research papers\(\def\hfill{\hskip 5em}\def\hfil{\hskip 3em}\def\eqno#1{\hfil {#1}}\)

Journal logoJOURNAL OF
SYNCHROTRON
RADIATION
ISSN: 1600-5775

Asymmetric response of electrical conductivity and V valence state to strain in cation-deficient Sr1–yVO3 ultrathin films based on absorption measurements at the V L2- and L3-edges

CROSSMARK_Color_square_no_text.svg

aFujian Provincial Key Laboratory of Semiconductors and Applications, Collaborative Innovation Center for Optoelectronic Semiconductors and Efficient Devices, Department of Physics, Xiamen University, Xiamen 361005, People's Republic of China, and b1 MESA+ Institute for Nanotechnology, University of Twente, Enschede, The Netherlands
*Correspondence e-mail: meng.wu@xmu.edu.cn, hqwang@xmu.edu.cn

Edited by R. W. Strange, University of Essex, UK (Received 14 March 2019; accepted 16 May 2019; online 12 July 2019)

The correlation between electronic properties and epitaxial strain in a cation-deficient system has rarely been investigated. Cation-deficient SrVO3 films are taken as a model system to investigate the strain-dependent electrical and electronic properties. Using element- and charge-sensitive soft X-ray absorption, V L-edge absorption measurements have been performed for Sr1–yVO3 films of different thicknesses capped with 4 u.c. (unit cell) SrTiO3 layers, showing the coexistence of V4+ and V5+ in thick films. A different correlation between V valence state and epitaxial strain is observed for Sr1–yVO3 ultrathin films, i.e. a variation in V valence state is only observed for tensile-strained films. Sr1–yVO3 thin films are metallic and exhibit a thickness-driven metal–insulator transition at different critical thicknesses for tensile and compressive strains. The asymmetric response of electrical conductivity to strain observed in cation-deficient Sr1–yVO3 films will be beneficial for functional oxide electronic devices.

1. Introduction

Perovskite transition metal oxides show fascinating physical properties attractive for both device applications and theoretical investigations. Among these materials, non-stoichiometry can easily be formed and indeed is one of the crucial parameters in determining the novel properties. For instance, oxygen vacancies provide an excellent platform for tuning the electrical, magnetic and optical properties in titanates (Tarun et al., 2013[Tarun, M. C., Selim, F. A. & McCluskey, M. D. (2013). Phys. Rev. Lett. 111, 187403.]), manganites (Ma et al., 2016[Ma, J., Zhang, Y., Wu, L., Song, C., Zhang, Q., Zhang, J., Ma, J. & Nan, C.-W. (2016). MRS Commun. 6, 354-359.]), cobaltates (Stemmer et al., 2001[Stemmer, S., Jacobson, A. J., Chen, X. & Ignatiev, A. (2001). J. Appl. Phys. 90, 3319-3324.]) and iron-based multiferroic compounds (Farokhipoor & Noheda, 2011[Farokhipoor, S. & Noheda, B. (2011). Phys. Rev. Lett. 107, 127601.]). The coupling between strain and oxygen vacancies has been emphasized in a review article recently that showed the strongly correlated features in a diversity of perovskite systems which might be beneficial for information and energy technologies (Herklotz et al., 2017[Herklotz, A., Lee, D., Guo, E.-J., Meyer, T. L., Petrie, J. R. & Lee, H. N. (2017). J. Phys. Condens. Matter, 29, 493001.]). Another form of non-stoichiometry with cation deficiencies and/or an excess of oxygen atoms behaves differently, which can for instance induce a p-type conductivity as observed in manganite (Skjarvo et al., 2016[Skjaervø, S. H., Wefring, E. T., Nesdal, S. K., Gaukås, N. H., Olsen, G. H., Glaum, J., Tybell, T. & Selbach, S. M. (2016). Nat. Commun. 7, 13745.]), an enhanced Tc and more complicated physics as reported in high-Tc cuprates (Pan et al., 2001[Pan, S. H., O'Neal, J. P., Badzey, R. L., Chamon, C., Ding, H., Engelbrecht, J. R., Wang, Z., Eisaki, H., Uchida, S., Gupta, A. K., Ng, K.-W., Hudson, E. W., Lang, K. M. & Davis, J. C. (2001). Nature, 413, 282-285.]), or novel interfacial properties as reported in LaTiO3+δ and LaNiO3 heterostructures (Cao et al., 2016[Cao, Y., Liu, X., Kareev, M., Choudhury, D., Middey, S., Meyers, D., Kim, J.-W., Ryan, P. J., Freeland, J. & Chakhalian, J. (2016). Nat. Commun. 7, 10418.]). However, the correlation between electronic properties and epitaxial strain in a cation-deficient and/or extra oxygen-doped system has rarely been investigated, so this will be addressed in our present study.

The SrVO3 (SVO) parent compound shows an ideal cubic crystal symmetry of lattice constant 3.843 Å. The correlated SVO thin films exhibit dimension-controlled metallic quantum well structures and a metal–insulator transition below a critical thickness (Yoshimatsu et al., 2011[Yoshimatsu, K., Horiba, K., Kumigashira, H., Yoshida, T., Fujimori, A. & Oshima, M. (2011). Science, 333, 319-322.], 2013[Yoshimatsu, K., Sakai, E., Kobayashi, M., Horiba, K., Yoshida, T., Fujimori, A., Oshima, M. & Kumigashira, H. (2013). Phys. Rev. B, 88, 115308.]). The low resistivity of an SVO film as well as the excellent structural and chemical compatibility between SVO and other iso­structural compounds show its potential application as the electrode layer for functional perovskite oxide devices (Moyer et al., 2013[Moyer, J. A., Eaton, C. & Engel-Herbert, R. (2013). Adv. Mater. 25, 3578-3582.]). However, experimental synthesis meets difficulties in the following aspects. Firstly, the partially filled V 3d shell allows a set of different valence states ranging from V5+ to V2+, which can form a number of single- and multi-valence vanadium oxides with various physical and spectroscopic properties (Wu et al., 2018[Wu, M., Zheng, J.-C. & Wang, H.-Q. (2018). Phys. Rev. B, 97, 245138.]; Sawa et al., 2002[Sawa, H., Ninomiya, E., Ohama, T., Nakao, H., Ohwada, K. J., Murakami, Y., Fujii, Y., Noda, Y., Isobe, M. & Ueda, Y. (2002). J. Phys. Soc. Jpn, 71, 385-388.]; Yoshino et al., 2017[Yoshino, T., Okawa, M., Kajita, T., Dash, S., Shimoyama, R., Takahashi, K., Takahashi, Y., Takayanagi, R., Saitoh, T., Ootsuki, D., Yoshida, T., Ikenaga, E., Saini, N. L., Katsufuji, T. & Mizokawa, T. (2017). Phys. Rev. B, 95, 075151.]). Secondly, the oxide ceramic target for SVO film deposition is usually single-phase polycrystalline Sr2V2O7 which requires extremely reducing conditions to stabilize V4+. Therefore, non-stoichiometry can easily be created for thin-film engineering related to SVO, similar to another well studied early transition metal compound LaTiO3 using an La2Ti2O7 target (Ohtomo et al., 2002[Ohtomo, A., Muller, D. A., Grazul, J. L. & Hwang, H. Y. (2002). Appl. Phys. Lett. 80, 3922-3924.]). Therefore, the understanding of the electrical properties of SVO thin films deserves particular attention. The easy oxidation of V ions to V5+ in SVO films brings the possibility of investigating the interplay between cation deficiency and other degrees of freedom.

In this article, we present a study of the electrical conductivity and valence state of cation-deficient Sr1–yVO3 thin films capped with 4 u.c. (unit cell) SrTiO3 (STO) layers. Different epitaxial strain-dependent studies show an asymmetric correlation between the V valence state and the strain effect in ultrathin Sr1–yVO3 films. We also provide a detailed analysis of the electrical transport properties of the strain- and thickness-dependent Sr1–yVO3 films. Our present studies shed light on understanding the electrical and electronic properties of functional oxides, as well as designing and optimizing functional oxide devices.

2. Experimental methods and results

2.1. Sample deposition and structural characterization

The Sr1–yVO3 thin films were deposited on STO, LaAlO3 (LAO) and LaSrAlO4 (LSAO) substrates by pulsed laser deposition from a single-phase Sr2V2O7 polycrystalline target, using a KrF excimer with a laser energy density of 2 mJ cm−2 and a frequency of 5 Hz. The temperature was increased at a rate of 10 K min−1 and stabilized at 1053 K to prevent surface contamination. An ultrahigh-vacuum pressure of 10−8 Torr (1 Torr = 133.322 Pa) is a prerequisite for film deposition in an extremely reducing environment. The materials were deposited at a substrate temperature of 983 K with an Ar flow at a pressure of 2 × 10−2 Torr. A nonreactive Ar atmosphere has been demonstrated as an effective strategy for high-quality SVO thin-film deposition at low oxygen pressure (Wang et al., 2018[Wang, J., Rijnders, G. & Koster, G. (2018). Appl. Phys. Lett. 113, 223103.]; Mirjolet et al., 2019[Mirjolet, M., Sánchez, F. & Fontcuberta, J. (2019). Adv. Funct. Mater. 29, 1808432.]), which influences the propagation of the plasma plume through reducing the kinetic energy of the species therein. After film growth, the temperature was lowered at a constant rate of 10 K min−1 to room temperature while the pressure was kept unchanged. All the deposited thin films were capped with 4 u.c. STO to prevent the influence of surface oxidation kinetics and interface energies (Ohtomo et al., 2002[Ohtomo, A., Muller, D. A., Grazul, J. L. & Hwang, H. Y. (2002). Appl. Phys. Lett. 80, 3922-3924.]). The lattice parameters of the substrate and the lattice mismatch are shown in Table 1[link].

Table 1
Lists of composition, (pseudo-)cubic or tetragonal lattice constants csub and asub of the substrate, the lattice mismatch calculated by (asubaSVO)/aSVO using bulk SVO with a lattice constant a = b = c = 3.843 Å, the measured out-of-plane lattice constant of an SVO thin film cfilm, the measured in-plane lattice constant afilm and the relaxation R = (afilmmeas - asub)/(afilmrelax - asub), with afilmmeas and afilmrelax denoting the in-plane lattice constants of the strained and relaxed layers, respectively

The lattice parameters for 10 u.c. Sr1–yVO3 on LSAO were not measured due to the measurement limit of a typical in-house XRD setup.

SVO on csub (Å) asub (Å) Mismatch (%) cfilm (Å) afilm (Å) R (%)
STO (26 u.c.) 3.905 3.905 1.6 3.846 3.905 0
LAO (26 u.c.) 3.79 3.79 −1.4 3.94 3.822 60
LSAO (10 u.c.) 12.636 3.756        

In situ reflection high-energy electron diffraction (RHEED) is highly surface sensitive and was used to provide real-time monitoring of the film deposition process. The electron diffraction patterns of cation-deficient SVO thin films [Figs. 1[link](b)–1[link](c)] show broadened streaks, indicating a surface of monolayer roughness and a two-dimensional layer-by-layer crystal growth. The growth rate and precise control of the unit-cell thickness were determined from the RHEED oscillations, where each full period oscillation corresponds to the formation of a single atomic layer thickness.

[Figure 1]
Figure 1
RHEED images captured (a) before deposition, (b) after Sr1–yVO3 film deposition and (c) after deposition of the STO capping layers. (d) A RHEED oscillation pattern for a 26 u.c. Sr1–yVO3 film (10 nm) with a 4 u.c. STO capping layer deposited on an STO substrate.

Figs. 2[link](a)–2[link](b) show high-resolution X-ray diffraction scans along the specular direction for a 26 u.c. Sr1–yVO3 film deposited with tensile and compressive strains, respectively. Figs. 2[link](c)–2[link](d) show the corresponding reciprocal-space maps around the 013 reflection. From the read-off positions of the diffraction angles, the epitaxial films show average out-of-plane constants of 3.846 (8) and 3.94(2) Å for 26 u.c. Sr1–yVO3 deposited under tensile and compressive strains, respectively. The films deposited on STO are fully strained with the same in-plane lattice constant of 3.905 Å as the substrate. However, the 26 u.c. Sr1–yVO3 film deposited on compressively strained LAO is partially relaxed, i.e. with an in-plane lattice constant of 3.822 (5) Å compared with 3.79 Å for the LAO substrate. The relaxation R is defined to describe the strain state of Sr1–yVO3 thin films on STO and LAO as shown in Table 1[link], where R = 0 represents the fully strained state. We note that expansion of the out-of-plane lattice constant implies volume expansion of the deposited thin films, which might be related to octahedral tilts and the variation in the valence state for the transition metal ions, similar to what has been observed in ultrathin La0.7Sr0.3MnO3 films (Sandiumenge et al., 2013[Sandiumenge, F., Santiso, J., Balcells, L., Konstantinovic, Z., Roqueta, J., Pomar, A., Espinós, J. P. & Martínez, B. (2013). Phys. Rev. Lett. 110, 107206.]).

[Figure 2]
Figure 2
High resolution X-ray diffraction curves along the (002) direction for 26 u.c. Sr1–yVO3 thin films deposited on (a) an STO substrate and (b) an LAO substrate. (c) and (d) The corresponding reciprocal-space maps along the off-specular 013 reflection for Sr1–yVO3, respectively.

2.2. X-ray absorption measurements

Soft X-ray absorption measurements were performed on the 08U1A soft X-ray beamline at the Shanghai Synchrotron Radiation Facility, China. Data were obtained by collecting the photocurrent in total electron yield mode. All the absorption spectra presented here were subtracted with a linear background prior to the absorption edge, followed by normalization to the unit value at an energy above the absorption edges (i.e. E ≃ 528.2 eV).

The absorption spectra for 26 u.c. Sr1–yVO3 films are similar for both tensile and compressive strains in terms of the same excitation energies and spectral line shapes, as shown in Fig. 3[link]. The spectra are composed of L3 and L2 sets of peaks owing to the spin–orbit coupling of the V 2p levels. A coexistence of V5+ and V4+ in thick Sr1–yVO3 films is suggested from: (i) the absorption energies, which are comparable with the reference spectrum of Sr2V2O7 powder; and (ii) the existence of small fine structures around 516 eV as a feature only for V5+ but with less enhanced intensities. To give a further estimate of the amount of V5+ and V4+ contributing to the measured profiles, we fitted the absorption spectra using two Lorenztians. Since the absorption peak at the L2-edge shows a broad peak feature for both V5+ and V4+, we only consider the fit at the V L3-edge, as shown in Figs. 3[link](c)–3[link](d). The fitting results of the spectral deconvolution are composed of two chemical states, where the low- and high-energy features arise from V4+ and V5+ chemical states, respectively. From the peak areas of the two Lorenztians, we estimate a mixture of 37% V5+ and 63% V4+ for the 26 u.c. Sr1–yVO3 film under tensile strain, whereas a mixture of 47% V5+ and 53% V4+ is obtained for the film under compressive strain. The amounts of V5+ and V4+ obtained from the fitting results suggest y = 0.185 considering charge conservation for the Sr1–yVO3 films under tensile strain, whereas y = 0.235 for the film under compressive strain.

[Figure 3]
Figure 3
(a) The absorption spectra for Sr1–yVO3 films of different thicknesses from 26 to 5 u.c. on an STO substrate with tensile strain. A reference absorption spectrum from Sr2V2O7 powder measured during the same beam time is shown for comparison. (b) The thickness-dependent absorption spectra for Sr1–yVO3 films of 26 u.c. on an LAO substrate, and 10 u.c. and 5 u.c. on LSAO substrates. The absorption spectra are shifted vertically by 2 for clarity. (c) and (d) The fitting results of the absorption spectra at the V L3-edge for 26 u.c. Sr1–yVO3 films on an STO substrate with tensile strain and on LAO with a compressive strain, respectively.

Surprisingly, different thickness-dependent behaviours are observed for Sr1–yVO3films in different strain states. For Sr1–yVO3 films under tensile strain, the absorption spectrum of 5 u.c. shifts to low binding energies, i.e. the absorption edges lie ∼0.7 eV lower in energy, as shown in Fig. 3[link](a). Note that the 10 u.c. Sr1–yVO3 thin film on an STO substrate already shows a small indication of a shift to lower energy. Lu et al. (2018[Lu, Q., Bishop, S. R., Lee, D., Lee, S., Bluhm, H., Tuller, H. L., Lee, H. N. & Yildiz, B. (2018). Adv. Funct. Mater. 28, 1803024.]) reported a systematic change in the V L-edge absorption spectra from V2O5 to VO2 as a function of electric bias. Our absorption spectra of Sr1–yVO3 films show a comparable energy shift to low energy, suggesting a variation from a mixture of V5+ and V4+ in the 26 u.c. Sr1–yVO3 film to mainly V4+ in the 5 u.c. film. However, instead of a variation in the V valence state, we observe a change in the spectral line shape to narrow band features for Sr1–yVO3 films deposited on an LSAO substrate. The change in absorption spectral line shape lies at the same thickness, i.e. 10 u.c., as the critical thickness of Sr1–yVO3 films undergoing the metal-to-insulator transition shown in next section. This is similar to the change in the spectral line shape resolved in the thickness-dependent metal–insulator transition in rare earth nickelate thin films (Wu et al., 2015[Wu, M., Benckiser, E., Audehm, P., Goering, E., Wochner, P., Christiani, G., Logvenov, G., Habermeier, H.-U. & Keimer, B. (2015). Phys. Rev. B, 91, 195130.]).

2.3. DC transport measurements

Fig. 4[link] shows the temperature-dependent resistivity curves for Sr1–yVO3 thin films of different thicknesses and strains measured using the standard four-probe method. The Sr1–yVO3 film deposited on an STO substrate with tensile strain shows a metal–insulator transition at 5 u.c., whereas the compressive-strained Sr1–yVO3 film on an LSAO substrate shows an insulating behaviour at 10 u.c. For the metallic phase of Sr1–yVO3 with both tensile and compressive [Fig. 4[link](a)] strains, the resistivity curves deviate from a linear and temperature-dependent form related to the electron–phonon interaction (i.e. ρT). The temperature-dependent resistivity curves for metallic Sr1–yVO3 films are similar to the results reported for a Ca1–xSrxVO3 solid solution (Inoue et al., 1998[Inoue, I. H., Goto, O., Makino, H., Hussey, N. E. & Ishikawa, M. (1998). Phys. Rev. B, 58, 4372-4383.]), SVO thin films (Gu et al., 2014[Gu, M., Wolf, S. A. & Lu, J. (2014). Adv. Mater. Interfaces, 1, 1300126.]) and a system of SVO layers sandwiched by insulating LaVO3 layers (Li et al., 2015[Li, Q.-R., Major, M., Yazdi, M. B., Donner, W., Dao, V. H., Mercey, B. & Lüders, U. (2015). Phys. Rev. B, 91, 035420.]), which were understood through a T2 dependence attributed to a strong electron–electron interaction up to room temperature. Similarly, the resistivity curves of our investigated Sr1–yVO3 films can be fitted well with ρT2, as shown in Fig. 4[link](a).

[Figure 4]
Figure 4
(a)–(b) Temperature-dependent resistivity curves for Sr1–yVO3 thin films of 26 u.c. on STO, 26 u.c. on LAO and 10 u.c. on STO substrates. The black lines in panel (a) show the fit with ρT2 and those in panel (b) exhibit the fit with the small polaronic conduction model. (c) The ρT relationship for compressively strained Sr1–yVO3 films on LSAO. (d) The metal–insulator transition of 5 u.c. Sr1–yVO3 on an STO substrate.

Moreover, we note that the striking T2 temperature-dependent resistivity behaviours (up to room temperature) are analogous to the results observed in the neighbouring doped titanates, where the transport behaviours were originally discussed in terms of strong electron–electron correlations, and then by the small polaronic conduction model (e.g. in La2/3TiO3 and La1–yTiO3+δ systems; Gariglio et al., 2001[Gariglio, S., Seo, J. W., Fompeyrine, J., Locquet, J.-P. & Triscone, J.-M. (2001). Phys. Rev. B, 63, 161103.]; Jung et al., 2000[Jung, W. H., Wakai, H., Nakatsugawa, H. & Iguchi, E. (2000). J. Appl. Phys. 88, 2560-2563.]). Given the similar coexistence of 3d0 and 3d1 electronic configurations in cation-deficient Sr1–yVO3 films as in La2/3TiO3 and La1–yTiO3+δ systems, we tried to fit the resistivity curves using the small polaronic conduction model. Interestingly, the temperature-dependent resistivity relationship can be equally well fitted by a small polaronic transport, as shown in Fig. 4[link](b).

The resistivity curves can be described by the following equation, considering only one low-lying optical mode ω0 in the small polaronic conduction model (Gariglio et al., 2001[Gariglio, S., Seo, J. W., Fompeyrine, J., Locquet, J.-P. & Triscone, J.-M. (2001). Phys. Rev. B, 63, 161103.]):

[\rho(T) = \rho _{0} + {{C} \over {\sinh^{2} (\hbar\omega _{0}/2k_{\rm B}T)}} , \eqno(1)]

with an identical value of [\hbar\omega_0/k_{\rm B}] = 80 K, where kB is the Boltzmann constant. The other two parameters are: C = 0.00115 mΩ cm and ρ0 = 0.072 mΩ cm for 26 u.c. Sr1–yVO3 on STO, C = 0.00115 mΩ cm and ρ0 = 0.209 mΩ cm for 10 u.c. on STO, and C = 0.00098 mΩ cm and ρ0 = 0.108 mΩ cm for 26 u.c. Sr1–yVO3 on an LAO substrate.

We note that the phonon frequency obtained from this fit has the same value as [\hbar\omega_0/k_{\rm B}] = 80 K resolved in doped La1–yTiO3+δ (Gariglio et al., 2001[Gariglio, S., Seo, J. W., Fompeyrine, J., Locquet, J.-P. & Triscone, J.-M. (2001). Phys. Rev. B, 63, 161103.]) and doped La1–xCaxMnO3 systems (Zhao et al., 2000[Zhao, G., Smolyaninova, V., Prellier, W. & Keller, H. (2000). Phys. Rev. Lett. 84, 6086-6089.]), which suggests the existence of a soft mode related to the tilting/rotation of oxygen octahedra which is strongly coupled to the carriers (Zhao et al., 2000[Zhao, G., Smolyaninova, V., Prellier, W. & Keller, H. (2000). Phys. Rev. Lett. 84, 6086-6089.]). Further studies from either photoemission (Fujimori et al., 1996[Fujimori, A., Bocquet, A. E., Morikawa, K., Kobayashi, K., Saitoh, T., Tokura, Y., Hase, I. & Onoda, M. (1996). J. Phys. Chem. Solids, 57, 1379-1384.]) or optical spectroscopic measurements (Nucara et al., 2008[Nucara, A., Maselli, P., Del Bufalo, M., Guidi, M. C., Garcia, J., Orgiani, P., Maritato, L. & Calvani, P. (2008). Phys. Rev. B, 77, 064431.]) would be useful to identify the optical phonon mode. However, the small polaron band which is expected to appear near the Fermi level in the photoemission measurements might be smeared out due to the broad valence band of V 3d states (Yoshimatsu et al., 2011[Yoshimatsu, K., Horiba, K., Kumigashira, H., Yoshida, T., Fujimori, A. & Oshima, M. (2011). Science, 333, 319-322.]). Our thin films with thicknesses less than 26 u.c. and with 4 u.c. STO capping layers are challenging for optical measurements with typical in-house setups. It is therefore difficult to justify the intrinsic conducting mechanism here and it needs further investigation with different experimental designs.

Compared with the metallic behaviour of 10 u.c. Sr1–yVO3 films under tensile strain, the Sr1–yVO3 film of the same thickness on an LSAO substrate shows a strong insulating behaviour where the resistivity is beyond the measurement limit below 110 K, as shown in Fig. 4[link](c). This insulating behaviour cannot be fitted with either a thermally activated conduction model ([\ln \rho \propto 1/T]) or the activated hopping model of small polarons [[\ln(\rho/T) \propto 1/T]], which can instead be fitted reasonably well by a two-dimensional variable-range hopping (VRH) model [[\ln \rho \propto T^{-1/3}], inset of Fig. 4[link](c)] and a three-dimensional VRH model ([\ln \rho \propto T^{-1/4}]) down to 160 K. The reasonable fitting of the resistivity curves using the VRH model suggests that the conduction model is governed by the localization of charge carriers and the hopping of carriers between different localized states. The localized bands are also visualized in the aforementioned X-ray absorption measurements. The resistivity measurement for 5 u.c. Sr1–yVO3 on an LSAO substrate is above the measurement limit at room temperature. The Sr1–yVO3 thin film with a thickness of 5 u.c exhibits a metal–insulator transition where the high-temperature regime shows a linear temperature dependence typical for metallic behaviour, i.e. ρT shown as the black fit line in Fig. 4[link](d). The resistivity value of the 5 u.c. Sr1–yVO3 film on an STO substrate is lower than that of 10 u.c. Sr1–yVO3 on an STO substrate, which is opposite to the typical tendency expected for reducing SVO film thicknesses (Gu et al., 2014[Gu, M., Wolf, S. A. & Lu, J. (2014). Adv. Mater. Interfaces, 1, 1300126.]; Fouchet et al., 2016[Fouchet, A., Allain, M., Bérini, B., Popova, E., Janolin, P.-E., Guiblin, N., Chikoidze, E., Scola, J., Hrabovsky, D., Dumont, Y. & Keller, N. (2016). Mater. Sci. Eng. B, 212, 7-13.]). This might be related to the variation in stoichiometry, as suggested by the V L2/L3-edge absorption spectroscopy analysis with a change of V valence state in ultra-thin 5 u.c. SVO films. The insulating regime is difficult to fit with the thermal or small polaronic activated conduction mechanism, or the VRH models.

2.4. EDX measurements

To investigate further for the presence of cation deficiencies, we provide local energy-dispersive X-ray (EDX) measurements for Sr1–yVO3 thin films deposited on different substrates, which probe the composition of individual elements in the thin films.

Fig. 5[link] displays the EDX spectra of Sr1–yVO3 thin films on (a) STO, (b) LAO and (c) LSAO substrates, while the corresponding atomic percentages of each individual element are listed in Table 2[link]. Due to the existence of Sr in both the investigated thin films and the substrates such as STO and LSAO, the relative abundance of Sr content in the Sr1–yVO3 thin films can be roughly obtained by subtracting the atomic percentage of the substrate with an estimate of 1:1 for the substrates, i.e. Sr:Ti = 1:1 for the STO substrate and Sr:Al = 1:1 for the LSAO susbtrate. The atomic percentage of Sr in Sr1–yVO3 is denoted Srsubtr after subtracting the contribution of the substrate. The ratios between Srsubtr and V for each sample are shown in the last column of Table 2[link]. The EDX measurements show the relative atomic deficiency of strontium for all the Sr1–yVO3 thin films under both compressive and tensile strains, which is consistent with the presence of Sr deficiencies and the existence of V5+ in the investigated thin films as discussed above.

Table 2
The relative atomic percentages of individual elements for Sr1–yVO3 thin films deposited on different substrates

For each sample, the EDX measurements were performed twice at different regions with a scanned area of ∼50 µm. The corresponding EDX spectra for the first measurement are shown in Fig. 5[link].

Substrate O Ti V Sr Al Srsubtr:V
STO1st 57.90 20.65 0.47 20.98   0.33:0.47
STO2nd 58.22 20.65 0.74 20.78   0.53:0.74
LAO1st 95.51   2.63 1.96   1.96:2.63
LAO2nd 95.38   3.02 1.60   1.60:3.02
LSAO1st 61.24   0.84 19.30 18.62 0.68:0.84
LSAO2nd 61.50   0.46 19.12 18.92 0.2:0.46
[Figure 5]
Figure 5
EDX spectra of Sr1–yVO3 thin films on (a) STO, (b) LAO and (c) LSAO substrates.

3. Discussion and conclusions

In contrast with the existence of oxygen vacancies expressed in the form ABO3–δ in perovskite oxides, the perovskite structure is close-packed in terms of its large oxygen anion backbone structure which is unlikely to leave additional room for oxygen interstitials. Charge-sensitive X-ray absorption measurements at the V L-edge show the coexistence of V5+ and V4+ in our investigated Sr1–yVO3 films. Creating cation deficiencies is required to compensate the V5+ valence state from the stoichiometric point of view.

In cation-deficient Sr1–yVO3 thin films capped with 4 u.c. STO layers, we observe different responses of the stoichiometric environment to epitaxial strains in ultrathin films, i.e. a variation in the V chemical valence state is only resolved for Sr1–yVO3 ultrathin films under tensile strain. Similar to the usage of the formation energy of oxygen vacancies (ΔEvac) in understanding the relationship between strain and oxygen vacancies in many perovskite compounds, we use the formation energy of cation deficiencies to describe our results. Our results indicate a larger cation-deficient formation energy for Sr1–yVO3 ultrathin films under tensile strain. This is opposite to the decrease in ΔEvac as the magnitude of the tensile strain increases. We further note that the interaction between elastic strain and ΔEvac is nontrivial (Herklotz et al., 2017[Herklotz, A., Lee, D., Guo, E.-J., Meyer, T. L., Petrie, J. R. & Lee, H. N. (2017). J. Phys. Condens. Matter, 29, 493001.]), i.e. both a linear reduction in ΔEvac as the strain increases (Marrocchelli et al., 2015[Marrocchelli, D., Sun, L. & Yildiz, B. (2015). J. Am. Chem. Soc. 137, 4735-4748.]) and a nearly quadratic dependence (Choi et al., 2015[Choi, S.-Y., Kim, S.-D., Choi, M., Lee, H.-S., Ryu, J., Shibata, N., Mizoguchi, T., Tochigi, E., Yamamoto, T., Kang, S.-J. L. & Ikuhara, Y. (2015). Nano Lett. 15, 4129-4134.]; Iglesias et al., 2017[Iglesias, L., Sarantopoulos, A., Magén, C. & Rivadulla, F. (2017). Phys. Rev. B, 95, 165138.]) have been proposed. Therefore, the observed asymmetric response of the cation-deficiency formation energy to elastic strain deserves further detailed theoretical calculations.

The resistivity curves of metallic Sr1–yVO3 films clearly deviate from a linear temperature dependence, which can be understood by a T2 dependence attributed to strong electron–electron correlations as reported previously. However, the resistivity curves can be fitted equally well by a small polaronic transport mechanism. The small polaronic conduction model as the dominant scattering mechanism has been proposed in doped titanates (Gariglio et al., 2001[Gariglio, S., Seo, J. W., Fompeyrine, J., Locquet, J.-P. & Triscone, J.-M. (2001). Phys. Rev. B, 63, 161103.]; Jung et al., 2000[Jung, W. H., Wakai, H., Nakatsugawa, H. & Iguchi, E. (2000). J. Appl. Phys. 88, 2560-2563.]) involving the coexistence of 3d0 and 3d1 electronic configurations, which is exactly the same situation as the Sr1–yVO3 films investigated here. Our studies draw attention to the electron–phonon interaction as a possible dominant mechanism for transport behaviour.

Furthermore, the Sr1–yVO3 films undergo metal–insulator transitions at different critical thicknesses under different strains. The strain-dependent response of electrical resistivity varies in different systems as reported in the literature. Taking nickel oxide thin films as an example, although asymmetric responses of electrical resistivity to compressive and tensile strains were reported in 12 nm PrNiO3 films (Hepting et al., 2014[Hepting, M., Minola, M., Frano, A., Cristiani, G., Logvenov, G., Schierle, E., Wu, M., Bluschke, M., Weschke, E., Habermeier, H.-U., Benckiser, E., Le Tacon, M. & Keimer, B. (2014). Phys. Rev. Lett. 113, 227206.]) and in 6 nm NdNiO3 films (Mundet et al., 2018[Mundet, B., Jareño, J., Gazquez, J., Varela, M., Obradors, X. & Puig, T. (2018). Phys. Rev. Mater. 2, 063607.]), symmetric responses to compressive and tensile strains were also observed in several thin-film systems. For instance, the low-temperature upturn in the resistivity curves for both tensile and compressive strained cases can be described by the same model for 6 nm LaNiO3 films (Moon et al., 2011[Moon, E. J., Gray, B. A., Kareev, M., Liu, J., Altendorf, S. G., Strigari, F., Tjeng, L. H., Freeland, J. W. & Chakhalian, J. (2011). New J. Phys. 13, 073037.]). The metal–insulator transition temperatures change with different magnitudes for tensile and compressive strains, but the low-temperature region of the resistivity curves could in all cases be fitted with the VRH model in 60 nm PrNiO3 films (Dan et al., 2016[Yao, D., Shi, L., Zhou, S., Liu, H., Wang, Y., Zhao, J. & Li, Y. (2016). J. Phys. D Appl. Phys. 49, 125301.]) and in 12 nm NdNiO3 films (Xiang et al., 2013[Xiang, P.-H., Zhong, N., Duan, C.-G., Tang, X. D., Hu, Z. G., Yang, P. X., Zhu, Z. Q. & Chu, J. H. (2013). J. Appl. Phys. 114, 243713.]). The asymmetric response of electrical conductivity to strain in Sr1–yVO3 ultrathin films, in particular for the 5 u.c. Sr1–yVO3 film under tensile strain, might be related to the variation in V valence states as resolved from the V L2/L3-edge absorption measurements.

The strain-induced tuning of the Sr/V ratio offers guidance for thin-film synthesis in other multivalent compounds. The asymmetric response of the electrical conductivity and cation-deficiency formation energy to epitaxial tensile and compressive strain are beneficial to understand the defect chemistry of functional oxide electronic devices, to investigate the novel physical properties arising from interfacial charge transfer in thin films and heterostructures, and to engineer potential electronic applications.

Acknowledgements

We acknowledge the provision of synchrotron radiation beamtime and the support given by L. J. Zhang at the Shanghai Synchrotron Radiation Facility.

Funding information

Funding for this research was provided by: National Natural Science Foundation of China (grant No. 11704317); China Postdoctoral Science Foundation (grant No. 2016M602064); Fundamental Research Funds for Central Universities (grant No. 20720160020).

References

First citationCao, Y., Liu, X., Kareev, M., Choudhury, D., Middey, S., Meyers, D., Kim, J.-W., Ryan, P. J., Freeland, J. & Chakhalian, J. (2016). Nat. Commun. 7, 10418.  Web of Science CrossRef PubMed Google Scholar
First citationChoi, S.-Y., Kim, S.-D., Choi, M., Lee, H.-S., Ryu, J., Shibata, N., Mizoguchi, T., Tochigi, E., Yamamoto, T., Kang, S.-J. L. & Ikuhara, Y. (2015). Nano Lett. 15, 4129–4134.  Web of Science CrossRef CAS PubMed Google Scholar
First citationFarokhipoor, S. & Noheda, B. (2011). Phys. Rev. Lett. 107, 127601.  Web of Science CrossRef PubMed Google Scholar
First citationFouchet, A., Allain, M., Bérini, B., Popova, E., Janolin, P.-E., Guiblin, N., Chikoidze, E., Scola, J., Hrabovsky, D., Dumont, Y. & Keller, N. (2016). Mater. Sci. Eng. B, 212, 7–13.  Web of Science CrossRef CAS Google Scholar
First citationFujimori, A., Bocquet, A. E., Morikawa, K., Kobayashi, K., Saitoh, T., Tokura, Y., Hase, I. & Onoda, M. (1996). J. Phys. Chem. Solids, 57, 1379–1384.  CrossRef CAS Web of Science Google Scholar
First citationGariglio, S., Seo, J. W., Fompeyrine, J., Locquet, J.-P. & Triscone, J.-M. (2001). Phys. Rev. B, 63, 161103.  Web of Science CrossRef Google Scholar
First citationGu, M., Wolf, S. A. & Lu, J. (2014). Adv. Mater. Interfaces, 1, 1300126.  Web of Science CrossRef Google Scholar
First citationHepting, M., Minola, M., Frano, A., Cristiani, G., Logvenov, G., Schierle, E., Wu, M., Bluschke, M., Weschke, E., Habermeier, H.-U., Benckiser, E., Le Tacon, M. & Keimer, B. (2014). Phys. Rev. Lett. 113, 227206.  Web of Science CrossRef PubMed Google Scholar
First citationHerklotz, A., Lee, D., Guo, E.-J., Meyer, T. L., Petrie, J. R. & Lee, H. N. (2017). J. Phys. Condens. Matter, 29, 493001.  Web of Science CrossRef PubMed Google Scholar
First citationIglesias, L., Sarantopoulos, A., Magén, C. & Rivadulla, F. (2017). Phys. Rev. B, 95, 165138.  Web of Science CrossRef Google Scholar
First citationInoue, I. H., Goto, O., Makino, H., Hussey, N. E. & Ishikawa, M. (1998). Phys. Rev. B, 58, 4372–4383.  Web of Science CrossRef CAS Google Scholar
First citationJung, W. H., Wakai, H., Nakatsugawa, H. & Iguchi, E. (2000). J. Appl. Phys. 88, 2560–2563.  Web of Science CrossRef CAS Google Scholar
First citationLi, Q.-R., Major, M., Yazdi, M. B., Donner, W., Dao, V. H., Mercey, B. & Lüders, U. (2015). Phys. Rev. B, 91, 035420.  Web of Science CrossRef Google Scholar
First citationLu, Q., Bishop, S. R., Lee, D., Lee, S., Bluhm, H., Tuller, H. L., Lee, H. N. & Yildiz, B. (2018). Adv. Funct. Mater. 28, 1803024.  Web of Science CrossRef Google Scholar
First citationMa, J., Zhang, Y., Wu, L., Song, C., Zhang, Q., Zhang, J., Ma, J. & Nan, C.-W. (2016). MRS Commun. 6, 354–359.  Web of Science CrossRef CAS Google Scholar
First citationMarrocchelli, D., Sun, L. & Yildiz, B. (2015). J. Am. Chem. Soc. 137, 4735–4748.  Web of Science CrossRef CAS PubMed Google Scholar
First citationMirjolet, M., Sánchez, F. & Fontcuberta, J. (2019). Adv. Funct. Mater. 29, 1808432.  Web of Science CrossRef Google Scholar
First citationMoon, E. J., Gray, B. A., Kareev, M., Liu, J., Altendorf, S. G., Strigari, F., Tjeng, L. H., Freeland, J. W. & Chakhalian, J. (2011). New J. Phys. 13, 073037.  Web of Science CrossRef Google Scholar
First citationMoyer, J. A., Eaton, C. & Engel-Herbert, R. (2013). Adv. Mater. 25, 3578–3582.  Web of Science CrossRef CAS PubMed Google Scholar
First citationMundet, B., Jareño, J., Gazquez, J., Varela, M., Obradors, X. & Puig, T. (2018). Phys. Rev. Mater. 2, 063607.  Web of Science CrossRef Google Scholar
First citationNucara, A., Maselli, P., Del Bufalo, M., Guidi, M. C., Garcia, J., Orgiani, P., Maritato, L. & Calvani, P. (2008). Phys. Rev. B, 77, 064431.  Web of Science CrossRef Google Scholar
First citationOhtomo, A., Muller, D. A., Grazul, J. L. & Hwang, H. Y. (2002). Appl. Phys. Lett. 80, 3922–3924.  Web of Science CrossRef CAS Google Scholar
First citationPan, S. H., O'Neal, J. P., Badzey, R. L., Chamon, C., Ding, H., Engelbrecht, J. R., Wang, Z., Eisaki, H., Uchida, S., Gupta, A. K., Ng, K.-W., Hudson, E. W., Lang, K. M. & Davis, J. C. (2001). Nature, 413, 282–285.  Web of Science CrossRef PubMed CAS Google Scholar
First citationSandiumenge, F., Santiso, J., Balcells, L., Konstantinovic, Z., Roqueta, J., Pomar, A., Espinós, J. P. & Martínez, B. (2013). Phys. Rev. Lett. 110, 107206.  Web of Science CrossRef PubMed Google Scholar
First citationSawa, H., Ninomiya, E., Ohama, T., Nakao, H., Ohwada, K. J., Murakami, Y., Fujii, Y., Noda, Y., Isobe, M. & Ueda, Y. (2002). J. Phys. Soc. Jpn, 71, 385–388.  Web of Science CrossRef ICSD CAS Google Scholar
First citationSkjaervø, S. H., Wefring, E. T., Nesdal, S. K., Gaukås, N. H., Olsen, G. H., Glaum, J., Tybell, T. & Selbach, S. M. (2016). Nat. Commun. 7, 13745.  Web of Science PubMed Google Scholar
First citationStemmer, S., Jacobson, A. J., Chen, X. & Ignatiev, A. (2001). J. Appl. Phys. 90, 3319–3324.  Web of Science CrossRef CAS Google Scholar
First citationTarun, M. C., Selim, F. A. & McCluskey, M. D. (2013). Phys. Rev. Lett. 111, 187403.  Web of Science CrossRef PubMed Google Scholar
First citationWang, J., Rijnders, G. & Koster, G. (2018). Appl. Phys. Lett. 113, 223103.  Web of Science CrossRef Google Scholar
First citationWu, M., Benckiser, E., Audehm, P., Goering, E., Wochner, P., Christiani, G., Logvenov, G., Habermeier, H.-U. & Keimer, B. (2015). Phys. Rev. B, 91, 195130.  Web of Science CrossRef Google Scholar
First citationWu, M., Zheng, J.-C. & Wang, H.-Q. (2018). Phys. Rev. B, 97, 245138.  Web of Science CrossRef Google Scholar
First citationXiang, P.-H., Zhong, N., Duan, C.-G., Tang, X. D., Hu, Z. G., Yang, P. X., Zhu, Z. Q. & Chu, J. H. (2013). J. Appl. Phys. 114, 243713.  Web of Science CrossRef Google Scholar
First citationYao, D., Shi, L., Zhou, S., Liu, H., Wang, Y., Zhao, J. & Li, Y. (2016). J. Phys. D Appl. Phys. 49, 125301.  Web of Science CrossRef Google Scholar
First citationYoshimatsu, K., Horiba, K., Kumigashira, H., Yoshida, T., Fujimori, A. & Oshima, M. (2011). Science, 333, 319–322.  Web of Science CrossRef CAS PubMed Google Scholar
First citationYoshimatsu, K., Sakai, E., Kobayashi, M., Horiba, K., Yoshida, T., Fujimori, A., Oshima, M. & Kumigashira, H. (2013). Phys. Rev. B, 88, 115308.  Web of Science CrossRef Google Scholar
First citationYoshino, T., Okawa, M., Kajita, T., Dash, S., Shimoyama, R., Takahashi, K., Takahashi, Y., Takayanagi, R., Saitoh, T., Ootsuki, D., Yoshida, T., Ikenaga, E., Saini, N. L., Katsufuji, T. & Mizokawa, T. (2017). Phys. Rev. B, 95, 075151.  Web of Science CrossRef Google Scholar
First citationZhao, G., Smolyaninova, V., Prellier, W. & Keller, H. (2000). Phys. Rev. Lett. 84, 6086–6089.  Web of Science CrossRef PubMed CAS Google Scholar

© International Union of Crystallography. Prior permission is not required to reproduce short quotations, tables and figures from this article, provided the original authors and source are cited. For more information, click here.

Journal logoJOURNAL OF
SYNCHROTRON
RADIATION
ISSN: 1600-5775
Follow J. Synchrotron Rad.
Sign up for e-alerts
Follow J. Synchrotron Rad. on Twitter
Follow us on facebook
Sign up for RSS feeds