Fabrication of La, -,Sr,CoO, -thin layers on porous supports by a polymeric sol-gel process Chunhua Chen,? Henny J. M. Bouwmeester," Henk Kruidhof, Johan E. ten Elshof and Anthony J. Burggraaf Laboratory of Inorganic Materials Science, Faculty of Chemical Technology, University of Twente, P.O. Box 21 7, 7500 AE Enschede, the Netherlands A polymeric sol-gel process was developed to fabricate porous thin layers of the perovskite-type Lao.3Sro.7Co03 for membrane applications. A spin-coating technique was used for deposition of the layer on porous a-and y-Al,03 supports. Both supported and non-supported membranes were characterized by means of thermal analysis, X-ray powder diffraction, particle size analysis, scanning electron microscopy and permporometry. The main phase formed upon heating in air above 400 "C is cubic perovskite, although traces of SrCO, and of an unknown phase can be observed up to 800 "C.Above this temperature, strong chemical interaction occurs between the deposited perovskite layer and the support material.The effects of the sintering temperature and the type of the La-precursor on the pore-size distribution were investigated. The perovskite-type oxides La, -,Sr,Co03 -(LSCO) have attracted much attention in recent years because of their potential uses as electrode materials in solid oxide fuel cells (SOFCs),'v2 gas sensors3 and oxygen separation membrane^.^ For these applications, the use of thin layers of LSCO is beneficial because it can decrease the electrode resistance, favour the miniaturization of a gas sensor and optimize the oxygen permeability. Furthermore, if the layer is porous, it can also increase the triple-phase boundary in SOFC electrodes and the active surface area in gas-sensing elements. For oxygen separation membranes the thin barrier layer should be dense, i.e.free from connected-through porosity, and crack free. The layer needs to be supported by a porous substrate if its thickness is less than ca. 150 pm in order to obtain sufficient mechanical strength. The successful application of the electro- chemical vapour deposition (ECVD) technique for the prep- aration of dense thin layers of yttria-stabilized zirconia (YSZ) on porous alumina supports has been reported.'^^ However, it is not easy to find a good substrate material for LSCO layers owing to its high reactivity with many substrate materials at high temperatures and thermal mismatch.It is therefore con- sidered advantageous to coat a substrate with a thin porous layer of LSCO, as an intermediate layer, prior to deposition of a dense top layer with similar or adapted composition. In the present paper, the preparation and characterization of such porous (intermediate) LSCO layers are described. The depos- ition of a dense layer on top of these will be a subject for future research. Sol-gel methods can be divided into colloidal and polymeric types. Although both approaches have been applied success- fully in the preparation of ceramic porous membranes, the polymeric approach is more often used for microporous and multi-component systems.798 The polymeric route is supposed to produce an inorganic network (gel) by gelation which results in an appropriately viscous sol preventing the compo- nents from penetration into the pores of the support.Nevertheless, there are a few reports on the deposition of thin oxide layers on porous supports by this polymeric sol-gel process.' In many cases crack-free films can be obtained when the thickness is less than 0.5 pm." In this study, a polymeric sol-gel process has been developed for the fabrication of our perovskite Lao.3Sro,7Co0, layers. Two types of porous alum- t Present address: Laboratory of Applied Inorganic Chemistry, Delft University of Technology, Julianalaan 136, 2628 BL Delft, the Netherlands.ina supports were used, both with well defined sharp pore-size distributions. As suitable metal alkoxide precursors of cobalt, strontium and lanthanum were not available, we first synthe- sized methoxyethoxides of these three elements with hydrolys- abilities suitable for easy manipulation. Details will be reported elsewhere." A spin-coating technique was adopted to prepare the thin LSCO layers. Experimenta1 The synthesis procedure is schematically shown in Fig. 1. Methoxyethoxide solutions of cobalt, strontium and lantha- num were synthesized separately according to the method developed in our laboratory and described elsewhere." Two kinds of lanthanum precursors were used in this study for comparison.The first is a distilled methoxyethanol (MOE) solution of hydrated lanthanum nitrate, as shown in Fig. 1. The second precursor is lanthanum methoxyethoxide synthe- sized from the reaction between lanthanum nitrate and stron- dissolved distilled, dissolved, in MOE repeat Fig. 1 Sol-gel La, -.Sr,CoO, mixed solution . I partial hydrolysis yjspin coating Ifilm drying and firing perovs kite synthesis flow chart for the preparation of -6 perovskite films J. Muter. Chem., 1996, 6(5), 815-819 815 tium methoxyethoxide Solutions containing La, Sr and Co were mixed in a 0 3 0 7 1 molar ratio to a final concentration of ca 0 2 mol dm-, Co The mixed solution was heated at reflux in a three-necked flask under pure nitrogen at 125°C for 3-5 h and subsequently cooled to room temperature to serve as a stock precursor solution If well sealed, the solution has a shelf-life of several months A few millilitres of the stock precursor solution was trans- ferred into a small glass bottle MOE and distilled water were added, under stirring, in order to partially hydrolyse the precursor solution into a sol Usually, the volume ratios of the precursor solution methoxyethanol water were kept at 1 1 03 The viscosity of the sol increased with time due to gelation About 1-3 h after the aforementioned addition, the support was coated with the sol A few droplets of the sol were dripped onto the middle of a porous alumina substrate (diameter 12mmx2mm) which was held in a spin coater (model PMlOlD, Headway Research Inc, Garland, TX) placed in a class 100 clean hood (Down-flow unit, The Netherlands) Following this procedure, however, the layer thus obtained was usually thicker at the centre, and thinner towards the edge Therefore, the whole surface was flooded with the sol prior to spinning, in order to produce a uniform layer All results reported here refer to films prepared by the latter procedure The spinning was performed for 40 s with a speed of 4000 rpm The porous substrates used in the experiments were made of a-Al,O, with a pore size of ca 160nm, and lanthanum- doped y-Al,03 with a pore size of ca 17 nm l2 ', Both types of substrates were heat-treated at temperature > 1100 "C before spin coating in order to obtain a thermally stable pore size The supported gel layers were dried in air at room tempera- ture for several hours and successively heated to either 500, 600, 700 or 800°C for 5 h with a heating/cooling rate of 30°C h-l Sometimes the aforementioned steps were repeated to increase the layer thickness The surface and cross-section morphologies of 800 "C sintered samples were examined with a scanning electron microscope (SEM, Hitachi S800) Elemental analysis was performed on some selected areas by energy dispersive analysis by X-rays (EDAX) Direct analysis by standard X-ray diffraction (XRD) was not possible due to the fact that the layers were too thin to obtain reasonable intensities To enable XRD analysis, some samples were pre- pared with very thick layers, both on a-A1203 and y-A1203 supports, by using a highly concentrated sol [derived from La(NO,),] in the spinning process After sintering at 800°C for 5 h, some powder was scratched from the surface of the sintered layer for a step-scanning XRD analysis (Philips PW 1710, Ni-filtered Cu-Ka radiation) with a 28 step of 0 02" and a counting time of 5 s per interval In addition, a non-supported membrane was prepared by drying a partially hydrolysed sol in a glass dish under ambient conditions into a xerogel The resulting gel was crushed into a fine powder This powder was characterized by a TG-DSC analyser (Stanton Redcroft PL-STA 625) in the temperature range 25-1300°C at a heating/cooling rate of 10"Cmin-l in flowing air A step scanning XRD analysis with a 20 step of 0 04" (5 s per interval) was also applied to the powder samples heat treated at 600, 700 and 800 "C A laser diffraction particle- size distribution analyser (HORIBA, LA-500) was used to analyse the non-heat-treated xerogel powder after ultrasonic dispersion in distilled water The pore-size distribution of deposited layers was examined by means of permporometry In this method, a mixture of cyclohexane vapour, oxygen and nitrogen is used to flush one side of a supported membrane which is sealed with O-rings in a diffusion cell Initially, the gas mixture is saturated with cyclohexane so that all the pores in the membrane become filled with liquid cyclohexane owing to capillary condensation 816 J Mater Chem, 1996, 6(5),815-819 No gas can permeate through the membrane at this stage By decreasing stepwise the relative pressure of the cyclohexane in the mixture, pores with radii equal to or larger than the corresponding Kelvin radii are opened and oxygen diffuses through the membrane to the other side The relation between the accumulated oxygen permeability and the relative pressure of cyclohexane provides information about the pore-size distri- bution of the membrane under consideration For further details on the principles and experimental set-up of this technique, see ref 12 Results and Discussion Non-supported membrane Fig 2 shows the TG-DSC curves of the La, 3Sro ,Coo,-, gel powder A senes of mass losses and several exothermic peaks appear below 400"C, which can be attnbuted to the evapor- ation and burn-out of organic components (such as residual solvent MOE, acetate groups and unhydrolysed alkoxy groups) and dehydration of hydroxides existing in the gel The large exothermic peak around 400"C corresponds to the formation of the perovskite phase, as was confirmed by XRD Based on the XRD results, discussed below, the small mass loss and the very small endothermic peak between 600 and 750°C can be attnbuted to the decomposition of residual SrCO, No chemi-cal reaction nor any phase transition is observed above 800 "C The DSC peaks between 1000 and 1200 "C are due to baseline instabilities The initial state of the gel powder is assumed to be either highly amorphous or poorly crystalline Fig 3 shows the XRD patterns of gel powders calcined at 600,700 and 800 "C It can be seen that the cubic perovskite-type La, $r0 ,COO,-, phase is the main phase14 The peak widths in the pattern of the powder calcined at 600°C indicate a low level of crystallinity 105 95 -85 s Y 5 75E 65 55 45 0 200 400 600 800 100012001400 TI"C Fig.2 TG-DSC curves of the La, 3Sro ,COO, gel powder 30 35 40 45 50 55 60 2 eldegrees Fig.3 XRD patterns of the non-supported Lao,Sr,,CoO, mem-branes calcined at (a) 600, (b) 700 and (c) 800°C Lines belonging to the cubic perovskite phase are indexed The asterisks indicate SrC0, Unidentified lines are marked by open circles or small particle size.From the change of width of the corresponding diffraction peaks with increasing temperature, it can be concluded that either the crystallinity improves or the particle size increases. The XRD pattern of the powder calcined at 600 "C contains several peaks which belong to SrCO,. Upon increasing the calcination temperature to 700-800 "C, these peaks almost disappear. Apparently, most of the SrCO, that is present in the sample is decomposed below 700 "C. The latter temperature is much lower than the onset temperature for SrCO, decompo- sition reported in ref. 15. This is probably due to the high reactivity of the sol-gel derived powders. This result is also in agreement with the DSC-TG curves and thus the small endothermic peak between 600 and 750°C can be attributed to the decomposition of SrCO,.A few unidentified diffraction peaks were found in the XRD patterns of powders calcined at 700 and 800°C. A metastable tetragonal phase has been observed previously in the La, -,Sr,CoO, system.14 Some of the unidentified peaks, for instance the peak at 28 cu. 44", may be attributed to this metastable phase. Supported membrane The morphology of the supported membranes was studied by SEM. The SEM pictures shown in Fig. 4 were taken after sintering the samples at 800°C. In the cross-section of a y-Al,O,-supported sample, shown in Fig. 4(u), three regions can be distinguished. At the bottom, the porous a-Al,O, phase can be seen. The intermediate layer with a thickness of cu.2.5 pm consists of porous y-Al,O,. The top layer is the perovskite phase, having a thickness of ca. 300nm, and was Fig. 4 SEM images of the supported La,, ,SrO 7C003 single-layer coated (a)-(e) and triple-layer coated (f),(8)membranes sintered at 800 "C for 5 h. (a)-(f) supported on porous y-Al,O,, (8)supported on porous a-Al,O,. (a)cross-section, La(NO,), as precursor (10000x); (b)La(NO,), as precursor (100O00x); (c) La(MOE), as precursor (100000x); (d) La(NO,), as precursor (10000 x); (e) La(MOE), as precursor (loo00 x),(f)La(NO,), as precursor (100000 x); (g)La(NO,), as precursor (100000 x). J. Muter. Chern., 1996, 6(5), 815-819 817 I h yAIpO, Supported v) a-AI,O, supported 30 35 40 45 50 55 60 2Bldegrees Fig.5 XRD patterns of powders scratched from the supported La, 3Sr0$003 membranes after sintenng at 800°C for 5 h Reflections belonging to cubic perovskite are indexed Open circles indicate A1,0,, while asterisks indicate CoAI,O, I 1025I: 0 5 10 15 20 25 30---. pore radudnrn Fig. 6 Pore-sue distnbution of the supported La, 3Sr0 7C003 mem-branes prepared with La( NO3), as precursor matenal Sintenng temperatures (a)500, (b) 600 and (c) 700 "C found to exhibit a mirror-like appearance The layers obtained are porous, which can be clearly seen from Fig 4(b)-(e) The grain size is estimated as ca 30-50 nm EDAX indicated that the composition of the surface is approximately the same as that in the sol This implies that selective penetration of components into the pores, giving rise to a non-stoichiometric composition, did not occur within the EDAX experimental error 818 J Muter Chem, 1996,6(5), 815-819 The layer prepared with La(NO,), as the lanthanum precur- sor was found to be somewhat inhomogeneous in its strontium distribution by EDAX The bright area in Fig 4(d) is rich in strontium This is probably due to the fact that during the reflux of the mixed precursor solution, La(N03), reacted with Sr(OCH,CH,OCH,), resulting in La(OCH2CH20CH,), and Sr(N03), Since the La Co ratio used in this study was 3 7, this implies that when all La(NO,), was converted to La(OCH,CH,OCH,), ,the strontium precursor consisted of a mixture of Sr(OCH,CH,OCH,), and Sr(NO,), Although an inorganic salt of one component can be used as a precursor in a multi-component polymeric sol-gel process,16 a certain extent of inhomogeneity can occur when salt mixtures are used This may be due to the precipitation of one the salts during drying and gelation as a result of the difference in the solubility of the salts The layer that was prepared with Sr(OCH,CH,OCH,), as the strontium precursor is found to be homogeneous, but more porous, which can be seen by comparing Fig 4(d) and 4(e) The multiple-coated membranes, shown in Fig 4(f) and (g),exhibit denser morphologies than the single-coated ones This is especially clear for the a-Al,O,- supported layer in Fig 4(g) Gas-tightness measurements indi- cated that the obtained layer was still porous A fully dense supported layer can be obtained only when sintering occurs by a viscous mechanism Even though the existence of an i11- defined liquid phase is assumed to promote densification of bulk ceramics of La, ,S~,COO,-~,~~the present results suggest that densification is retarded by the interaction with the support material Fig 5 shows the XRD patterns of powders scratched from a-Al,O,- and y-Al,O,-supported La, 3Sro ,COO, -membranes sintered at 800°C The results are essentially similar to those obtained for the unsupported membranes, the main phase formed being cubic perovskite Diffraction peaks from Al,O, and a diffraction peak from CoA120, can also be seen in the patterns The latter compound may result from the interaction between the La, 3Sro layers and the A1,0, supports during sintenng The spinel CoA120, was also found to be formed in an investigation of CoO-A1,03 catalysts '* Although fully densified La, -,Sr,CoO, layers could possibly result at sintering temperatures >800°C, in view of the high sinter- ability of the perovskite phase, alumina is not suitable as support material because of its strong chemical interaction with La, 3Sro 7C00,-s A search for other, more inert, porous supports is deemed necessary to develop dense structured layers Fig 6 shows estimates of the pore-size distribution of some y-Al,O,-supported perovskite membranes sintered at different temperatures and prepared with different lanthanum precur- sors The results can be used only in a qualitative manner owing to the badly developed accumulated oxygen per- meability plots It can be seen from Fig 6(u)-(c) that the average active pore size of a perovskite membrane tends to increase with increasing sintenng temperature Increasing the sintering temperature also broadens the pore-size distnbution It should be noted that the estimated average pore radius of the membrane sintered at 700°C is cu 12 nm, which is larger than the average pore radius of La-doped y-Al,O, (8 5 nm) This may be caused by the large error in the estimated pore size In addition, the interaction between the deposited layer and the alumina support, as confirmed by XRD, may play a role The membranes sintered at 400 and 800°C could not be charactenzed by the permporometry technique, probably because their pore sizes are beyond the range of this technique (1-50 nm) Conclusions A polymeric sol-gel process was developed to fabricate thin La, 3Sro &oO, layers on porous alumina supports Partially hydrolysed mixed precursor solutions in methoxyethanol were spin-coated on porous ct-and y-A1203 supports to form gel layers The perovskite phase was formed slightly above 400 "C Most of the residual strontium carbonate existing in the gel layers could be decomposed by heating to 600-700°C When the preparation conditions were carefully controlled, the formed layers were smooth and crack free with an esti- mated thickness of ca 300nm after a single coating step The layer was found to be more homogeneous when La(OCH2CH20CH,), was used as the lanthanum precursor instead of La(NO,), The pore size of layers tended to increase with increasing sintering temperature Sintering had to be performed at tem- peratures below 800°C in order to prevent chemical inter- actions between the deposited layer and the alumina support, resulting in the formation of spinel-type CoA1,0, The alumina-supported Lao 3Sro ,COO, layers developed in this study may be used either as intermediate layers onto which a dense layer of the same material can be applied (eg by vapour deposition) or as porous membranes for high- temperature liquid or gas separation This work was performed in the framework of a joint pro- gramme between the Academia Sinica and the Dutch Academy of Science (KNAW) and was supported financially in part by the KNAW We are indebted to Professor Meng Guangyao (Department of Materials Science and Engineering, University of Science and Technology of China) for his support in stimulating this project P Fransen (MESA, University of Twente), H W Brinkman and B de Boer are sincerely acknowledged for their technical assistance References 1 0 Yamamoto, Y Takeda, R Kanno and M Noda, Solid State Ionzcs, 1987,22,241 2 A Mackor, T P M Koster, J G Kraaijkamp, J Gerretsen and J P G M van Eijk, in Proc 2nd Int Symp on Solid Oxide Fuel Cells, Athens, Greece, ed F Grosz, P Zegers, S C Singhal and 0 Yamamoto, Commission of the European Communities, Luxumbourg, 1991, p 463 3 Y Yamamura, Y Ninomiya and S Sekido, in Proc Int Meeting on Chemical Sensors, Tokyo, 1983, p 187 4 Y Teraoka, T Nobunaga, K Okamoto, N Miura and N Yamazoe, Solid State Ionics, 1991,48,207 5 Y S Lin and A J Burggraaf, AIChE J, 1992,38,445 6 G Z Cao, H W Brinkman, J Meijennk, K J de Vries and A J Burggraaf, J Am Ceram Soc ,1993,76,2201 7 C J Brinker, T L Ward, R Sehgal, N K Raman, S L Hietala, D M Smith, D-W Hua and T J Headley, J Muter Sci, 1993,77, 165, W F Maier, I-C Tilgner, M Wiedorn, H-C KO, A Ziehfreund and R Sell, Adv Muter, 1993,5730 8 C D E Lakeman and D A Payne, J Am Ceram Soc, 1992,75, 3091, G Moore, S Kramer and G Kordas, Muter Lett, 1989, 7, 415, K Okuwada, M Imai and K Kakuno, Jpn J Appl Phys, 1989,28, L1271 9 T W Kueper, S J Visco and L C De Jonghe, Solid State Ionics, 1992,52,251 10 C J Brinker and G W Scherer, Sol-Gel Science The Physics and Chemistry of Sol-Gel Processing, Academic Press, Boston, 1990 11 C H Chen, H Kruidhof, H J M Bouwmeester and A J Burggraaf, to be published 12 G Z Cao, J Meijennk, H W Brinkman and A J Burggraaf, J Membr Sci ,1993,83,221 13 Y S Lin and A J Burggraaf, J Am Ceram SOC,1991,74,219 14 J Kirchnerova and D B Hibbert, J Muter Sci , 1993,28,5800 15 M J Scholten, J Schoonman, J C van Miltenburg and H A J Oonk, Solid State Ionics, 1993,61, 83 16 P Griesmar, G Papin, C Sanchez and J Livage, J Muter Scz Lett, 1990, 9, 1288, S-F Ho, L C Klein and R Caracciolo, J Non-Cryst Solids, 1990, 120, 267, H Murakami, S Yaegashi, J Nishino, Y Shiohara and S Tanaka, Jpn J Appl Phvs, 1990, 29, L445 17 H Kruidhof and H J M Bouwmeester, unpublished results 18 P Arnoldy and J A Moulijn, J Catal, 1985,93,38 Paper 5/06083D, Received 14th September, 1995 J Muter Chem, 1996, 6(5),815-819 819