Epitaxial Strain-Controlled Ionic Conductivity in Li-Ion Solid Electrolyte

Mar 19, 2015 - Recent Developments in Oxide-Based Ionic Conductors: Bulk Materials, Nanoionics, and Their Memory Applications. Tao Wan , Lepeng Zhang ...
2 downloads 0 Views 3MB Size
Article pubs.acs.org/crystal

Epitaxial Strain-Controlled Ionic Conductivity in Li-Ion Solid Electrolyte Li0.33La0.56TiO3 Thin Films Jie Wei,† Daisuke Ogawa,† Tomoteru Fukumura,*,†,‡ Yasushi Hirose,†,‡ and Tetsuya Hasegawa†,‡ †

Department of Chemistry, School of Science, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan CREST, Japan Science and Technology Agency (JST), 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan



Downloaded via UNIV OF NEW ENGLAND on November 25, 2018 at 03:30:33 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

S Supporting Information *

ABSTRACT: Ionic conductive Li0.33La0.56TiO3 (LLT) epitaxial thin films were grown on perovskite SrTiO3 (100), NdGaO3 (110), and (LaAlO3)0.3-(SrAl0.5Ta0.5O3)0.7 (100) single crystal substrates by pulsed laser deposition. The use of Li-rich Li0.84La0.56TiO3+δ target together with an optimized laser fluence resulted in the growth of phase pure LLT thin films with high growth rate of 2 nm/min. The a-axis and c-axis oriented films were selectively grown by choosing the substrates. Ionic conductivity at room temperature of LLT epitaxial film on NdGaO3 (110) substrate was close to that of bulk previously reported, representing the highly crystalline quality. In addition, the unequally strained lattice due to different inplane lattice constants of orthorhombic NdGaO3 substrate resulted in laterally anisotropic ionic conductivity with different activation energy perpendicular to NdGaO3 [11̅0] and [001], 6.7 × 10−4 S·cm−1 with 0.34 eV and 4.3 × 10−4 S·cm−1 with 0.36 eV, respectively. This result suggests that the lattice engineering can provide a way to control Li ionic conduction. metal element.16,17 During the deposition process in the case of PLD, for example, Li content in the film is significantly reduced in comparison with that in the target particularly for high laser fluence. This is possibly due to stronger resputtering18 and/or expanded plume19 effects on Li in comparison with that on much heavier La and Ti (Supporting Information (SI) Figure S1a). In order to avoid the significant Li-deficiency accompanied by secondary phase, quite low laser fluence was used, resulting in very low growth rate.11 In this study, we report on the epitaxial thin film growth of Li0.33La0.56TiO3 on perovskite oxide substrates with various lattice mismatches by PLD. Use of Li-rich target significantly enhanced the growth rate. The varied lattice mismatch with optimized deposition condition led to the orientation control of phase pure LLT epitaxial thin films with varied strain. Ionic conductivity measurement was successfully performed for LLT on NdGaO3 (110) substrate, showing the highest value among epitaxial thin film studies. The different degrees of strain in LLT along the laterally orthogonal direction on the orthorhombic NdGaO3 substrate led to laterally anisotropic ionic conduction due to the different cell size for ionic conduction path.

1. INTRODUCTION Lithium ion solid electrolyte has attracted great interest for its potential applications in all solid state Li ion batteries. Compared to conventional organic solvent electrolyte, Li ion solid electrolyte is more stable and inflammable, contributing to safety concern and compatibility with various environments. Perovskite lithium lanthanum titanate Li3xLa2/3−xTiO3 (LLT, 0.05 < x < 0.167) has attracted much attention because of its high ionic conductivity owing to the A-site vacancies (Figure 1a).1 The highest ionic conductivity in Li0.33La0.56TiO3 at room temperature of ∼10−3 S·cm−1 is comparable to that in conventional organic solvent electrolyte.1 Various studies of LLT on its microstructure,2 A-site ordering,3,4 and improvement of ionic conductivity by F-substitution5,6 have been reported by using bulk specimens. Also, various thin film growth methods have been applied for amorphous or polycrystalline thin films, such as sol−gel method,7 pulsed laser deposition,8,9 and atomic layer deposition.10 The demand for fabrication of highly crystalline thin film and heterostructure has recently stimulated epitaxial thin film growth of LLT by pulsed laser deposition (PLD)11,12 and RF sputtering.13 Indeed, a cathode-electrolyte epitaxial heterostructure has already been demonstrated with LLT epitaxial thin film.14 In addition, the application of tensile and compressive strains will control the cell size (i.e., the bottleneck size of ionic conduction path) (Figure 1b,c), possibly leading to the artificial control of ionic conductivity in ambient condition, in contrast with previous bulk study under hydrostatic pressure.15 For the thin film growth, however, control of Li content in the film is usually difficult because Li is the lightest © 2015 American Chemical Society

2. EXPERIMENTAL SECTION Li-rich LLT targets with nominal compositions of Li0.84La0.56TiO3+δ were fabricated by solid state reaction. Prescribed amount of Li2CO3, Received: December 18, 2014 Revised: March 3, 2015 Published: March 19, 2015 2187

DOI: 10.1021/cg501834s Cryst. Growth Des. 2015, 15, 2187−2191

Article

Crystal Growth & Design

3. RESULT AND DISCUSSION Figure 2 shows out-of-plane 2θ−θ patterns of LLT epitaxial thin films on LSAT (100) substrates as a function of the laser

Figure 2. Out-of-plane 2θ−θ XRD patterns for LLT thin films on LSAT (100) substrates, grown at substrate temperature of 950 °C in oxygen pressure of 50 mTorr with different laser fluences.

Figure 1. (a) Schematic crystal structure for Li3xLa2/3−xTiO3 (LLT). Schematic structure of epitaxial LLT thin films with (b) tensile strain and (c) compressive strain. Red dashed square denotes a bottleneck formed by four oxygen atoms. Blue arrow indicates the conduction path of Li ion.

fluence at fixed substrate temperature and oxygen pressure. With the laser fluence of 1.2 and 1.3 J/cm2, the LLT 004 peak was clearly seen together with a Li-rich spinel Li4Ti5O12 impurity phase. With the laser fluence of 1.5 J/cm2, on the other hand, the LLT 003 and 004 peaks became more intense, but a pyrochlore La2Ti2O7 impurity phase appeared. Only with the laser fluence of 1.4 J/cm2 was a phase pure Li0.33La0.56TiO3 thin film obtained with the most intense XRD peaks. The observed Li4Ti5O12 and La2Ti2O7 secondary phases were probably caused by the excess Li and deficient Li in the films, respectively. The strong dependence of Li content on laser fluence was confirmed in X-ray photoemission spectroscopy of amorphous films (SI Figure S1a). These results represent the importance of the Li supply amount for the growth of phase pure LLT film. The decreased laser fluence prevents the reduction of Li supply at the expense of growth rate (SI Figure S1b). This is why the Li-rich LLT target was used to realize high growth rate in this study. Figure 3a shows out-of-plane 2θ−θ pattern of 120-nm-thick LLT thin films on STO, LSAT, and NGO substrates after optimizing laser fluence of 1.4 J/cm2, substrate temperature Ts = 950 °C, and oxygen pressure pO2 = 30 mTorr. It is noted that the growth rate of 3 × 10−3 nm/pulse is faster than those in previous studies.11,12 The film orientation on STO (100) substrates was along the a-axis with two domains, where the domain structure was confirmed by φ-scan (SI Figure S2). The film orientations on LSAT (100) and NGO (110) substrates were along the c-axis with single domain. FWHM of LLT 004 or 200 peak in rocking curve on STO, LSAT, and NGO was 0.06°, 0.03°, and 0.05°, respectively (Figure 3b), representing the highly crystalline quality of these epitaxial thin films. The film surface morphology on STO was a step and terrace structure with approximately one unit cell height observed by atomic force microscope (Figure 3c), suggesting the layer-bylayer growth of the film. From the reciprocal space mapping (Figure 4), the epitaxial relation of LLT thin film on each substrate was determined as LLT (100) // STO (100) with LLT [001] // STO [001] and

TiO2, and as-dried La2O3 powders (Wako Pure Chemical) were mixed and calcined at 850 °C for 12 h in air. Subsequently, the powders were pressed into pellets under 40 MPa and sintered at 1200 °C for 12 h in air. LLT thin films were fabricated by PLD (Pascal Co. Ltd.) with KrF excimer laser (λ = 248 nm, Coherent COMPexPro 50). Deposition time and ablation frequency were 1 h and 10 Hz, respectively. Laser fluence was varied from 0.5 to 2 J/cm2 by using a half-wave attenuator module. Laser spot was 1.0 mm2 at the target surface. Distance between target and substrate was 50 mm. In order to control the lattice strain of LLT with the perovskite parameter ap = 0.3875 nm (2ap3 is the volume of primitive cell of tetragonal perovskite Li0.33La0.56TiO3, where c/2 ≥ a ∼ ap), cubic SrTiO3 (100) (STO: a = 0.3905 nm), cubic (LaAlO3)0.3-(SrAl0.5Ta0.5O3)0.7 (100) (LSAT: a = 0.3868 nm), and orthogonal NdGaO3 (110) (NGO: a = 0.5431 nm, b = 0.5499 nm, c = 0.771 nm) single crystals were chosen as substrates. STO (100) substrates were treated in BHF solutions as purchased and annealed at 1050 °C for 2 h in order to obtain a step and terrace surface morphology, while LSAT (100) and NGO (110) substrates were annealed at 1000 °C for 2 h for a flatter surface. Substrate temperature at 700−1000 °C and oxygen pressure of 1−100 mTorr were investigated to optimize the crystallinity. Thickness of LLT thin films was measured by stylus profilometer (Dektak XT, Bruker). The lattice parameters and crystallinity were evaluated by X-ray diffraction (XRD) with Cu Kα radiation (D8 Discover with Gadds, Bruker). Surface morphology was observed by atomic force microscopy (SPI4000, Hitachi High Technologies). Impedance spectroscopy of the LLT films was performed for Li ion conductivity measurement by using impedance analyzer (Solartron 1260) in a prober system in air, where Au/Ti comb electrodes with length/width = 5900 μm/20 μm were formed on the LLT films with photolithography and e-beam evaporation. In order to remove possible organic contaminations from the photolithography process, oxygen plasma ashing and preheating at 150 °C for 10 min was performed before impedance measurement. Thin film was dissolved in a mixture of 60% HNO3 (2 mL) and 35% HCl (3 mL) at 100 °C for 90 min as spectrometry solution sample for composition measurements with the inductively coupled plasma− optical emission spectrometry for La and Ti contents and flameless graphite furnace atomic absorption spectroscopy for Li content. 2188

DOI: 10.1021/cg501834s Cryst. Growth Des. 2015, 15, 2187−2191

Article

Crystal Growth & Design

Figure 5. Schematic orientation for LLT thin films (green) on different substrates (gray) with (a) tensile strain and (b) compressive strain. (c) Out-of-plane lattice constant (left axis, red) and cell volume calculated as tetragonal perovskite structure (right axis, blue) as a function of in-plane lattice constant of epitaxial LLT thin film on each substrate.

Figure 3. (a) Out-of-plane 2θ−θ patterns of LLT epitaxial thin films on SrTiO3 (100) (green), LSAT (100) (blue), and NdGaO3 (110) (red) substrate with optimized condition. Asterisks denote peaks from the substrates. (b) Rocking curves of LLT 200 on STO (100) (green), LLT 004 on LSAT (100) (blue), and LLT 004 on NGO (110) (red). (c) Atomic force microscope image of LLT thin film surface on STO (100).

smaller strain along the in-plane direction, resulting in a 00l orientation. This tendency is consistent with previous reports on epitaxial LLT thin films.11 The variable range of the out-ofplane lattice constants is much wider than that of bulk LLT from 0.3872 to 0.3877 nm by varying x in Li3xLa2/3−xTiO3 (0.05 < x < 0.167) as seen in Figure 5c.20 This result represents the lattice flexibility against the epitaxial strain, which is an alternative method to control the bottleneck size for applying chemical pressure by element substitution. It is noted that the distortion of LLT on NGO was laterally anisotropic due to the orthorhombic NGO (110) substrate; thus, the LLT on NGO was distorted to not be tetragonal like LLT on STO and LSAT, but to be orthorhombic. The composition of LLT film on NGO (110) was approximately stoichiometric as Li0.32La0.54TiO3. This result evidenced a significant loss of Li during the PLD process as was seen in previous studies.9 This result proves that the Li-rich target is a key factor for highly crystalline and stoichiometric LLT thin film, taking into account previous studies (SI Table S1). Li ionic conductivity was measured for 120-nm-thick LLT thin film on NGO substrate with 10 × 10 mm2 area. Several comb electrodes were formed along NGO [001] and NGO [11̅0] directions on the same sample. Here, LLT thin film was an orthogonally distorted perovskite structure, so that we define a- and b-axes of LLT as along NGO [001] and NGO [11̅0], respectively (ap = 0.3855 nm and bp = 0.3864 nm). In-plane Li ionic conductivity was measured at different temperatures with the range of 298−393 K along a- and b-axes, corresponding to ionic conduction path across two bottleneck bc- and ac-planes, respectively. Complex impedance curve showed semicircles at different temperatures with the frequency range of 10 Hz to 10 MHz (Figure 6a), which can be described as a parallel circuit of a resistance and a capacitance. The double layer capacitance was evaluated from the Randle circuit model.21 The double layer capacitance of LLT epitaxial film across bc-plane at room temperature was calculated to be 3 × 10−12 F. This impedance

Figure 4. Reciprocal space mappings of LLT thin films on (a) SrTiO3 (100), (b) LSAT (100), and (c) NdGaO3 (110).

LLT [001] // STO [010] on STO, LLT (001) // LSAT (100) with LLT [100] // LSAT [010] on LSAT, and LLT (001) // NGO (110) with LLT [100] // NGO [001] and LLT [010] // NGO [1−10] on NGO. Figure 4 indicates that in-plane lattice constants of LLT thin films on STO (100), LSAT (100), and NGO (110) were the same as those of substrates, ranging from 0.3855 to 0.3905 nm. This result represents their coherent epitaxial growth, resulting in their completely strained lattices. Figure 5a,b shows schematic orientation of LLT under tensile and compressive strain. Comparing the lattice parameters of substrates and tetragonal perovskite LLT, one can find that aSTO > cLLT/2 ≥ aLLT > aLSAT > apNGO. On the STO substrate, tensile strain was applied on the LLT lattice so that ac-plane had a smaller strain along the in-plane direction, resulting in an h00 orientation. On LSAT and NGO substrates, compressive strain was applied on the LLT lattice, so that the ab-plane had a 2189

DOI: 10.1021/cg501834s Cryst. Growth Des. 2015, 15, 2187−2191

Article

Crystal Growth & Design

NGO (110) are larger along the a-axis than those along the baxis (see inset of Figure 6b). Therefore, the bottleneck for the Li ionic conduction path across the ac-plane is smaller than that across the bc-plane, indicating that larger compressive strain results in a lower conductivity with higher activation energy due to more contracted bottleneck, being consistent with previous study using external physical or chemical pressure.15 It is noted that the effect of the biaxial strain on the activation energy is more significant than that of isotropic static pressure. The equivalent pressure caused by in-plane biaxial strain in LLT on NGO (110) along the a- and b-axes was estimated to be 1.03 and 0.57 GPa according to the elasticity equation, in which the Young’s modulus and lattice constant of LLT are 200 GPa23 and 0.3875 nm, respectively. The relative change of activation energy under the biaxial strain (1/Ea)·(dEa/dP) is 12%/GPa, which is almost twice as large as that under isotropic static pressure study of bulk sample.15 Instead of compressive strain, tensile strain can be applied to LLT thin film on substrates with larger lattice constant, such as STO. It is expected that the activation energy in the tensile strained thin film would be lower than that of bulk due to the expanded bottleneck (Figure 1b). However, it was reported that the STO interfacial layer beneath LLT thin film was reduced and electrically conducting, hampering appropriate ionic conductivity measurement.11 Indeed, impedance measurements of LLT thin films on insulating STO and LSAT substrates were failed in this study, probably due to electrically conducting (La,Sr)TiO3 interfacial layer generated by Srintermixing from STO and LSAT substrates. Accordingly, a suitable buffer layer on STO and LSAT substrates is necessary to prohibit the generation of conducting interfacial layer for ionic conductivity measurement. The strain controllable LLT thin film provides a new approach to control ionic conductivity, possibly useful for the application of all solid epitaxial devices using solid electrolyte LLT.

Figure 6. (a) Complex impedance curves of LLT thin film on NGO (110) measured across the bc-plane. Inset shows the enlarged impedance spectra at high temperatures. The low frequency part measured at only room temperature is shown for convenience. Purple curve shows a fitting curve using the Randle circuit (SI Figure S3). (b) Arrhenius plot of in-plane ionic conductivity of LLT thin film across ac- and bc-planes. Inset shows schematic of Li ionic conduction across each bottleneck.

4. CONCLUSION We successfully fabricated high-quality Li0.33La0.56TiO3 epitaxial thin films on STO (100), LSAT (100), and NGO (110) substrates by PLD. Use of Li-rich target resulted in the phase pure LLT thin film with very high growth rate of 2 nm/min. Orientation of LLT thin films was controlled by altering inplane strain: a-axis orientated LLT with two domains on STO (100) substrate and c-axis oriented LLT with single domain on LSAT (100) and NGO (110) substrates. The ionic conductivity is comparable with that of bulk. In addition, anisotropic ionic conductive behavior was observed in LLT on NGO substrate due to the orthorhombic distortion of LLT lattice. In-plane ionic conductivity of LLT thin film on NGO (110) substrate was 6.7 × 10−4 S·cm−1 at room temperature with an activation energy of 0.34 eV perpendicular to NGO [1−10] direction, while those were 4.3 × 10−4 S·cm−1 and 0.36 eV perpendicular to NGO [001] direction. The activation energy of Li ionic conduction increased with an increase in compressive strain, corresponding to contracted ionic conduction path, which has a similar tendency to pressure and substitution effect in bulk study. Epitaxial strain is an effective method to control Li ionic conduction. High and straincontrolled Li ionic conductivity demonstrated in this study pose a possibility for implementation of all epitaxial solid state lithium ion battery structure.

response was mainly contributed from intragrain, since capacitances of grain boundary and intragrain in bulk polycrystalline LLT were on the order of 10−9 F and 10−12 F, respectively.1 Ionic conductivity across the bc-plane at room temperature (σ298 K(bc)) was 6.7 × 10−4 S·cm−1 with an activation energy (Ea) of 0.34 eV, where σ298 K was one order higher than LLT(001) epitaxial thin film on NGO (110) (3.5 × 10−5 S·cm−1)11 and also higher than Li0.17La0.61TiO3 (111) epitaxial thin film on α-Al2O3 (0001) (3.76 × 10−4 S·cm−1) in previous reports.12 Ea is similar to that of bulk polycrystalline LLT (0.33−0.40 eV). This significant improvement of ionic conductivity could be attributed to the compensated Lideficiency and high crystallinity via use of Li-rich target with laser fluence adjustment because Li-deficiency and grain boundary are the main factors to reduce the ionic conductivity significantly.2,22 In-plane anisotropic ionic conduction was observed across ac- and bc-planes. σ298 K(ac) was 4.3 × 10−4 S·cm−1 with Ea(ac) = 0.36 eV, while σ298 K(bc) was 6.7 × 10−4 S·cm−1 with Ea(bc) = 0.34 eV (Figure 6b). Considering the lattice constants of bulk cubic Li0.33La0.56TiO3 (ap = 0.3875 nm), the lattice mismatch and compressive strain of the orthogonal LLT thin film on 2190

DOI: 10.1021/cg501834s Cryst. Growth Des. 2015, 15, 2187−2191

Article

Crystal Growth & Design



(21) Barsoukov, E.; Macdonald, J. R., Eds.Impedance spectroscopy: theory, experiment, and applications, 2nd ed.; John Wiley & Sons: New York, 2005. (22) Inaguma, Y.; Itoh, M. Solid State Ionics 1996, 86−88, 257−260. (23) Cho, Y.-H.; Wolfenstine, J.; Rangasamy, E.; Kim, H.; Choe, H.; Sakamoto, J. J. Mater. Sci. 2012, 47, 5970−5977.

ASSOCIATED CONTENT

S Supporting Information *

XPS Li 1s peaks of amorphous LLT and growth rate of epitaxial LLT on STO (100) substrate. φ-scan of LLT (101) on STO (100). Summary of PLD conditions in previous epitaxial LLT studies. This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. TEL: +81-3-58417595. FAX: +81-3-5841-4603. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS We would like to thank S. Sakae for technical support in chemical composition measurements. X-ray photoemission spectroscopy, impedance measurement, and photolithography process were conducted in Research Hub for Advanced Nano Characterization and in Center for Nano Lithography & Analysis, The University of Tokyo, supported by MEXT, Japan. This research was in part supported by JSPS through the NEXT Program initiated by CSTP (GR029) and JSPS Grantin-Aid for Scientific Research (26105002).



REFERENCES

(1) Inaguma, Y.; Chen, L.; Itoh, M.; Nakamura, T.; Uchida, T.; Ikuta, H.; Wakihara, M. Solid State Commun. 1993, 86, 689−693. (2) Gao, X.; Fisher, C. A. J.; Kimura, T.; Ikuhara, Y. H.; Kuwabara, A.; Moriwake, H.; Oki, H.; Tojigamori, T.; Kohama, K.; Ikuhara, Y. J. Mater. Chem. A 2014, 2, 843−852. (3) Harada, Y.; Hirakoso, Y.; Kawai, H.; Kuwano, J. Solid State Ionics 1999, 121, 245−251. (4) Okumura, T.; Ina, T.; Orikasa, Y.; Arai, H.; Uchimoto, Y.; Oqumi, Z. J. Mater. Chem. 2011, 21, 10195−10205. (5) Okumura, T.; Yokoo, K.; Fukutsuka, T.; Uchimoto, Y.; Saito, M.; Amezawa, K. J. Power Sources 2009, 189, 536−538. (6) Okumura, T.; Ina, T.; Orikasa, Y.; Arai, H.; Uchimoto, Y.; Oqumi, Z. J. Mater. Chem. 2011, 21, 10061−10068. (7) Kitaoka, K.; Kozuka, H.; Hashimoto, T.; Yoko, T. J. Mater. Sci. 1997, 32, 2063−2070. (8) Maqueda, O.; Sauvage, F.; Laffont, L.; Martínez-Sarrión, M. L.; Mestres, L.; Baudrin, E. Thin Solid Films 2008, 516, 1651−1655. (9) Kim, D. H.; Imashuku, S.; Wang, L.; Shao-Horn, Y.; Ross, C. A. J. Cryst. Growth 2013, 372, 9−14. (10) Aaltonen, T.; Alnes, M.; Nilsen, Ola.; Costelle, L.; Fjellvåga, H. J. Mater. Chem. 2010, 20, 2877−2881. (11) Ohnishi, T.; Takada, K. Solid State Ionics 2012, 228, 80−82. (12) Kim, S.; Hirayama, M.; Cho, W.; Kim, K.; Kobayashi, T.; Kaneko, R.; Suzuki, K.; Kanno, R. CrystEngComm 2014, 16, 1044− 1049. (13) Morales, M.; Laffez, P.; Chateigner, D.; Vickridge, I. Thin Solid Films 2002, 418, 119−128. (14) Kim, S.; Hirayama, M.; Suzuki, K.; Kanno, R. Solid State Ionics 2013, 262, 578−581. (15) Inaguma, Y.; Yu, J.; Shan, Y.-J.; Itoh, M.; Nakamura, T. J. Electrochem. Soc. 1995, 142, L8−L11. (16) Ohnishi, T.; Hang, B. T.; Xu, X.; Osada, M.; Takada, K. J. Mater. Res. 2010, 25, 1886−1889. (17) Ohnishi, T.; Takada, K. Appl. Phys. Express 2012, 5 (055502), 1−3. (18) Schou, J. Appl. Surf. Sci. 2009, 255, 5191−5198. (19) Saenger, K. L. J. Appl. Phys. 1991, 70, 5629−5635. (20) Stramare, S.; Thangadurai, V.; Weppner, W. Chem. Mater. 2003, 15, 3974−3990. 2191

DOI: 10.1021/cg501834s Cryst. Growth Des. 2015, 15, 2187−2191