- Email: [email protected]
Tape casted thin films of solid electrolyte Lithium-Lanthanum-Titanate
Solid State Ionics 328 (2018) 25–29
Contents lists available at ScienceDirect
Solid State Ionics journal homepage: www.elsevier.com/locate/ssi
Tape casted thin films of solid electrolyte Lithium-Lanthanum-Titanate ⁎
T
Felix Schröckert , Nikolas Schiffmann, Ethel C. Bucharsky, Karl G. Schell, Michael J. Hoffmann Institute for Applied Materials – Ceramic Materials and Technologies, Karlsruhe Institute of Technology (KIT), Haid-und-Neu-Str. 7, 76131 Karlsruhe, Germany
A B S T R A C T
Phase-pure solid electrolyte Lithium-Lanthanum-Titanate (LLTO) has been prepared by solid-state-synthesis. Thin sheets of LLTO have been successfully prepared via tape casting and subsequent sintering. A tape casting slurry composition has been developed. The influencing parameters in slurry composition are presented. These are dispersant concentration, ratio of binder to plasticizer and total amount of binder and plasticizer. Furthermore it was found that drying conditions significantly influence tape casting results. Sintered tapes were compared to massive reference samples regarding density, phase-purity, microstructure and conductivity. Our results show that sintered tapes and massive samples exhibit comparable properties.
1. Introduction Batteries with solid electrolyte promise to overcome the disadvantages related to liquid electrolytes. There are various materials that are suitable for this application, one of them is the perovskite Li3xLa2/3−xTiO3 (LLTO) [1–3]. In literature bulk conductivities of 10−4 S/cm are reported for x = 0.1–0.12 [4,5], whereas the total conductivity is restricted by the less conductive grain boundaries [6,7]. A challenge for the implementation of solid electrolytes in batteries is the production of thin sheets. Vapor deposition methods are capable of producing thin layers [8], but are often expensive or limited in their scalability. In this study this problem is addressed by implementing the tape casting technology to cast thin tapes of LLTO. Tape casting is an established industrial production process [9,10] that can easily be scaled up from lab scale [11]. Jimenez et al. have already pursued an approach to fabricate thick films of LLTO via tape casting [12]. An interesting perspective for the application of a tape cast NASICON-type solid electrolyte Li1.3Al0.5Nb0.2Ti1.3(PO4)3 in a lithium-air battery was recently demonstrated by Nemori et al. [13]. For successful tape casting the slurry composition and its processing are essential [14]. This article reports on the slurry development and adjustments to the fabrication process for this specific problem. Green tapes are cast and then sintered to dense samples. The sintered tapes are characterized and compared to massive samples.
mixed and attrition milled with zirconia milling balls (2 mm) in isopropanol. In a previous study the mass loss due to the water binding properties of La2O3 was determined up to 1200 °C in order to weigh in correctly [15]. The powder was placed in alumina crucibles and calcined at 950 °C for 8 h. After calcination the powder was milled again in an attrition mill. The particle size of the derived powder was determined by laser diffraction (CILAS 1064, France), phase purity was verified by X-ray diffraction (XRD, D8 Advance, Bruker). XRD patterns were derived in a 2θ range from 22° to 90° with an increment of 0.01°. Cylindrical reference samples were prepared by uniaxial dry pressing and subsequent cold-isostatic pressing at 400 MPa (dimensions of sintered samples: diameter 17 mm, height 2 mm). The preparation of the tape casting slurry followed the given scheme (adapted from [16]): 1) Dissolution of dispersant (Zschimmer & Schwarz KM 3014) in ethanol 2) Addition of LLTO powder 3) Homogenization in ball mill for 24 h (10 mm zirconia balls) 4) Addition of plasticizer (polyethylene glycol 400, Merck KGaA, Germany) and binder (polyvinyl butyral Mowital B 45H, Kuraray Europe GmbH, Germany) 5) Homogenization in ball mill for 24 h and in dual asymmetrical centrifuge (SpeedMixer DAC 700.2 VAC-P, Hauschild Engineering, Germany) with degassing
2. Experimental The LLTO powder was prepared by a solid state synthesis. The starting materials Li2CO3 (purity > 99%, Alfa Aesar GmbH & Co KG, Germany), TiO2 (purity > 99.9%,Sigma-Aldrich Chemie GmbH, Germany) and La2O3 (purity > 99.5%, Merck KGaA, Germany) were ⁎
To develop a suitable tape casting slurry several recipe and process parameters were studied, these were: optimal dispersant concentration, ideal binder-plasticizer ratio, influence of total amount of binder and plasticizer at a fixed ratio and the influence of drying conditions on the green tape quality. The optimal dispersant concentration was
Corresponding author. E-mail address: [email protected] (F. Schröckert).
https://doi.org/10.1016/j.ssi.2018.10.028 Received 29 August 2018; Received in revised form 19 October 2018; Accepted 25 October 2018 Available online 18 November 2018 0167-2738/ © 2018 Elsevier B.V. All rights reserved.
Solid State Ionics 328 (2018) 25–29
F. Schröckert et al.
determined by viscosity measurements (HAAKE MARS 60, Thermo Scientific, coaxial cylinder geometry) of the suspension (only ethanol, dispersant and powder). Several dispersant concentrations with a fixed solids load (cV = 20%) were studied. Viscosities were measured by testing a shear rate ramp from 1 to 100 1/s, results are shown for shear rates of 10, 50 and 100 1/s since the values for possible shear rates in the casting gap lie within this interval. For the investigation of the binder-plasticizer ratio and the total amount of binder and plasticizer the slurries were prepared according to the above mentioned scheme. Their viscosity was then studied with a cone-plate geometry (2° angle). For evaluation of green tape quality the slurry was cast on polymer substrates (Cerapeel Q1 (S), Toray, Japan) on a laboratory film applicator (AB3000, TQC). The casting gap was 200 μm and the speed of the casting head was 5 mm/s. To study the influence of drying speed the cast tapes were dried under various conditions. Slow drying was achieved by covering the wet tape with a lid so that the atmosphere beneath the lid was enriched with evaporating solvent. Drying under ambient conditions without covering led to a moderate drying rate. Fast drying was realized by laying the substrate with the still wet tape on a heated plate (T = 70 °C). The green tapes and the reference samples were sintered at various temperatures for 1 h, the heating rate was 3 K/min. For sintering the tapes were placed on an alumina plate and covered with an alumina crucible. After sintering, the volume of the tapes was determined by measuring their area with a digital microscope (Keyence VHX-6000, Japan) and their mean thickness with a mechanical measuring sensor. For thickness determination 30 measurements in total were performed on each sample: ten measurements at three different sites of the sample. From these three sites the average and standard deviation was calculated. Additionally He-pycnometry (AccuPyc 1330, Micromeritics GmbH, Germany) has been carried out to verify the results. The density of massive samples was determined by the Archimedes method. Phase-purity of the sintered samples was checked by XRD. Tapes were placed on modeling clay and were aligned in height and angle by pressing perpendicular to the measuring plane with an axial press. Microstructural analysis was carried out by Electron backscatter diffraction (EBSD, XFlash and Esprit 2.0, Bruker, Germany). For local element analysis Energy-dispersive X-ray spectroscopy (EDS, Quantax, Bruker) was employed. The conductivity of the samples was measured by impedance spectroscopy (VersaSTAT 4, Princeton Applied Research) in potentiostatic mode with an amplitude of 10 mV in the frequency range of 10−3 to 106 Hz, the contacted area was sputtered with gold.
Fig. 1. XRD analysis on LLTO powder, tapes and massive samples.
Fig. 2. Dependence of viscosity on dispersant concentration.
3. Results and discussion LLTO powders have been successfully synthesized; in Fig. 1 the result of XRD analysis is shown. Measured diffractograms are in good agreement with the reference pattern. After milling a median particle size of d50 ≈ 200 nm was measured. 3.1. Slurry development To determine the optimal dispersant concentration the viscosity of suspensions with 1, 2 and 3 wt% (percent by weight) dispersant with respect to powder weight were studied (Fig. 2). The suspension with 2 wt% shows a minimum at all studied shear rates. Since the dispersant influences the particle-particle interaction and thereby the viscosity we can conclude that the optimal dispersant concentration can be found at the minimum of viscosity. In further results and discussion the dispersant concentration will be set to 2 wt%. In order to determine the appropriate amount of binder relative to plasticizer a study with varying binder-plasticizer ratios has been carried out; ratios were 3:1, 2:1 and 1:1. As can clearly been seen in Fig. 3 the addition of more plasticizer significantly reduces viscosity. Since all other components of the slurry remain the same, this behavior can be attributed to the interaction of the plasticizer with the binder [17]. The
Fig. 3. Influence of the binder-plasticizer ratio on the viscosity at different shear rates.
26
Solid State Ionics 328 (2018) 25–29
F. Schröckert et al.
polymer load. 3.2. Study of drying conditions Until now drying was performed under ambient conditions. Even though the slurry formulation has already been adjusted some warping of the green tapes could still be observed. Variation of the drying speed showed that the drying rate has significant influence on this effect. Drying of the tapes can be limited by two factors: evaporation of solvent at the surface and diffusion of solvent to the surface, whereby diffusion is usually the limiting factor [19,21]. If evaporation at the surface is faster than elsewhere the slurry concentration throughout the tape does not remain homogeneous [19]. If the evaporation and diffusion rate are the same no concentration gradient should be observed. By saturating the atmosphere with solvent the evaporation rate at the surface is also reduced, whereas the diffusion rate should remain unchanged. However, the result was that the slowly dried tapes show even more pronounced warping than those dried at ambient conditions. Accelerated drying was carried out by placing the substrate with the still wet, freshly cast tape on a heated plate. This process increases the diffusion rate. Tapes that were dried in this manner did not exhibit warping any more. This shows that the drying process is of significant importance for the green tape quality. By heating from below the transport of solvent to the surface and thereby the drying process as a whole is accelerated. We can think of this as a quasi-instantaneous conservation of the perfectly homogeneous distribution of components in the freshly sheared slurry.
Fig. 4. Influence of the total amount of binder and plasticizer on the viscosity at different shear rates.
measured viscosities are all within the limits for tape casting slurries [10]. Nevertheless the tapes derived from these slurries show great differences in green and sintered condition. The slurry 3:1 has large agglomerates that can already be seen in the slurry but become even more visible after casting and drying of the green tape. Examination by EDS has shown that these clumps consist of binder that has obviously not been fully dissolved. The slurry 1:1 has a much lower viscosity, just slightly above the considered minimum value of 1 Pas [10]. It showed a better processability than 3:1 and produced more homogeneous green tapes, clumps in the slurry were no longer observed. This and the significantly reduced viscosity illustrate the interaction of binder and plasticizer [17]. Nevertheless the tapes warped during the sintering process. The slurry 2:1 was found to be a practicable compromise, since it showed a mixture of the above described phenomena but neither of them to such pronounced extends. With a fixed ratio of binder to plasticizer the influence of their total amount on viscosity and tape quality has been studied. The amount of binder and plasticizer can simplified be considered the polymer-load, since these two components make up for the most of polymers in the slurry. In Fig. 4 the results of viscosity measurements are displayed. It becomes clear, that an increased polymer-load of the slurry results in an increased viscosity. When casting the two slurries with viscosities of around 1 Pas and below cracks in the green tapes could be observed. It is assumed, that the total amount of non-volatile components is not sufficient to form a continuous tape. As the slurry dries, i.e. the solvent evaporates, the remaining slurry becomes thicker. Eventually the green tape forms as the whole arrangement shrinks. The solvent can only evaporate at the (upper) surface of the slurry, on the lower side evaporation is hindered by the substrate on which the slurry is cast onto. This means, that the thickening process takes place at the upper side first [18,19]. One can think of this as floating ice floes. In the further drying process some solvent will still travel through the already dry top layer, but evaporation in the spaces between the floes, which will eventually become cracks, should be much easier. The crack formation disappears with increased polymer load. At high polymer loads warping of the green tapes to the side of the substrate was observed. A possible explanation for this phenomenon is that when the drying process starts, the dry top layer is already in its final configuration. The lower layer of the tape forms later, since it takes longer to remove the solvent from there. The bottom layer dries thereby under fixation of the casting substrate [20]. When the tape is removed from the substrate this constraint is removed and compensated by deformation. At a polymer load of 15 wt% a practicable compromise was found. No cracks were observed and the warping problem was less visible than at the higher
3.3. Sintering and characterization Tapes and massive samples were sintered at 950, 1000, 1050 and 1100 °C for 1 h in air. After sintering their densities were measured. Massive samples were measured employing the Archimedes method. The densities of the tapes were determined volumetrically. From Fig. 5 can be concluded that densification of both massive samples and tapes is possible from 1050 °C. For the tape sample sintered at 1100 °C the density has also been measured by He-pycnometry (relative density greater 98%). The result proves the validity of the density measurements by volume. The thickness of the sintered tapes was around 30 μm. Below 1050 °C the density of the tapes is lower than the density of the massive samples. This can be explained by the empty volume that is created by the outgassing polymeric components during heating. The relative densities (relative to the theoretic density of LLTO) of green bodies were just below 60% for both green tapes and pressed massive samples. For the green tapes the relative density is actually slightly
Fig. 5. Dependency of relative density on sintering temperature. 27
Solid State Ionics 328 (2018) 25–29
F. Schröckert et al.
Fig. 6. Microstructure of massive samples and tapes sintered at 1100 °C in air.
overestimated since the spaces between the LLTO particles are filled with polymer which has a density greater than air, which fills the spaces in the massive samples. This density difference is not accounted for in the calculation of the density of green tapes. When the tapes are heated up the polymeric components burn and leave empty spaces behind [22]. As long as these spaces are not densified, the density – which is measured by volume – must obviously be lower than in the massive samples. On the other hand the circumstance that sintering with insufficient green body packing still results in full densification can be taken as evidence that tape casting – as sensitive a process it might be in other aspects – is somewhat tolerant to suboptimal green body packing. In Fig. 1 can be seen, that phase-purity was remained throughout the whole process. Sintered samples, massive and tapes, are shown for a sintering temperature of 1100 °C. EBSD studies on samples sintered at 1100 °C revealed that the microstructure in tapes is comparable to that found in massive samples (Fig. 6). Note that the microstructural picture of the massive sample was derived from a polished cross section of the sample but the picture of the tape displays an EBSD study that was performed at the surface of the tape without prior mechanical treatment. The fairly difficult preparation of a polished tape cross section made this compromise necessary. The slightly rough nature of the surface explains the black spots which are simply caused by shading of the beam. Nevertheless one can judge from these pictures that the microstructure does not differ by far, the median equivalent grain diameter is around 0.5 μm for both
microstructures. Impedance measurements (Fig. 7) on the sintered electrolytes reveal the following behavior: In the frequency range from 102 Hz to 106 Hz the dependence of the conductivity on the frequency of both samples is very similar; at lower frequencies their conductivities differ in value, but their courses over frequency remain similar. This can also be concluded from Fig. 8, which shows a Nyquist plot of the measured specific impedances: at frequencies of 300 Hz and higher the values are in good agreement, below 300 Hz there is a difference. The diagram in Fig. 8 can be divided into different regimes which are attributed to different transport phenomena, as stated by Inaguma et al. [23]. At high frequencies the resistivity of the bulk is measured. This is shown in the inset of Fig. 8. In Fig. 7 an asymptotic behavior of the conductivity towards higher frequencies can be seen and thereby the bulk conductivity can be estimated to 4–7·10−4 S/cm at room temperature. The grain boundary resistivity can be measured at intermediate frequencies [24]. In the LLTO tape there is a gradual transition from grain boundary conduction to diffusion below 300 Hz. From there down to the lowest measured frequency the resistivity of the tape is higher than the resistivity of the massive sample. In the massive sample a sharp transition from grain boundary conduction to diffusion can be seen at 50 Hz. From
Fig. 7. Conductivities of tape and massive sample as a function of frequency.
Fig. 8. Nyquist plot of specific impedances of LLTO-tape and massive sample. 28
Solid State Ionics 328 (2018) 25–29
F. Schröckert et al.
these measurements we conclude that the transport mechanism in LLTO has not changed fundamentally, since the bulk conductivities of both samples are the same. This is plausible since the material itself has not changed, which was verified by XRD-analysis (Fig. 1). The obvious difference in conductivity must therefore be explained differently. The difference in resistivity starts to evolve at 300 Hz in the regime of grain boundary conduction. In the regime of diffusion the resistivities of the two samples exhibit again a similar dependence on frequency. From this we conclude that the cause for the different resistivity originates from hindered grain boundary conduction. This could be caused by a delayed grain boundary development during sintering. As mentioned above the density of massive samples is higher than the density of tapes up to temperatures of around 1000 °C. It is obvious that the densification in tapes starts from a lower level than in massive samples, so the formation of grain boundaries may not quite reach the quality grade of those in massive samples. Residual carbon from binder burnout [25] could also play a role. The identification of the exact cause of the lower conductivity of tapes at low frequencies and its correction should be investigated in further studies. Conclusively it can yet be stated that the conductivity of tapes and massive samples is comparable.
(2009) 911–916, https://doi.org/10.1016/j.ssi.2009.03.022. [4] H. Kawai, J. Kuwano, Lithium ion conductivity of a-site deficient perovskite solid solution La(0.67−x)Li3TiO3, J. Electrochem. Soc. 141 (1994) 78–79. [5] Y. Inaguma, L. Chen, M. Itoh, T. Nakamura, Candidate compounds with perovskite structure for high lithium ionic conductivity, Solid State Ionics 70–71 (1994) 196–202, https://doi.org/10.1016/0167-2738(94)90309-3. [6] C.W. Ban, G.M. Choi, The effect of sintering on the grain boundary conductivity of lithium lanthanum titanates, Solid State Ionics 140 (2001) 285–292, https://doi. org/10.1016/S0167-2738(01)00821-9. [7] Y. Inaguma, J. Yu, T. Katsumata, M. Itoh, Lithium lanthanum ion conductivity lithium titanate in a perovskite single crystal, J. Ceram. Soc. Jpn. 550 (1997) 548–550. [8] J. Song, All solid-state thin film batteries, in: N.J. Dudney, W.C. West, J. Nanda (Eds.), (Hrsg.), Handb. solid state Batter., World Scientific, New Jersey; Hong Kong, 2015, pp. 593–625. [9] M. Jabbari, R. Bulatova, A.I.Y. Tok, C.R.H. Bahl, E. Mitsoulis, J.H. Hattel, Ceramic tape casting: a review of current methods and trends with emphasis on rheological behaviour and flow analysis, Mater. Sci. Eng., B 212 (2016) 39–61, https://doi.org/ 10.1016/j.mseb.2016.07.011. [10] A. Roosen, Foliengießen: Verfahren zur Herstellung planarer und dreidimensionaler keramischer Strukturen, Das Keramiker-Jahrb., 1998, pp. 42–53. [11] J. Will, A. Mitterdorfer, C. Kleinlogel, D. Perednis, L.J. Gauckler, Fabrication of thin electrolytes for second-generation solid oxide fuel cells, Solid State Ionics 131 (2000) 79–96, https://doi.org/10.1016/S0167-2738(00)00624-X. [12] R. Jiménez, A. del Campo, M.L. Calzada, J. Sanz, S.D. Kobylianska, S.O. Solopan, A.G. Belous, Lithium La0.57Li0.33TiO3 perovskite and Li1.3Al0.3Ti1.7 (PO4)3 LiNASICON supported thick films electrolytes prepared by tape casting method, J. Electrochem. Soc. 163 (2016) A1653–A1659, https://doi.org/10.1149/2. 0881608jes. [13] H. Nemori, X. Shang, H. Minami, S. Mitsuoka, M. Nomura, H. Sonoki, Y. Morita, D. Mori, Y. Takeda, O. Yamamoto, N. Imanishi, Aqueous lithium-air batteries with a lithium-ion conducting solid electrolyte Li1.3Al0.5Nb0.2Ti1.3(PO4)3, Solid State Ionics 317 (2018) 136–141, https://doi.org/10.1016/j.ssi.2018.01.020. [14] N.G. Howatt, R.G. Breckenridge, J.M. Brownlow, Fabrication of thin ceramic sheets for capacitors, J. Am. Ceram. Soc. 30 (1947) 237–242, https://doi.org/10.1111/j. 1151-2916.1947.tb18889.x. [15] K.G. Schell, F. Lemke, E.C. Bucharsky, A. Hintennach, M.J. Hoffmann, Microstructure and mechanical properties of Li0.33La0.567TiO3, J. Mater. Sci. 659 (2016) 279–287, https://doi.org/10.1016/j.matlet.2008.11.044. [16] H. Hellebrand, Tape casting, in: R.J. Brook (Ed.), (Hrsg.), Process. Ceram, WILEYVCH Verlag GmbH & Co KGaA, 1996, pp. 190–265. [17] J. Böhnlein-Mauß, W. Sigmund, G. Wegner, W.H. Meyer, F. Heßel, K. Seitz, A. Roosen, The function of polymers in the tape casting of alumina, Adv. Mater. 4 (1992) 73–81, https://doi.org/10.1002/adma.19920040203. [18] Z. Fu, A. Roosen, Shrinkage of tape cast products during binder burnout, J. Am. Ceram. Soc. 98 (2015) 20–29, https://doi.org/10.1111/jace.13270. [19] M. Jabbari, J. Hattel, Modeling coupled heat and mass transfer during drying in tape casting with a simple ceramics–water system, Dry. Technol. 34 (2016) 244–253, https://doi.org/10.1080/07373937.2015.1045602. [20] C.J. Martinez, J.A. Lewis, Rheological, structural, and stress evolution of aqueous Al2O3: latex tape-cast layers, J. Th. 85 (2002) 2409–2416. [21] G.W. Scherer, Theory of drying, J. Am. Ceram. Soc. 73 (1990) 3–14, https://doi. org/10.1111/j.1151-2916.1990.tb05082.x. [22] M.J. Cima, J.A. Lewis, A.D. Devoe, Binder distribution in ceramic greenware during thermolysis, J. Am. Ceram. Soc. 72 (1989) 1192–1199, https://doi.org/10.1111/j. 1151-2916.1989.tb09707.x. [23] Y. Inaguma, C. Liquan, M. Itoh, T. Nakamura, T. Uchida, H. Ikuta, M. Wakihara, High ionic conductivity in lithium lanthanum titanate, Solid State Commun. 86 (1993) 689–693, https://doi.org/10.1016/0038-1098(93)90841-A. [24] E.C. Bucharsky, K.G. Schell, A. Hintennach, M.J. Hoffmann, Preparation and characterization of sol-gel derived high lithium ion conductive NZP-type ceramics Li1 + xAlxTi2 − X(PO4)3, Solid State Ionics 274 (2015) 77–82, https://doi.org/10. 1016/j.ssi.2015.03.009. [25] H. Yan, W.R. Cannon, D.J. Shanefield, Evolution of carbon during burnout and sintering of tape-cast aluminum nitride, J. Am. Ceram. Soc. 76 (1993) 166–172, https://doi.org/10.1111/j.1151-2916.1993.tb03702.x.
3.4. Summary and conclusions Phase-pure LLTO was derived through solid-state-synthesis and a tape casting slurry was developed. The influence of dispersant concentration, binder-plasticizer ratio and total amount of binder and plasticizer has been studied. It was found that a dispersant concentration of 2 wt% is sufficient for a well dispersed suspension. The binderplasticizer ratio was set to 2:1. Slurries with a total amount of binder and plasticizer of 15 wt% produced tapes of high quality. Furthermore the drying conditions were studied. Fast drying yielded the best tapes. The tapes and massive samples as a reference were sintered at various temperatures. Good densification could be observed at sintering temperatures of 1050 °C and above. Dense samples and tapes were characterized regarding their phase-purity, microstructure and conductivity. At low frequencies the conductivity of a LLTO-tape is lower than that of a massive sample. Nevertheless tapes and massive samples exhibit comparable properties at the whole. Acknowledgement The authors gratefully acknowledge financial support provided by Deutsche Forschungsgemeinschaft (DFG contract number HO1165/181). References [1] S. Stramare, V. Thangadurai, W. Weppner, Lithium lanthanum titanates: a review, Chem. Mater. 15 (2003) 3974–3990, https://doi.org/10.1021/cm0300516. [2] V. Thangadurai, W. Weppner, Recent progress in solid oxide and lithium ion conducting electrolytes research, Ionics 12 (2006) 81–92, https://doi.org/10.1007/ s11581-006-0013-7. [3] P. Knauth, Inorganic solid Li ion conductors: an overview, Solid State Ionics 180
29
Contents lists available at ScienceDirect
Solid State Ionics journal homepage: www.elsevier.com/locate/ssi
Tape casted thin films of solid electrolyte Lithium-Lanthanum-Titanate ⁎
T
Felix Schröckert , Nikolas Schiffmann, Ethel C. Bucharsky, Karl G. Schell, Michael J. Hoffmann Institute for Applied Materials – Ceramic Materials and Technologies, Karlsruhe Institute of Technology (KIT), Haid-und-Neu-Str. 7, 76131 Karlsruhe, Germany
A B S T R A C T
Phase-pure solid electrolyte Lithium-Lanthanum-Titanate (LLTO) has been prepared by solid-state-synthesis. Thin sheets of LLTO have been successfully prepared via tape casting and subsequent sintering. A tape casting slurry composition has been developed. The influencing parameters in slurry composition are presented. These are dispersant concentration, ratio of binder to plasticizer and total amount of binder and plasticizer. Furthermore it was found that drying conditions significantly influence tape casting results. Sintered tapes were compared to massive reference samples regarding density, phase-purity, microstructure and conductivity. Our results show that sintered tapes and massive samples exhibit comparable properties.
1. Introduction Batteries with solid electrolyte promise to overcome the disadvantages related to liquid electrolytes. There are various materials that are suitable for this application, one of them is the perovskite Li3xLa2/3−xTiO3 (LLTO) [1–3]. In literature bulk conductivities of 10−4 S/cm are reported for x = 0.1–0.12 [4,5], whereas the total conductivity is restricted by the less conductive grain boundaries [6,7]. A challenge for the implementation of solid electrolytes in batteries is the production of thin sheets. Vapor deposition methods are capable of producing thin layers [8], but are often expensive or limited in their scalability. In this study this problem is addressed by implementing the tape casting technology to cast thin tapes of LLTO. Tape casting is an established industrial production process [9,10] that can easily be scaled up from lab scale [11]. Jimenez et al. have already pursued an approach to fabricate thick films of LLTO via tape casting [12]. An interesting perspective for the application of a tape cast NASICON-type solid electrolyte Li1.3Al0.5Nb0.2Ti1.3(PO4)3 in a lithium-air battery was recently demonstrated by Nemori et al. [13]. For successful tape casting the slurry composition and its processing are essential [14]. This article reports on the slurry development and adjustments to the fabrication process for this specific problem. Green tapes are cast and then sintered to dense samples. The sintered tapes are characterized and compared to massive samples.
mixed and attrition milled with zirconia milling balls (2 mm) in isopropanol. In a previous study the mass loss due to the water binding properties of La2O3 was determined up to 1200 °C in order to weigh in correctly [15]. The powder was placed in alumina crucibles and calcined at 950 °C for 8 h. After calcination the powder was milled again in an attrition mill. The particle size of the derived powder was determined by laser diffraction (CILAS 1064, France), phase purity was verified by X-ray diffraction (XRD, D8 Advance, Bruker). XRD patterns were derived in a 2θ range from 22° to 90° with an increment of 0.01°. Cylindrical reference samples were prepared by uniaxial dry pressing and subsequent cold-isostatic pressing at 400 MPa (dimensions of sintered samples: diameter 17 mm, height 2 mm). The preparation of the tape casting slurry followed the given scheme (adapted from [16]): 1) Dissolution of dispersant (Zschimmer & Schwarz KM 3014) in ethanol 2) Addition of LLTO powder 3) Homogenization in ball mill for 24 h (10 mm zirconia balls) 4) Addition of plasticizer (polyethylene glycol 400, Merck KGaA, Germany) and binder (polyvinyl butyral Mowital B 45H, Kuraray Europe GmbH, Germany) 5) Homogenization in ball mill for 24 h and in dual asymmetrical centrifuge (SpeedMixer DAC 700.2 VAC-P, Hauschild Engineering, Germany) with degassing
2. Experimental The LLTO powder was prepared by a solid state synthesis. The starting materials Li2CO3 (purity > 99%, Alfa Aesar GmbH & Co KG, Germany), TiO2 (purity > 99.9%,Sigma-Aldrich Chemie GmbH, Germany) and La2O3 (purity > 99.5%, Merck KGaA, Germany) were ⁎
To develop a suitable tape casting slurry several recipe and process parameters were studied, these were: optimal dispersant concentration, ideal binder-plasticizer ratio, influence of total amount of binder and plasticizer at a fixed ratio and the influence of drying conditions on the green tape quality. The optimal dispersant concentration was
Corresponding author. E-mail address: [email protected] (F. Schröckert).
https://doi.org/10.1016/j.ssi.2018.10.028 Received 29 August 2018; Received in revised form 19 October 2018; Accepted 25 October 2018 Available online 18 November 2018 0167-2738/ © 2018 Elsevier B.V. All rights reserved.
Solid State Ionics 328 (2018) 25–29
F. Schröckert et al.
determined by viscosity measurements (HAAKE MARS 60, Thermo Scientific, coaxial cylinder geometry) of the suspension (only ethanol, dispersant and powder). Several dispersant concentrations with a fixed solids load (cV = 20%) were studied. Viscosities were measured by testing a shear rate ramp from 1 to 100 1/s, results are shown for shear rates of 10, 50 and 100 1/s since the values for possible shear rates in the casting gap lie within this interval. For the investigation of the binder-plasticizer ratio and the total amount of binder and plasticizer the slurries were prepared according to the above mentioned scheme. Their viscosity was then studied with a cone-plate geometry (2° angle). For evaluation of green tape quality the slurry was cast on polymer substrates (Cerapeel Q1 (S), Toray, Japan) on a laboratory film applicator (AB3000, TQC). The casting gap was 200 μm and the speed of the casting head was 5 mm/s. To study the influence of drying speed the cast tapes were dried under various conditions. Slow drying was achieved by covering the wet tape with a lid so that the atmosphere beneath the lid was enriched with evaporating solvent. Drying under ambient conditions without covering led to a moderate drying rate. Fast drying was realized by laying the substrate with the still wet tape on a heated plate (T = 70 °C). The green tapes and the reference samples were sintered at various temperatures for 1 h, the heating rate was 3 K/min. For sintering the tapes were placed on an alumina plate and covered with an alumina crucible. After sintering, the volume of the tapes was determined by measuring their area with a digital microscope (Keyence VHX-6000, Japan) and their mean thickness with a mechanical measuring sensor. For thickness determination 30 measurements in total were performed on each sample: ten measurements at three different sites of the sample. From these three sites the average and standard deviation was calculated. Additionally He-pycnometry (AccuPyc 1330, Micromeritics GmbH, Germany) has been carried out to verify the results. The density of massive samples was determined by the Archimedes method. Phase-purity of the sintered samples was checked by XRD. Tapes were placed on modeling clay and were aligned in height and angle by pressing perpendicular to the measuring plane with an axial press. Microstructural analysis was carried out by Electron backscatter diffraction (EBSD, XFlash and Esprit 2.0, Bruker, Germany). For local element analysis Energy-dispersive X-ray spectroscopy (EDS, Quantax, Bruker) was employed. The conductivity of the samples was measured by impedance spectroscopy (VersaSTAT 4, Princeton Applied Research) in potentiostatic mode with an amplitude of 10 mV in the frequency range of 10−3 to 106 Hz, the contacted area was sputtered with gold.
Fig. 1. XRD analysis on LLTO powder, tapes and massive samples.
Fig. 2. Dependence of viscosity on dispersant concentration.
3. Results and discussion LLTO powders have been successfully synthesized; in Fig. 1 the result of XRD analysis is shown. Measured diffractograms are in good agreement with the reference pattern. After milling a median particle size of d50 ≈ 200 nm was measured. 3.1. Slurry development To determine the optimal dispersant concentration the viscosity of suspensions with 1, 2 and 3 wt% (percent by weight) dispersant with respect to powder weight were studied (Fig. 2). The suspension with 2 wt% shows a minimum at all studied shear rates. Since the dispersant influences the particle-particle interaction and thereby the viscosity we can conclude that the optimal dispersant concentration can be found at the minimum of viscosity. In further results and discussion the dispersant concentration will be set to 2 wt%. In order to determine the appropriate amount of binder relative to plasticizer a study with varying binder-plasticizer ratios has been carried out; ratios were 3:1, 2:1 and 1:1. As can clearly been seen in Fig. 3 the addition of more plasticizer significantly reduces viscosity. Since all other components of the slurry remain the same, this behavior can be attributed to the interaction of the plasticizer with the binder [17]. The
Fig. 3. Influence of the binder-plasticizer ratio on the viscosity at different shear rates.
26
Solid State Ionics 328 (2018) 25–29
F. Schröckert et al.
polymer load. 3.2. Study of drying conditions Until now drying was performed under ambient conditions. Even though the slurry formulation has already been adjusted some warping of the green tapes could still be observed. Variation of the drying speed showed that the drying rate has significant influence on this effect. Drying of the tapes can be limited by two factors: evaporation of solvent at the surface and diffusion of solvent to the surface, whereby diffusion is usually the limiting factor [19,21]. If evaporation at the surface is faster than elsewhere the slurry concentration throughout the tape does not remain homogeneous [19]. If the evaporation and diffusion rate are the same no concentration gradient should be observed. By saturating the atmosphere with solvent the evaporation rate at the surface is also reduced, whereas the diffusion rate should remain unchanged. However, the result was that the slowly dried tapes show even more pronounced warping than those dried at ambient conditions. Accelerated drying was carried out by placing the substrate with the still wet, freshly cast tape on a heated plate. This process increases the diffusion rate. Tapes that were dried in this manner did not exhibit warping any more. This shows that the drying process is of significant importance for the green tape quality. By heating from below the transport of solvent to the surface and thereby the drying process as a whole is accelerated. We can think of this as a quasi-instantaneous conservation of the perfectly homogeneous distribution of components in the freshly sheared slurry.
Fig. 4. Influence of the total amount of binder and plasticizer on the viscosity at different shear rates.
measured viscosities are all within the limits for tape casting slurries [10]. Nevertheless the tapes derived from these slurries show great differences in green and sintered condition. The slurry 3:1 has large agglomerates that can already be seen in the slurry but become even more visible after casting and drying of the green tape. Examination by EDS has shown that these clumps consist of binder that has obviously not been fully dissolved. The slurry 1:1 has a much lower viscosity, just slightly above the considered minimum value of 1 Pas [10]. It showed a better processability than 3:1 and produced more homogeneous green tapes, clumps in the slurry were no longer observed. This and the significantly reduced viscosity illustrate the interaction of binder and plasticizer [17]. Nevertheless the tapes warped during the sintering process. The slurry 2:1 was found to be a practicable compromise, since it showed a mixture of the above described phenomena but neither of them to such pronounced extends. With a fixed ratio of binder to plasticizer the influence of their total amount on viscosity and tape quality has been studied. The amount of binder and plasticizer can simplified be considered the polymer-load, since these two components make up for the most of polymers in the slurry. In Fig. 4 the results of viscosity measurements are displayed. It becomes clear, that an increased polymer-load of the slurry results in an increased viscosity. When casting the two slurries with viscosities of around 1 Pas and below cracks in the green tapes could be observed. It is assumed, that the total amount of non-volatile components is not sufficient to form a continuous tape. As the slurry dries, i.e. the solvent evaporates, the remaining slurry becomes thicker. Eventually the green tape forms as the whole arrangement shrinks. The solvent can only evaporate at the (upper) surface of the slurry, on the lower side evaporation is hindered by the substrate on which the slurry is cast onto. This means, that the thickening process takes place at the upper side first [18,19]. One can think of this as floating ice floes. In the further drying process some solvent will still travel through the already dry top layer, but evaporation in the spaces between the floes, which will eventually become cracks, should be much easier. The crack formation disappears with increased polymer load. At high polymer loads warping of the green tapes to the side of the substrate was observed. A possible explanation for this phenomenon is that when the drying process starts, the dry top layer is already in its final configuration. The lower layer of the tape forms later, since it takes longer to remove the solvent from there. The bottom layer dries thereby under fixation of the casting substrate [20]. When the tape is removed from the substrate this constraint is removed and compensated by deformation. At a polymer load of 15 wt% a practicable compromise was found. No cracks were observed and the warping problem was less visible than at the higher
3.3. Sintering and characterization Tapes and massive samples were sintered at 950, 1000, 1050 and 1100 °C for 1 h in air. After sintering their densities were measured. Massive samples were measured employing the Archimedes method. The densities of the tapes were determined volumetrically. From Fig. 5 can be concluded that densification of both massive samples and tapes is possible from 1050 °C. For the tape sample sintered at 1100 °C the density has also been measured by He-pycnometry (relative density greater 98%). The result proves the validity of the density measurements by volume. The thickness of the sintered tapes was around 30 μm. Below 1050 °C the density of the tapes is lower than the density of the massive samples. This can be explained by the empty volume that is created by the outgassing polymeric components during heating. The relative densities (relative to the theoretic density of LLTO) of green bodies were just below 60% for both green tapes and pressed massive samples. For the green tapes the relative density is actually slightly
Fig. 5. Dependency of relative density on sintering temperature. 27
Solid State Ionics 328 (2018) 25–29
F. Schröckert et al.
Fig. 6. Microstructure of massive samples and tapes sintered at 1100 °C in air.
overestimated since the spaces between the LLTO particles are filled with polymer which has a density greater than air, which fills the spaces in the massive samples. This density difference is not accounted for in the calculation of the density of green tapes. When the tapes are heated up the polymeric components burn and leave empty spaces behind [22]. As long as these spaces are not densified, the density – which is measured by volume – must obviously be lower than in the massive samples. On the other hand the circumstance that sintering with insufficient green body packing still results in full densification can be taken as evidence that tape casting – as sensitive a process it might be in other aspects – is somewhat tolerant to suboptimal green body packing. In Fig. 1 can be seen, that phase-purity was remained throughout the whole process. Sintered samples, massive and tapes, are shown for a sintering temperature of 1100 °C. EBSD studies on samples sintered at 1100 °C revealed that the microstructure in tapes is comparable to that found in massive samples (Fig. 6). Note that the microstructural picture of the massive sample was derived from a polished cross section of the sample but the picture of the tape displays an EBSD study that was performed at the surface of the tape without prior mechanical treatment. The fairly difficult preparation of a polished tape cross section made this compromise necessary. The slightly rough nature of the surface explains the black spots which are simply caused by shading of the beam. Nevertheless one can judge from these pictures that the microstructure does not differ by far, the median equivalent grain diameter is around 0.5 μm for both
microstructures. Impedance measurements (Fig. 7) on the sintered electrolytes reveal the following behavior: In the frequency range from 102 Hz to 106 Hz the dependence of the conductivity on the frequency of both samples is very similar; at lower frequencies their conductivities differ in value, but their courses over frequency remain similar. This can also be concluded from Fig. 8, which shows a Nyquist plot of the measured specific impedances: at frequencies of 300 Hz and higher the values are in good agreement, below 300 Hz there is a difference. The diagram in Fig. 8 can be divided into different regimes which are attributed to different transport phenomena, as stated by Inaguma et al. [23]. At high frequencies the resistivity of the bulk is measured. This is shown in the inset of Fig. 8. In Fig. 7 an asymptotic behavior of the conductivity towards higher frequencies can be seen and thereby the bulk conductivity can be estimated to 4–7·10−4 S/cm at room temperature. The grain boundary resistivity can be measured at intermediate frequencies [24]. In the LLTO tape there is a gradual transition from grain boundary conduction to diffusion below 300 Hz. From there down to the lowest measured frequency the resistivity of the tape is higher than the resistivity of the massive sample. In the massive sample a sharp transition from grain boundary conduction to diffusion can be seen at 50 Hz. From
Fig. 7. Conductivities of tape and massive sample as a function of frequency.
Fig. 8. Nyquist plot of specific impedances of LLTO-tape and massive sample. 28
Solid State Ionics 328 (2018) 25–29
F. Schröckert et al.
these measurements we conclude that the transport mechanism in LLTO has not changed fundamentally, since the bulk conductivities of both samples are the same. This is plausible since the material itself has not changed, which was verified by XRD-analysis (Fig. 1). The obvious difference in conductivity must therefore be explained differently. The difference in resistivity starts to evolve at 300 Hz in the regime of grain boundary conduction. In the regime of diffusion the resistivities of the two samples exhibit again a similar dependence on frequency. From this we conclude that the cause for the different resistivity originates from hindered grain boundary conduction. This could be caused by a delayed grain boundary development during sintering. As mentioned above the density of massive samples is higher than the density of tapes up to temperatures of around 1000 °C. It is obvious that the densification in tapes starts from a lower level than in massive samples, so the formation of grain boundaries may not quite reach the quality grade of those in massive samples. Residual carbon from binder burnout [25] could also play a role. The identification of the exact cause of the lower conductivity of tapes at low frequencies and its correction should be investigated in further studies. Conclusively it can yet be stated that the conductivity of tapes and massive samples is comparable.
(2009) 911–916, https://doi.org/10.1016/j.ssi.2009.03.022. [4] H. Kawai, J. Kuwano, Lithium ion conductivity of a-site deficient perovskite solid solution La(0.67−x)Li3TiO3, J. Electrochem. Soc. 141 (1994) 78–79. [5] Y. Inaguma, L. Chen, M. Itoh, T. Nakamura, Candidate compounds with perovskite structure for high lithium ionic conductivity, Solid State Ionics 70–71 (1994) 196–202, https://doi.org/10.1016/0167-2738(94)90309-3. [6] C.W. Ban, G.M. Choi, The effect of sintering on the grain boundary conductivity of lithium lanthanum titanates, Solid State Ionics 140 (2001) 285–292, https://doi. org/10.1016/S0167-2738(01)00821-9. [7] Y. Inaguma, J. Yu, T. Katsumata, M. Itoh, Lithium lanthanum ion conductivity lithium titanate in a perovskite single crystal, J. Ceram. Soc. Jpn. 550 (1997) 548–550. [8] J. Song, All solid-state thin film batteries, in: N.J. Dudney, W.C. West, J. Nanda (Eds.), (Hrsg.), Handb. solid state Batter., World Scientific, New Jersey; Hong Kong, 2015, pp. 593–625. [9] M. Jabbari, R. Bulatova, A.I.Y. Tok, C.R.H. Bahl, E. Mitsoulis, J.H. Hattel, Ceramic tape casting: a review of current methods and trends with emphasis on rheological behaviour and flow analysis, Mater. Sci. Eng., B 212 (2016) 39–61, https://doi.org/ 10.1016/j.mseb.2016.07.011. [10] A. Roosen, Foliengießen: Verfahren zur Herstellung planarer und dreidimensionaler keramischer Strukturen, Das Keramiker-Jahrb., 1998, pp. 42–53. [11] J. Will, A. Mitterdorfer, C. Kleinlogel, D. Perednis, L.J. Gauckler, Fabrication of thin electrolytes for second-generation solid oxide fuel cells, Solid State Ionics 131 (2000) 79–96, https://doi.org/10.1016/S0167-2738(00)00624-X. [12] R. Jiménez, A. del Campo, M.L. Calzada, J. Sanz, S.D. Kobylianska, S.O. Solopan, A.G. Belous, Lithium La0.57Li0.33TiO3 perovskite and Li1.3Al0.3Ti1.7 (PO4)3 LiNASICON supported thick films electrolytes prepared by tape casting method, J. Electrochem. Soc. 163 (2016) A1653–A1659, https://doi.org/10.1149/2. 0881608jes. [13] H. Nemori, X. Shang, H. Minami, S. Mitsuoka, M. Nomura, H. Sonoki, Y. Morita, D. Mori, Y. Takeda, O. Yamamoto, N. Imanishi, Aqueous lithium-air batteries with a lithium-ion conducting solid electrolyte Li1.3Al0.5Nb0.2Ti1.3(PO4)3, Solid State Ionics 317 (2018) 136–141, https://doi.org/10.1016/j.ssi.2018.01.020. [14] N.G. Howatt, R.G. Breckenridge, J.M. Brownlow, Fabrication of thin ceramic sheets for capacitors, J. Am. Ceram. Soc. 30 (1947) 237–242, https://doi.org/10.1111/j. 1151-2916.1947.tb18889.x. [15] K.G. Schell, F. Lemke, E.C. Bucharsky, A. Hintennach, M.J. Hoffmann, Microstructure and mechanical properties of Li0.33La0.567TiO3, J. Mater. Sci. 659 (2016) 279–287, https://doi.org/10.1016/j.matlet.2008.11.044. [16] H. Hellebrand, Tape casting, in: R.J. Brook (Ed.), (Hrsg.), Process. Ceram, WILEYVCH Verlag GmbH & Co KGaA, 1996, pp. 190–265. [17] J. Böhnlein-Mauß, W. Sigmund, G. Wegner, W.H. Meyer, F. Heßel, K. Seitz, A. Roosen, The function of polymers in the tape casting of alumina, Adv. Mater. 4 (1992) 73–81, https://doi.org/10.1002/adma.19920040203. [18] Z. Fu, A. Roosen, Shrinkage of tape cast products during binder burnout, J. Am. Ceram. Soc. 98 (2015) 20–29, https://doi.org/10.1111/jace.13270. [19] M. Jabbari, J. Hattel, Modeling coupled heat and mass transfer during drying in tape casting with a simple ceramics–water system, Dry. Technol. 34 (2016) 244–253, https://doi.org/10.1080/07373937.2015.1045602. [20] C.J. Martinez, J.A. Lewis, Rheological, structural, and stress evolution of aqueous Al2O3: latex tape-cast layers, J. Th. 85 (2002) 2409–2416. [21] G.W. Scherer, Theory of drying, J. Am. Ceram. Soc. 73 (1990) 3–14, https://doi. org/10.1111/j.1151-2916.1990.tb05082.x. [22] M.J. Cima, J.A. Lewis, A.D. Devoe, Binder distribution in ceramic greenware during thermolysis, J. Am. Ceram. Soc. 72 (1989) 1192–1199, https://doi.org/10.1111/j. 1151-2916.1989.tb09707.x. [23] Y. Inaguma, C. Liquan, M. Itoh, T. Nakamura, T. Uchida, H. Ikuta, M. Wakihara, High ionic conductivity in lithium lanthanum titanate, Solid State Commun. 86 (1993) 689–693, https://doi.org/10.1016/0038-1098(93)90841-A. [24] E.C. Bucharsky, K.G. Schell, A. Hintennach, M.J. Hoffmann, Preparation and characterization of sol-gel derived high lithium ion conductive NZP-type ceramics Li1 + xAlxTi2 − X(PO4)3, Solid State Ionics 274 (2015) 77–82, https://doi.org/10. 1016/j.ssi.2015.03.009. [25] H. Yan, W.R. Cannon, D.J. Shanefield, Evolution of carbon during burnout and sintering of tape-cast aluminum nitride, J. Am. Ceram. Soc. 76 (1993) 166–172, https://doi.org/10.1111/j.1151-2916.1993.tb03702.x.
3.4. Summary and conclusions Phase-pure LLTO was derived through solid-state-synthesis and a tape casting slurry was developed. The influence of dispersant concentration, binder-plasticizer ratio and total amount of binder and plasticizer has been studied. It was found that a dispersant concentration of 2 wt% is sufficient for a well dispersed suspension. The binderplasticizer ratio was set to 2:1. Slurries with a total amount of binder and plasticizer of 15 wt% produced tapes of high quality. Furthermore the drying conditions were studied. Fast drying yielded the best tapes. The tapes and massive samples as a reference were sintered at various temperatures. Good densification could be observed at sintering temperatures of 1050 °C and above. Dense samples and tapes were characterized regarding their phase-purity, microstructure and conductivity. At low frequencies the conductivity of a LLTO-tape is lower than that of a massive sample. Nevertheless tapes and massive samples exhibit comparable properties at the whole. Acknowledgement The authors gratefully acknowledge financial support provided by Deutsche Forschungsgemeinschaft (DFG contract number HO1165/181). References [1] S. Stramare, V. Thangadurai, W. Weppner, Lithium lanthanum titanates: a review, Chem. Mater. 15 (2003) 3974–3990, https://doi.org/10.1021/cm0300516. [2] V. Thangadurai, W. Weppner, Recent progress in solid oxide and lithium ion conducting electrolytes research, Ionics 12 (2006) 81–92, https://doi.org/10.1007/ s11581-006-0013-7. [3] P. Knauth, Inorganic solid Li ion conductors: an overview, Solid State Ionics 180
29