ISSN 0036�0236, Russian Journal of Inorganic Chemistry, 2014, Vol. 59, No. 5, pp. 424–430. © Pleiades Publishing, Ltd., 2014. Original Russian Text © G.B. Kunshina, O.G. Gromov, E.P. Lokshin, V.T. Kalinnikov, 2014, published in Zhurnal Neorganicheskoi Khimii, 2014, Vol. 59, No. 5, pp. 583–588. 424 Substituted lithium titanophosphates having the NASICON structure of the general formula Li1 + xAlxTi2 – x(PO4)3 (LATP) are candidate materials for use in all�solid�state lithium�ion batteries for their high lithium�ion conductivities [1–3]. Attempts have recently been undertaken at creating composite cath� odes and anodes based on Li1 + xAlxTi2 – x(PO4)3 [4]. In order for composite (active material + electrolyte + carbonaceous material) electrolytes to be manufac� tured, electrolyte powders with minimal particle sizes are required to provide an enhanced cycling perfor� mance of batteries as a result of the formation of a uni� form electroconducting grid [5]. Submicron�sized powders are also required for manufacturing dense films using new ceramic sputtering (aerosol deposi� tion) technology for LATP�based thin�film batteries [6]. It follows that powders having minimal particle sizes and prepared at moderate temperatures are required for many applications of electrolytes. One possible way to prepare submicron�sized pow� ders is to use solutions as precursors of a solid electro� lyte. Various variant syntheses of LATP solid electro� lyte from solutions are found in the literature [7–10]. Duluard et al. [9] prepared single�phase LATP as a result of sintering, within a narrow temperature range (750–850°C), of a precursor prepared by precipita� tion from an ethanolic solution containing titanium alkoxide Ti(OC3H7)4 and aluminum alkoxide Al(OC4H9)3 and as the initial reagents. Freeze drying for several days was required for avoiding the agglom� eration of powders prepared by coprecipitation from ethanolic solution which contained titanium and alu� minum alkoxides [8, 9]. The preferable titanium�containing component to be used in the sol–gel process in most cases consists of alkoxides Ti(OC4H9)4 and Ti(OC3H7)4, which are very moisture sensitive, and high�cost and hardly available organic compounds: Al(OC4H9)3, Al(OC3H7)3, CH3COOLi ⋅ 2H2O, n�C4H9OH, РО(OC4H9)4, СН3СОСН2СОСН3, C2H5OH, and C3H7OH [11, 12]. The difficulties encountered in implementing sol–gel synthesis arise from the low solubilities of phosphates in alcoholic solutions and the formation of hydrolysis products of titanium alkoxides in the pres� ence of water. Schroeder et al. [13] carried out a method similar to the sol–gel synthesis of LATP in the following manner: concentrated HNO3 was added to enhance the solubility of NH4H2PO4 in ethanol, then acetylacetone СН3СОСН2СОСН3 was poured to sta� bilize Ti4+ in the complex and to avoid precipitation of Ti(OCH2(CH2)2CH3)4. The final product as probed by X�ray powder diffraction contained impurities of AlPO4 and TiP2O7, which might be responsible for a moderate ionic conductivity of the solid electrolyte. A significant weakness of the proposed synthesis methods consist in the use of toxic and flammable solvents such as methanol, methylcellosolve СН3ОСН2СН2ОН, and acetylacetone СН3СОСН2СОСН3, which com� plicates carrying out technological operations and decreases environmental safety. Therefore, it is topical to develop an accessible sol– gel method for preparing single�phase LATP having high ionic conductivity that would use nontoxic reagents. The goal of this work was to prepare LATP electro� lyte using a modified sol–gel process intended for manufacturing submicron�sized electrolyte powders. Sol–Gel Synthesis of Li1.3Al0.3Ti1.7(PO4)3 Solid Electrolyte G. B. Kunshina, O. G. Gromov, E. P. Lokshin, and V. T. Kalinnikov Tananaev Institute of Chemistry and Technology of Rare Elements and Mineral Resources, Kola Scientific Center, Russian Academy of Sciences, ul. Fersmana 14, Apatity, Murmansk oblast, 184200 Russia e�mail:
[email protected] Received April 10, 2013 Abstract—A modified sol–gel process was studied as applied to synthesize a lithium�conducting solid elec� trolyte of composition Li1.3Al0.3Ti1.7(PO4)3 (LATP) using water�soluble salts Al(NO3)3 ⋅ 9H2O, LiNO3 ⋅ 3H2O, and (NH4)2HPO4 and a titanium(IV) citrate complex. As�synthesized samples were characterized using X�ray powder diffraction, DSC/TG, SEM, and impedance spectroscopy. Sintering of as�synthesized amorphous powders at 700°C was found to yield LATP with crystallite sizes of 42–48 nm. Ionic conductivity of the electrolyte measured in the frequency range 25–106 Hz in disks having 86–90% density that were sin� tered at 1000°C was (3–4) × 10–4 S/cm. Temperature�dependent ionic conductivity was studied in the range 25–200°C. The activation energy of conduction was determined for LATP. DOI: 10.1134/S0036023614050118 SYNTHESIS AND PROPERTIES OF INORGANIC COMPOUNDS RUSSIAN JOURNAL OF INORGANIC CHEMISTRY Vol. 59 No. 5 2014 SOL–GEL SYNTHESIS 425 EXPERIMENTAL Of the variant sol–gel methods, we chose to modify Peccini’s method in order to synthesize LATP; this method comprises dissolving cationic precursors in aqueous solutions of citric acid to form metal chelate complexes [14]. The initial reagents used were avail� able chemicals: LiNO3 ⋅ 3H2O (pure for analysis), Al(NO3)3 ⋅ 9H2O (chemically pure), (NH4)2HPO4 (pure for analysis), TiO2 (anatase), 30–35% H2O2 (Russian State Standard (GOST) 177�88), HF, HNO3, NH4OH, C6H8O7 ⋅ H2O (chemically pure), and ethylene glycol. A distinctive feature of the syn� thesis method from the state�of�art methods consists in that the titanium�containing component is freshly precipitated hydrous titanium hydroxide, which is well soluble in HNO3 and whose preparation we described earlier in detail [10]. As�synthesized LATP solid elec� trolyte powders were characterized by X�ray powder diffraction, DSC/TG, and ionic and electron conduc� tivity measurements. Chemical composition was monitored by inductively coupled plasma mass spec� trometry on an ELAN 9000 DRC�e quadrupole ana� lyzer. Phase composition was determined using a DRON�2 diffractometer (graphite�monochromated СuK α radiation); phase identification was performed with the use of JCPDS databases. A DTA experiment was performed on a NETZSCH STA 409 PC/PG syn� chronous thermal analyzer in the range 25–1000°C under an argon atmosphere. Specific surface areas of powders were determined by thermal nitrogen desorp� tion on a TriStar II 3020 specific area analyzer, which is capable of measuring porosity parameters. Crystal� lite morphology was studied using a SEM LEO�420 scanning electron microscope. An average particle size was calculated from d = К/Sρ, (1) where d is average particle diameter; К is the particle form factor, equal to six for spherical and cubic parti� cles; S is specific surface area; and ρ is density of the material. Conductivity studies comprised studying the vari� ance in the complex impedance of LATP samples (flat capacitor geometry; Ag electrodes) in the frequency range 25–106 Hz on an E7�20 immittance meter. The electrolyte synthesis scheme was as shown in Fig. 1. Freshly precipitated hydrous titanium hydrox� ide TiO2 ⋅ xH2O [15] was dissolved in HNO3 (65 wt %). To the resulting nitrate solution, concentrated C6H8O7 solution was poured to form a titanium(IV) citrate complex; then concentrated aqueous solutions of Al(NO3)3 ⋅ 9H2O, LiNO3 ⋅ 3H2O, and (NH4)2HPO4 were added in the stoichiometric proportion that cor� responded to Li1.3Al0.3Ti1.7(PO4)3. The molar ratio [C6H8O7] : [Ti 4+ + Al3+ + Li+] was 4 : 1. In order to retain sparingly soluble phosphates in solution, a mix� ture of H2O2 (30 wt %) and HNO3 was added in the proportion 2 : 1, and then ethylene glycol was added in an amount required for gel to be formed (25–30 vol %). The resulting solution, which contained 75–80 g/L LATP, was heated under stirring with a magnetic stir� rer at 150°C for 2 h. Etherification and polymerization between ethylene glycol and citric acid gave rise to a thick, viscose mass, white in color, in which metal ions were distributed uniformly. The gel was pyrolyzed in the range 300–450°C to yield a brown, amorphous, and highly disperse product. RESULTS AND DISCUSSION Figure 2 displays the results of thermal analysis for the precursor of LATP obtained after drying at 150°C. The major weight loss (∼55%) occurred in the range Fig. 1. Scheme of the citrate sol–gel process for preparing LATP. HF TiO2 (Anatase) Dissolutionе Solution H2TiF4 NH4OHconc Precipitation Filtering the precipitate Precipitate TiO2 · xH2O HNO3 Citric acid LiNO3 · 3H2O Titanium(IV) citrate complex Al(NO3)3 · 9H2O (NH4)2HPO4 Mixing in stoichiometric proportions HNO3 + H2O2 Ethylene glycol Heating the solution under stirring Etherification at 150–200°С, gel formation Pyrolysis at 350–450°C Calcining amorphous LATP precursor at 500–700°C Compacting LATP disks Sinteringdisks at 800–1000°C LATP conductivity measurement 426 RUSSIAN JOURNAL OF INORGANIC CHEMISTRY Vol. 59 No. 5 2014 KUNSHINA et al. below 350°C and was associated with the dehydration and subsequent decomposition of the precursor with evolution of nitrogen oxides. The total weight loss was 78%. The diffuse exotherm at 605°C can serve as evi� dence that the precursor components reacted with each other to produce crystalline LATP. Weight loss at temperatures above 600°C was due to carbon burning out. The combination of DTA and X�ray powder dif� fraction data indicates that the crystalline electrolyte can be prepared by calcining the amorphous precursor at temperatures higher than 600°C. Figure 3 displays X�ray diffraction patterns for LATP precursors calcined for 1 h in air at various tem� peratures. One can clearly see that distinct major peaks corresponding to a LiTi2(PO4)3 phase (accord� ing to JCPDS card 35�0754) appeared on the back� ground of an amorphous halo after the precursor was calcined at 500°C (curve 1). These reflections were used to identify samples, since a partial substitution of Ti4+ cations by Al3+ in the LiTi2(PO4)3 gives rise to Li1 + хAlхTi2 – х(PO4)3 (0 ≤ х ≤ 0.4) solid solutions and does not alter the phase composition of the resulting materials [16, 17]. When the temperature increased to 600°C, the sample lost its amorphism but an impurity phase of lithium metatitanate Li2TiO3 was present in minor amount. After the precursor was calcined at 700°C, the X�ray diffraction pattern featured only sharp reflections from crystalline LiTi2(PO4)3 (curve 3). According to X�ray powder diffraction, the as�synthe� sized electrolyte was a single�phase solid solution hav� ing LiTi2(PO4)3 structure. A further increase in precur� sor calcination temperature to 1000°C only enhanced the degree of crystallinity and did not cause formation of minor phases. The correspondence of chemical composition to Li1.3Al0.3Ti1.7(PO4)3 was verified by mass spectrometry (Table 1). After the precursor was calcined at 700°C, an X�ray diffraction pattern of a single�phase LATP sample was recorded on an XRD�6000 Shimadzu diffractometer in the range 15° < 2θ < 40° in order to calculate coher� ent scattering lengths (CSLs). The CSL sizes (Dhkl) were calculated by the Selyakov and Scherrer relation� ship (2) where λ is СuК α radiation length (0.154178 nm), βhkl is physical broadening of the diffraction peak, and θ is diffraction angle. The major lines in X�ray spectra were fitted by Gaussians. Physical broadening was calculated from βhkl = β – s, 0.94 ,hkl hkl D λ= β θcos 100 90 80 70 60 50 40 30 900600500400100 700 800300200 m, % T, °C TG DTG DTA 138 245 605 392 Fig. 2. Results of thermal analysis for the sol–gel precursor of LATP obtained after drying at 150°C. Table 1. Chemical compositions of LATP samples Sample Content, % Li2O Al2O3 TiO2 P2O5 LATP (theor.) 5.06 3.99 35.42 55.52 No. 101 5.27 4.01 33.91 55.82 No. 279 5.04 3.99 33.18 54.81 RUSSIAN JOURNAL OF INORGANIC CHEMISTRY Vol. 59 No. 5 2014 SOL–GEL SYNTHESIS 427 where β is full width at half�height of the X�ray diffrac� tion peak, and s is instrumental broadening (0.09° ± 0.01° 2θ). The reference used to determine the instrumental broadening was metallic zinc. The peaks chosen to calculate coherent scattering lengths were those hav� ing maximal intensities from crystallographic planes (113) and (104) with reflection angles 2θ = 24.8° and 21.2°, respectively. An inspection of diffraction peak broadenings indicates that the primary crystallite size of the LATP upon calcination at 700°C was 48 nm according to peak (113) (2θ = 24.8°) and 42 nm according to peak (104) (2θ = 21.2°). The values of Dhkl calculated for the two characteristic diffraction peaks are almost coincident. Table 2 displays the specific surface areas of LATP powders depending on the precursor sintering temper� ature and the particle sizes calculated from relation� ship (1). A maximal specific surface area (35 m2/g) was found in porous LATP samples after the precursor was calcined at 600°C for 1 h (the average particle size was 58 nm). In this case, burning out of the organic com� ponent gave rise to micropores, which collapsed upon further sintering, so that the specific surface areas of LATP powders decreased considerably (to 8 m2/g). The average particle size upon sintering the electrolyte powder at 700°C was 245 nm. Electron micrographs (Fig. 4) show how the microstructure changes upon sintering the sol–gel precursor at various temperatures. From the SEM images, it follows that primary crystallites are aggre� 5040302010 2θ, deg 1 2 3 (0 12 ) (1 04 ) (1 13 ) (2 02 ) (0 24 ) (2 11 ) (1 16 ) (3 00 ) (2 20 ) (2 23 ) (1 34 ) (0 42 ) (0 12 ) (1 04 ) (1 13 ) (2 02 ) (0 24 ) (2 11 ) (1 16 ) (3 00 ) (2 23 ) ( 13 4) (0 42 ) (0 12 ) (1 04 ) (1 13 ) (0 24 ) (3 00 ) * * Fig. 3. X�ray diffraction patterns of LATP precursor samples calcined for 1 h in air at various temperatures: (1) 500, (2) 600, and (3) 700°C; Li2TiO3 reflections are marked with asterisks. 428 RUSSIAN JOURNAL OF INORGANIC CHEMISTRY Vol. 59 No. 5 2014 KUNSHINA et al. gated to a considerable extent. The particle sizes cal� culated from (1) exceed the CSLs derived from X�ray dif� fraction line broadenings, and this fact may indicate that crystallites are grown together to form one particle. In order to measure ionic conductivity of LATP, the electrolyte molding powder prepared at 700°C was pressed into cylinder�shaped disks 12 mm in diameter and 2–3 mm high to be then sintered at 1000°C for 1 h (sintered disks reached densities of up to 86–90% X�ray density as measured by hydrostatic weighing). The end faces of disks were coated with silver paste to form electrodes. Figure 5 displays an electrochemical impedance spectrum for a sintered LATP disk on a complex plane at 25°C. The spectrum has two branches: the high�frequency branch describes the properties of the sample, and the low�frequency one describes the effect of electrode processes [19]. The conductivity value was calculated by extrapolating the high�frequency portion of the impedance spectrum to the resistance axis. The room�temperature value of conductivity was (3–4) × 10–4 S/cm. Figure 6 shows the temperature�dependent ionic conductivity for a sintered LATP disk in the range 25– 200°C. The slope of the linear plot in the Arrhenius Table 2. Specific surface areas and particles sizes of LATP powders T, °C Ssp, m 2/g dpart, nm CSL 500 9.44 215 600 35.15 58 700 8.29 245 42–48 ρ = 2.95 g/cm3 [18]. 400 nm 400 nm 1 μm400 nm (а) (b) (c) (d) Fig. 4. Electron micrographs of (a–c) LATP precursors calcined at (a) 600°C, (b), 700°C and (c) 800°C and of (d) a LATP disk sintered at 1000°C for use in ionic conductivity measurements. RUSSIAN JOURNAL OF INORGANIC CHEMISTRY Vol. 59 No. 5 2014 SOL–GEL SYNTHESIS 429 coordinates was used to calculate the activation energy Еа of conductivity of LATP from (3) where σо is frequency factor, Еa is the activation energy of the electrolyte, R is universal gas constant, and Т is absolute temperature. The activation energy Eа was calculated as 0.21 eV, which correlates with the values known for LATP [20]. ( )−σ = σ a0 exp ,ERT In summary, we have studied a citrate sol–gel pro� cess for producing a lithium�conducting solid electro� lyte LATP in submicron�sized powders from low�cost initial reagents, so as to simplify the synthesis appre� ciably. A single�phase product free of nonconducting TiP2O7 and AlPO4 impurities, was formed as a result of calcining the precursor at 700°C for 1 h. Ionic con� ductivities in LATP disks calcined at 1000°C for 1 h were (3–4) × 10–4 S/cm. ACKNOWLEDGMENTS This study was supported by a grant for leading sci� entific school (grant no. NSh 1937.2012.3). REFERENCES 1. P. Birke, F. Salam, S. Doring, and W. Weppner, Solid State Ionics 118, 149 (1999). 2. X. M. Wu, J. L. Liu, R. S. Li, et al., Russ. J. Electro� chem. 47, 980 (2011). 3. A. Aboulaich, R. Bouchet, G. Delaizir, et al., Adv. Energy Mater. 1, 179 (2011). 4. M. Kotobuki, Y. Isshiki, H. Munakata, and K. Kana� mura, Electrochim. Acta 55, 6892 (2010). 5. L. Liu, M. Zhou, G. Wang, et al., Electrochim. Acta 70, 136 (2012). 6. D. Popovici, H. Nagai, S. Fujishima, and J. Akedo, J. Am. Ceram. Soc. 94, 3847 (2011). 7. L. Huang, Z. Wen, M. Wu, et al., J. Power Sources 196, 6943 (2011). 150 120 90 60 30 15012090300 4 2 1086420 60 Z '' × 10–3, Ω 3 × 103 T = 400 K 50 50 T = 300 K 3 × 103 ω Z ' × 10–3, Ω Fig. 5. Electrochemical impedance spectrum for a sintered LATP disk (S = 0.89 cm2, h = 0.29 cm) on a complex plane at 300 and 400 K (figures denote frequency in Hz). –2.2 –2.4 –2.6 –2.8 –3.0 –3.2 –3.4 3.63.22.82.42.0 logσ [S/cm] 1000/T, K–1 Fig. 6. Temperature�dependent ionic conductivity in LATP. 430 RUSSIAN JOURNAL OF INORGANIC CHEMISTRY Vol. 59 No. 5 2014 KUNSHINA et al. 8. M. Cretin and P. Fabry, J. Eur. Ceram. Soc. 19, 2931 (1999). 9. S. Duluard, A. Paillassa, L. Puech, et al., J. Eur. Ceram. Soc. 33, 1145 (2013). 10. G. B. Kunshina, O. G. Gromov, E. P. Lokshin, and V. T. Kalinnikov, Inorg. Mater. 49, 95 (2013). 11. X. M. Wu, J. L. Liu, S. Chen, et al., Ionics, 827 (2010). 12. M. Kotobuki, Proceedings of the 2nd International Con� ference on Electrical, Electronics and Civil Engineering (ICEECE'2012) (Singapore, 2012), p. 165. 13. M. Schroeder, S. Glatthaar, and J. R. Binder, Solid State Ionics 201, 49 (2011). 14. C. R. Mariappan, C. Galven, M.�P. Crosnier�Lopez, et al., J. Solid State Chem. 179, 450 (2006). 15. G. B. Kunshina, V. V. Efremov, and E. P. Lokshin, Russ. J. Electrochem. 49, 725 (2013). 16. H. Aono, E. Sugimoto, Y. Sadaoka, et al., J. Electro� chem. Soc. 137, 1023 (1990). 17. T. S. Sysoeva, E. A. Asabina, V. I. Pet’kov, and V. S. Kurazhkovskaya, Russ. J. Inorg. Chem. 54, 829 (2009). 18. J. L. Narváez�Semanate and A. C. M. Rodrigues, Solid State Ionics 181, 1197 (2010). 19. A. K. Ivanov�Shits and I. V. Murin, Solid State Ionics (St. Petersburg Univ. Press, St. Petersburg, 2000), Vol. 1 [in Russian]. 20. N. V. Kosova, E. T. Devyatkina, and D. I. Osintsev, Zh. Strukt. Khim. 45, 144 (2004). Translated by O. Fedorova