Ind. Eng. Chem. Res. 2010, 49, 2895–2901
2895
Effects of Sintering on the Properties of La0.6Sr0.4Co0.2Fe0.8O3-δ Perovskite Hollow Fiber Membranes Xiaoyao Tan,*,†,‡ Zhigang Wang,†,‡ and Kang Li§ School of EnVironmental and Chemical Engineering, Tianjin Polytechnic UniVersity, Tianjin 300160, China, School of Chemical Engineering, Shandong UniVersity of Technology, Zibo 255049, China, and Department of Chemical Engineering and Technology, Imperial College London, South Kensington, London SW7 2AZ, United Kingdom
La0.6Sr0.4Co0.2Fe0.8O3-δ (LSCF) hollow fiber membrane precursors were prepared by spinning a starting suspension containing 68.75 wt % LSCF powders, 6.25 wt % polyethersulfone (PESf), and 25.0 wt % N-methyl2-pyrrolidinone (NMP) at room temperature using deionized water and tape water as the internal and external coagulants, respectively. High temperature sintering was carried out in a range of 1200-1500 °C to study the influences of the sintering process on the properties of the LSCF hollow fiber membranes including the microstructure, crystalline phase, mechanical strength, as well as the oxygen permeability. Mechanical strength of the LSCF hollow fibers increased with increasing sintering temperature and reached a maximum of 115 MPa at 1500 °C sintering temperature. To obtain gastight LSCF hollow fiber membranes, the sintering temperature must be higher than 1250 °C, and the sintering time must be longer than 2 h. However, higher than 1350 °C sintering temperature would facilitate the formation of sulfate impurity phases, resulting in noticeable reduction of oxygen permeation flux. The optimum sintering temperature should be around 1300 °C, and the sintering time should be within the range of 2-4 h to obtain the gastight and high flux LSCF hollow fiber membranes. 1. Introduction Since Teraoka et al.1 first reported the SrCo0.8Fe0.2O3-δ oxygen permeable membrane, the mixed ionic and electronic conducting oxides have attracted much interest due to their potential applications in air separation, oxy-fuel combustion, and partial oxidation of hydrocarbons into value-added products.1-5 Most research was focused on the development of new perovskite materials with higher oxygen permeability and better stability.5,6 However, for the practical applications, the mixed conducting membrane must also have high mechanical strength, which is closely related to the sintering process, so as to assemble membrane modules. Among the various types of mixed conducting membranes, La1-xSrxCo1-yFeyO3-δ (LSCF) perovskite is one of the most important oxygen selective membrane materials because of its outstanding mechanical and chemical stability, although it exhibits a relatively low oxygen permeability.5-8 It is well-known that the microstructure of the membranes plays a predominant role in their transport properties. For ceramic membranes, the microstructure depends not only on the preparation method but also on the sintering procedures. Several groups worldwide have investigated the effects of microstructure on the oxygen permeation properties of mixed conducting membranes.9-17 They have obtained very different conclusions for different membranes or even for the same compositions. For example, for the LaCoO3-δ,9 La0.3Sr0.7CoO3-δ,10 Ba0.5Sr0.5Fe0.8Zn0.2O3-δ,11 La0.6Sr0.4Co0.2Fe0.8O3-δ,12 and Ba0.5Sr0.5Co0.8Fe0.2O3-δ13 membranes, the increase of grain size by increasing sintering temperature leads to an enhanced permeation, but for the SrFe0.2Co0.8O3-δ,14 La0.5Sr0.5FeO3-δ,15 La0.1Sr0.9Co0.9Fe0.1O3-δ,16 and La0.6Sr0.4Fe0.9Ga0.1O3-δ17 membranes, the oxygen permeation flux decreased noticeably with * To whom correspondence should be addressed. Tel.: 86-5332786292. Fax: 86-533-2786292. E-mail:
[email protected]. † Tianjin Polytechnic University. ‡ Shandong University of Technology. § Imperial College London.
increasing grain size induced by the increase of sintering temperature. This indicated whether the grain boundaries act as high diffusivity paths or barriers for oxygen transport depends on the type of crystalline solids that adjoin the interfaces. Therefore, the influence of sintering on the permeation properties of mixed conducting membranes is complicated and cannot be generalized. It depends not only on the chemical nature of membrane materials and microstructures but also on the ratecontrolling step of the oxygen permeation process (surface exchange kinetics or bulk diffusion). For instance, in the surface exchange-controlled processes, an increase in oxygen permeation is expected with decreasing grain size because the oxygen exchange coefficient increases significantly when the average grain size on the membrane surface decreases.11,18 In recent years, the combined phase inversion/sintering technique has been extensively applied to fabricate ceramic hollow fiber membranes.19-23 The hollow fiber configuration exhibits many advantages over the planar or tubular membranes in particular such as high surface area/volume ratio and facile high-temperature sealing. More importantly, because the membrane cross section of the hollow fiber membranes prepared through the phase inversion process is asymmetric (i.e., a thin separating layer integrated with a substrate), the resistance to permeation is therefore low. As a result, the hollow fiber membranes have more potential than other configurations to meet commercial targets in air separation units. As compared to the relatively homogeneous ceramics manufactured using the normal pressing or casting method, the sintering process of the ceramic hollow fiber membranes prepared by a phase inversion method is more complex because of their special asymmetric structure.24,25 However, up to now, all of the studies on the influence of sintering on the microstructure and the permeation properties of mixed conducting membranes were carried out on the basis of the disk membranes prepared by the pressing method.12 In this work, we presented a detailed investigation of the influence of sintering on the microstructure and the
10.1021/ie901403u 2010 American Chemical Society Published on Web 02/18/2010
2896
Ind. Eng. Chem. Res., Vol. 49, No. 6, 2010
properties of the phase inversion-induced LSCF hollow fiber membranes. Such a study is helpful to identify the optimum sintering conditions of the production of LSCF hollow fiber membranes for oxygen separation by the phase inversion/ sintering technique. 2. Experimental Section 2.1. Materials. Sr(NO3)2 (AR), La(NO3)3 · 6H2O (AR), Co(NO3)3 · 6H2O (>99%), and Fe(NO3)3 · 9H2O (AR) were purchased from Kermel Chem Inc., Tianjin, China, and were used as the metallic precursors for the preparation of La0.6Sr0.4Co0.2Fe0.8O3-δ (LSCF) powders. Citric acid (>99%, Ajax) and ethylene glycol (AR, Longjili, Tianjin) were used as the complexing agents. Nitric acid and ammonium hydroxide were used to adjust the pH of the starting solution. Polyethersulfone (PESf) [(Radel A-300), Ameco Performance, USA] and N-methyl-2-pyrrolidone (NMP) [AR grade, >99.8%, Kermel Chem Inc., Tianjin, China] were used to prepare the spinning suspension. 2.2. Preparation of LSCF Powders and Hollow Fiber Membranes. LSCF powders were prepared through a solgel-combustion process.26 Stoichiometric amounts of Sr(NO3)2, La(NO3)3 · 6H2O, Co(NO3)3 · 6H2O, and Fe(NO3)3 · 9H2O were dissolved in distilled water. Citric acid and ethylene glycol in quantities of 3 times the desired LSCF product were then added under magnetic stirring until they were dissolved completely. The pH of the mixture was adjusted to be 3-4 using nitric acid and ammonium hydroxide to avoid precipitation. Subsequently the solution was stirred at 70 °C on a hot plate for 5-10 h to form a transparent brown sol. Further heating was conducted under continuous stirring until a viscous gel was formed. As the temperature was increased to around 300 °C, autocombustion took place to form a fluffy black powder (LSCF powder precursor). Under an air flow, the powder precursor was calcined at 800 °C for 3 h to remove the residual carbon and to form the desired structure. Prior to the preparation of spinning suspensions, the resultant powders were ball-milled for 48 h, followed by sieving through a sieve of 200-mesh or 24 µm sieve-pore diameter. LSCF hollow fiber membranes were prepared by the phase inversion and sintering technique at room temperature.27 The spinning suspension consisted of 68.75 wt % LSCF powders, 6.25 wt % PESf, and 25.0 wt % NMP. A spinneret with the orifice diameter/ inner diameter of 3.0/1.2 mm was applied to form the hollow fiber precursors. Deionized water and tap water were used as the internal and external coagulants, respectively. The hollow fiber precursors were preserved in a water bath for 48 h to complete solidification. Sintering was conducted in a high temperature furnace after the hollow fiber precursors were dry-treated at ambient temperature. The sintering temperature range studied in this work was 1200-1500 °C within which the hollow fibers can be sintered into its dense form. The heating rate was of 1-3 °C min-1 applied during the entire sintering process. 2.3. Characterization. The gas-tightness of the sintered hollow fiber membranes was examined through a gas permeation measurement that was described elsewhere.28 Nitrogen was used as the test gas. The densities of the hollow fibers were measured by the Archimedes method in water. Morphology and microstructures of the hollow fiber membranes were observed with scanning electron microscopy (SEM) (FEI Sirion200, The Netherlands). Gold sputter coating was performed on the samples under vacuum before the measurements. Crystalline phases of the membranes were determined
by X-ray diffraction (BRUKER D8 Advance, Germany) using Cu-KR radiation (λ ) 0.15404 nm). The fibers were ground into fine powders prior to the XRD measurements. Continuous scan mode was used to collect 2θ data from 20° to 80° with a 0.02° sampling pitch and a 2° min-1 scan rate. The X-ray tube voltage and current were set at 40 kV and 30 mA, respectively. The mechanical strength of the hollow fibers was measured on a three-point bending instrument (Instron model 5544) with a crosshead speed of 0.5 mm min-1. Hollow fiber samples were fixed on the sample holder at a distance of 32 mm. The bending strength, σF, was calculated from the following equation: σF )
8FLD π(D4 - d4)
(1)
where F is the measured force at which fracture takes place; L, D, and d are the length (32 mm), the outer diameter, and the inner diameter of the fiber sample, respectively. The values of outer diameter (D) and inner diameter (d) were obtained from the SEM graphs. Three or four fiber samples were taken for each sintering temperature to measure mechanical strength. The difference between the measurements was within 11%, and the averaged value was used to present the mechanical strength of the hollow fiber corresponding to the sintering temperature. Oxygen permeation properties of the LSCF membranes were measured using a single fiber permeation cell with details described elsewhere.29 Only those fibers confirmed to be gastight at room temperature in advance were used for the permeation measurement. Air was fed in the shell side and helium passed through the fiber lumen to collect the permeate gas. The composition of the permeate gas was measured online using a gas chromatograph (Agilent 6890N) fitted with a 5 Å molecular sieve column (Φ3 mm × 3 m) and a TCD detector. The overall oxygen permeation flux of the hollow fiber was calculated by: JO2 )
FtyO2 Am
or JO2 )
1 FHeyO2 · Am 1 - yO2
(2)
where Am is the effective membrane area for oxygen permeation, Am ) πDmL, in which Dm is the average logarithmic diameter of the hollow fiber, Dm ) (D - d)/ln(D/d); Ft and FHe are the volumetric flow rates of the effluent stream and of the sweep gas, respectively; and yO2 is the oxygen fraction in permeate stream. All of the values of oxygen permeation fluxes or other gas flow rates in this study were calculated at the conditions of the standard temperature and pressure (STP). The results obtained are average values of three measurements, and their differences are less than 5%. 3. Results and Discussion To acquire the exact knowledge of relationships between sintering, microstructure, and the properties of LSCF hollow fiber membranes, all of the hollow fibers discussed in this work were made from the same dope composition. In addition, because the gastight hollow fiber membranes can only be obtained by sintering at higher than 1250 °C, the sintering temperatures studied in this study ranged within 1200-1500 °C.
Ind. Eng. Chem. Res., Vol. 49, No. 6, 2010
2897
Figure 1. SEM photographs of the LSCF hollow fiber membranes sintered at 1300 °C for 4 h: (a) cross section; (b) fiber wall; (c) outer surface; and (d) inner surface.
3.1. Microstructure. The overall microstructure of the gastight LSCF hollow fiber membranes that were sintered at 1300 °C for 4 h can be shown in Figure 1. As can be seen, the resultant hollow fiber membrane has an asymmetric structure formed during the phase inversion process, which is attributed to the rapid precipitation occurring at the outer-layer resulting in thin and dense skin layer and to the slow precipitation giving the porous sublayer structure. As compared to the membranes prepared by the extrusion or pressing method, the phase inversion-derived hollow fiber membranes possess a large number of macro voids inside the membrane wall. Therefore, the sintering temperature to achieve gas-tightness for the hollow fiber membranes would be much higher than those derived by other methods. For example, to obtain gastight hollow fiber membranes, the sintering temperature must be higher than 1250 °C, but for the pressing-derived disk membranes, the sintering temperature to achieve gas-tightness could be as low as 1000 °C.12 Although there are some pores still present on both the inner and the outer surfaces, as can be seen in Figure 1c and d, they did not affect the gas-tightness property of the membrane. This implies the gas-tightness of the hollow fibers was mainly achieved due to the central dense layer of the fiber wall marked by the rectangle shape in Figure 1b. Figure 2 displays the surface morphology of the LSCF hollow fiber membranes sintered at the temperatures between 1200 and 1500 °C for duration of 4 h. As can be seen from Figure 2a, the membrane surface calcined at 1200 °C is composed of granular ceramic particles of 0.28-0.91 µm that have been partially calcined. Smaller particles have disappeared, and the dominated grain size (>90%) is within 0.41-0.74 µm. However, the membrane surface at this stage is still very porous. The gas permeation test also showed that such hollow fibers were not gastight. When the sintering temperature was increased to 1250 °C (Figure 2b), obvious coalescence of granular ceramic particles takes place, and the number of pores on the membrane surfaces was reduced noticeably. Although there were still some pores present on the membrane surface, the gas permeation test indicated that these hollow fibers were gastight. This indicated that the middle dense layer takes more effect on the gas-tightness of the membranes. Further increasing the sintering temperature
Figure 2. Surface morphology of the LSCF hollow fiber membranes sintered at (a) 1200 °C; (b) 1250 °C; (c) 1275 °C; (d) 1300 °C; (e) 1350 °C; (f) 1400 °C; (g) 1450 °C; and (h) 1500 °C for 4 h.
2898
Ind. Eng. Chem. Res., Vol. 49, No. 6, 2010
Figure 3. Average grain size area of the LSCF hollow fiber membranes as a function of sintering temperature (sintering time ) 4 h).
Figure 5. SEM photographs of the LSCF hollow fiber membrane surfaces sintered at 1300 °C for (a) 2 h; (b) 4 h; (c) 6 h; and (d) 8 h.
Figure 4. Effect of sintering temperature on the density, shrinkage, and mechanical strength of the LSCF hollow fiber membranes (sintering time ) 4 h).
to 1275 °C led to the increase of grain size to 0.51-2.15 µm with the majority ranging within 0.63-1.72 µm (Figure 2c). Some large pores on the membrane surfaces were preserved even if the sintering temperature was increased to 1300 °C, as shown in Figure 2d. Under this temperature the membrane surfaces became smooth with distinct grain boundaries, indicating noticeable fusion has taken place. The grain size was increased to 0.57-2.67 µm with most of the grains within the range of 0.83-2.13 µm. When the sintering temperature was increased to 1350 °C (Figure 2e), all of the pores on the membrane surfaces have disappeared, and the grain size increased to 0.96-4.26 µm with the majority at 1.40-3.01 µm. Interestingly, a closing pore was observed on the membrane surface, as marked by the red circle in Figure 2e. With the sintering temperature being further increased to 1400, 1450, and 1500 °C (Figure 2f-h), the grain size was developed remarkably to 1.62-6.53, 3.71-15.38, and 4.57-25.67 µm with the majority at 2.9-5.1, 7.0-13.0, and 13.0-20.0 µm, respectively. Figure 3 plots the average grain size, which was calculated from the average grain area, as a function of sintering temperature in the form of Arrhenius equation. As the sintering temperature was increased from 1200 to 1500 °C, the average grain size increased from 0.22 to 8.51 µm. The activation energy for the grain growth obtained from the slope of the line is 270.2 kJ mol-1. Figure 4 shows the apparent density and shrinkage of the hollow fibers as a function of sintering temperature. As can be seen, both the apparent density and the shrinkage of the hollow fibers changed little within the whole sintering temperature range except at 1200 °C. This implies that the macrostructure of the hollow fibers would not be changed noticeably with the increase of sintering temperature after the hollow fiber was sintered into gastight. Figure 5 shows the SEM pictures of surface morphology of the LSCF hollow fiber membranes sintered at 1300 °C for different dwelling times. Noteworthy is that the membranes would not be gastight if the sintering time was less than 1.0 h. It can be seen from Figure 5a that the membrane surface has
Figure 6. XRD patterns of the LSCF hollow fiber membranes sintered at (a) 1200 °C; (b) 1250 °C; (c) 1300 °C; (d) 1350 °C; (e) 1400 °C; (f) 1450 °C; and (g) 1500 °C for 4 h (p, perovskite; ), SrSO4).
been sintered with isolated pores even if the sintering time was only 2 h. As the sintering time was prolonged, the grains were enlarged, but the pores on the membrane surfaces would not disappear completely (Figure 5b-d). As compared to the sintering temperature, the sintering time has less influence on the microstructure of the LSCF hollow fiber membranes. 3.2. Crystalline Structure. Figure 6 depicts the XRD patterns of the LSCF hollow fiber membranes sintered under air for 4 h at different temperatures. It can be seen that the hollow fiber samples sintered at temperatures lower than 1350 °C possess the cubic perovskite crystalline structure (indicated by “p”), and no intermediate phases were identified. This implies that no phase transition occurred from 1200 to 1350 °C during the sintering process. However, after the sintering temperature was increased to 1400 °C, some additional peaks (indicated by “)”) to the perovskite phase appeared on the XRD patterns.
Ind. Eng. Chem. Res., Vol. 49, No. 6, 2010
2899
Figure 8. Mechanical strength of the LSCF hollow fiber membranes as a function of sintering temperature (sintering time: 4 h).
Figure 7. XRD patterns of the LSCF hollow fiber membranes sintered at 1300 °C for (a) 2 h; (b) 4 h; (c) 6 h; and (d) 8 h.
This indicated that there are some new phases, one of which was identified to be SrSO4 (PDF reference code 73-0529), that have formed at higher temperatures. Moreover, the fact that the intensity of the additional peaks increased with sintering temperature indicated that the amount of the impurity phases increased with sintering temperature. The formation of strontium sulfate phase may be attributed to the interaction of the minor amount of SO2 with the surface of the perovskite membrane.30,31 In addition to the trace amount of SO2 presented in atmospheric air, the origin of sulfur to form strontium sulfate phase might also come from the decomposition of polyethersulfone, which was used as the polymer binder in spinning the fiber precursors. At lower temperatures (