Proton conducting perovskite hollow fibre membranes with surface catalytic modification for enhanced hydrogen separation

G Model

ARTICLE IN PRESS

JECS-10463; No. of Pages 9

Journal of the European Ceramic Society xxx (2016) xxx–xxx

Contents lists available at www.sciencedirect.com

Journal of the European Ceramic Society journal homepage: www.elsevier.com/locate/jeurceramsoc

Proton conducting perovskite hollow fibre membranes with surface catalytic modification for enhanced hydrogen separation Jian Song a , Jian Kang b , Xiaoyao Tan a,∗ , Bo Meng c , Shaomin Liu b,∗ a State Key Laboratory of Separation Membranes and Membrane Processes, Department of Chemical Engineering, Tianjin Polytechnic University, Tianjin 300387, China b Department of Chemical Engineering, Curtin University, Perth, WA 6102, Australia c School of Chemical Engineering, Shandong University of Technology, Zibo 255049, China

a r t i c l e

i n f o

Article history: Received 13 June 2015 Received in revised form 26 December 2015 Accepted 6 January 2016 Available online xxx Keywords: Perovskite membrane Hollow fibre Hydrogen separation Surface modification

a b s t r a c t BaCe0.85 Tb0.05 Co0.10 O3−ı (BCTCo) perovskite hollow fibre membranes were fabricated by a combined phase-inversion and sintering technique. The hollow fibre surfaces were modified by coating Ni or Pd particles. Hydrogen permeation fluxes at 700–1000 ◦ C can be improved due to the surface modification from the original 0.009–0.164 mL(STP) cm−2 min−1 to 0.018–0.269 mL cm−2 min−1 for the Ni-coated membranes with maximum improvement by 64%, and to 0.1–0.42 mL cm−2 min−1 for the Pd-loaded membranes with maximum enhancement by 155%, respectively. Loading of the catalyst on the hollow fibre outer surface is better than on the inner surface, but coating on both sides may enhance the hydrogen permeation most effectively. The permeation enhancement depends on both the catalyst loading amount and its structure, which can be controlled by the plating conditions. The optimal Pd loading and coverage should be around 0.667 mg cm−2 and 82%, respectively for maximizing the permeation improvement. © 2016 Elsevier Ltd. All rights reserved.

1. Introduction Hydrogen is not only an important raw material for chemical and petrochemical industries, but also a potential clean fuel for advanced energy technologies to mitigate the global climate change caused by the excessive green-house gas emissions from conventional thermal power plants. However, pure hydrogen gas does not exist as other natural fuel resources like coal, oil or methane in nature but has to be produced from available hydrogencontaining compounds by means of reforming reactions, where the downstream separation and purification contributes most part of the production costs. Membrane technology based on the Pd or Pd-alloy membranes represents a promising method to separate hydrogen because they not only have high permeability and selectivity but also can be directly integrated into membrane reactors for a variety of dehydrogenation reactions so as to combine reaction and separation in one unit [1–3]. However, the high cost of the Pd or Pd-alloy membranes resulting from the noble metal Pd and the

∗ Corresponding authors. E-mail addresses: [email protected] (X. Tan), [email protected] (S. Liu).

complex procedure of the membrane synthesis have impeded their wide applications. In recent years, high temperature proton conducting ceramic membranes (HTPCMs), typically based on SrCeO3 , BaCeO3 , SrZrO3 and BaZrO3 perovskites in a general form of ABO3 have attracted considerable interest for their potential applications in hydrogen pumps (separators), fuel cells, gas sensors and catalytic membrane reactors [4–8]. However, as hydrogen separation membranes, the HTPCMs have very high selectivity (up to 100% in theory) but generally exhibit very low hydrogen permeation fluxes, i.e. in the order of magnitude of 10−3 –10−2 mL(STP) cm−2 min−1 [9] in the thickness of 1 mm, which is far lower than the flux value of commercial interest, 1.0 mL cm−2 min−1 . This suggests that the permeability of the HTPCMs still needs to be improved for commercial applications. To this end, efforts have been placed on the material development like optimizing the perovskite compositions of SrCe(1−x) Mx O(3−0.5x) or BaCe(1−x) Mx O(3−0.5x) using a better dopant M3+ like Y, Yb, Sm, Eu, Tm, Tb, Ti, V, Cr, Mn, Co, Ni, Cu, Al, Ga, and In to achieve high proton and electronic conductivities [10–16]. On the other hand, engineering considerations should also be given to improve the hydrogen flux for a certain material with a fixed composition. It is known that hydrogen permeation through the HTPCMs involves hydrogen disassociation and

http://dx.doi.org/10.1016/j.jeurceramsoc.2016.01.006 0955-2219/© 2016 Elsevier Ltd. All rights reserved.

Please cite this article in press as: J. Song, et al., Proton conducting perovskite hollow fibre membranes with surface catalytic modification for enhanced hydrogen separation, J Eur Ceram Soc (2016), http://dx.doi.org/10.1016/j.jeurceramsoc.2016.01.006

G Model JECS-10463; No. of Pages 9 2

ARTICLE IN PRESS J. Song et al. / Journal of the European Ceramic Society xxx (2016) xxx–xxx

incorporation with oxygen vacancies on the membrane surfaces (superficial activation) followed by the hydroxyl ion transport on interstitial lattice sites (bulk diffusion) [17]. Accordingly, the hydrogen permeation flux through the HTPCMs can be limited by the overall resistance from the surface reaction and the bulk diffusion. In order to improve the hydrogen flux through the HTPCMs, an effective way is to reduce the membrane thickness, which can be achieved via the asymmetric membrane structure where a thin HTPCM dense film is formed on a porous substrate [11,18,19]. For instance, Cheng et al. prepared asymmetric SrCe0.95 Tm0.05 O3−ı (SCTm) membranes consisting of a dense thin film and a thick porous support of the same material by the conventional dry pressing method [18]. The hydrogen permeation flux for the 150 ␮m thick disc-shaped membrane attained up to 0.126 mL cm−2 min−1 at 900 ◦ C when using 10% H2 –He as the feed gas and air as the sweep gas, respectively. This flux has been increased by a factor of 4 compared to that obtained from 1.6 mm thick SCTm membrane. Hamakawa et al. prepared dense SrCe0.95 Yb0.05 O3−ı (SCYb) thin films on porous SrZr0.95 Y0.05 O3−ı substrate by the spin-coating method [11]. H2 permeation rate through 2 ␮m SCYb film reached up to 13.44 mL cm−2 min−1 at 677 ◦ C, which is 500 times higher than the conventional 1-mm-thick membranes in the same material. However, the composite membranes normally suffer problems such as preparation complexity and difficulty in scaling up, since the multi-step membrane preparation is very time-consuming and costly. Moreover, the dense layer of the composite structure tends to crack or peel off from the support during firing processes because of their different thermal expansion behaviors. In the last decade, the immersion induced phase inversion technique has been widely applied in the fabrication of asymmetric perovskite hollow fibre membranes [20–22]. Since the separation layer and the porous support are formed from the same ceramic material in one-step, the cracks or peeling off between the two structural layers can be avoided. For these hollow fibre membranes, the wall thickness generally ranges between 0.3–0.5 mm, but the real effective separation thickness is much less than the apparent wall thickness due to the presence of asymmetric structure. Accordingly, such hollow fibres generally have much higher permeability than thick disk-shaped or tubular membranes [23]. As the membrane thickness decreases, the relative limiting effect of the surface activation processes will increase or even become the rate-limiting step. In this case, the surface activation kinetics have to be promoted to further improve the permeation flux. This can be achieved by increasing the surface area [24] or by coating a porous catalytic layer [25–34]. Mather et al. [35] studied the hydrogen permeability of Sr0.97 Ce0.9 Yb0.1 O3−ı (SCYb) membrane with a porous Pt catalytic layer facing the hydrogen feed side. The hydrogen flux at 804 ◦ C reached a maximum of 0.048 mL cm−2 min−1 employing 10% H2 –90% N2 feed and Ar as the sweep gas, over one order of magnitude higher than that obtained on membranes of similar thickness without surface modification [36]. In this paper, some cost-effective metal catalysts like Ni or Pd rather than expensive Pt have been investigated to improve the BaCe0.85 Tb0.05 Co0.10 O3−ı (BCTCo) hollow fibre membrane surface reaction kinetics and subsequently to enhance the hydrogen flux. The proton conducting perovskite hollow fibre membranes were fabricated by the combined phase-inversion and sintering technique. Ni has been widely used as the catalyst for dehydrogenation/hydrogenation reactions and Pd is a well-known membrane material for hydrogen permeation, thus we hypothesize that Ni or Pd would have catalytic effect to promote the hydrogen dissociation/association leading to the hydrogen flux improvement. The hydrogen permeation behavior through the original BCTCo membrane and the Ni/Pd-coated membranes was both experimentally and theoretically studied under various conditions.

2. Experimental 2.1. Fabrication of BCTCo perovskite powders and hollow fibre membranes BaCe0.85 Tb0.05 Co0.1 O3−ı (BCTCo) perovskite powders were prepared through a sol–gel method using ethylenediaminetetraacetic acid (EDTA) and citric acid (CA) as the chelating agents. Ba(NO3 )2 , Ce(NO3 )3 ·6H2 O, Tb(NO3 )3 ·6H2 O, Co(NO3 )2 ·6H2 O, all in analytical grades, were used as the raw materials for metal ion sources. The details of the synthesis process were described elsewhere [22]. The powder precursor was calcined at 800 ◦ C for 4 h to eliminate the organic components and ball-milled for 10 h in an agate jar with ethanol as the medium followed by sieving through a sifter of 200mesh to exclude agglomerates. BaCe0.85 Tb0.05 Co0.1 O3−ı (BCTCo) hollow fibre membranes were fabricated through the combined phase inversion and sintering technique with the details described elsewhere [36,37]. The spinning solution consisted of 61.54 wt% BCTCo calcined powders, 7.69 wt% polysulphone (PSU) (Udel® P3500, Solvay) as the polymer binder, 30.77 wt% 1-methyl-2-pyrrolidinone (NMP) (AR Grade, >99.8%, Shandong Qingyun Changxin Chemical Science-Tech. Co., Ltd., China) as the solvent. For each batch of the hollow fibre, approximately 300 g of BCTCo powder would be required. A spinneret with the orifice diameter/inner diameter of 3.0/1.5 mm was used for spinning hollow fibre precursors. De-ionized water and tap water were used as the internal and the external coagulants, respectively. The hollow fibre precursors were immersed in water for 24 h to fully consolidate the hollow fibre structure. The dried hollow fibre precursors were sintered at 1350 ◦ C for 4 h to form dense ceramic structure. The gas-tight property of the hollow fibres for subsequent surface modification and hydrogen permeation were measured through the nitrogen gas permeation test described elsewhere [38]. 2.2. Pd-coating by chemical plating Pd-catalyst was deposited on the surface of the BCTCo hollow fibres by electroless plating. Prior to deposition, the gas-tight BCTCo hollow fibre sample with the length of around 25 cm was cleaned with deionized water and ethanol successively. It was then immersed in a 2 g L−1 SnCl2 acidic solution at 45 ◦ C for 4 min for sensitization, and then in a 0.2 g L−1 PdAc2 solution at 45 ◦ C for another 4 min for Pd seeding. A copious amount of DI water was used to rinse the fibre between the immersion operations. The seeded hollow fibre was immersed in 40 mL plating solution with the composition of 4.4 g L−1 Pd(Ac)2 , 37.2 g L−1 EDTA-2Na and 128.5 mL L−1 NH3 ·H2 O at 45 ◦ C. N2 H4 aqueous solution of 12 mL L−1 was injected into the plating solution under stirring to form a Pd coating based on the following equation 2Pd2+ + N2 H4 → 2Pd0 + 4H+ + N2 . The plating time was 30 min for each addition of 120 ␮L N2 H4 aqueous solution. The amount of Pd loading was controlled by the amount of N2 H4 used and the plating time. A final heat-treatment was conducted at 120 ◦ C in air for 12 h to fix the Pd deposition on the hollow fibre surfaces. In order to form a dense Pd film on the hollow fibre surfaces, 100 mL plating solution was applied, and the N2 H4 aqueous solution was 1200 ␮L which was added in eight stages. The overall reducing time was about 240 min for the formation of a dense Pd film. Table 1 summarizes the plating conditions for the preparation of samples. 2.3. Ni-coating by brushing Ni-catalyst layers were applied to the hollow fibre surfaces via a slurry brushing and sintering process. For the preparation of Ni-coating slurry, nickel nitrate (Ni(NO3 )2 ·6H2 O) was ball-

Please cite this article in press as: J. Song, et al., Proton conducting perovskite hollow fibre membranes with surface catalytic modification for enhanced hydrogen separation, J Eur Ceram Soc (2016), http://dx.doi.org/10.1016/j.jeurceramsoc.2016.01.006

G Model

ARTICLE IN PRESS

JECS-10463; No. of Pages 9

J. Song et al. / Journal of the European Ceramic Society xxx (2016) xxx–xxx

3

Table 1 Experimental results of the BCTCo and Pd/Ni hollow fibre membranes with different coatings. SampleNo.

Single/both side

Coating conditionsa

Act. Energyb (kJ mol−1 )

Flux increment

1 2 3 4 5 6 7 8 9 10 11 12

Original Inside Outside Both Both Both Both Both Both Inside Outside Both

No coating Pd/40 mL-P/120 ␮L-R/30 min Pd/40 mL-P/120 ␮L-R/30 min Pd/40 mL-P/120 ␮L-R/30 min Pd/40 mL-P/120 ␮L-R/10 min Pd/40 mL-P/120 ␮L-R/20 min Pd/40 mL-P/240 ␮L-R/60 min Pd/40 mL-P/360 ␮L-R/120 min Pd/100 mL-P/1200 ␮L-R/240 min Ni/brush Ni/brush Ni/brush

100.42 94.49 93.17 90.50 97.21 93.26 87.79 86.12 93.00 96.33 95.09 92.84

0% 71% 87% 132% 23% 84% 151% 155% 75% 39% 54% 64%

a

Pd coating solution: P and R stand for plating solution and N2 H4 aqueous solution, respectively. The calculation of activation energy is based on the permeation measurement between 700 and 1000 ◦ C. The flux increment is based on the permeation flux value measured at 1000 ◦ C. b

milled together with ethanol, 2-butanone and triethanolamine (TEA) for 24 h. The additives including diethylene glycol (DEG) and polyvinylbutyral (PVB) were then added, followed by another 24 h ball-milling treatment. The final slurry has a composition of 0.58 wt% DEG, 0.86 wt% TEA, 1.73 wt% PVB, 27.67 wt% nickel nitrate, 23.05 wt% ethanol and 46.11 wt% butanone. The prepared slurry was brushed onto the surface of the BCTCo hollow fibres, followed by heat-treatment at 1050 ◦ C with a programmed temperature ramp of 5 ◦ C min−1 to convert the coating layer to oxide (NiO). Before the hydrogen permeation test, the BCTCo membranes coated with the slurry layer were reduced to Ni from NiO in hydrogen containing atmosphere at 700 ◦ C for 90 min. 2.4. Hydrogen permeation measurement Hydrogen permeation properties of the hollow fibre membranes were measured in a self-made permeation cell described elsewhere [39]. The hollow fibre was completely gas-tight and had a length of around 25 cm. The permeation cell was positioned in a specially designed furnace having a 5 cm effective heating length. In operation, the H2 –He mixture in 1:1 molar ratio was fed to the shell side while nitrogen as the sweep gas was introduced in the fibre lumen to collect the permeated hydrogen. The flow rates of the H2 –He feed and the N2 sweep gas were controlled by mass flow controllers (D07-7B, Beijing Sevenstar Electronics Co., Ltd., China) which were calibrated with a soap bubble flow meter. The gas flow rates were also measured with a soap bubble flow meter and calibrated to standard conditions. The compositions of the permeate gas were analysed by a gas chromatograph (Agilent 6890N) equipped with a 5A molecular sieve column (10 feet × 2.1 mm) and a TCD detector. Prior to collecting data, the coated catalysts were activated in 50% H2 atmosphere at 700 ◦ C for 1.5 h. The hydrogen permeation flux was calculated by [37]: JH2 =

 Vt  yH2 − yHe Am

(1)

where Vt is the flow rate of the permeate stream; yH2 and yHe are the hydrogen and helium fractions in the permeated stream, respectively, and Am is the effective membrane area for hydrogen permeation, which was calculated by Eq. (2). Am =

(dout − din )Le ln(dout /din )

(2)

where, Le is the effective fibre length for hydrogen permeation (5 cm in this work) and dout , din are the outer and inner diameter of the hollow fibre membrane, respectively. The helium concentration appears in Eq. (1) if a minor leakage occurs during the permeation, and hence the amount of hydrogen due to the leakage has to be

deducted so as to obtain the net permeation flux. Also noteworthy is that the hydrogen permeation beyond the central heating zone was not taken into account since its contribution towards the overall hydrogen flux is much lower than the central part. 2.5. Characterization The crystal phase of the BCTCo hollow fibre was obtained by Xray diffraction with Cu K-␣ radiation ( = 0.154178 nm). The scan was made from 20 to 80◦ with a 0.1◦ min−1 scan rate at 40 kV. The morphology and microstructures of the hollow fibre membranes were studied with scanning electron microscopy (SEM) on FEI Sirion 200 (The Netherlands) or HITACHI S-4800 (Japan). The morphology and microstructure of the BCTCo powder prepared by crushing the hollow fiber membrane were also investigated by a transmission electron microscopy (TEM, JEOL JEM-200CX). 3. Results and discussion 3.1. Structure of the hollow fibre membranes Fig. 1 shows the SEM micrograph of the BCTCo hollow fibre membranes sintered at 1350 ◦ C for 4 h. As can be seen, an asymmetric structure consisting of a dense layer on the outer surface and a porous layer near the inner surface of the hollow fibres was formed due to features of the polymer phase inversion from liquid state to precipitation [40]. The outer and inner diameters of the hollow fibres can be measured from Fig. 1(a) to be around 1.70 and 1.08 mm, respectively. The outside dense layer has a thickness around 132 ␮m (Fig. 1(b)), which ensures the required gas-tightness of the membrane for hydrogen permeation only through the ionic conduction. This gas-tight property of the resultant hollow fibres was also confirmed by the nitrogen leaking test. Meanwhile, both the inside and outer surfaces of the hollow fibre seem to be very dense as observed in Fig. 1(c) and (d). The ceramic particles with maximum grain size around 8 ␮m are tightly connected to each other with clear crystal boundaries. The macro voids formed during the phase inversion were transferred into isolated pores and trapped inside the wall of the hollow fibre. The existence of isolated macro-voids is a disadvantage for hydrogen permeation as the ionic diffusion needs a continuous material phase, the formation of which results in a narrowed pathway for ionic diffusion. The sintered fibres were crushed and the crystal structure of the crushed fibres was probed by XRD and TEM analysis. The pure perovskite phase is confirmed by the XRD spectra in Fig. 2(a) where the characteristic diffraction peaks of the perovskite phase are located at (2) 28.74◦ (0 0 2), 41.10◦ (0 2 2), 50.92◦ (2 1 3), and 59.54◦ (4 2 2), respectively. Fig. 2(b) is the XRD pattern of the hollow fibre after

Please cite this article in press as: J. Song, et al., Proton conducting perovskite hollow fibre membranes with surface catalytic modification for enhanced hydrogen separation, J Eur Ceram Soc (2016), http://dx.doi.org/10.1016/j.jeurceramsoc.2016.01.006

G Model JECS-10463; No. of Pages 9 4

ARTICLE IN PRESS J. Song et al. / Journal of the European Ceramic Society xxx (2016) xxx–xxx

Fig. 1. SEM micrographs of the BCTCo hollow fibre membrane (a) cross section; (b) fibre wall; (c) inner surface; (d) outer surface.

high temperature hydrogen permeation test, which indicates the high membrane stability in reducing atmosphere. A TEM image at low magnification as shown in Fig. 2(c) reveals one smaller particle of the crushed BCTCo fibre in size of 80 nm. The HRTEM image of this particle shows a lattice spacing of 0.287 nm, which can be indexed to the planar spacing (1 1 1) of BCTCo perovskite. Fig. 3 displays the Pd-coating on the outer surface of the BCTCo hollow fibre formed by the reduction of Pd precursor with N2 H4 solution. The Pd loading on the membrane surface is determined by the reaction rate related to the temperature and reactant concentrations for both Pd-salt and reducing agent (N2 H4 ) and the plating time. For the sample treated by plating with the addition of 120 ␮L H4 N2 aqueous solution in 40 mL plating solution, the Pd particles with an average size of approximately 200 nm are dispersed on the hollow fibre surface. But the Pd deposition is not homogeneous over the entire membrane surface, where some bare surface areas can be clearly observed (Fig. 3(a)). The Pd loading was measured by the weight increase after plating to be about 0.206 mg cm−2 . As the added N2 H4 aqueous solution was increased to 240 ␮L and the reaction time to 60 min, more Pd particles were deposited and aggregated with the loading of 0.371 mg cm−2 , and some areas were even covered by a thick Pd layer (Fig. 3(b)). A more uniform Pddeposition was achieved in the sample of Fig. 3(c), where more N2 H4 aqueous solution (360 ␮L) was added inside the Pd-precursor. In this case, the Pd loading was measured by the weight increase after plating to be about 0.667 mg cm−2 . When 100 mL plating solution and 1200 ␮L N2 H4 aqueous solution were applied and the plating time was extended to 240 min, a very dense Pd layer around 1.88 ␮m thickness was formed and integrated well with the hollow fibre membrane surface as shown in Fig. 3(d and e). The good integration between the membrane layer and the catalyst layer is of high importance to minimize the boundary resistance for surface exchange reactions and bulk diffusion [34]. From Fig. 3, the coverage of the Pd catalyst can be estimated to be around 44%, 63% and 82% for the 120 ␮L-, 240 ␮L- and 360 ␮L-N2 H4 solution derived hollow fibre membranes, respectively.

The Ni-coating deposited on the BCTCo hollow fibre surface by the brushing and sintering method is shown in Fig. 4. A porous NiO layer with a thickness of around 99.6 ␮m was well-integrated on the hollow fibre membrane surface (Fig. 4(a and b)). Compared to the Pd-coating, the thickness of the Ni-coating can be more flexibly controlled to ensure a homogeneous catalyst deposition without the cost consideration due to the cheaper material. The NiO layer could be reduced into a more proliferous Ni layer by hydrogen at elevated temperature, as shown in Fig. 4(c). Obviously, the porous structure favors the mass transfer for gas reactions. 3.2. Hydrogen permeation Hydrogen permeation measurements were conducted on the original and the modified BCTCo membranes with the results presented in Figs. 5–7 and Table 1. Fig. 5 shows the impact of Pd modification on one membrane surface side or both sides on the hydrogen fluxes where the chemical plating time to deposit the Pd particles was fixed at 30 min (samples 2–4 in Table 1). For comparison purpose, the hydrogen fluxes of the original hollow fibre (sample-1 in Table 1) were also plotted. All the flux measurement was based on feed and sweep flow rates of 30 mL/min. As anticipated, two general patterns can be observed from Fig. 5(A). All the hydrogen fluxes increased with raising the operation temperature since a higher operating temperature facilitates the hydrogen diffusion and surface exchange reactions. Another observation is that all coated samples exhibited higher flux values than the original membrane. For example, tested at 1000 ◦ C, the hydrogen fluxes through the original, inner-surface coated, outer-surface coated and bothsurface coated samples were in the rising order of 0.164, 0.281, 0.307 and 0.382 mL cm−2 min−1 , respectively. Further inspection gives that coating the Pd particles on the outside or inner surface improved the flux values in a quite similar efficiency; for instance at 1000 ◦ C, the flux increment was around 70–80% by the Pd coating either on outside or inside membrane surface. This is an important observation which indicates the hydrogen disassociation to protons and proton re-association to hydrogen molecule on the original

Please cite this article in press as: J. Song, et al., Proton conducting perovskite hollow fibre membranes with surface catalytic modification for enhanced hydrogen separation, J Eur Ceram Soc (2016), http://dx.doi.org/10.1016/j.jeurceramsoc.2016.01.006

G Model JECS-10463; No. of Pages 9

ARTICLE IN PRESS J. Song et al. / Journal of the European Ceramic Society xxx (2016) xxx–xxx

5

Fig. 2. XRD patterns of the BCTCo hollow fibre membrane (original membrane (a) and after hydrogen permeation (b)) and TEM images of the BCTCo particles from the crushed BCTCo fresh membrane (c scale bar 20 nm and d scale bar 5 nm).

BCTCo hollow fibre membrane surfaces poses similar resistances on the overall hydrogen permeation. To efficiently improve the surface reaction kinetics, catalyst coating should be performed on both membrane surfaces. As expected, the flux improvement at 1000 ◦ C by both-side-coatings was enhanced up to 132% much more than the 70–80% by single-side-coating. Fig. 5(B) shows Arrhenius plots of hydrogen fluxes through the hollow fiber samples 1–4 (Table 1). Applying the Arrhenius equation, activation energies for hydrogen transport through fiber samples 1–4 under the conditions of Fig. 5(A) are 100.42, 94.49, 93.17, 90.50 kJ/mol, respectively, in a very good trend that coated membrane (sample 2 or 3) has a lower activation energy than the original membrane and sample-4 with both-side coatings has the smallest activation energy thus resulting in the largest hydrogen flux improvement. The Pd loading on the membrane surface is controlled by varying the amount of N2 H4 addition or the plating time, thus its effect on the hydrogen permeation was investigated with results shown in Fig. 6. As can be seen, the flux through the surface-modified fibres was continuously improved by increasing the plating time from 10, 20 to 30 min for a constant N2 H4 addition, and by increasing the N2 H4 addition from 120 ␮L to 240 and 360 ␮L. However, once a dense Pd film is formed, the hydrogen flux declined instead.

For example, at the N2 flow rate of 30 mL min−1 and operated at 900 ◦ C, the hydrogen fluxes through the original and the modified membranes with different Pd loadings obtained by varying plating conditions were 0.063, 0.09, 0.115, 0.126, 0.176, 0.188 and 0.119 mL cm−2 min−1 , respectively. This, again, confirms that there is an optimum Pd loading to achieve the best catalytic efficiency; and the deposition of a fully dense Pd layer with large thickness on the membrane surface would be meaningless and should be avoided [41]. Although the hydrogen permeation in the Pd-loaded perovskite hollow fibre membranes can be promoted by the Pd film bulk to promote the dissociation of hydrogen into atomic hydrogen or by the Pd-ceramic-gas three phase boundaries to catalytically dissociate hydrogen molecules into protons and electrons, the latter is obviously playing a greater role in the permeation improvement. As shown in Table 1 and Fig. 6, in term of hydrogen flux, sample-8 derived by 360 ␮L-N2 H4 solution addition and 120 min plating time gave the best performance evidenced by the largest flux improvement of 155% at 1000 ◦ C. Looking back again at the SEM images in Fig. 3c, it is consistent that the modification carried out under this circumstances produced the most uniform microstructure with Pd catalyst attachment thus providing the hollow fibre with the largest flux values. For this sample, the Pd loading capacity

Please cite this article in press as: J. Song, et al., Proton conducting perovskite hollow fibre membranes with surface catalytic modification for enhanced hydrogen separation, J Eur Ceram Soc (2016), http://dx.doi.org/10.1016/j.jeurceramsoc.2016.01.006

G Model JECS-10463; No. of Pages 9 6

ARTICLE IN PRESS J. Song et al. / Journal of the European Ceramic Society xxx (2016) xxx–xxx

Fig. 3. SEM micrographs of the Pd-coatings on the outer surface of the BCTCo hollow fibre membrane formed under different conditions: (a) 120 ␮L N2 H4 solution in 40 mL Pd2+ solution for 30 min; (b) 240 ␮L N2 H4 solution in 40 mL Pd2+ solution for 60 min; (c) 360 ␮L N2 H4 solution in 40 mL Pd2+ solution for 120 min; (d, e) 1200 ␮L N2 H4 solution in 100 mL Pd2+ solution for 240 min.

Fig. 4. SEM micrographs of the BCTCo hollow fibre membranes modified by Ni-coating (a) cross-sectional view before H2 -reduction; (b) top-surface before H2 -reduction; (c) top-surface after H2 -reduction.

was about 0.667 mg cm−2 and the coverage around 82%, which can be looked on as a reference for optimally loading the Pd catalyst. In consistency with this positive effect on flux enhancement, the calculated activation energy of this sample (sample-8 in Table 1) was reduced to 86.12 kJ/mol, the lowest value among these Pd coated membranes. Pd particles attached on the membrane surface help to improve the surface reaction kinetics thus the hydrogen fluxes significantly increase for these Pd-coated membranes through a mechanism of hydrogen spill-over effect from Pd particle to the BCTCo membrane surface. Fig. 7 compares the hydrogen fluxes as a function of temperature through the original and Ni-modified membranes. Again, the Ni-coated membranes gave higher hydrogen fluxes than the original fibre at every operation temperature point no matter whether the membrane was coated on inner surface or outer surface; the membranes with both surfaces modified by Ni exhibited the highest improvement. For example, at 950 ◦ C, the hydrogen flux through the unmodified BCTCo hollow fibre membrane was 0.105 mL cm−2 min−1 ; after surface modification, the hydrogen flux rose to 0.146, 0.157, and 0.173 mL cm−2 min−1 with the enhancement factor of 39, 50 and 65% by the coated Ni catalyst on the inside, outside and both sides of the BCTCo fibres, respectively. The reduced activation energy for these Ni coated samples in Table 1 also implies relative easier hydrogen permeation process compare to the original blank hollow fibre membrane. Furthermore, just analogous to the Pd coating, the hollow fibre with Ni layer coated on

the outer surface has slightly higher permeation fluxes than coated on the inner surface, which may be caused by the different catalyst loading on two different surfaces provided by the hollow fibre geometry. The inner surface of the hollow fibre has the surface area less than 20–30% compared to the external surface; thus less Ni or Pd loading on inner surface is expected and therefore causing a slightly reduced efficiency to promote the hydrogen flux. When the membrane surface catalyst coating efficiently enhances the hydrogen surface reaction kinetics to pose an ignorable resistance and the electronic conductivity of the membrane material is sufficiently high, in such ideal cases, the hydrogen flux would be completely controlled by the proton bulk diffusion and the maximum theoretical hydrogen flux can be calculated on the basis of Wagner equation: 

JH2 =

PH RT i ln  2 2 16F L PH

(3)

2

where T is the absolute temperature, R is the ideal gas constant, F is the Faraday‘s constant, L is the thickness of the membrane, i ’ represents the proton conductivity of BCTCo, PH and PH” repre2 2 sent the hydrogen partial pressure at the feed and permeate side. It is assumed that the BCTCo perovskite has equivalent proton conductivities with BaCe0.9 Y0.1 O3−ı at similar operating temperatures. The maximum theoretical hydrogen fluxes were calculated with results showing in Table 2. As can be seen, the measured fluxes

Please cite this article in press as: J. Song, et al., Proton conducting perovskite hollow fibre membranes with surface catalytic modification for enhanced hydrogen separation, J Eur Ceram Soc (2016), http://dx.doi.org/10.1016/j.jeurceramsoc.2016.01.006

G Model

ARTICLE IN PRESS

JECS-10463; No. of Pages 9

J. Song et al. / Journal of the European Ceramic Society xxx (2016) xxx–xxx

7

Table 2 Calculation results of the maximum theoretical hydrogen permeation flux values and comparison with actual measurement in this work. Perovskite

Tem.(◦ C)

 i (S cm−1 )

Theoretical JH2 (mL min−1 cm−2 )a

Measured JH2 (blank membrane)

Measured JH2 (Ni-coated)

Measured JH2 (Pd-coated)

BCTCo BCTCo BCTCo

700 800 900

0.014 [42] 0.02 [42] 0.04 [43]

0.24 0.34 0.62

0.010 0.021 0.063

0.018 0.0425 0.109

0.0385 0.0853 0.205

The theoretical calculation is based on the membrane thickness around 150 ␮m; feed gas flow rate of 15 mL/min (H2 + N2 ) and sweep gas flow rate (N2 ) of 30 mL/min.

Hydrogen permeation flux, mL cm-2 min-1

0.4

0.3

Original Pd-inside Pd-outside Pd-both sides

0.3

A

0.1 0 650

750

850 950 o Temperature, C

1

1050

Original Pd-inside Pd-outside Pd-both sides

0.1

0.01

Original Ni-inside Ni-outside Ni-both sides

0.2

0.2

B

Hydrogen permeation flux, mL cm-2 min-1

a

A

0.1

0 650

0.85

0.95

1.05

-1

1000/T, K

850 950 o Temperature, C

1

0.001 0.75

B

0.85 1000/T, K

Fig. 5. Comparison of the different Pd modified position and the unmodified hollow fibre membranes in hydrogen permeation (chemical plating time = 30 min, feed flow rate = 30 mL min−1 , sweep gas flow rate = 30 mL min−1 ). (A) Plot of hydrogen flux vs. temperature; (B) Arrhenius plot of hydrogen flux against reciprocal of temperature.

Original p40-r120-t10 Pd-10min p40-r120-t20 Pd-20 min p40-r120-t30 Pd-30min p40-r240-t60 Pd-60 min p40-r360-t120 Pd-120 min -t240 p100-r1200 Pd-240 min

0.4

Hydrogen permeation flux, mL cm-2 min-1

0.3 0.2

A

0.1 0 650

750

850 950 o Temperature, C

1050

1

B

1050

Original Ni-inside Ni-outside Ni-both sides

0.1

0.01

0.001 0.75

750

0.95

1.05

-1

Fig. 7. Comparison of the Ni-coating modified and the unmodified hollow fibre membranes in hydrogen permeation. (A) Plot of hydrogen flux vs. temperature; (B) Arrhenius plot of hydrogen flux against reciprocal of temperature.

in this work are only 3.75–33.32% of the theoretically calculated values depending on the operating temperature and membrane surface conditions. In particular, the measured hydrogen flux of blank membrane without surface modification is only 3.75–15.86% of the theoretical flux values mirroring that the surface reaction does contribute a very big resistance to limit the overall hydrogen transport. As expected, the surface modification on both membrane sides by Ni or Pd pushed the real flux rates closer to the theoretically calculated. For example, at 900 ◦ C, the Ni or Pd-coated membrane exhibited 17.69 or 33.32% of the theoretical value, respectively, from which, we can see that there is still some space to further increase the hydrogen flux either by surface modification or molecular doping to increase its electronic conductivity, which may also limit the hydrogen permeating behavior.

0.1

0.01

0.001 0.75

Original p40-r120-t10 Pd-10min p40 -r120 -t20 Pd-20 min p40 -r120 -t30 Pd-30 min p40 -r240 -t60 Pd-60 min p40 -r360min -t120 Pd-120 00-t240 p100-r12 Pd-240 min 0.85

4. Conclusions

0.95

1.05

1000/T, K-1 Fig. 6. Comparison of hydrogen fluxes through the Pd particles-coated on both sides of BCTCo hollow fibre membranes with different plating conditions (e.g. p40-r120t10 represents 40 mL plating solution, 120 ␮L N2 H4 aqueous solution and 10 min plating time). (A) Plot of hydrogen flux vs. temperature; (B) Arrhenius plot of hydrogen flux against reciprocal of temperature.

On the basis of self-made perovskite powder, gas-tight BaCe0.85 Tb0.05 Co0.1 O3−˛ (BCTCo) perovskite hollow fibre membranes were fabricated by spinning the BCTCo-containing polymer solution followed by sintering at 1350 ◦ C for 4 h. In addition to the proton bulk diffusion, the surface exchange kinetics plays an important role in determining the overall hydrogen permeation flux through the perovskite membranes. The membrane surface reaction kinetics can be improved by coating additional porous Ni or Pd particles as the catalyst on the membrane surface; in particular, coating on both sides of the membrane surfaces is more effective than on the single-side. Coating of Pd particles on both

Please cite this article in press as: J. Song, et al., Proton conducting perovskite hollow fibre membranes with surface catalytic modification for enhanced hydrogen separation, J Eur Ceram Soc (2016), http://dx.doi.org/10.1016/j.jeurceramsoc.2016.01.006

G Model JECS-10463; No. of Pages 9

ARTICLE IN PRESS J. Song et al. / Journal of the European Ceramic Society xxx (2016) xxx–xxx

8

sides of the hollow fibers in 40 mL plating solution (containing Pd(Ac)2 (4.4 g/L), EDTA-2Na (37.2 g/L) and NH3 ·H2 O (128.5 mL/L)) with 360 ␮L-N2 H4 solution addition as the reducer had been performed for 120 min, which could increase the hydrogen flux from original value of 0.164–0.413 mL cm−2 min−1 at 1000 ◦ C with improvement by 155%. A larger Pd particle loading on the membrane surface by an extension of chemical plating did not effectively result in hydrogen flux increment due to the requirement of optimum catalyst-BCTCo-hydrogen triple-phase boundary area for gas exchange reactions. The optimal Pd loading and coverage should be around 0.667 mg cm−2 and 82%, respectively for maximizing the permeation improvement. Our results also verify that performing Ni coating on the membrane surface is also effective to increase the hydrogen flux rates through the BCTCo hollow fibre membranes. Given its wider availability and lower material cost than Pd, Ni can be considered as the practical catalyst for membrane surface modification to promote the hydrogen permeation. However, to achieve the best separation efficiency with Ni catalyst, much more work is still needed to optimize the coating processing factors like the particle size, surface coverage, and heat-treatment temperatures. The comparison of experimental results with theoretical values confirms that to further improve the hydrogen flux, efforts should be focused not only on surface modification but also on membrane thickness reduction in particular for high temperature operations.

Acknowledgements The authors gratefully acknowledge the research funding provided by the National Natural Science Foundation of China (NSFC, No. 21176187, 21476131), the Research Fund for the Doctoral Program of Higher Education of China (RFDP 20131201110007) and the Program for Changjiang Scholars and Innovative Research Team in University (PCSIRT) of Ministry of Education of China (Grand no. IRT13084). Dr. Liu acknowledges the financial support provided by the Australian Research Council through the Future Fellow Program (FT12100178).

References [1] R. Dittmeyer, V. Hollein, K. Daub, Membrane reactors for hydrogenation and dehydrogenation processes based on supported palladium, J. Mol. Catal. A: Chem. 173 (2001) 135–184. [2] S. Tosti, Overview of Pd-based membranes for producing pure hydrogen and state of art at ENEA laboratories, Int. J. Hydrogen Energy 35 (2010) 12650–12659. [3] F. Gallucci, E. Fernandez, P. Corengia, M. van Sint Annaland, Recent advances on membranes and membrane reactors for hydrogen production, Chem. Eng. Sci. 92 (2013) 40–66. [4] H. Iwahara, Technological challenges in the application of proton conducting ceramics, Solid State Ion. 77 (1995) 289–298. [5] H. Iwahara, Proton conducting ceramics and their applications, Solid State Ionics 86–88 (part 1) (1996) 9–15. [6] H. Iwahara, Y. Asakura, K. Katahira, M. Tanaka, Prospect of hydrogen technology using proton-conducting ceramics, Solid State Ion. 168 (2004) 299–310. [7] T. Norby, Y. Larring, Mixed hydrogen ion–electronic conductors for hydrogen permeable membranes, Solid State Ion. 136–137 (2000) 139–148. [8] G. Marnellos, S. Zisekas, M. Stoukides, Synthesis of ammonia at atmospheric pressure with the use of solid state proton conductors, J. Catal. 193 (2000) 80–87. [9] M.Y. Cai, S. Liu, K. Efimov, J. Caro, A. Feldhoff, H.H. Wang, Preparation and hydrogen permeation of BaCe0.95 Nd0.05 O3 -delta membranes, J. Membr. Sci. 343 (2009) 90–96. [10] X. Qi, Y.S. Lin, Electrical conducting properties of proton-conducting terbium-doped strontium cerate membrane, Solid State Ion. 120 (1999) 85–93. [11] S. Hamakawa, L. Li, A. Li, E. Iglesia, Synthesis and hydrogen permeation properties of membranes based on dense SrCe0.95 Yb0.05 O3−˛ thin films, Solid State Ion. 148 (2002) 71–81.

[12] S.J. Song, E.D. Wachsman, J. Rhodes, S.E. Dorris, U. Balachandran, Hydrogen permeability of SrCe1−x Mx O3−ı (x = 0.05, M = Eu, Sm), Solid State Ion. 167 (2004) 99–105. [13] D. Dionysiou, X. Qi, Y.S. Lin, G. Meng, D. Peng, Preparation and characterization of proton conducting terbium doped strontium cerate membranes, J. Membr. Sci. 154 (1999) 143–153. [14] X. Qi, Y.S. Lin, Electrical conduction and hydrogen permeation through mixed proton–electron conducting strontium cerate membranes, Solid State Ion. 130 (2000) 149–156. [15] J. Guan, S.E. Dorris, U. Balachandran, L. Meilin, Transport properties of SrCe0.95 Y0.05 O3−ı and its application for hydrogen separation, Solid State Ion. 110 (1998) 303–310. [16] X. Wei, Y.S. Lin, Protonic and electronic conductivities of terbium doped strontium cerates, Solid State Ion. 178 (2008) 1804–1810. [17] Y. Liu, X. Tan, K. Li, Mixed conducting ceramics for catalytic membrane processing, Catal. Rev. 48 (2006) 145–198. [18] S. Cheng, V.K. Gupta, J.Y.S. Lin, Synthesis and hydrogen permeation properties of asymmetric proton-conducting ceramic membranes, Solid State Ion. 176 (2005) 2653–2662. [19] S. Zhan, X. Zhu, B. Ji, W. Wang, X. Zhang, J. Wang, W. Yang, L. Lin, Preparation and hydrogen permeation of SrCe0.95 Y0.05 O3−ı asymmetrical membranes, J. Membr. Sci. 340 (2009) 241–248. [20] S. Liu, X. Tan, K. Li, R. Hughes, Preparation and characterisation of SrCe0.95 Yb0.05 O2.975 hollow fibre membranes, J. Membr. Sci. 193 (2001) 249–260. [21] X. Tan, Y. Liu, K. Li, Preparation of La0.6 Sr0.4 Co0.2 Fe0.8 O3−ı hollow fiber membranes for oxygen production by a phase-inversion/sintering technique, Ind. Eng. Chem. Res. 44 (2005) 61–66. [22] X. Tan, J. Song, X. Meng, B. Meng, Preparation and characterization of BaCe0.95 Tb0.05 O3−˛ hollow fibre membranes for hydrogen permeation, J. Eur. Ceram. Soc. 32 (2012) 2351–2357. [23] X. Tan, N. Liu, B. Meng, S. Liu, Morphology control of the perovskite hollow fibre membranes for oxygen separation using different bore fluids, J. Membr. Sci. 378 (2011) 308–318. [24] Z. Wang, H. Liu, X. Tan, Y. Jin, S. Liu, Improvement of the oxygen permeation through perovskite hollow fibre membranes by surface acid-modification, J. Membr. Sci. 345 (2009) 65–73. [25] F.M. Figueiredo, V.V. Kharton, A.P. Viskup, J.R. Frade, Surface enhanced oxygen permeation in CaTi1−x Fex O3−ı ceramic membranes, J. Membr. Sci. 236 (2004) 73–80. [26] Z. Chen, Z. Shao, R. Ran, W. Zhou, P. Zeng, S. Liu, A dense oxygen separation membrane with a layered morphologic structure, J. Membr. Sci. 300 (2007) 182–190. [27] K.S. Lee, S. Lee, J.W. Kim, S.K. Woo, Enhancement of oxygen permeation by La0.6 Sr0.4 CoO3−ı coating in La0.7 Sr0.3 Ga0.6 Fe0.4 O3−ı membrane, Desalination 147 (2002) 439–444. [28] Y. Wang, H. Hao, J. Jia, D. Yang, X. Hu, Improving the oxygen permeability of Ba0.5 Sr0.5 Co0.8 Fe0.2 O3−ı membranes by a surface-coating layer of GdBaCo2 O5+ı , J. Eur. Ceram. Soc. 28 (2008) 3125–3130. [29] S. Lee, K.S. Lee, S.K. Woo, J.W. Kim, T. Ishihara, D.K. Kim, Oxygen-permeating property of LaSrBFeO3 (B = Co, Ga) perovskite membrane surface-modified by LaSrCoO3 , Solid State Ion. 158 (2003) 287–296. [30] X. Tan, Z. Wang, H. Liu, S. Liu, Enhancement of oxygen permeation through La0.6 Sr0.4 Co0.2 Fe0.8 O3−ı hollow fibre membranes by surface modifications, J. Membr. Sci. 324 (2008) 128–135. [31] A. Thursfield, I.S. Metcalfe, Air separation using a catalytically modified mixed conducting ceramic hollow fibre membrane module, J. Membr. Sci. 288 (2007) 175–187. [32] A. Leo, S. Liu, J.C. Diniz da Costa, The enhancement of oxygen flux on Ba0.5 Sr0.5 Co0.8 Fe0.2 O3−ı (BSCF) hollow fibers using silver surface modification, J. Membr. Sci. 340 (2009) 148–153. [33] H. Pan, L. Li, X. Deng, B. Meng, X. Tan, K. Li, Improvement of oxygen permeation in perovskite hollow fibre membranes by the enhanced surface exchange kinetics, J. Membr. Sci. 428 (2013) 198–204. [34] C. Yacou, J. Sunarso, C.X.C. Lin, S. Smart, S. Liu, J.C. Diniz da Costa, Palladium surface modified La0.6 Sr0.4 Co0.2 Fe0.8 O3−ı hollow fibres for oxygen separation, J. Membr. Sci. 380 (2011) 223–231. [35] G.C. Mather, D. Poulidi, A. Thursfield, M.J. Pascual, J.R. Jurado, I.S. Metcalfe, Hydrogen-permeation characteristics of a SrCeO3 -based ceramic separation membrane: thermal, ageing and surface-modification effects, Solid State Ion. 181 (2010) 230–235. [36] X. Meng, N. Yang, J. Song, X. Tan, Z.-F. Ma, K. Li, Synthesis and characterization of terbium doped barium cerates as a proton conducting SOFC electrolyte, Int. J. Hydrogen Energy 36 (2011) 13067–13072. [37] J. Song, L. Li, X. Tan, K. Li, BaCe0.85 Tb0.05 Co0.1 O3−ı perovskite hollow fibre membranes for hydrogen/oxygen permeation, Int. J. Hydrogen Energy 38 (2013) 7904–7912. [38] X. Tan, Y. Liu, K. Li, Mixed conducting ceramic hollow fiber membranes for air separation, AIChE J. 51 (2005) 1991–2000. [39] B. Meng, Z. Wang, X. Tan, S. Liu, SrCo0.9 Sc0.1 O3−ı perovskite hollow fibre membranes for air separation at intermediate temperatures, J. Eur. Ceram. Soc. 29 (2009) 2815–2822. [40] Z.-L. Xu, F. Alsalhy Qusay, Polyethersulfone (PES) hollow fiber ultrafiltration membranes prepared by PES/non-solvent/NMP solution, J. Membr. Sci. 233 (2004) 101–111.

Please cite this article in press as: J. Song, et al., Proton conducting perovskite hollow fibre membranes with surface catalytic modification for enhanced hydrogen separation, J Eur Ceram Soc (2016), http://dx.doi.org/10.1016/j.jeurceramsoc.2016.01.006

G Model JECS-10463; No. of Pages 9

ARTICLE IN PRESS J. Song et al. / Journal of the European Ceramic Society xxx (2016) xxx–xxx

[41] J. Song, B. Meng, X. Tan, S. Liu, Surface-modified proton conducting perovskite hollow fibre membranes by Pd-coating for enhanced hydrogen permeation, Int. J. Hydrogen Energy 40 (2015) 6118–6127. ˜ [42] M. Amsif, D. Marrero-López, A. Magrasó, J. Pena-Martínez, J.C. Ruiz-Morales, P. ˜ Synthesis and characterisation of BaCeO3 -based proton conductors Núnez,

obtained from freeze-dried precursors, J. Eur. Ceram. Soc. 29 (2009) 131–138. [43] G. Ma, T. Shimura, H. Iwahara, Simultaneous doping with La3+ and Y3+ for Ba2+ - and Ce4+ -sites in BaCeO3 and the ionic conduction, Solid State Ion. 120 (1999) 51–60.

Please cite this article in press as: J. Song, et al., Proton conducting perovskite hollow fibre membranes with surface catalytic modification for enhanced hydrogen separation, J Eur Ceram Soc (2016), http://dx.doi.org/10.1016/j.jeurceramsoc.2016.01.006

9