, A.; Zurbuchen, M. A. A Thin Films Approach to Engineering Functionality into Oxides. J. Am. Ceram. Soc. 2008, 91, 2429−2454. (2) De Melo, O.; Vargas-Hernandez, C.; Hernandez-Calderon, I. Strain Relaxation During the Layer by Layer Growth of Cubic CdSe onto ZnSe. Appl. Phys. Lett. 2003, 82, 43−45. (3) Dodson, B. W.; Tsao, J. Y. Relaxation of Strained-Layer Semiconductor Structures via Plastic Flow. Appl. Phys. Lett. 1987, 51, 1325−1327. (4) Tsao, J. Y.; Dodson, B. W.; Picraux, S. T.; Cornelison, D. M. Critical Stresses for SixGe1‑x Strained-Layer Plasticity. Phys. Rev. Lett. 1987, 59, 2455−2458. (5) Frank, F. C.; van der Merwe, J. One-Dimensional Dislocations. I. Static Theory. Proc. R. Soc. London, Ser. A 1949, 198, 216−225. (6) Matthews, J. W.; Blakeslee, A. E. Defects in Epitaxial Multilayers. J. Cryst. Growth 1974, 27, 118−125. (7) Maree, P. M. J.; Barbour, J. C.; van der Veen, J. F.; Kavanagh, K. L.; Bulle-Lieuwma, C. W. T.; Viegers, M. P. A. Generation of Misfit Dislocaitons in Semiconductors. J. Appl. Phys. 1987, 62, 4413. (8) Tsao, J. Y.; Dodson, B. W.; Picraux, S. T.; Cornelison, D. M. Critical Streses for SixGe1-x Strained-Layer Plasticity. Phys. Rev. Lett. 1987, 59, 2455−2458. (9) Niu, F.; Meier, A. L.; Wessels, B. W. Epitaxial Growth and Strain Relaxation of MgO Thin Films on Si Grown by Molecular Beam Epitaxy. J. Vac. Sci. Technol. B 2006, 24, 2586−2591. (10) Peng, L. S.-J.; Xi, X. X.; Moeckly, B. H.; Alpay, S. P. Strain Relaxation During in situ Growth of SrTiO3 Thin Films. Appl. Phys. Lett. 2003, 83, 4592−4594. (11) Suzuki, T.; Nishi, Y.; Fujimoto, M. Analysis of Misfit Relaxation in Heteroepitaxial BaTiO3 Thin Films. Philos. Mag. A 1999, 79, 2461− 2483. (12) Merckling, C.; El-Kazzi, M.; Delhaye, G.; Favre-Nicolin, V.; Robach, Y.; Gendry, M.; Grenet, G.; Saint-Girons, G.; Hollinger, G. Strain Relaxation and Critical Thickness for Epitaxial LaAlO3 Thin Films Grown on SrTiO3 (001) Substrates by Molecular Beam Epitaxy. J. Cryst. Growth 2007, 306, 47−51. (13) Vailionis, A.; Boschker, H.; Siemons, W.; Houwman, E. P.; Blank, D. H. A.; Rijnders, G.; Koster, G. Misfit Strain Accommodation in Epitaxial ABO3 Perovskites: Lattice Rotations and Lattice Modulations. Phys. Rev. B: Condens. Matter Mater. Phys. 2011, 83, 064101. (14) Lee, K.; Baik, S. Ferroelastic Domain Structure and Switching in Epitaxial Ferroelectric Thin Films. Annu. Rev. Mater. Res. 2006, 36, 81−116. (15) Damodaran, A. R.; Liang, C.-W.; He, Q.; Peng, C.-Y.; Chang, L.; Chu, Y.-H.; Martin, L. W. Nanoscale Structure and Mechanism for Enhanced Electromechanical Response of Highly strained BiFeO3 Thin Films. Adv. Mater. 2011, 23, 3170−3175. (16) Morin, F. J. Oxides Which Show a Metal-to-Insulator Transition at the Neel Temperature. Phys. Rev. Lett. 1959, 3, 34−36. (17) Manning, T. D.; Parkin, I. P.; Clark, R. J. H.; Sheel, D.; Pemble, M. E.; Vernadou, D. Intelligent Window Coatings: Atmospheric Pressure Chemical Vapour Deposition of Vanadium Oxides. J. Mater. Chem. 2002, 12, 2936−2939. (18) Granqvist, C. G.; Lansaker, P. C.; Mlyuka, N. R.; Niklasson, G. A.; Avendano, E. Progress in Chromogenics: New Results for Electrochromic and Thermochromics Materials and Devices. Sol. Energy Mater. Sol. Cells 2009, 93, 2032−2039.

CONCLUSION In this work, we have demonstrated the ability to reduce the semiconductor-to-metal transition of VO2 to ∼44 °C by growing the film epitaxially upon a TiO2 buffer layer on a single-crystal M-cut Al2O3 substrate. The specific strain states of VO2/TiO2/Al2O3 heterostructures as a function of TiO2 growth temperature were examined with a combination of electrical transport and XRD RSM studies. Finally, surface morphology and interfacial atomic structure were studied in detail by AFM and TEM techniques. Our results show that the columnar microstructure with an 85 nm lateral size present in TiO2 films is responsible for the abnormal strain behavior found in these heterostructures. More specifically, the antiphase boundaries between TiO2 columns serve as ideal locations for the nucleation of dislocations. By optimizing the density of these boundaries, we allow the strain in the VO2 layers to partially relax, thereby sustaining intermediate values of epitaxial strain even with large film thicknesses. This allows the reduction of the Tc far below the bulk value even in thick films. Our results indicate that careful growth and control over the specific microstructure of the TiO2 buffer layer provides a useful way to tune the subsequent properties of the VO2 layer. Thus, these findings establish a potential road map for greater control over the SMT properties of VO2 by using a microstructured buffer layer as a template to direct the film’s strain and relaxation behavior. Taken altogether, these results comprise an important step toward the technological implementation of the VO2 system. There are, furthermore, implications for other oxide thin-film systems in which the relationship between defects, structure, and properties will be crucial for their continued scientific understanding and technological use. ASSOCIATED CONTENT

* Supporting Information S

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsami.6b13112. Optical transmittance and STEM EELS measurements of the interface are discussed and illustrated in Figures S1 and S2 (PDF)

■

ACKNOWLEDGMENTS

This work was funded by the Office of Naval Research (ONR) through the Naval Research Laboratory Basic Research Program (Grant No. N0001415WX00004). E.B. is a National Research Council Fellow at the Naval Research Laboratory. K.B. is an American Association for Engineering Education Postdoctoral Fellow at the Naval Research Laboratory.

■

■

■

Research Article

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Heungsoo Kim: 0000-0002-0181-8010 Notes

The authors declare no competing financial interest. G

DOI: 10.1021/acsami.6b13112 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

ACS Applied Materials & Interfaces

Research Article

crystals with Enhanced Thermochromic Performance. ACS Appl. Mater. Interfaces 2014, 6, 6555−6561. (37) Breckenfeld, E.; Shah, A. B.; Martin, L. W. Strain Evolution in Non-Stoichiometric Heteroepitaxial Thin-Film Perovskites. J. Mater. Chem. C 2013, 1, 8052−8059. (38) He, J. Q.; Vasco, E.; Dittmann, R.; Wang, R. H. Growth Dynamics and Strain Relaxation Mechanisms in BaTiO3 Pulsed Laser Deposited on SrRuO3/SrTiO3. Phys. Rev. B: Condens. Matter Mater. Phys. 2006, 73, 125413. (39) Chen, Y. B.; Matx, M. B.; Pan, X. Q. Microstructure and Strain Relaxation of Epitaxial PrScO3 Thin Films Grown on (001) SrTiO3 Substrates. Appl. Phys. Lett. 2007, 91, 031902. (40) Ding, Y.; Chen, J.; He, J.; Liu, M.; Ma, C.; Chen, C. Interface Structures and Strain Relaxation Mechanisms of Ferroelectric BaTiO3/ SrTiO3 Multilayers on (001) MgO Substrates. J. Cryst. Growth 2013, 383, 19−24. (41) Narayan, J.; Bhosle, V. M. Phase Transition and Critical Issues in Struture-Property Correlations of Vanadium Oxide. J. Appl. Phys. 2006, 100, 103524.

(19) Tan, X. G.; Yao, T.; Long, R.; Sun, Z. H.; Feng, Y. J.; Cheng, H.; Yuan, X.; Zhang, W. Q.; Liu, Q. H.; Wu, C. Z.; Xie, Y.; Wei, S. Q. Unraveling Metal-Insulator Transition Mechanism of VO2 Triggered by Tungsten Doping. Sci. Rep. 2012, 2, 466. (20) Booth, J. M.; Casey, P. S. Anisotropic Structure Deformation in the VO2 Metal-Insulator Transition. Phys. Rev. Lett. 2009, 103, 086402. (21) Holman, K. L.; McQueen, T. M.; Williams, A. J.; Klimczuk, T.; Stephens, P. W.; Zandbergen, H. W.; Xu, Q.; Ronning, F.; Cava, R. J. Insulator to Correlated Metal Transition in V1−xMoxO2. Phys. Rev. B: Condens. Matter Mater. Phys. 2009, 79, 245114. (22) Jeong, J.; Aetukuri, N.; Graf, T.; Schladt, T. D.; Samant, M. G.; Parkin, S. S. P. Suppression of Metal-Insulator Transition in VO2 by Electric Field−Induced Oxygen Vacancy Formation. Science 2013, 339, 1402−1405. (23) Fan, L. L.; Chen, S.; Wu, Y. F.; Chen, F. H.; Chu, W. S.; Chen, X.; Zou, C. W.; Wu, Z. Y. Growth and Phase Transition Characteristics of Pure M-phase VO2 Epitaxial Film Prepared by Oxide Molecular Beam Epitaxy. Appl. Phys. Lett. 2013, 103, 131914. (24) Zhang, S. X.; Kim, I. S.; Lauhon, L. J. Stoichiometry Engineering of Monoclinic to Rutile Phase Transition in Suspended Single Crystalline Vanadium Dioxide Nanobeams. Nano Lett. 2011, 11, 1443−1447. (25) Wei, J.; Ji, H.; Guo, W. H.; Nevidomskyy, A. H.; Natelson, D. Hydrogen Stabilization of Metallic Vanadium Dioxide in Single-Crystal Nanobeams. Nat. Nanotechnol. 2012, 7, 357−362. (26) Muraoka, Y.; Hiroi, Z. Metal−Insulator Transition of VO2 Thin Films Grown on TiO2 (001) And (110) Substrates. Appl. Phys. Lett. 2002, 80, 583−585. (27) Aetukuri, N. B.; Gray, A. X.; Drouard, M.; Cossale, M.; Gao, L.; Reid, A. H.; Kukreja, R.; Ohldag, H.; Jenkins, C. A.; Arenholz, E.; Roche, K. P.; Durr, H. A.; Samant, M. G.; Parkin, S. S. P. Control of The Metal−Insulator Transition in Vanadium Dioxide by Modifying Orbital Occupancy. Nat. Phys. 2013, 9, 661. (28) Nagashima, K.; Yanagida, T.; Tanaka, H.; Kawai, T. Stress Relaxation Effect on Transport Properties of Strained Vanadium Dioxide Epitaxial Thin Films. Phys. Rev. B: Condens. Matter Mater. Phys. 2006, 74, 172106. (29) Fan, L. L.; Chen, S.; Luo, Z. L.; Liu, Q. H.; Wu, Y. F.; Song, L.; Ji, D. X.; Wang, P.; Chu, W. S.; Gao, C.; Zou, C. W.; Wu, Z. Y. Strain Dynamics of Ultrathin VO2 Film Grown on TiO2 (001) and the Associated Phase Transition Modulation. Nano Lett. 2014, 14, 4036− 4043. (30) Kim, H.; Charipar, N.; Osofsky, M.; Qadri, S. B.; Piqué, A. Optimization of The Semiconductor-Metal Transition in VO 2 Epitaxial Thin Films as a Function of Oxygen Growth Pressure. Appl. Phys. Lett. 2014, 104, 081913. (31) Kim, H.; Charipar, N.; Breckenfeld, E.; Rosenberg, A.; Piqué, A. Active Terahertz Metamaterials Based on the Phase Transition of VO2 Thin Films. Thin Solid Films 2015, 596, 45−50. (32) Bayati, M. R.; Molaei, R.; Wu, F.; Budai, J. D.; Liu, Y.; Narayan, R. J.; Narayan, J. Correlation Between Structures and Semiconductorto-metal Transition Characteristics of VO2/TiO2/Sapphire Thin Film Heterostructures. Acta Mater. 2013, 61, 7805−7815. (33) Cao, J.; Ertekin, E.; Srinivasan, V.; Fan, W.; Huang, S.; Zheng, H.; Yim, J. W. L.; Khanal, D. R.; Ogletree, D. F.; Grossman, J. C.; Wu, J. Strain Engineering and One-Dimensional Organization of Metal− Insulator Domains in Single-Crystal Vanadium Dioxide Beams. Nat. Nanotechnol. 2009, 4, 732−737. (34) Booth, J. M.; Drumm, D. W.; Casey, P. S.; Smith, J. S.; Seeber, A. J.; Bhargava, S. K.; Russo, S. P. Correlating the Energetics and Atomic Motions of the Metal-Insulator Transition of M1 Vanadium Dioxide. Sci. Rep. 2016, 6, 26391. (35) Li, Y.; Ji, S.; Gao, Y.; Luo, H.; Jin, P. Modification of Mott Phase Transition Characteristics in VO2@TiO2 Core/Shell Nanostructures by Misfit-Strained Heteroepitaxy. ACS Appl. Mater. Interfaces 2013, 5, 6603−6614. (36) Li, D.; Li, M.; Pan, J.; Luo, Y.; Wu, H.; Zhang, Y.; Li, G. Hydrothermal Synthesis of Mo-Doped VO2/TiO2 Composite NanoH

DOI: 10.1021/acsami.6b13112 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX