Journal of Membrane Science 246 (2005) 173–180
Morphological study of supported thin Pd and Pd–25Ag membranes upon hydrogen permeation Y. Zhang∗ , M. Komaki, C. Nishimura Eco-Energy Materials Group, Ecomaterials Center, National Institute for Materials Science, Sengen 1-2-1, Tsukuba, Ibaraki 305-0047, Japan Received 16 June 2004; received in revised form 18 June 2004; accepted 2 September 2004 Available online 27 October 2004
Abstract Morphological change of Pd and Pd–25Ag membranes supported by V–15Ni alloy upon hydrogen permeation was investigated in the temperature range 423–673 K. The supported Pd–25Ag membrane exhibited higher resistance to hydrogen-induced cracking and grain growth than the supported Pd membrane. Long-term permeation of Pd–25Ag/V–15Ni composite membrane was carried out at 573 and 673 K for 200 h. There was no strong metallic interdiffusion between the Pd–25Ag membrane and the V–15Ni support after the long-term permeation at 573 K but small amounts of oxide had formed on the surface of Pd–25Ag membrane. Whisker and fissure-oxide morphologies were dominant on the exit and entrance side of the Pd–25Ag/V–15Ni composite membrane, respectively, accompanied by severe metallic interdiffusion after the long-term permeation at 673 K. AES and FE-SEM results revealed that metallic interdiffusion and selective oxidation of vanadium were responsible for the deterioration of Pd–25Ag membrane at 673 K. Hydrogenation–dehydrogenation of Pd and Pd–25Ag membranes supported by stainless steel and V–15Ni alloy were in situ examined by an optical microscope. The formation of hydride was uniform in the Pd/V–15Ni sample but localized in the Pd–25Ag/V–15Ni sample, suggesting that the hydrogen transfer through interface was strongly dependent on the composition of Pd alloy membranes. As for the stainless steel supported samples, both Pd and Pd–25Ag membranes had fractured. © 2004 Elsevier B.V. All rights reserved. Keywords: Hydrogen permeation; Palladium; Composite membranes
1. Introduction Palladium and its alloys are important for hydrogen separation membranes due to their high hydrogen permeability and excellent catalysis [1]. Since Pd is expensive and very limited in natural resource, composite membranes consisting of thin palladium alloy membranes supported by dense or porous substrates are of particular interest [2–7]. If supported by porous substrates, Pd alloy membrane determines total hydrogen selectivity and permeability of the composite membrane. There are numerous reports about the preparation and characterization of Pd alloy membranes on ∗ Corresponding author. Present address: MEMS and Packaging Group, Advanced Manufacturing Research Institute, National Institute of Advanced Industrial Science and Technology (AIST), Namiki 1-2-1, Tsukuba, Ibaraki 305-8564, Japan. Tel.: +81 29 861 7297; fax: +81 29 861 7167. E-mail address:
[email protected] (Y. Zhang).
0376-7388/$ – see front matter © 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.memsci.2004.09.002
porous supports [5–10]. Hydrogen selectivity of the composite membrane is no longer an issue if dense support is used. Nevertheless, hydrogen permeability of the dense supports becomes a main concern. Several candidate materials with high hydrogen permeability have been developing [2–4,11–15]. In fact, our recent works showed that Pd membrane has apparent negative effects on hydrogen permeation through dense composite membranes [16]. The negative effects of Pd membrane mainly result from its low resistance to hydrogeninduced cracking, and barrier effects of interface and surface on hydrogen permeation. The former could be improved by replacing Pd with Pd alloys such as Pd–Ag, Pd–Fe and Pd–Cu [17–20], whereas the barrier effects of interface and surface are difficult to be reduced. The barrier effects would become more and more severe with decreasing membrane thickness. Even for single-layer Pd alloy membranes, hydrogen permeability has been observed to decrease
174
Y. Zhang et al. / Journal of Membrane Science 246 (2005) 173–180
with the membrane thickness because of surface resistance [21–24]. The surface resistance through single-layer palladium membrane with a thickness of 5 m is equally important as diffusion resistance, whereas it becomes dominating (85%) at a thickness less than 1 m [24]. Ultra-thin dense composite membranes, however, are preferred in practical applications for high permeation flux. Many applications such as membrane reactors demand dense ultra-thin composite membranes with excellent performance at relatively low temperature. For instance, hydrogen separation membrane has to work at around 473 K in micro power systems for portable electronics [25,26]. Our previous work showed that hydrogen permeation through Pd–Ag/V–15Ni alloy composite membrane with 40 m thickness is surface reaction-limited at temperatures lower than 473 K [16]. Thus, it is necessary to investigate the surface and interface reactions of hydrogen permeation through palladium alloy membranes supported by dense substrates for practical applications. Unfortunately, it is difficult to examine the interface and surface reactions directly using gas permeation technology. There are few reports on this topic. Morphological examination could offer indirect but important information, particularly about surface reactions [27,28]. In this work, we investigated the hydrogen permeation of thin Pd and Pd–25Ag membranes with 100 nm thickness on V–15Ni alloy with 40 m thickness. We also prepared 100 nm thick Pd and Pd–25Ag membranes on 1 mm thick stainless steel and V–15Ni support, respectively, and their hydrogenation–dehydrogenation behavior was in situ examined under a homemade optical microscope.
2. Experimental Ingots of V–15Ni alloy were prepared by arc melting in an argon atmosphere, using pure vanadium (99.9%) and nickel (99.9%) as raw materials. The ingots were hot rolled, annealed and then cold rolled into strips of about 40 m thickness. The strips were annealed at 1573 K for 15 min under a vacuum of 4 × 10−4 Pa and rapidly cooled by argon flow. The strips were then cut into circular substrates with 12 mm in diameter and chemically cleaned using a solution of hydrofluoric acid:nitric acid:lactic acid = 1:1:1. The 100 nm thick Pd–25Ag and Pd membranes were prepared in a DC multi-target sputtering system with two separate Pd and Ag targets. Hydrogen permeation experiments were carried out using pure hydrogen in a conventional gas permeation apparatus. Short-term permeation was carried out under pressure of ∼1 × 105 Pa within the temperature range of 423–673 K at intervals of 50 K. The duration of short-term permeation at each temperature was about one and half hour. The long-term experiments were carried out at 573 and 673 K for 200 h under pressure of 4 × 104 Pa. Hydrogen permeability (Φ) was determined via the measurement of hydrogen permeation flux (J), inlet pressure (Pu ) and outlet pressure
Fig. 1. Schematic representation of in situ optical microscope.
(Pd ) at steady state, 1/2
J=
Φ(Pu
1/2
− Pd ) L
(1)
where L denotes the thickness of sample. The outlet pressure was neglected in this work because the outlet side was kept in vacuum during permeation experiment. FE-SEM and AES methods were used to examine the composite membranes before and after permeation. Fig. 1 shows a schematic of the homemade microscope in which morphological changes of sample during hydrogenation/dehydrogenation procedure was recorded by a CCD camera. Stainless steel and V–15Ni alloy substrates with about 1 mm thickness were used. V–15Ni alloy substrates with 1 mm thickness were prepared as described in our previous study [11,29]. The stainless steel substrates were directly machined from 1 mm thick sheets and polished by emery paper. The 100 nm thick Pd and Pd–25Ag membranes were prepared on the substrates using the aforementioned sputtering procedure. The experimental details in the microscope were as follows: (1) The chamber was evacuated to 5 × 10−4 Pa after loading of sample. (2) Sample was heated to 673 K by 10 K/min. (3) Pure hydrogen was introduced to 1.0 × 105 Pa and kept for 10 min; then evacuated. (4) Sample was cooled to 573, 473 and 423 K by 10 K/min, respectively, and Step (3) was repeated. After the above experiments, samples were taken out and examined in FE-SEM.
3. Results and discussion 3.1. Hydrogen permeation The short-term permeation experiments showed that there was no obvious difference between hydrogen permeability of the Pd/V–15Ni and Pd–25Ag/V–15Ni composite membranes at temperatures higher than 473 K. The Pd–25Ag/V–15Ni
Y. Zhang et al. / Journal of Membrane Science 246 (2005) 173–180
175
Fig. 2. Relationship between permeation flux and permeation duration of Pd–25Ag/V–15Ni composite membranes.
composite membrane had higher hydrogen permeability than the Pd/V–15Ni composite membrane at temperatures lower than 473 K, which had been discussed in detail in our previous work [16,29]. Fig. 2 shows hydrogen permeation flux through the Pd–25Ag/V–15Ni composite membranes as function of time during the long-term permeation. There was no apparent decline for the curve of 573 K, whereas the permeation flux decreased to less than 70% of the initial value after 200 h permeation at 673 K. During the long-term permeation at 673 K, baking treatment was carried out in order to reactivate the Pd–25Ag membrane. The surface contaminations of palladium alloy could be removed during the baking treatment by the oxidation–reduction of palladium metal [30]. The value of permeation flux at 673 K increased greatly just after the baking treatment but dropped very rapidly again. The baking treatment exhibited a temporary-reactivation effect on the Pd–25Ag membrane. In fact, with continued permeation, such temporary-reactivation effect had decreased. It indicated that the supported Pd–25Ag membrane had deteriorated permanently at 673 K.
Fig. 3. AES surface analysis of the supported Pd–25Ag membranes: (a) as-prepared; after 200 h permeation at (b) 573 K and (c) 673 K, respectively.
Fig. 4. AES depth profile of the supported Pd–25Ag membranes: (a) asprepared; after 200 h permeation at (b) 573 K and (c) 673 K, respectively.
Fig. 5. SEM images of as-prepared Pd and Pd–25Ag membranes.
176
Y. Zhang et al. / Journal of Membrane Science 246 (2005) 173–180
Fig. 3 shows AES surface analysis of the supported Pd– 25Ag membrane as prepared, after the long-term permeation at 573 and 673 K, respectively. After the long-term permeation at 573 K, there was a small peak of oxygen. After the long-term permeation at 673 K, the intensities of peaks of Ag and Pd element reduced greatly and there were strong peaks of oxygen and vanadium, suggesting the formation of vanadium-rich oxide. Fig. 4 shows AES depth profiles of the Pd–25Ag/V–15Ni composite membrane as prepared, and after the long-term permeation at 573 and 673 K, respectively. The metallic interdiffusion was so weak at 573 K that the depth profile of supported Pd–25Ag membrane was still wave-like after 200 h permeation. The wave-like depth profile resulted from the rotating of targets during sputtering [16]. Fig. 4(c) shows that severe metallic interdiffusion had occurred between the Pd–25Ag membrane and the V–15Ni alloy support at 673 K. There were vanadium enrichments on the membrane surface after the long-term permeation at 673 K. 3.2. SEM observation Fig. 5 shows SEM images of as-sputtered Pd and Pd–25Ag membranes. XRD analysis indicated that both membranes
had single-phase palladium structure. The Pd–25Ag membrane had smaller grains than the Pd membrane. The average grain size of the Pd–25Ag and Pd membrane was about 10 and 20 nm in diameter, respectively. Fig. 6 shows SEM images of the Pd/V–15Ni and Pd–25Ag/V–15Ni composite membranes after short-term hydrogen permeation. It was difficult to identify grain size of the supported Pd–25Ag membrane here. Some particles could be seen on the supported Pd–25Ag membrane at the exit side of hydrogen permeation. Apparent grain growth had occurred in the supported Pd membrane. Grain growth was more pronounced in the Pd membrane at exit side than at entrance side. Fig. 6(d) also shows that the grain growth of the Pd membrane was not uniform. Some grains were as large as 0.5 m in diameter while others were only about 0.1 m in diameter. Phenomenon of grain growth has also been observed in single Pd–25Ag membrane [31]. It was unclear what influence the grain growth of palladium alloy membranes would have on hydrogen permeation, but a thermally stable membrane is always desired in practical applications. Pd–25Ag membrane seemed more thermally stable than Pd membrane. The Pd membrane was cracked at entrance side mainly along grain boundary. Fig. 7 shows representative SEM images of the Pd–25Ag/V–15Ni composite membrane after long-term
Fig. 6. SEM images of the composite membranes after the short-term hydrogen permeation in the temperature range of 423–673 K: (a) and (c) Pd–25Ag/V–15Ni; (b) and (d) Pd/V–15Ni. (a) and (b) Hydrogen feed side; (c) and (d) hydrogen outlet side.
Y. Zhang et al. / Journal of Membrane Science 246 (2005) 173–180
177
Fig. 7. Representative SEM images of Pd–25Ag/V–15Ni composite membranes after the long-term hydrogen permeation at 573 (a) and (c) and 673 K (b) and (d). (a) and (b) Hydrogen feed side; (c) and (d) hydrogen outlet side.
permeation at 573 and 673 K. Fig. 7(a) and (c) shows that there was fissure-like morphology on the surface of Pd–25Ag/V–15Ni composite membrane. It indicated that some oxides had formed, which was consistent with the AES surface analysis (see Fig. 3(b)). We, however, did not find the presence of vanadium using AES and FE-SEM. Then the fissure morphology was mainly due to the oxidation of Pd–25Ag alloy. Since Fig. 2 shows that the permeation flux did not decline during permeation, the oxidation of Pd–25Ag alloy might have formed after the permeation process. Fig. 7(b) indicates that a layer of whisker-like oxide had formed at the entrance side of the Pd–25Ag/V–15Ni composite membrane after long-term permeation at 673 K. There was fissure morphology at the exit side, suggesting formation of oxides. The AES results demonstrated that vanadium and oxygen were enriched on the surface of composite membrane after long-term permeation at 673 K. It could be inferred that the whisker and fissure morphology mainly resulted from the formation of vanadium-rich oxides. Thus, we could draw the conclusions that selective oxidation of vanadium had occurred during long-term permeation at 673 K. The vanadium-rich oxide layer reduced the effective area for
hydrogen permeation and permeation flux in turn. Then, the deterioration of Pd–25Ag coated composite membranes was not only due to metallic interdiffusion between the Pd–25Ag membrane and V–15Ni alloy support but also selective oxidation of vanadium. Selective oxidation of vanadium could explain the aforementioned temporary-reactivation effect of the baking treatment at 673 K as follows. As vanadium has higher affinity to oxygen than palladium and silver, it would be oxidized selectively when its concentration is higher than a critical value. Generally, the critical concentration of a metal for its selective oxidation is about 10–20 at.% [32]. Fig. 4(c) shows that the vanadium concentration in the supported Pd–25Ag membrane was rather low. It seemed that the selective oxidation of vanadium could not occur. If environment is reductive, the critical value, however, will decrease to only several percentages and even lower [32]. In this work, all the samples were in reductive environment during hydrogen permeation. It was possible for vanadium to be selectively oxidized at low concentrations. The effective area for hydrogen permeation would be thus reduced with the formation of vanadium oxides and permeation flux decreased in return. It is known that the oxide layer formed selectively
178
Y. Zhang et al. / Journal of Membrane Science 246 (2005) 173–180
Fig. 8. Optical images of Pd–25Ag membrane (a)–(c) and Pd membrane (d)–(f) supported by V–15Ni alloy during hydrogenation/dehydrogenation procedure under the in situ microscope.
in reductive environment would be destroyed by introduction of oxygen or air [32]. During the baking treatment, the introducing of air would destroy the layer of vanadium-rich oxide formed selectively. The effective area for hydrogen permeation thus increased. When hydrogen was introduced again, permeation flux reduced because of the selective oxidation of vanadium. 3.3. Hydrogen microscope observation Under the optical microscope, morphological change of samples was recorded by CCD camera. Representative images were shown in Fig. 8. There was no visible change in the V–15Ni alloy supported samples until the sample temperature decreased to 423 K. There were some localized deformation in the Pd–25Ag/V–15Ni sample (see Fig. 8(b)) and several stripes in the Pd/V–15Ni sample at 423 K (see Fig. 8(e)), respectively. After evacuated, these morphologies had disappeared but there was some crack-like morphology left in the
Pd/V–15Ni sample. It indicated that hydrides had formed in all the V–15Ni alloy supported samples at 423 K. Fig. 8(e) shows that although the hydride formation in the Pd/V–15Ni sample was uniform, its depth was shallower than that in the Pd–25Ag/V–15Ni sample. There was no visible morphological change for the Pd/stainless steel and Pd–25Ag/stainless steel under the hydrogen microscope. All the samples were examined under FE-SEM soon after they were taken out from the optical microscope. Fig. 9(a) and (b) shows that many pores had formed in the Pd membrane supported by V–15Ni alloy, whereas the supported Pd–25Ag membrane on V–15Ni alloy was still dense and compact. It indicated that Pd membrane had fractured but Pd–25Ag membrane did not during the hydrogenation/dehydrogenation. A slight grain growth was identified in the supported Pd–25Ag membrane under higher magnifications. As the formation of hydride had occurred in all the V–15Ni supported samples, it was difficult to understand why the Pd/V–15Ni sample had fractured. One possibility was that the formation of hy-
Y. Zhang et al. / Journal of Membrane Science 246 (2005) 173–180
179
Fig. 9. SEM images of the supported Pd (a) and (c) and Pd–25Ag membranes (b) and (d). (a) and (b) V–15Ni alloy supports; (c) and (d) stainless steel supports.
dride in the Pd membrane was limited in the area near to surface. Fig. 9(c) and (d) show that pore-like and particle-like morphologies were dominant in both supported Pd and Pd–25Ag membranes on stainless steel. Another character of Fig. 9(c) and (d) was that the surface morphology of supported Pd alloy membranes was dependent on the grain orientation of stainless steel. The above results indicated that palladium alloy membranes and metal supports had complicated effects on each other. As hydrogen solubility of stainless steel is very low, we could regard V–15Ni and stainless steel as representative supports with high affinity and low affinity to hydrogen, respectively. It was understandable that palladium alloy membrane would crack or become porous during hydrogenation/dehydrogenation if supported by materials with low hydrogen affinity such as stainless steel. If supported by materials with high affinity to hydrogen, Pd alloy membrane with higher hydrogen solubility had higher resistance to hydrogen-induced failure. Pd alloy membranes also had complicated effects on hydrogenation–dehydrogenation of metal supports. The formation mode of hydride in metal supports, i.e. V–15Ni alloy here, was dependent on the composition of
Pd alloy membranes, suggesting that the distribution of hydrogen atoms in the surface layer and beneath in V–15Ni alloy supports was dependent on the Pd alloy membranes. In other words, hydrogen transfer from Pd alloys to V–15Ni support seems dependent on the Pd alloy membranes. In Pd/V–15Ni sample, the distribution of hydrogen in the side of V–15Ni support seems uniform but the depth that hydrogen could reach was shorter. In Pd–25Ag/V–15Ni sample, the hydrogen distribution in the side of V–15Ni was not uniform but the hydrogen diffusion depth was deeper than that of the Pd/V–15Ni sample. It indicated that the hydrogen content in the V–15Ni support adjacent to interface was higher in case of Pd–25Ag than that for Pd membrane. Localized hydride formation in the Pd–25Ag/V–15Ni sample was unclear.
4. Conclusion Through morphology study in this work, it was found that the selective oxidation of vanadium and metallic interdiffusion were responsible for the deterioration of Pd–25Ag/V–15Ni composite membranes at 673 K. The selective oxidation of vanadium could explain that Pd alloy
180
Y. Zhang et al. / Journal of Membrane Science 246 (2005) 173–180
membranes supported by vanadium group metals always deteriorated at a faster rate by orders of magnitude than the prediction only based on metallic interdiffusion. There were complicated bilateral effects between thin Pd alloy membranes and their support metals during the hydrogenation/dehydrogenation procedure. Pd alloy membrane with higher hydrogen solubility had significant effects on hydride formation in support metals and improved their resistance to hydrogen-induced cracking. Inversely, support metals with very low hydrogen solubility could make the supported Pd alloy membranes subject to hydrogen-induced cracking. In fact, support metals with too high hydrogen solubility, e.g. Nb, are not applicable for the dense composite membranes because they would crack during hydrogen permeation due to too high hydrogen solubility [33]. It indicated that support metals with hydrogen solubility compatible with Pd alloy membranes are necessary. Searching such support metals is on going in our group. At present, we focus our efforts on vanadium-based alloys. Acknowledgements This study was supported by the Research Funds for Atomic Energy from Ministry of Education, Culture, Sports, Science and Technology. We also thank Mr. T. Kimura and Mr. K. Nishida for AES analysis, and Mr. Awane for FE-SEM observation. References [1] S.N. Paglieri, J.D. Way, Innovations in palladium membrane research, Sep. Purif. Meth. 31 (1) (2002) 1. [2] R.E. Buxbaum, T.L. Marker, Hydrogen transport though non-porous membranes of palladium-coated niobium, tantalum and vanadium, J. Membr. Sci. 85 (1993) 29. [3] T.S. Moss, N.M. Peachey, R.C. Snow, R.C. Dye, Multilayer metal membranes for hydrogen separation, Int. J. Hydrogen Energy 23 (2) (1998) 99. [4] N.M. Peachey, R.C. Snow, R.C. Dye, Composite Pd/Ta metal membranes for hydrogen separation, J. Membr. Sci. 111 (1996) 123. [5] S. Uemiya, N. Sato, H. Ando, Y. Kude, T. Matsuda, K. Kikuchi, Separation of hydrogen through palladium thin film supported on a porous glass tube, J. Membr. Sci. 56 (1991) 303. [6] B. McCool, G. Xomeritakis, Y.S. Lin, Composition control and hydrogen permeation characteristics of sputter deposited palladium–silver membranes, J. Membr. Sci. 161 (1999) 67. [7] W. Li, R. Liang, Hughes, Fabrication of defect-free Pd/␣-Al2 O3 composite membranes for hydrogen separation, Thin Solid Films 350 (1999) 106. [8] J. Shu, B.P.A. Grandjean, E. Ghali, S. Kaliaguine, Simultaneous deposition of Pd and Ag on porous stainless steel by electroless plating, J. Membr. Sci. 77 (1993) 181. [9] J. O’Brien, R. Hughes, J. Hisek, Pd/Ag membranes on porous alumina substrates by unbalanced magnetron sputtering, Surf. Coat. Technol. 142–144 (2001) 253. [10] W. Li, R. Liang, Hughes, Characterization and permeation of palladium/stainless steel composite membranes, J. Membr. Sci. 149 (1998) 259. [11] M. Nishimura, S. Komaki, M. Hwang, Amano, V–Ni alloy membranes for hydrogen purification, J. Alloys Comp. 330–332 (2002) 902.
[12] Nishimura, M. Komaki, M. Amano, Hydrogen permeation characteristics of vanadium–molybdenum alloys, Trans. Mater. Res. Soc. Jpn. 18B (1994) 1273. [13] Y. Zhang, T. Ozaki, M. Komaki, C. Nishimura, Hydrogen permeation characteristics of vanadium–aluminium alloys, Scripta Mater. 47 (2002) 601. [14] S. Hara, K. Sakaki, N. Itoh, H.M. Kimura, K. Asami, A. Inoue, An amorphous alloy membrane without noble metals for gaseous hydrogen separation, J. Membr. Sci. 164 (2000) 289. [15] K. Hashi, K. Ishikawa, T. Matsuda, K. Aoki, Hydrogen permeation characteristics of multi-phase Ni–Ti–Nb alloys, J. Allo. Comp. 368 (2004) 215. [16] Y. Zhang, T. Ozaki, M. Komaki, C. Nishimura, Hydrogen permeation of Pd–Ag alloy coated V–15Ni composite membrane: effects of overlayer composition, J. Membr. Sci. 224 (2003) 81. [17] B. McCool, G. Xomeritakis, Y.S. Lin, Composition control and hydrogen permeation characteristics of sputter deposited palladium–silver membranes, J. Membr. Sci. 161 (1999) 67. [18] K.J. Bryden, J.Y. Ying, Nanostructured palladium–iron membranes for hydrogen separation and membrane hydrogenation reactions, J. Membr. Sci. 203 (2002) 29. [19] F. Roa, M.J. Block, J.D. Way, The influence of alloy composition on the H2 flux of composite Pd–Cu membranes, Desalination 147 (2002) 411. [20] F. Roa, J.D. Way, R.L. McCormick, S.N. Pagliere, Preparation and characterization of Pd–Cu composite membranes for hydrogen separation, Chem. Eng. J. 93 (2003) 11. [21] J.P. Collins, J.D. Way, Preparation and characterization of a composite palladium–ceramic membrane, Ind. Eng. Chem. Res. 32 (1993) 3006. [22] Y. Li, Y.T. Cheng, Hydrogen diffusion and solubility in palladium thin films, Int. J. Hydrogen Energy 21 (4) (1996) 281. [23] R.C. Hurlbert, J.O. Konecny, Diffusion of hydrogen through palladium, J. Chem. Phys. 34 (2) (1961) 655. [24] A. Criscuoli, A. Basile, E. Drioli, O. Loiacono, An economic feasibility study for water gas shift membrane reactor, J. Membr. Sci. 181 (2001) 21. [25] S.V. Karnik, M.K. Hatalis, M.V. Kothare, Towards a palladium micro-membrane the water gas shift reaction: microfabrication approach and hydrogen purification results, J. Microelectro. Syst. 12 (1) (2003) 93. [26] H.D. Tong, J.W. Berenschot, M.J. De Beor, J.G.E. Gardeniers, H. Wensink, H.V. Jansen, W. Nijdam, M.C. Elwenspoek, F.C. Gielens, C.J.M. van Rijn, Microfabrication of palladium–silver alloy membranes for hydrogen separation, J. Microelectro. Syst. 12 (5) (2003) 622. [27] D.W. Lee, Y.G. lee, S.E. Nam, S.K. Ihm, K.H. Lee, Study on the variation of morphology and separation behavior of the stainless steel and supported membranes at high temperature, J. Membr. Sci. 220 (2003) 137. [28] J. Shu, B.E.W. Bondondo, B.P.A. Grandjean, S. Kaliguine, Morphological changes of Pd–Ag membranes upon hydrogen permeation, J. Mater. Sci. Lett. 16 (1997) 294. [29] Y. Zhang, M. Komaki, C. Nishimura, Hydrogen permeation and diffusion of Pd and Pd–25Ag-coated V–15Ni alloy composite membranes, Int. J. Hydrogen Energy, submitted for publication. [30] R.G. Musket, Effects of contamination on the interaction of hydrogen gas with palladium: a review, J. Less-Common Met. 45 (1976) 173. [31] S. Tosti, A. Adrover, A. Basile, V. Camilli, G. Chiappetta, V. Violante, Characterization of thin wall Pd–Ag rolled membranes, Int. J. Hydrogen Energy 28 (2003) 105. [32] P. Kofstod, High-temperature Oxidation of Metals, Wiley, New York, 1966. [33] S. Tosti, Supported and laminated Pd-based metallic membranes, Int. J. Hydrogen Energy 28 (2003) 1445.