Visualization of local phase transition behaviors near dislocations in epitaxial VO2/TiO2 thin films

Visualization of local phase transition behaviors near dislocations in epitaxial VO2/TiO2

thin films

Ahrum Sohn, Teruo Kanki, Hidekazu Tanaka, and Dong-Wook Kim

Citation: Applied Physics Letters 107, 171603 (2015); doi: 10.1063/1.4934943

View online: http://dx.doi.org/10.1063/1.4934943

View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/107/17?ver=pdfcov Published by the AIP Publishing

Articles you may be interested in

Transport properties of ultra-thin VO2 films on (001) TiO2 grown by reactive molecular-beam epitaxy Appl. Phys. Lett. 107, 163101 (2015); 10.1063/1.4932123

Spinodally decomposed nanostructures in a TiO2–VO2 crystal APL Mater. 3, 062508 (2015); 10.1063/1.4919764

Light scattering by epitaxial VO2 films near the metal-insulator transition point J. Appl. Phys. 117, 184304 (2015); 10.1063/1.4921057

Improved metal-insulator-transition characteristics of ultrathin VO2 epitaxial films by optimized surface preparation of rutile TiO2 substrates

Appl. Phys. Lett. 104, 081918 (2014); 10.1063/1.4866037

Direct observation of giant metallic domain evolution driven by electric bias in VO2 thin films on TiO2(001) substrate

Appl. Phys. Lett. 101, 243118 (2012); 10.1063/1.4772211

Visualization of local phase transition behaviors near dislocations in

epitaxial VO

/TiO

thin films

AhrumSohn,1TeruoKanki,2HidekazuTanaka,2and Dong-WookKim1,a)

1

Department of Physics, Ewha Womans University, Seoul 120-750, South Korea

2

The Institute of Scientific and Industrial Research, Osaka University, Osaka 567-0047, Japan

(Received 26 July 2015; accepted 19 October 2015; published online 28 October 2015)

We investigated local phase transition behaviors in epitaxial VO2/TiO2thin films using

variable-temperature Kelvin probe force microscopy while spanning the metal–insulator transition (MIT). Fully strained thin films were almost free of grain boundaries. In contrast, thicker films had cracks (dislocations) caused by strain relaxation. The surface area fraction of the insulating phase near the dislocations was higher than that in other regions. Thicker films have complicated domain patterns; hence, the three-dimensional percolation model properly described the MIT behaviors. In contrast, the two-dimensional percolation model well explained the transition behaviors of uniformly strained thinner films.VC _{2015 AIP Publishing LLC.}

[http://dx.doi.org/10.1063/1.4934943]

VO2 shows a tetragonal-monoclinic structural phase

transition (SPT) as well as a metal–insulator transition (MIT) near room temperature.1–15Due to this unique feature, both Mott transition and Peierls instability have been considered as key theoretical components in explaining this intriguing phase transition phenomenon.1–9 The structural distortion, occurring during the SPT, affects the orbital occupancy which directly changes the electron-electron correlation.6–9 VO6octahedra undergo distortion during the SPT; the

result-ing change in the V4þ-V4þdistance varies the bandwidth of Vd orbitals.6–10Thus, strain states of VO2can significantly

modify the MIT behaviors, as reported in many thin film studies; additionally, strain states can be controlled by varying the substrates and/or thickness of the samples.7–15 Thick films, especially samples with a large lattice mis-match, may have nonuniform strain states and contain struc-tural defects.10–14Local phase transition behaviors may vary depending on the lattice spacing at specific locations and the interaction with nearby regions. Thus, contributions from all nanoscopic domains exhibiting diverse transition behaviors determine the overall macroscopic physical properties (e.g., MIT temperature and resistivity) of the thin films.

Ultrathin epitaxial VO2 films on the TiO2 (001)

sub-strates can have large-sized grains, due to the small lattice mismatch (0.86%).7–15 _{As the thickness increases, the}

strain relaxation in the VO2/TiO2heterostructures is known

to create cracks and dislocations.11,12 Cracks deform the VO2lattice in their vicinity, which may alter the local phase

transition behaviors. Recently, we reported that measure-ments of the local surface work function using Kelvin probe force microscopy (KPFM) could distinguish metallic and insulating domains in 15-nm-thick VO2/TiO2 thin films.15

KPFM has a spatial resolution of tens of nm and has been used to probe the nonuniform electronic property of a sample.16Thus, the KPFM measurements allow us to exam-ine the domain configurations near the cracks to provide

information on how structural defects modify the local phase transition behaviors of VO2.

In this letter, we investigated the domain patterns of 15-, 30-, and 45-nm-thick VO2/TiO2 thin films, while spanning

the phase transition. Local work function maps obtained from the variable-temperature KPFM measurements allowed us to visualize the evolution of metallic and insulating domain configurations. Fully strained 15-nm-thick films were almost free of grain boundaries. In contrast, 30- and 45-nm-thick films had cracks (dislocations) generated by strain relaxation, on top of that ridges were located. The surface area fraction of insulating domains near the ridges was higher than that in surrounding areas at elevated temperatures. This implies that less strain may stabilize the insulating states. Additionally, these results highlight the complexity of the domain pattern in 30- and 45-nm-thin films and suggest that the three-dimensional (3D) percolation model be used to represent thicker films. In contrast, the two-dimensional (2D) percolation model well-explained the MIT behavior of the 15-nm-thick film.

VO2 thin films were grown on rutile TiO2 (001)

sub-strates by pulsed laser deposition (ArF Excimer laser, k: 193 nm; repetition rate: 2 Hz; energy fluence: 10 mJ cm2), at 430C in an oxygen atmosphere of 1.0 Pa using a V2O5

pellet as a target. High-quality epitaxial films were prepared; detailed structural properties of the thin films can be obtained from previous reports.11,12

KPFM measurements were conducted using an atomic force microscopy (AFM) system (XE-100, Park Systems Co.) with a glove box, as illustrated in Fig.1(a).15 Samples were cooled to 285 K and heated to 355 K by a Peltier device. Pt-coated Si cantilevers were used in non-contact mode for both topography and surface work function meas-urements. Immediately after each KPFM measurement, the work function of the tip was calibrated with a highly ordered pyrolytic graphite (HOPG, SPI Supplies) reference sample to determine the work function of the sample. During heating and cooling cycles, the resistance of the sample was meas-ured by bonding Al wires to the sample and connecting them

a)

E-mail: [email protected]

to a semiconductor parameter analyzer (4156B, Hewlett Packard) (see the schematic diagram in Fig.1(a)). The wires were connected together in a short circuit while performing the KPFM measurements. Water adsorbates can form elec-tric dipoles at the sample surface and hinder measurement of the inherent sample work function.17To avoid such artifacts, our sample underwent heat treatment at 373 K for 30 min in a dry N2atmosphere, prior to the measurements; the N2gas

ambient was maintained throughout the experiments.15 The temperature dependences of both measured resistance and work function did not show any noticeable differences dur-ing repeated heatdur-ing and cooldur-ing cycles, confirmdur-ing the reproducibility of the data.

Figures 1(b)–1(d) show the AFM images of the VO2/TiO2 thin films with three different thicknesses: 15,

30, and 45 nm. The 15-nm-thick film had a smooth surface (root-mean-square [rms] roughness: 0.4 nm). In contrast, the 30- and 45-nm-thick films showed ridges (height 3 nm) on their surfaces, as reported earlier.11,12_{The ridges}

of the 30-nm-film were straight and parallel. The ridges of the 45-nm-film were wider and less straight than those in the 30-nm-film. The surface area fractions of ridges were 16.9% and 24.7% of the scanned area in the 30- and 45-nm-thick films, respectively. High resolution AFM images showed that cracks existed in the middle of the ridges.11 High-resolution transmission electron microscopy images showed that dislocations were present just below the cracks in our VO2/TiO2 epitaxial thin films.12 Nagashima et al.

reported that the critical thickness for crack creation was 15 nm, consistent with our surface morphology images in Figs.1(b)–1(d).11

Figures 2(a)–2(c) show the normalized resistance and work function of the 15-, 30-, and 45-nm-thick VO2/TiO2

thin films as a function of temperature, respectively; the resistance at 285 K was set to 1 for normalization, and the work function was averaged over a 2 2 lm2area. The 15-nm-thick film showed an abrupt change in resistance at the transition temperature (TC) of300 K. The TCwas less

than the bulk value, because the tensile strain can shorten the c-axis length of VO2 and trigger phase transition at lower

temperatures.6–15 As the sample thickness increases, TC

shifts toward higher temperatures, which indicates strain relaxation.6–14 Thicker samples showed a step-like change and gradual variation in their resistance values (Figs. 2(b)

and 2(c)); this suggests that nonuniform strain states may induce complicated phase transition behaviors in the 30- and 45-nm-thick films.11

Figures2(a)–2(c)also show the thickness dependence of the work function change while spanning the MIT. The work function at low temperatures was higher than that at high temperatures; this was attributed to alteration of the elec-tronic structure during the phase transition. The work

function difference at low and high temperatures is 0.1–0.2 eV, which is about ten times larger than the standard deviation of the measured work function data. It should also be noted that the thicker films exhibited a wider hysteresis width in the work function versus temperature plots. Recent studies reported that the electronic band structures and orbital occupancy of VO2 depended on its lattice

parame-ters;7–9this suggests that strain states in the localized regions determine the electronic structures and temperature depend-ence in the specific region. Therefore, the average work function of thick samples with nonuniform strain states would be expected to exhibit broader hysteresis.

Figures3(a)–3(c) show the work function maps of the 15-, 30-, and 45-nm-thick VO2/TiO2films obtained at three

different temperatures. Figure 3(a) shows that two distinct regions with high and low work functions exist for all of the measurement temperatures. An intermediate work function

FIG. 1. (a) Schematic illustration of the variable temperature Kelvin probe force microscopy system with a glove box. Surface topographic images of (b) 15-, (b) 30-, and (c) 45-nm-thick VO2/

TiO2 thin films. The scale bars are

500 nm.

FIG. 2. Normalized resistance (black circles) and work function (red squares) as a function of temperature for (a) 15-, (b) 30-, and (c) 45-nm-thick VO2/TiO2thin films. Filled and open symbols are for heating and

cool-ing, respectively.

171603-2 Sohn et al. Appl. Phys. Lett. 107, 171603 (2015)

region appeared at the boundaries between the high and low work function regions. In our earlier work, we showed that local surface work function maps of a 15-nm-film could be used to reveal the spatial distribution of metallic and insulating domains during phase transition.15 The estimated metallic fraction indicated that the 2D percolation model well-explained the temperature dependence of the resistance. The 30- and 45-nm-thick films have ridges, as shown in Figs. 2(c) and 2(d). These two samples consist of regions with two distinct work functions, similar to the 15-nm-thick film. This suggests that the metallic and insulating domains can be identified in all three samples. From Figs.3(a)–3(c), it should be noted that the size of the low work function regions (metallic domains) in the thicker samples was smaller than that of the thinner samples. Kawatani et al. reported that the thicker VO2/TiO2 films had smaller

domains, based on the optical microscopy investigations.12 Taken together, the results from this study show that the sur-face work function estimated by the KPFM measurements well-represents the local electronic phase of VO2thin films.

At elevated temperatures, the surface area fraction of the insulating phase near the ridges (red regions in the color maps in Figs. 3(a)–3(c)) was higher than that in other regions. For example, most of the insulating domains were located on the ridges of the 30-nm-thick film at 355 K, as shown in Fig.3(b). As discussed earlier, strain relaxation in the epitaxial VO2/TiO2 thin films occurs via crack

forma-tion.11–14 The local region near the ridges undergoes less

tensile strain; hence, the insulating phase can be stabilized even at higher temperatures. The ridges in the 45-nm-thick film were wider and closer to neighboring ridges, compared with those in the 30-nm-thick film. This suggests that the spatial relationship between the ridges and strain-relaxed region in the 45-nm-thick film is not straightforward. Despite such complicated behaviors, a large portion of the ridges were covered with the insulating domains, even at 355 K (Fig. 3(c)). The work function maps showed elonga-tion features along the tip scan direcelonga-tion (x-direcelonga-tion in Figs.3(a)–3(c)), which might be caused by tip-sample inter-action. Such scan-direction-related artifacts, however, did not affect the estimation of insulating domain fractions of the samples.18

While spanning the MIT, both metallic and insulating domains were evident in the VO2thin films. As a result,

elec-trical transport in the film can be explained by the percola-tion model.3,15 According to the percolation model, the conductivity of the sample, r, is a function of the metallic fraction, PM; specifically, r / (PM – PC)t, where PC is the

percolation threshold andt is the critical exponent. The uni-versal values arePC¼ 0.45 and t ¼ 1.4 for the 2D percolation

conduction model, and 0.15 <PC< 0.17 and t¼ 2.0 for the

3D percolation model.3,15

Figure4shows the inverse of the normalized resistance (1/[normalized resistance]/r) as a function of PMestimated

from the work function maps. The 1/[resistance] versus PM

plots can be fitted by the percolation model, with PC andt

extracted from the best-fit curve (lines of Fig. 4).PC of the

15-nm-thick film (0.36) was much larger than those of the 30-nm-thick (0.17) and 45-30-nm-thick (0.15) films;t for the 15-nm-thick film was 1.46, whereas for both the 30- and 45-nm-15-nm-thick filmst was 2.00. These analyses show that the 2D percolation model well-explains the relationship between the metallic do-main fraction and the sample resistivity of the 15-nm-thick VO2/TiO2film.15In contrast, the 3D model was appropriate

for the 30- and 45-nm-thick films. The 15-nm-thick VO2/TiO2

samples were free of grain boundaries, with whole layers under uniform tensile strain. Thus, the electronic domains were homogeneous along the thickness direction. Thicker films underwent nonuniform strain relaxation, as evidenced by local crack formation. However, the strain states near the surface may be different from those at the film–substrate

FIG. 3. Work function maps obtained at 285, 325, and 355 K for (a) 15-, (b) 30-, and (c) 45-nm-thick VO2/TiO2films. These images were obtained from

the regions indicated by the yellow rectangles in Figs.1(b)–1(d). The white lines indicate the region of which height is 1.5 nm from the flat surface (1/ 2 of the ridge height).

FIG. 4. Inverse of the normalized resistance as a function of the metallic fractionPMduring heating. The solid lines fit the measured data.

interface.9Therefore, multiple domains may exist along the thickness direction. Consequently, the domain configurations in the thick films were more complex and required the 3D per-colation model for their representation.

In summary, we investigated the evolution of metallic and insulating domain configurations in the epitaxial VO2/

TiO2thin films, based on the nanoscopic surface work

func-tion measurements. Strain relaxafunc-tion generated cracks (dislo-cations) and surface ridges in the 30- and 45-nm-thick films. The surface area fraction of insulating domains on the ridges was higher than that in other regions, while spanning the phase transition. At elevated temperatures, the results showed that less strain near the dislocations could stabilize the insulating states. The 30- and 45-nm-thin films exhibited 3D percolation behavior, whereas the 15-nm-thick film showed 2D behavior. This suggests that the strain relaxation in thicker films induced a nonuniform strain state in the film, resulting in complicated domain configurations.

This work was supported by a Grant-in-Aid for Scientific Research A (No. 26246013) and a Grant-in-Aid for Scientific Research B (No. 25286058) from the Japan Society for the Promotion of Science (JSPS). A.S. and D.K. acknowledge the support from Basic Science Research Program and the Quantum Metamaterials Research Center (QMMRC) through the National Research Foundation of Korea Grant (Nos. 2013R1A1A2063744 and 2015001948).

1

F. J. Morin,Phys. Rev. Lett.3, 34 (1959).

2

H.-T. Kim, Y. W. Lee, B.-J. Kim, B.-G. Chae, S. J. Yun, K.-Y. Kang, K.-J. Han, K.-J. Yee, and Y.-S. Lim,Phys. Rev. Lett.97, 266401 (2006).

3_{Y. J. Chang, J. S. Yang, Y. S. Kim, D. H. Kim, and T. W. Noh,Phys. Rev. B}

76, 075118 (2007).

4_{A. Sharoni, J. G. Ramırez, and I. K. Schuller, Phys. Rev. Lett. 101,}

026404 (2008).

5

A. C. Jones, S. Berweger, J. Wei, D. Cobden, and M. B. Raschke,Nano Lett.10, 1574 (2010).

6

J. Laverock, A. R. H. Preston, D. Newby, Jr., K. E. Smith, S. Sallis, L. F. J. Piper, S. Kittiwatanakul, J. W. Lu, S. A. Wolf, M. Leandersson, and T. Balasubramanian,Phys. Rev. B86, 195124 (2012).

7

N. B. Aetukuri, A. X. Gray, M. Drouard, M. Cossale, L. Gao, A. H. Reid, R. Kukreja, H. Ohldag, C. A. Jenkins, E. Arenholz, K. P. Roche, H. A. D€urr, M. G. Samant, and S. S. P. Parkin,Nat. Phys.9, 661 (2013).

8_{N. F. Quackenbush, J. W. Tashman, J. A. Mundy, S. Sallis, H. Paik, R.}

Misra, J. A. Moyer, J.-H. Guo, D. A. Fischer, J. C. Woicik, D. A. Muller, D. G. Schlom, and L. F. J. Piper,Nano Lett.13, 4857 (2013).

9

L. L. Fan, S. Chen, Z. L. Luo, Q. H. Liu, Y. F. Wu, L. Song, D. X. Ji, P. Wang, W. S. Chu, C. Gao, C. W. Zou, and Z. Y. Wu,Nano Lett.14, 4036 (2014).

10

Y. Muraokaa and Z. Hiroi,Appl. Phys. Lett.80, 583 (2002).

11

K. Nagashima, T. Yanagida, H. Tanaka, and T. Kawai,Phys. Rev. B74, 172106 (2006).

12

K. Kawatani, T. Kanki, and H. Tanaka, Phys. Rev. B 90, 054203 (2014).

13_{M. K. Liu, M. Wagner, E. Abreu, S. Kittiwatanakul, A. McLeod, Z. Fei,}

M. Goldflam, S. Dai, M. M. Fogler, J. Lu, S. A. Wolf, R. D. Averitt, and D. N. Basov,Phys. Rev. Lett.111, 096602 (2013).

14

T. Yajima, Y. Ninomiya, T. Nishimura, and A. Toriumi,Phys. Rev. B91, 205102 (2015).

15_{A. Sohn, T. Kanki, K. Sakai, H. Tanaka, and D.-W. Kim, Sci. Rep.5,}

10417 (2015).

16

S. Sadewasser and T. Glatze, Kelvin Probe Force Microscopy: Measuring and Compensating Electrostatic Forces (Springer, Heidelberg, 2012).

17

H. Kim, S. Hong, and D.-W. Kim, Appl. Phys. Lett. 100, 022901 (2012).

18

See supplementary material at http://dx.doi.org/10.1063/1.4934943 for scan-direction dependence of the work function maps.

171603-4 Sohn et al. Appl. Phys. Lett. 107, 171603 (2015)