Processing and oxygen permeability of asymmetric ferrite-based ceramic membranes

Processing and oxygen permeability of asymmetric ferrite-based ceramic membranes

Available online at www.sciencedirect.com Solid State Ionics 179 (2008) 61 – 65 www.elsevier.com/locate/ssi Processing and oxygen permeability of as...

770KB Sizes 1 Downloads 98 Views

Available online at www.sciencedirect.com

Solid State Ionics 179 (2008) 61 – 65 www.elsevier.com/locate/ssi

Processing and oxygen permeability of asymmetric ferrite-based ceramic membranes A.V. Kovalevsky a,⁎, V.V. Kharton a , F.M.M. Snijkers b , J.F.C. Cooymans b , J.J. Luyten b , J.R. Frade a a

Department of Ceramics and Glass Engineering, CICECO, University of Aveiro, 3810-193 Aveiro, Portugal b Materials Department, Flemish Institute for Technological Research (VITO), 2400 Mol, Belgium

Abstract In order to assess membrane architecture-related effects on the oxygen transport through ferrite-based mixed conductors, a series of planar ceramic membranes with dense layers made of the dual-phase (SrFeO3 − δ)0.7(SrAl2O4)0.3 composite were appraised at 1023–1223 K under oxidizing conditions. The asymmetric membranes with porous La0.5Sr0.5FeO3 − δ and (SrFeO3 − δ)0.7(SrAl2O4)0.3 supports were fabricated by a two-stage compaction procedure using various pore-forming additives and sintering at 1623–1723 K. Analysis of the oxygen permeation fluxes through model symmetric membranes showed significant limiting role of the composite surface oxygen exchange. At temperatures above 1173 K, a substantially improved performance was observed for asymmetric self-supported composite membranes with the dense layer thickness of 0.12 mm, surface-modified with (SrFeO3 − δ)0.7(SrAl2O4)0.3–Pt mixture. At 1023–1173 K, higher oxygen fluxes were achieved using perovskitetype La0.5Sr0.5FeO3 − δ as the porous support material. © 2007 Elsevier B.V. All rights reserved. Keywords: Oxygen permeation; Mixed ionic–electronic conductor; Asymmetric ceramic membrane; Ferrite; Composite

1. Introduction Technologies for high-purity oxygen separation from air and partial oxidation of light hydrocarbons using dense mixedconducting membranes have high potential for the gas and energy industries [1,2]. However, the materials exhibiting highest oxygen permeation are often unstable in reducing atmospheres, such as CH4 conversion products. At the same time, the phases thermodynamically stable in the whole necessary range of oxygen partial pressures, p(O2), possess a low oxygen diffusivity [1,3,4]. While the search for novel materials with improved stability and transport properties is still a major challenge, research effort was also directed towards optimization of the membrane architecture. One promising concept relates to the asymmetric configuration comprising a thin dense highly-permeable layer applied on a porous support [2,5–8]. Depending upon thickness of the gas-tight layer in such membranes, the performance-determining factors ⁎ Corresponding author. Tel.: +351 234 370263; fax: +351 234 425300. E-mail address: [email protected] (A.V. Kovalevsky). 0167-2738/$ - see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.ssi.2007.12.027

may be associated with bulk ambipolar conductivity, surface exchange, and/or oxygen diffusion in the porous layers. Proper control of the contributions caused by the latter two processes makes it possible to achieve kinetic stabilization of the mixedconducting materials under high oxygen chemical potential gradients in combination with substantially high oxygen fluxes [1,4]. Analysis of experimental data on the oxygen permeability, thermal expansion and stability of known mixed conductors (e.g. [1,4,9–12] and references cited) suggests that an optimum combination of these properties may be expected for Fe-containing materials, particularly for the perovskite-related compounds. The present work was therefore centered on the fabrication and testing of model ferrite-based asymmetric membranes in order to assess the effects related to the membrane architecture. 2. Experimental The powder of La0.5Sr0.5FeO3 − δ was synthesized by solidstate reaction from high-purity SrCO3, Fe(NO3)3·9H2O and La2O3, which was previously annealed in air at 1473 K for 2 h

62

A.V. Kovalevsky et al. / Solid State Ionics 179 (2008) 61–65

Fig. 1. SEM micrographs of (SrFeO3 − δ)0.7(SrAl2O4)0.3–La0.5Sr0.5FeO3 − δ (A,B) and (SrFeO3 − δ)0.7(SrAl2O4)0.3-based asymmetric membranes (C,D). DL, S and AL correspond to dense (SrFeO3 − δ)0.7(SrAl2O4)0.3 layer, porous La0.5Sr0.5FeO3 − δ or (SrFeO3 − δ)0.7(SrAl2O4)0.3 support, and activating layer, respectively.

to remove adsorbates. After thermal decomposition of nitrate and carbonate components, the reaction was conducted in air at 1370 K for 25 h with multiple intermediate grindings. In the case of (SrFeO3 − δ)0.7(SrAl2O4)0.3 composite, a commercial powder prepared by the combustion spray pyrolysis (Praxair Surface Technologies, Seattle) was used. A brief clarification of the materials choice and description of the processing conditions for porous supports and asymmetric membranes are given below. To analyze the pore size distribution by mercury intrusion porosimetry and dilatometric studies, a series of porous ceramics were prepared using the same processing conditions as for asymmetric membranes. Also, in order to assess the factors inherent to asymmetric configuration, additional symmetric (SrFeO3 − δ)0.7(SrAl2O4)0.3 membranes with density 92–96% of theoretical were fabricated by uniaxial pressing at 80–200 MPa and subsequent sintering in air at 1623 K for 5 h. General characterization included X-ray diffraction (XRD) analysis, scanning electron microscopy coupled with energydispersive spectroscopy (SEM/EDS), picnometry, dilatometry, gas-tightness control and determination of the steady-state oxygen permeation fluxes; description of the experimental procedures and equipment is found elsewhere (Refs. [4,11–16] and references cited). For all membranes studied in this work, zero level of physical leakages was confirmed under the total pressure gradient of 2–4 atm. The data on the oxygen permeation presented below correspond to the membrane feed-side oxygen partial pressure (p2) equal to 0.21 atm (atmospheric air). In

order to distinguish and/or to decrease the limiting effects of exchange processes, several membranes were surface-modified. The surface activation procedures included deposition of porous layer consisting of (SrFeO3 − δ)0.7(SrAl2O4)0.3–Pt mixture (50:50 wt.%), sintering at 1373 K for 2 h, impregnation with Pr(NO3)3 5H2O solution in ethanol, drying and final annealing at 1393 K. In the case of asymmetric membranes, only the dense layer surface was modified via the deposition of activating layer (AL, Fig. 1A,C); in the course of oxygen permeability measurements, this surface was placed at the permeate side where the oxygen partial pressure (p1) is lower than atmospheric. 3. Results and discussion Table 1 lists the values of oxygen permeation fluxes, thermal expansion coefficients (TECs) and approximate stability limits in reducing atmospheres for selected ferrite-based mixed conductors. Due to insufficient concentration of the ionic charge carriers and unfavorable structural features, brownmillerite-type CaFe0.5Al0.5O2.5 + δ and intergrowth Sr4Fe6O13 ± δ demonstrate the lowest level of oxygen permeability. Perovskitetype phases containing cobalt are among the most permeable materials, but also possess unfavorably high thermal expansion and low thermodynamic stability in reducing conditions, hampering their direct application in the reactors for partial oxidation of hydrocarbons. Among ferrites, the highest permeation

A.V. Kovalevsky et al. / Solid State Ionics 179 (2008) 61–65

63

Table 1 Oxygen permeation fluxes, average TECs and low-p(O2) stability limits of selected ferrite-based ceramic membrane materials Membrane material

SrFe0.2Co0.8O3 − δ La0.5Sr0.5FeO3 − δ SrFe0.7Al0.3O3 − δ (SrFeO3 − δ)0.7(SrAl2O4)0.3 La0.3Sr0.7Fe0.8Ti0.2O3 − δ La0.8Sr0.2Fe0.8Co0.2O3 − δ Sr0.97Fe0.8Ti0.2O3 − δ Sr4Fe6O13 ± δ CaFe0.5Al0.5O2.5 + δ

Phase composition⁎

C R C C + SA C R C L B

j(O2)⁎⁎, mol/(cm2 × s) −7

4.2 × 10 8.7 × 10− 8 5.7 × 10− 8 4.2 × 10− 8 2.7 × 10− 8 3.2 × 10− 9 1.1 × 10− 7 1.7 × 10− 10 8.6 × 10− 11

Average thermal expansion coefficient in air T, K

α × 106, K− 1

300–700/800–1100 350–950/950–1310 370–920/920–1220 350–920/950–1230 400–790/790–1260 300–1050 300–780/780–1040 770–1100 370–850/930–1300

18.8/29.4 12.4/23.7 15.4/23.0 13.2/24.7 13.6/21.7 12.9 13.8/27.0 10.8 16.7/12.6

Low-p(O2) stability limits at 1173 K (atm)

Ref.

9.9 × 10− 13 ∼ 5 × 10− 19 ⁎⁎⁎ 4.2 × 10− 19 – 3.0 × 10− 19 – – 2.5 × 10− 17 5.1 × 10− 18

[17] [13,18–20] [15] [14] [21] [22] [16] [23] [12]

⁎B, SA, C and R correspond to the brownmillerite, monoclinic SrAl2O4, cubic perovskite and rhombohedrally-distorted perovskite phases, respectively; L relates to orthorhombic intergrowth structure with the space group Iba2. ⁎⁎Oxygen permeation flux through 1.00 mm thick membrane under the oxygen partial pressure gradient of 0.21/0.066 atm. ⁎⁎⁎The values was roughly estimated as an average between the stability limits for SrFeO3 − δ and La0.5Sr0.5FeO3 − δ at 1173 K.

fluxes are observed for strontium-deficient Sr0.97Fe(Ti)O3 − δ [16], SrFe(Al)O3 − δ [15] and La0.5Sr0.5FeO3 − δ [13], having disordered perovskite-type lattices where the oxygen diffusion pathways form a three-dimensional network. At the same time, these materials display a substantial thermodynamic stability at low p(O2), which, in turn, may be improved using suitable membrane architectures to provide kinetic stabilization under operation conditions [1,4]. The results of dilatometric studies (Table 1) show that the ferrite-based materials with maximum oxygen permeability, such as Sr0.97Fe0.8Ti0.2O3 − δ, exhibit also excessively high TECs. This correlation is in agreement with the phenomenological theory of ionic transport [24] and can be explained by the increase in the point defects mobility when lattice expands. The extremely high expansion of Sr0.97Fe(Ti)O3 − δ makes it rather impossible to use these materials to form a thermally-stable dense layer in asymmetric membranes. Two other promising parent compositions, SrFe0.7Al0.3O3 − δ and La0.5Sr0.5FeO3 − δ, exhibit significantly lower and quite similar TECs (Table 1). Moreover, further optimization of SrFe(Al)O3 − δ is possible by varying the Fe:Al concentration ratio and increasing the Sr deficiency [15]. Previous works [14,25] showed that combination of strontiumdeficient Sr1 − xFe(Al)O3 − δ perovskite and monoclinic SrAl2O4 phases makes it possible to decrease thermal expansion and to improve mechanical strength of the ceramics, whereas the oxygen permeation is governed predominantly by the perovskite component. A promising combination of ionic transport and thermomechanical properties was found for (SrFeO3 − δ)0.7(SrAl2O4)0.3 composite [25], for which the composition of perovskite phase is close to Sr0.9Fe0.8Al0.2O3 − δ [14]. Consequently, perovskite-type La0.5Sr0.5FeO3 − δ and (SrFeO3 − δ)0.7(SrAl2O4)0.3 composite were selected as possible candidate materials for asymmetric membranes. Notice that dense composite ceramics can be sintered at temperatures approximately 150 K lower with respect to lanthanum–strontium ferrite [13,14,25]. The latter is, again, beneficial for processing of asymmetric membranes with dense (SrFeO3 − δ)0.7(SrAl2O4)0.3 layer supported by porous La0.5Sr0.5 FeO3 − δ, enabling co-sintering of these materials and preserving a high porosity of the support.

The planar asymmetric membranes were fabricated by uniaxial compaction in two steps. A part of La0.5Sr0.5FeO3 − δ or (SrFeO3 − δ)0.7(SrAl2O4)0.3 powder containing pore-forming agent (graphite or maize starch, correspondingly) was introduced to a mold and uniaxially compacted at 20–50 MPa. Then a known amount of milled (SrFeO3 − δ)0.7(SrAl2O4)0.3 powder without pore-forming agent was added on the top of the first compact, and the whole structure was pressed at 65–150 MPa. The choice of the composition and amount of the pore-forming agent for each material was based on empirical tests aimed to obtain the porous supports having appropriate mechanical strength, optimal microstructure and low resistance to the gas flow. The compacted structures were then sintered at 1623– 1723 K for 1–2 h to produce the asymmetric membranes, where (SrFeO3 − δ)0.7(SrAl2O4)0.3 dense layer is supported by porous La0.5Sr0.5FeO3 − δ or (SrFeO3 − δ)0.7(SrAl2O4)0.3 ceramics. For the sake of simplicity, the above membranes are further denoted as SFSA-LSF and SFSA-2, respectively. The average open porosity of the supporting layers, determined by the mercury intrusion porosimetry, was 20–30%. As expected, the dilatometric studies demonstrated that dense and porous (SrFeO3 − δ)0.7(SrAl2O4)0.3 composite ceramics exhibit very similar expansion upon heating (Fig. 2). The minor difference in the thermal expansion at 900–1000 K (b0.06%) results from a faster equilibration of the porous sample with atmospheric oxygen, thus providing a locally higher chemical contribution to the ΔL/L0 values in this temperature range under continuous heating conditions. At higher temperatures, the difference in oxygen nonstoichiometry values becomes insignificant; both porous and dense materials display again a linear behavior, almost independent of porosity. Similar behavior of the dense and porous samples was also observed for La0.5Sr0.5FeO3 − δ. The difference in thermal expansion between dense (SrFeO3 − δ)0.7(SrAl2O4)0.3 and La0.5Sr0.5FeO3 − δ ceramics is, however, slightly higher. For SFSA-LSF membranes, the latter may provoke formation of microcracks in the vicinity of dense layer/porous support boundary (Fig. 1B). Note that no analogous similar effects were observed in the case of SFSA-2

64

A.V. Kovalevsky et al. / Solid State Ionics 179 (2008) 61–65

Fig. 2. Dilatometric curves of dense and porous (SrFeO3 − δ)0.7(SrAl2O4)0.3 and La0.5Sr0.5FeO3 − δ ceramics in air.

(Fig. 1C). Therefore, in order to suppress crack propagation in the membrane structure, the dense and porous layers in SFSALSF membranes were both made thicker compared to SFSA-2. This approach still makes it possible to estimate the impact of membrane architecture on the permeation-limiting steps, as described below. The thicknesses of dense and porous layers determined from SEM micrographs were 0.5 and 1.4 mm for SFSA-LSF, and 0.12 and 0.42 mm for SFSA-2, respectively. The data on oxygen permeation through 0.60 mm thick (SrFeO3 − δ)0.7(SrAl2O4)0.3 ceramics with and without surface modification (Fig. 3) unambiguously indicate that the overall transport is strongly affected by exchange processes at the membrane/gas interface. In particular, the permeation flux through the surface-activated membrane at 1123 K and p1 = 6.6 × 10− 2 atm is more than 2.7 times higher than that for non-modified ceramics. This difference increases with increasing

Fig. 3. Temperature dependence of the oxygen permeation fluxes through symmetric and asymmetric ferrite-based membranes. The estimated activation energies for oxygen permeation are given in the legends.

p(O2) gradient and with decreasing temperature (Fig. 4), which indicates, in particular, that the activation energy (Ea) for surface oxygen exchange is higher than that for the bulk ambipolar conductivity. Similar behavior is well-known for numerous mixed-conducting materials (e.g. [1,2] and references therein]). The positive influence of porous (SrFeO3 − δ)0.7(SrAl2O4)0.3–Pt layers originates from both the relatively high catalytic activity of platinum and effective enlargement of the membrane surface. The activation leads to a substantial decrease in the apparent Ea values, from 132 down to 93 kJ/mol at 1073–1223 K (Fig. 3). It is noteworthy also that the exchange limitations to oxygen transport may completely inhibit positive effects expected on decreasing thickness of the membrane dense layers. A considerable improvement in the oxygen permeation fluxes at 1073–1173 K was achieved when using the SFSALSF asymmetric membrane concept (Figs. 3 and 4). The corresponding activation energy decreased down to 71 kJ/ mol at 1073–1223 K and became lower than that for 0.60 mm thick symmetric composite membrane after surface modification. The activation with (SrFeO3 − δ)0.7(SrAl2O4)0.3–Pt mixture seems, hence, not effective enough to suppress surface limitations; porous La0.5Sr0.5FeO3 − δ has a higher catalytic activity in the

Fig. 4. Oxygen partial pressure dependence of the oxygen permeation fluxes through 0.60 mm thick (SrFeO3 − δ)0.7(SrAl2O4)0.3 symmetric and asymmetric SFSA-LSF (ddense = 0.50 mm, dporous = 1.40 mm) and SFSA-2 (ddense = 0.12 mm, dporous = 0.42 mm) membranes.

A.V. Kovalevsky et al. / Solid State Ionics 179 (2008) 61–65

intermediate-temperature range. On heating up to 1223 K, SFSA-LSF and symmetric surface-modified membranes with comparable thicknesses of the dense layers display similar oxygen fluxes. This suggests an improved performance of the Ptcontaining activating agent and a higher contribution of the bulk ion diffusion. The latter also explains high oxygen permeation fluxes through SFSA-2 membrane at 1223 K (Fig. 4). Nonetheless, the influence of other contributions to the overall oxygen-transport kinetics remains significant. In particular, at 1223 K the fluxes through SFSA-2 are 2.0–2.2 times higher than those for the symmetric surface-modified membrane; for permeability governed exclusively by bulk ion diffusion in the dense layer, a 5-fold increase should be observed. Finally, the results obtained in this work confirm that a high performance of ferrite-based membranes may be achieved using the asymmetric concept. Despite relatively large thickness of the dense layers, La0.5Sr0.5FeO3 − δ supported (SrFeO3 − δ)0.7(SrAl2O4)0.3 membrane performs at a considerably good level at 1073–1173 K, reaching values close to the ideal intrinsic materials performance. For (SrFeO3 − δ)0.7 (SrAl 2O4) 0.3-based asymmetric membrane, a reasonable improvement of transport properties was only observed at temperatures above 1150 K. For the intermediate-temperature range, further optimization of architecture and processing conditions for SFSA-LSF asymmetric membrane is still necessary in order to avoid mechanical stresses produced by the minor thermal expansion mismatch, simultaneously decreasing thicknesses of the dense and porous components. One possible approach for these goals relates to the introduction of intermediate layers with graded porosity and preferable pore orientation. Furthermore, for both SFSA-LSF and SFSA-2 membranes, the challenge of improving of the surface exchange kinetics still remains; the efforts should be focused not only on the search for novel catalytically active materials, but also on optimization of the surface morphology and/or microstructure (e.g. by ultrasonic treatment in the solutions containing appropriate components, etc.) or incorporation of nano-sized catalyst particles into the pores of supporting layer. Acknowledgements This work was supported by the FCT, Portugal (projects POCI/CTM/58570/2004 and SFRH/BPD/15003/2004). Experimental assistance made by F. Maxim and A. Markov is gratefully acknowledged.

65

References [1] H.J.M. Bouwmeester, A.J. Burggraaf, Dense ceramic membranes for oxygen separation, in: A.J. Burggraaf, L. Cot (Eds.), Fundamentals of Inorganic Membrane Science and Technology, Elsevier Science, Amsterdam, 1996, p. 435. [2] A. Thursfield, I.S. Metcalfe, J. Mater. Chem. 14 (2004) 2475. [3] T. Nakamura, G. Petzow, L.J. Gauckler, Mater. Res. Bull. 14 (1979) 649. [4] V.V. Kharton, A.A. Yaremchenko, A.A. Valente, V.A. Sobyanin, V.D. Belyaev, G.L. Semin, S.A. Veniaminov, E.V. Tsipis, A.L. Shaula, J.R. Frade, J. Rocha, Solid State Ionics 176 (2005) 781. [5] M.F. Carolan, P.N. Dyer, US Patent 5534471 (1996). [6] G. Etchegoyen, T. Chartier, P. Del-Gallo, J. Eur. Ceram. Soc. 26 (2006) 2807. [7] W. Jin, S. Li, P. Huang, N. Xu, J. Shi, J. Membr. Sci. 185 (2001) 237. [8] L.M. van der Haar, H. Verweij, J. Membr. Sci. 180 (2000) 147. [9] H.U. Anderson, Solid State Ionics 52 (1992) 33. [10] B.A. van Hassel, T. Kawada, N. Sakai, H. Yokokawa, M. Dokiya, H.J.M. Bouwmeester, Solid State Ionics 66 (1993) 295. [11] V.V. Kharton, A.A. Yaremchenko, A.V. Kovalevsky, A.P. Viskup, E.N. Naumovich, P.F. Kerko, J. Membr. Sci. 163 (1999) 307. [12] V.V. Kharton, I.P. Marozau, N.P. Vyshatko, A.L. Shaula, A.P. Viskup, E.N. Naumovich, F.M.B. Marques, Mater. Res. Bull. 38 (2003) 773. [13] E.V. Tsipis, M.V. Patrakeev, V.V. Kharton, A.A. Yaremchenko, G.C. Mather, A.L. Shaula, I.A. Leonidov, V.L. Kozhevnikov, J.R. Frade, Solid State Sci. 7 (2005) 355. [14] V.V. Kharton, A.V. Kovalevsky, A.A. Yaremchenko, F.M.M. Snijkers, J.F.C. Cooymans, J.J. Luyten, A.A. Markov, J.R. Frade, F.M.B. Marques, J. Solid State Electrochem. 10 (2006) 663. [15] V.V. Kharton, A.L. Shaula, F.M.M. Snijkers, J.F.C. Cooymans, J.J. Luyten, A.A. Yaremchenko, A.A. Valente, E.V. Tsipis, J.R. Frade, F.M.B. Marques, J. Rocha, J. Membr. Sci. 252 (2005) 215. [16] V.V. Kharton, A.V. Kovalevsky, A.P. Viskup, J.R. Jurado, F.M. Figueiredo, E.N. Naumovich, J.R. Frade, J. Solid State Chem. 156 (2001) 437. [17] V.V. Kharton, A.V. Kovalevsky, M. Avdeev, E.V. Tsipis, M.V. Patrakeev, A.A. Yaremchenko, E.N. Naumovich, J.R. Frade, Chem. Mater. 19 (2007) 2027. [18] T. Nakamura, G. Petzow, L.J. Gauckler, Mater. Res. Bull. 14 (1979) 649. [19] T. Katsura, K. Kitayama, T. Sugihara, N. Kimizuka, Bull. Chem. Soc. Jpn. 48 (1975) 1809. [20] M.V. Patrakeev, I.A. Leonidov, V.L. Kozhevnikov, V.V. Kharton, Solid State Sci. 6 (2004) 907. [21] A.A. Yaremchenko, V.V. Kharton, A.L. Shaula, M.V. Patrakeev, F.M.B. Marques, J. Eur. Ceram. Soc. 25 (2005) 2603. [22] V.V. Kharton, A.V. Kovalevsky, A.P. Viskup, A.L. Shaula, F.M. Figueiredo, E.N. Naumovich, F.M.B. Marques, Solid State Ionics 160 (2003) 247. [23] M.Yu. Avdeev, M.V. Patrakeev, V.V. Kharton, J.R. Frade, J. Solid State Electrochem. 6 (2002) 217. [24] V.N. Chebotin, Physical Chemistry of Solids, Khimiya, Moscow, 1982. [25] A.A. Yaremchenko, V.V. Kharton, A.L. Shaula, F.M.M. Snijkers, J.F.C. Cooymans, J.J. Luyten, F.M.B. Marques, J. Electrochem. Soc. 153 (2006) J50.