- Email: [email protected]
The hydrogen permeability of Cu–Zr binary amorphous metallic membranes and the importance of thermal stability
Journal of Membrane Science 489 (2015) 264–269
Contents lists available at ScienceDirect
Journal of Membrane Science journal homepage: www.elsevier.com/locate/memsci
The hydrogen permeability of Cu–Zr binary amorphous metallic membranes and the importance of thermal stability Tianmiao Lai, Huidan Yin, Mary Laura Lind n School for Engineering of Matter, Transport and Energy, Arizona State University, Tempe, AZ 85287-6106, USA
art ic l e i nf o
a b s t r a c t
Article history: Received 30 January 2015 Received in revised form 11 March 2015 Accepted 22 March 2015 Available online 30 April 2015
We synthesized three compositions of amorphous metallic Cu–Zr (Zr ¼37, 54, 60 at%) membranes by splat quenching. We measured the thermal properties with differential scanning calorimetry (DSC). We tested hydrogen permeability of the membranes before and after Pd coatings. Experimentally, we found that the hydrogen permeability of the alloys is lower than that predicted by existing modeling results; we hypothesize this may be the result of oxides formed on the surface of our membranes. The formation of oxides is composition dependent. The permeability change during the testing time indicates that hydrogen has promoted structural change at a temperature lower than the glass transition temperature (Tg). Our original experimental results show that thermal stability is critical for hydrogen separation applications in amorphous metallic membranes. & 2015 Elsevier B.V. All rights reserved.
Keywords: Metallic membranes Cu–Zr amorphous metals Hydrogen permeability Surface oxides Stability
1. Introduction In the United States, 95% of the industrially produced hydrogen is from natural gas reforming [1]. Pd-based alloys are the benchmark metallic membrane materials for hydrogen separation. Pdbased membranes have pure hydrogen permeability of 10 8 mol m 1 s 1 Pa 0.5 and infinite H2/N2 selectivity at temperatures up to 600 K [2–4]. However, at 1 bar and at temperatures below 573 K, pure palladium suffers an α/β palladium hydride phase transformation [5]. The lattice expansion mismatch between the α and β phases results in mechanical failure after long time exposures to hydrogen. The high cost of palladium also limits widespread application, so researchers are interested in lower cost metallic materials [6]. Amorphous metals are one alternative to replace palladium-based membranes [2]. Density functional theory calculations and quantum Monte Carlo simulations have demonstrated that amorphous metallic films can have hydrogen permeability close to pure Pd, and predict less hydrogen-induced expansion compared to their crystalline counterparts [7]. Experimentally, Hara et al. found that the amorphous alloy Zr36Ni64, without a Pd surface coating, has hydrogen permeability only one order of magnitude lower than Pd77Ag23 at 623 K [8].
n
Corresponding author. Tel.: þ 1 480 727 8613; fax: þ 1 480 727 9321. E-mail address: [email protected] (M.L. Lind).
http://dx.doi.org/10.1016/j.memsci.2015.03.098 0376-7388/& 2015 Elsevier B.V. All rights reserved.
Amorphous metals are formed through rapid cooling of a molten alloy. The resulting amorphous metallic alloys have no long-range crystalline order [9]. The critical cooling rate is strongly dependent on the alloy composition or glass forming ability (GFA) of the composition. To further improve the GFA and hydrogen permeability of binary Zr–Ni alloys, ternary alloys in which Zr has been substituted partially by other elements like Nb, Ta and Ti have been investigated [10–13]. Paglieri et al. and Yamamura et al. found that for multi-component Zr-based amorphous alloy compositions, higher concentrations of Zr yielded higher hydrogen permeability [10,14]. This may be a result of the strong affinity of Zr for hydrogen combined with the amorphous structure of the alloys [15,16]. However, severe hydrogen embrittlement rapidly occurred in alloys with a Zr concentration higher than 40% [14,17]. Thermal stability is an important parameter in membrane performance. A significant limitation to potential use of amorphous metallic membranes for high-temperature hydrogen separations is related to their metastability. Because these materials are thermodynamically metastable, they possess a super cooled liquid region (SCLR) which is bound by the glass transition (Tg) (on the low end) and crystallization temperatures (Tx) (on the high end) [18]. It is well established that holding an amorphous metal at the Tg or at temperatures within the SCLR for long times results in crystallization [19]. For membrane separation applications, the hydrogen permeability is known to decrease after crystallization of the amorphous metal [20]. Therefore, it is preferable to use amorphous metals with a Tg that is higher than the temperature at which the process is being performed, in order to maintain the
T. Lai et al. / Journal of Membrane Science 489 (2015) 264–269
amorphous structure. Additionally, the presence of hydrogen can impact the thermal stability of amorphous metals by promoting crystallization [21]. Dini et al. found that electrolytically hydrogenated amorphous Cu–Zr alloys have lower Tx than as-cast amorphous Cu–Zr alloys [22]. The Cu–Zr binary amorphous alloy system has drawn interest because of its good mechanical properties and low cost [23,24]. Xu et al. and Zhang et al. found a series of Cu–Zr binary compositions that are bulk metallic glass formers [25,26]. Calculation of the Time–Transformation–Temperature (TTT) curve is a method to predict the GFA of amorphous alloys [23]. Ge et al. used the CALPHAD method to estimate TTT curves of Cu–Zr alloys [23]. They found that the critical cooling rate to form amorphous alloys within the Cu–Zr system is between 4.32 102–2.63 104 K/s. They calculated that the Tg of Cu–Zr alloys ranges from 620 K to 750 K [23]. This makes Cu–Zr alloys suitable for hydrogen separation at operation temperatures near 573 K. The specific composition Zr54Cu46 has critical casting thicknesses up to 2 mm (Tg:696 K, Tx:746 K) [25]. Hao et al. used first principles density functional theory to calculate the hydrogen permeability of a series of binary amorphous metals [27]. Their results predicted that Zr54Cu46 could have hydrogen permeability comparable to pure Pd at 600 K [27]. In this paper, we measured the experimental hydrogen permeability of Cu–Zr amorphous free-standing membranes. We investigated alloys with three different Zr concentrations. Our experimentally measured hydrogen permeabilities deviated from the result calculated by Hao et al. [27]. We hypothesize that surface oxides created resistance for hydrogen transport into the bulk material. We found that hydrogen permeability decreased after 1–2 h during the permeability test. We hypothesize that this is the result of hydrogen induced crystallization. We also discuss the importance of thermal stability under hydrogen exposure. Our results point to the importance of the thermal properties and thermal stability of amorphous alloys relative to the hydrogen permeation testing temperature.
2. Experiments 2.1. Materials We purchased raw metals (Cu and Zr) from Alfa-Aesar with purity of 99.999%. We cleaned the raw metals with ultrasonication in acetone and then in ethanol before we weighed the raw metals and arc melted them (Edmund Buehler, model AM) in an argon atmosphere into 15 g buttons. We chose three compositions with atomic percentages of Zr of 37, 54 and 60. During the arc-melting process we flipped the buttons twice to ensure complete alloying of the metals. 2.2. Membrane synthesis We used a splat quencher (Edmund Buehler, model Buehler Splat Quencher) to synthesize the membranes. The thickness of the membranes is 45 mm 75 mm and the diameter of the membranes is 20–25 mm. The splat quencher reaches cooling rates of up to 106 K/s [23], which is capable of synthesizing amorphous Cu–Zr alloys according to the GFA predicted by Ge et al. [28]. We deposited palladium on the surface of the membranes to promote hydrogen dissociation through direct current (DC) sputtering with a Lesker Supersystem II. We annealed the membranes at 523 K for half an hour in a vacuum oven (10 kPa) to promote Pd bonding to the amorphous alloy. Without annealing, the Pd layer delaminated from the membrane during hydrogen permeation testing.
265
2.3. Characterizations 2.3.1. Structure We analyzed the extent of amorphicity or crystallinity of the membranes with X-ray diffraction (XRD, Philips PANalytical X'Pert Pro Cu Kα at 40 kV and 40 mA). Hydrogen permeation tests only used membranes that exhibited an amorphous broad peak. After palladium deposition and annealing, the samples are still amorphous, as crystalline peaks present in the XRD patterns are only from the Pd coating. 2.3.2. Surface depth profile We used Rutherford Backscattering Spectroscopy (RBS) with a 2 MeV He þincident beam to determine the thickness of the deposited Pd layer and analyzed the RBS data using the RUMP software package (Rutherford Backscattering Spectroscopy Analysis Package, Genplot, Cortland, OH). We performed X-ray photoelectron spectroscopy (XPS VG ESCALAB 220i) with an Al monochromatic source Kα (12 KV, 65 W, hν ¼1486.6 eV). Binding energy was calibrated with pure Ag with Ag 3d5/2 at 368.2 eV [29] and we used Casa XPS software to analyze the chemical shifts of elements. 2.3.3. Thermal properties We used differential scanning calorimetry (DSC, TA instrument Q20) with a heating rate of 20 K/min and nitrogen flow rate of 50 ml/min to investigate the thermal properties of the amorphous membranes. The TA instrument software calculated the Tg and Tx (Universal Analysis processing software). 2.3.4. Hydrogen permeability test We measured hydrogen permeability in our custom-built crossflow testing system [30,31]. Copper gaskets sealed our selfsupporting membranes into the stainless steel permeation module. Prior to use, we first annealed the copper gaskets in air at 703 K for 15 min, next polished them up to 2400 grit sand paper, and finally ultrasonically cleaned them to optimize their sealing properties. The leakage rate of the copper gaskets did not exceed 5% of the tested hydrogen flow rate at 573 K. The effective permeation area of the membranes sealed into the permeation module was 1.82 cm2. We used a muffle furnace to heat the module to 573 K with a ramp rate of about 16 K/min. K-type thermocouples monitored the temperatures at both the exterior of the module and the interior of the module near the membranes. During the initial ramp to operating temperature we flushed N2 on the feed side of the membrane and Ar on the permeate side. Mass flow controllers regulated the flow rate of all gases. Nitrogen and argon were ultra-high purity grade (99.999%), and a hydrogen generator (VWR International, Model H2PEM-100-L) provided hydrogen with purity of 99.99% þ. After reaching the desired temperature, we shut off the nitrogen feed purge and introduced hydrogen and pressurized the cell. A back pressure regulator controlled the feed pressure to 140 kPa. A pressure gauge displayed the pressure of the permeate side, which we maintained at 1 atm at all times. We used an Agilent 7890 gas chromatograph (GC) to monitor the concentration of gases in the permeate stream. The total testing time was 5–6 h . We calibrated the GC with customized gas standards from Air‐Products. We confirmed the permeate flow rate with a soap bubble flow meter. We calculated permeability using Sieverts' law J¼
P 0:5 P H P 0:5 L t
ð1Þ
In Eq. (1), J [mol s 1 m 2] is the flux of hydrogen, P [mol m 1 s 1 Pa 0.5] is the permeability, t [m] is the thickness
266
T. Lai et al. / Journal of Membrane Science 489 (2015) 264–269
of the membrane; PH [Pa] is hydrogen partial pressure of the feed side, and PL [Pa] is hydrogen partial pressure of the permeate side.
3. Results and discussion 3.1. Structure Fig. 1 shows a photograph of the as-cast membrane. The membranes have a diameter of approximately 20–25 mm with an average thickness of 45 mm 75 mm. Fig. 2 shows the XRD patterns of as-cast membranes and after Pd deposition and annealing. The as-cast XRD spectra only shows a broad peak centered around 381; no sharp (crystalline) peaks exist in the ascast membrane, confirming the amorphous structure. After Pd deposition and annealing, the broad amorphous peak remained and the sharp peaks matching the crystalline XRD pattern of Pd appear. No other crystalline peaks appeared after deposition, indicating that the annealing did not significantly change the amorphous structure. The RBS measurements indicated that the thickness of the Pd layer was about 120 nm on both sides of the membranes. As mentioned before, annealing is necessary to promote the adhesion of Pd with the membranes.
3.2. Thermal properties Table 1 lists the Tg, Tx and ΔT of the three compositions we investigated. The Tgs are comparable to both the theoretical and experimental results [23,28,32,33]. As the concentration of Zr increased, Tg and Tx decreased. The Tgs do not change after Pd deposition and annealing. The minimum size of the SCLR is 49 K, which indicates that these three compositions have good glass forming ability. According to previous research on Cu–Zr systems, the three compositions we chose can be formed into bulk metallic glass [25,26,33]. We chose 573 K to perform the hydrogen permeability test, which is at least 63 K lower than the minimum Tg of the alloys we used. Our previous work has shown that it is possible to anneal an amorphous Zr-based alloy in an inert environment at 50 K below the measured Tg for 8.5 h and then cycle it to the Tg without changing the ultrasonically measured alloy properties [24]. We performed stability tests of Zr54Cu46 with DSC. We annealed samples for two hours at 580, 633 and 653 K under nitrogen. After holding at these temperatures, we performed a regular DSC scan (from the holding temperature to 773 K, with a heating rate of 20 K/s). The samples annealed at 580 and 633 K have the same DSC curves (including Tg, Tx and heat of crystallization) as the as-cast samples. However, we found that after annealing the samples at temperatures 11 K less than the measured Tg of the as-cast alloy (i.e. annealing at 653 K), the heat of crystallization is slightly different from the as-cast samples. 3.3. Hydrogen permeability and thermal stability Fig. 3 shows the permeability of uncoated and coated Cu–Zr membranes at 573 K with a feed of pure hydrogen at a pressure of 140 kPa (we kept the permeate side of the membrane at atmosphere pressure with an Ar sweep). There is no significant difference in the permeabilities of uncoated and palladium coated Cu63Zr37 and Cu46Zr54. However, there is a significant difference in the permeability of the uncoated and palladium coated Cu40Zr60. Table 1 Measured Tg , Tx and ΔT of amorphous metals.
Fig. 1. A picture showing an as-cast Cu40Zr60 amorphous piece.
Fig. 2. XRD pattern of Cu46Zr54 after casting and after Pd-deposition and annealing.
Composition
Tg (K)
Tx (K)
ΔT (K)
Cu63Zr37 Cu46Zr54 Cu40Zr60
732 664 640
788 713 690
56 49 50
Fig. 3. Hydrogen permeability of uncoated and Pd-coated amorphous metallic membranes.
T. Lai et al. / Journal of Membrane Science 489 (2015) 264–269
There appears to be no correlation between the Zr-content of the membranes and the overall hydrogen permeability. Fig. 4 shows the RBS spectra of as-synthesized membranes (before coating with Pd and before permeation testing). The penetration depth of RBS can exceed 10 mm with a resolution limit of 3 nm [34]. The RBS test enables us to collect the depth profiles of the surface region of the tested samples [34]. Our RUMP simulation results show that the oxides layers are present on the surface of the samples. In Fig. 4, a RUMP simulation shows that peaks around 1.65 MeV (Channel 330) indicate the presence of zirconium oxides and a peak around 1.5 MeV (Channel 300) indicates the presence of copper–zirconium oxides. In Fig. 4(a), Cu63Zr37 samples have zirconium oxide and copper–zirconium oxide layers of total thickness about 130 nm. In Fig. 4(b), there is a zirconium oxide layer approximately 8 nm thick and layers of copper–zirconium oxides approximately 10 nm thick on the surface of as-synthesized Cu46Zr54. In Fig. 4(c), there is about 5 nm of zirconium oxides on the surface of Cu40Zr60. We also performed XPS which has a penetration depth of a few nanometers [35]. Our XPS results (not shown here for brevity) indicate that all the three compositions have small amounts of zirconium and copper oxides
267
present on the surface. The presence of surface oxides on the Cu63Zr37 (surface oxides 130 nm thick) and Cu46Zr54 (surface oxides 18 nm thick) amorphous metallic membranes explain the insignificant difference in permeability between uncoated and Pdcoated samples. This also explains why the Pd-coated Cu46Zr54 membranes exhibit a slightly lower average permeability (although not one that is statistically different) than the uncoated Cu46Zr54 membranes. Although Pd effectively dissociates hydrogen, the surface oxides on the amorphous metallic membranes may act as a transport barrier, counteracting the catalytic effect of the Pd. In the case of Cu40Zr60, there is limited, if any, oxide ( 5 nm thick) present on the membrane surfaces, therefore we still see the catalytic effect of the Pd in increasing the hydrogen permeability. Thin layers (10s of nanometers) of surface oxide could readily result in large resistance to the hydrogen transport because hydrogen transport through dense metallic membranes is by a dissociation–solution–diffusion process [36]. It is very difficult to eliminate all oxides in the synthesis of a metallic glass [37,38]. Thermodynamically, it is favorable for oxides to migrate to the surfaces of a metallic glass [39]. Zirconium-oxides have a larger negative heat of formation than
Fig. 4. RBS spectrum of as-synthesized (a) Cu63Zr37, (b) Cu46Zr54 and (c) Cu40Zr60 (before coating and before hydrogen permeation testing). The peaks in the circles indicate the presence of oxides.
Fig. 5. (a) Permeability vs time during the hydrogen permeation test of a Pd-coated Cu63Zr37 sample (b) XRD pattern of an uncoated Cu63Zr37 membrane permeate side after completion of the hydrogen permeation test.
268
T. Lai et al. / Journal of Membrane Science 489 (2015) 264–269
copper-oxides (at 298 K the heats of formation of oxides are CuO: 156.1 KJ/mol, Cu2O: 170.7 KJ/mol, ZrO2: 1097.5 KJ/mol [40]). However, according to the RUMP simulation, we found that the composition with the highest concentration of Zr (Cu40Zr60) has the least oxides on the surface. Sen et al. found that the Cu–Zr alloy with higher Zr concentration has a lower surface oxidation rate both in crystalline and glassy states [41]. This could be attributed to the different growth rate of different oxides which is determined by the Cu/Zr composition ratio as shown in Fig. 4. Hickman et al. found that between 573 K and 773 K only copper oxides can form in Zr20Cu80 [42]. Both reports by Sen et al. [42] and Hickman et al. [41] agree with our observations. Also, it is possible that the surface composition of the as-cast membranes is different than the bulk composition. Because Cu has a lower surface energy (1.79 J/m) than Zr (2.0 J/m) [38], copper has the tendency to migrate to the surface of the alloy. Fig. 5(a) shows the hydrogen permeability of a Pd-coated Zr37Cu63 membrane as a function of time during the permeation test. Fig. 5(b) shows the XRD spectra of the permeate side of Cu63Zr37 after the completion of the hydrogen permeation test. The system permeation error is within 10%. As Fig. 5 shows, after two hours of testing, the permeability starts to decrease and the permeability continues to decrease through the end of the test ( 6 h). This may be the result of the amorphous structure relaxation [22]. The onset of crystallization may happen sometime during the third hour of testing (as indicated by the decrease in permeability seen around 2.5 h of testing). Although we used a testing temperature of 573 K, which is 63 K lower than the Tg of Cu63Zr37, we hypothesize that with the presence of hydrogen, the structure relaxation and formation of hydride may left shift the TTT curve of Cu–Zr alloys (the nose moves to a lower temperature) which accelerated the crystallization of amorphous membranes [21,22]. Crystallization of an amorphous alloy further decreases the hydrogen permeability [10,11]. The crystallization of Cu–Zr may be the result of the formation of hydrides. As shown in Fig. 5(b), there are peaks corresponding to zirconium hydrides and zirconium oxides after permeability test. Zr has a large, negative hydride heat of formation [15]. The strong affinity of Zr for hydrogen facilitates hydrogen absorption of the alloy with higher concentration of Zr [10,14]. But the formation of zirconium hydride promotes crystallization and is detrimental to the thermal stability.
4. Conclusions We synthesized Cu–Zr (Zr at%¼37, 54, 60) free-standing, amorphous metallic membranes by splat quenching and verified their amorphous nature with XRD. Palladium catalytic coatings promoted hydrogen dissociation and recombination. We measured hydrogen permeability of the amorphous free-standing membranes at a feed hydrogen pressure of 140 kPa. The hydrogen permeability was lower than that predicted by the first-principle calculations. The observed permeability was explained by oxide formation on the surface. The variation in oxide thickness may be the result of the formation of different oxides as a function of the Cu/Zr composition ratio and their different growth kinetics. The stability tests and the permeability tests indicate that the presence of hydrogen accelerated crystallization which leads to significant permeability degradation. Overall, the thermal stability of amorphous metallic membranes is very important for hydrogen separation applications.
Acknowledgments We wish to acknowledge the financial support of Ira A. Fulton Schools of Engineering. Acknowledgment is made to the Donors of
the American Chemical Society Petroleum Research Fund for support of this research (award # 52461-DNI10). We gratefully acknowledge the use of facilities with the LeRoy Eyring Center for Solid State Science at Arizona State University. We also would like to acknowledge Dr. William L. Johnson, Pamela Albertson, and George Kaltenboeck at the California Institute of Technology for facilitating the use of their arc-melter and the splat quencher. We would like to acknowledge Dr. Jerry Y.S. Lin and his students for help in designing the hydrogen permeation testing system. We would also like to acknowledge Fred Pena at ASU for his extensive help in building and maintaining the hydrogen permeation testing system. We sincerely thank Dr. David Sholl and postdoc Dr. Zhu in Georgia Institute of Technology for discussions about the permeability results.
References [1] Hydrogen production, (2013). Available from: 〈http://energy.gov/eere/fuel cells/natural-gas-reforming〉. [2] M.D. Dolan, N.C. Dave, A.Y. Ilyushechkin, L.D. Morpeth, K.G. McLennan, Composition and operation of hydrogen-selective amorphous alloy membranes, J. Membr. Sci. 285 (1–2) (2006) 30–55. [3] N.W. Ockwig, T.M. Nenoff., Membranes for hydrogen separation, Chem. Rev. 107 (10) (2007) 4078–4110. [4] S.N. Paglieri, J.D. Way., Innovations in palladium membrane research, Sep. Purif. Methods 31 (1) (2002) 1–169. [5] F.A. Lewis., The palladium hydrogen system – structures near phase-transition and critical-points, Int. J. Hydrog. Energy 20 (7) (1995) 587–592. [6] C. Nishimura, M. Komaki, M. Amano., Hydrogen permeation characteristics of vanadium–nickel alloys, JIM, Mater. Trans. 32 (1991) 501–507. [7] S. Hao, D.S. Sholl, Comparison of first principles calculations and experiments for hydrogen permeation through amorphous ZrNi and ZrNiNb films, J. Membr. Sci. 350 (1-2) (2010) 402–409. [8] 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 (1-2) (2000) 289–294. [9] W.L. Johnson., Bulk glass-forming metallic alloys: science and technology, MRS Bull. 24 (10) (1999) 42–56. [10] S.N. Paglieri, N.K. Pal, M.D. Dolan, S.M. Kim, W.M. Chien, J. Lamb, et al., Hydrogen permeability, thermal stability and hydrogen embrittlement of Ni–Nb–Zr and Ni–Nb–Ta–Zr amorphous alloy membranes, J. Membr. Sci. 378 (1–2) (2011) 42–50. [11] S.N. Paglieri, N.K. Pal, S.M. Kim, M.D. Dolan, W.M. Chien, J. Lamb, et al., Ni–Nb– Zr–(Ta, Co) amorphous alloy membranes for hydrogen separation, in: Abstracts of Papers of the American Chemical Society, 2011, p. 242. [12] M.D. Dolan, S. Hara, N.C. Dave, K. Haraya, M. Ishitsuka, A.Y. Ilyushechkin, et al., Thermal stability, glass-forming ability and hydrogen permeability of amorphous Ni64Zr36 xMx (M ¼Ti, Nb, Mo, Hf, Ta or W) membranes, Sep. Purif. Technol. 65 (3) (2009) 298–304. [13] S. Hara, N. Hatakeyama, N. Itoh, H.M. Kimura, A. Inoue, Hydrogen permeation through palladium-coated amorphous Zr–M–Ni (M ¼Ti, Hf) alloy membranes, Desalination 144 (1–3) (2002) 115–120. [14] S. i. Yamaura, M. Sakurai, M. Hasegawa, K. Wakoh, Y. Shimpo, M. Nishida, et al., Hydrogen permeation and structural features of melt-spun Ni–Nb–Zr amorphous alloys, Acta Mater. 53 (13) (2005) 3703–3711. [15] R. Griessen, A. Driessen., Heat of formation and band structure of binary and ternary metal hydrides, Phys. Rev. B 30 (8) (1984) 4372–4381. [16] A.J. Maeland, L.E. Tanner, G.G. Libowitz., Hydrides of metallic glass alloys, J. Less Common Met. 74 (2) (1980) 279–285. [17] S. Yamaura, Y. Shimpo, H. Okouchi, M. Nishida, O. Kajita, H. Kimura, et al., Hydrogen permeation characteristics of melt-spun Ni–Nb–Zr amorphous alloy membranes, Mater. Trans. 44 (9) (2003) 1885–1890. [18] E.S. Park, J.H. Na, D.H. Kim, Abnormal behavior of supercooled liquid region in bulk-forming metallic glasses, J. Appl. Phys. 108 (5) (2010) 053515-1–053515-9. [19] S.I. Yamaura, S. Nakata, H. Kimura, A. Inoue., Hydrogen permeation of Ni–Nb– Zr metallic glasses in a supercooled liquid state, Mater. Trans. 47 (12) (2006) 2991–2996. [20] K.W. Park, J.P. Ahn, H.K. Seok, Y.C. Kim., Relationship between activation energy for hydrogen permeation and hydrogen permeation properties of amorphous Cu50Zr50 and Cu65Zr35 membranes, Intermetallics 19 (12) (2011) 1887–1890. [21] N. Eliaz, D. Eliezer., An overview of hydrogen interaction with amorphous alloys, Adv. Perform. Mater. 6 (1) (1999) 5–31. [22] K. Dini, R.A. Dunlap., Crystallization and hydrogen absorption in amorphous Cu60Zr40 and Cu50Zr50, J. Phys. F: Met. Phys. 15 (2) (1985) 273–277. [23] L. Ge, X. Hui, E.R. Wang, G.L. Chen, R. Arroyave, Z.K. Liu., Prediction of the glass forming ability in Cu–Zr binary and Cu–Zr–Ti ternary alloys, Intermetallics 16 (1) (2008) 27–33.
T. Lai et al. / Journal of Membrane Science 489 (2015) 264–269
[24] M.L. Lind, G. Duan, W.L. Johnson., Isoconfigurational elastic constants and liquid fragility of a bulk metallic glass forming alloy, Phys. Rev. Lett. 97 (1) (2006) 015501. [25] D.H. Xu, G. Duan, W.L. Johnson., Unusual glass-forming ability of bulk amorphous alloys based on ordinary metal copper, Phys. Rev. Lett. 92 (2004) 24. [26] A. Zhang, D. Chen, Z. Chen., Bulk metallic glass-forming region of Cu–Zr binary and Cu–Zr based multicomponent alloy systems, J. Alloy. Compd. 477 (1–2) (2009) 432–435. [27] S.Q. Hao, D.S. Sholl., Computational prediction of durable amorphous metal membranes for H2 purification, J. Membr. Sci. 381 (1–2) (2011) 192–196. [28] L. Ge, X.D. Hui, G.L. Chen, Z.K. Liu., Prediction of the glass-forming ability of Cu-Zr binary alloys, Acta Phys.-Chim. Sin. 23 (6) (2007) 895–899. [29] R. Nyholm, N. Mårtensson., Experimental core-hole ground state energies for the elements 41Nb to 52Te, Solid State Commun. 40 (3) (1981) 311–314. [30] V. Jayaraman, Y.S. Lin, M. Pakala, R.Y. Lin., Fabrication of ultrathin metallic membranes on ceramic supports by sputter-deposition, J. Membr. Sci. 99 (1) (1995) 89–100. [31] T. Lai, M.L. Lind., Heat treatment driven surface segregation in Pd77Ag23 membranes and the effect on hydrogen permeability, Int. J. Hydrog. Energy 40 (1) (2015) 373–382. [32] J.S.C. Jang, C.C. Koch., The glass transition temperature in amorphous Cu1 xZrx alloys synthesized by mechanical alloying/milling, Scr. Metall. 23 (10) (1989) 1805–1810. [33] D.H. Xu, B. Lohwongwatana, G. Duan, W.L. Johnson, C. Garland., Bulk metallic glass formation in binary Cu-rich alloy series – Cu100 xZrx (x ¼34, 36, 38.2,
[34]
[35] [36] [37]
[38] [39] [40] [41]
[42]
269
40 at%) and mechanical properties of bulk Cu64Zr36 glass, Acta Mater. 52 (9) (2004) 2621–2624. S.H. Sie, Rutherford backscattering spectrometry and nuclear reaction analysis, Surface Analysis Methods in Materials Science, Springer, Verlag Berlin Heidelberg New York, 2003. R. Schulz, K. Samwer, W.L. Johnson., Kinetics of phase separation in Cu50Zr50 metallic glasses, J. Non-Cryst. Solids 61–62 (Part 2) (1984) 997–1002. T.L. Ward, T. Dao., Model of hydrogen permeation behavior in palladium membranes, J. Membr. Sci. 153 (2) (1999) 211–231. A. Gebert, K. Buchholz, A. Leonhard, K. Mummert, J. Eckert, L. Schultz., Investigations on the electrochemical behaviour of Zr-based bulk metallic glasses, Mater. Sci. Eng.: A 267 (2) (1999) 294–300. J. Schroers., Processing of bulk metallic glass, Adv. Mater. 22 (14) (2010) 1566–1597. A. Gebert, P.F. Gostin, L. Schultz., Effect of surface finishing of a Zr-based bulk metallic glass on its corrosion behaviour, Corros. Sci. 52 (5) (2010) 1711–1720. I. Ihsan Barin, Thermochemical Data of Pure Substances, VCH, Weinheim, Germany, 1989. P. Sen, D.D. Sarma, R.C. Budham, K.L. Chopra, C.N.R. Rao., An electron spectroscopic study of the surface oxidation of glassy and crystalline Cu–Zr alloys, J. Phys. F: Metal Phys. 14 (2) (1984) 565. J.W. Hickman, E.A. Gulbransen., Oxide films formed on titanium, zirconium, and their alloys with nickel, copper, and cobalt, Anal. Chem. 20 (2) (1948) 158–165.
Contents lists available at ScienceDirect
Journal of Membrane Science journal homepage: www.elsevier.com/locate/memsci
The hydrogen permeability of Cu–Zr binary amorphous metallic membranes and the importance of thermal stability Tianmiao Lai, Huidan Yin, Mary Laura Lind n School for Engineering of Matter, Transport and Energy, Arizona State University, Tempe, AZ 85287-6106, USA
art ic l e i nf o
a b s t r a c t
Article history: Received 30 January 2015 Received in revised form 11 March 2015 Accepted 22 March 2015 Available online 30 April 2015
We synthesized three compositions of amorphous metallic Cu–Zr (Zr ¼37, 54, 60 at%) membranes by splat quenching. We measured the thermal properties with differential scanning calorimetry (DSC). We tested hydrogen permeability of the membranes before and after Pd coatings. Experimentally, we found that the hydrogen permeability of the alloys is lower than that predicted by existing modeling results; we hypothesize this may be the result of oxides formed on the surface of our membranes. The formation of oxides is composition dependent. The permeability change during the testing time indicates that hydrogen has promoted structural change at a temperature lower than the glass transition temperature (Tg). Our original experimental results show that thermal stability is critical for hydrogen separation applications in amorphous metallic membranes. & 2015 Elsevier B.V. All rights reserved.
Keywords: Metallic membranes Cu–Zr amorphous metals Hydrogen permeability Surface oxides Stability
1. Introduction In the United States, 95% of the industrially produced hydrogen is from natural gas reforming [1]. Pd-based alloys are the benchmark metallic membrane materials for hydrogen separation. Pdbased membranes have pure hydrogen permeability of 10 8 mol m 1 s 1 Pa 0.5 and infinite H2/N2 selectivity at temperatures up to 600 K [2–4]. However, at 1 bar and at temperatures below 573 K, pure palladium suffers an α/β palladium hydride phase transformation [5]. The lattice expansion mismatch between the α and β phases results in mechanical failure after long time exposures to hydrogen. The high cost of palladium also limits widespread application, so researchers are interested in lower cost metallic materials [6]. Amorphous metals are one alternative to replace palladium-based membranes [2]. Density functional theory calculations and quantum Monte Carlo simulations have demonstrated that amorphous metallic films can have hydrogen permeability close to pure Pd, and predict less hydrogen-induced expansion compared to their crystalline counterparts [7]. Experimentally, Hara et al. found that the amorphous alloy Zr36Ni64, without a Pd surface coating, has hydrogen permeability only one order of magnitude lower than Pd77Ag23 at 623 K [8].
n
Corresponding author. Tel.: þ 1 480 727 8613; fax: þ 1 480 727 9321. E-mail address: [email protected] (M.L. Lind).
http://dx.doi.org/10.1016/j.memsci.2015.03.098 0376-7388/& 2015 Elsevier B.V. All rights reserved.
Amorphous metals are formed through rapid cooling of a molten alloy. The resulting amorphous metallic alloys have no long-range crystalline order [9]. The critical cooling rate is strongly dependent on the alloy composition or glass forming ability (GFA) of the composition. To further improve the GFA and hydrogen permeability of binary Zr–Ni alloys, ternary alloys in which Zr has been substituted partially by other elements like Nb, Ta and Ti have been investigated [10–13]. Paglieri et al. and Yamamura et al. found that for multi-component Zr-based amorphous alloy compositions, higher concentrations of Zr yielded higher hydrogen permeability [10,14]. This may be a result of the strong affinity of Zr for hydrogen combined with the amorphous structure of the alloys [15,16]. However, severe hydrogen embrittlement rapidly occurred in alloys with a Zr concentration higher than 40% [14,17]. Thermal stability is an important parameter in membrane performance. A significant limitation to potential use of amorphous metallic membranes for high-temperature hydrogen separations is related to their metastability. Because these materials are thermodynamically metastable, they possess a super cooled liquid region (SCLR) which is bound by the glass transition (Tg) (on the low end) and crystallization temperatures (Tx) (on the high end) [18]. It is well established that holding an amorphous metal at the Tg or at temperatures within the SCLR for long times results in crystallization [19]. For membrane separation applications, the hydrogen permeability is known to decrease after crystallization of the amorphous metal [20]. Therefore, it is preferable to use amorphous metals with a Tg that is higher than the temperature at which the process is being performed, in order to maintain the
T. Lai et al. / Journal of Membrane Science 489 (2015) 264–269
amorphous structure. Additionally, the presence of hydrogen can impact the thermal stability of amorphous metals by promoting crystallization [21]. Dini et al. found that electrolytically hydrogenated amorphous Cu–Zr alloys have lower Tx than as-cast amorphous Cu–Zr alloys [22]. The Cu–Zr binary amorphous alloy system has drawn interest because of its good mechanical properties and low cost [23,24]. Xu et al. and Zhang et al. found a series of Cu–Zr binary compositions that are bulk metallic glass formers [25,26]. Calculation of the Time–Transformation–Temperature (TTT) curve is a method to predict the GFA of amorphous alloys [23]. Ge et al. used the CALPHAD method to estimate TTT curves of Cu–Zr alloys [23]. They found that the critical cooling rate to form amorphous alloys within the Cu–Zr system is between 4.32 102–2.63 104 K/s. They calculated that the Tg of Cu–Zr alloys ranges from 620 K to 750 K [23]. This makes Cu–Zr alloys suitable for hydrogen separation at operation temperatures near 573 K. The specific composition Zr54Cu46 has critical casting thicknesses up to 2 mm (Tg:696 K, Tx:746 K) [25]. Hao et al. used first principles density functional theory to calculate the hydrogen permeability of a series of binary amorphous metals [27]. Their results predicted that Zr54Cu46 could have hydrogen permeability comparable to pure Pd at 600 K [27]. In this paper, we measured the experimental hydrogen permeability of Cu–Zr amorphous free-standing membranes. We investigated alloys with three different Zr concentrations. Our experimentally measured hydrogen permeabilities deviated from the result calculated by Hao et al. [27]. We hypothesize that surface oxides created resistance for hydrogen transport into the bulk material. We found that hydrogen permeability decreased after 1–2 h during the permeability test. We hypothesize that this is the result of hydrogen induced crystallization. We also discuss the importance of thermal stability under hydrogen exposure. Our results point to the importance of the thermal properties and thermal stability of amorphous alloys relative to the hydrogen permeation testing temperature.
2. Experiments 2.1. Materials We purchased raw metals (Cu and Zr) from Alfa-Aesar with purity of 99.999%. We cleaned the raw metals with ultrasonication in acetone and then in ethanol before we weighed the raw metals and arc melted them (Edmund Buehler, model AM) in an argon atmosphere into 15 g buttons. We chose three compositions with atomic percentages of Zr of 37, 54 and 60. During the arc-melting process we flipped the buttons twice to ensure complete alloying of the metals. 2.2. Membrane synthesis We used a splat quencher (Edmund Buehler, model Buehler Splat Quencher) to synthesize the membranes. The thickness of the membranes is 45 mm 75 mm and the diameter of the membranes is 20–25 mm. The splat quencher reaches cooling rates of up to 106 K/s [23], which is capable of synthesizing amorphous Cu–Zr alloys according to the GFA predicted by Ge et al. [28]. We deposited palladium on the surface of the membranes to promote hydrogen dissociation through direct current (DC) sputtering with a Lesker Supersystem II. We annealed the membranes at 523 K for half an hour in a vacuum oven (10 kPa) to promote Pd bonding to the amorphous alloy. Without annealing, the Pd layer delaminated from the membrane during hydrogen permeation testing.
265
2.3. Characterizations 2.3.1. Structure We analyzed the extent of amorphicity or crystallinity of the membranes with X-ray diffraction (XRD, Philips PANalytical X'Pert Pro Cu Kα at 40 kV and 40 mA). Hydrogen permeation tests only used membranes that exhibited an amorphous broad peak. After palladium deposition and annealing, the samples are still amorphous, as crystalline peaks present in the XRD patterns are only from the Pd coating. 2.3.2. Surface depth profile We used Rutherford Backscattering Spectroscopy (RBS) with a 2 MeV He þincident beam to determine the thickness of the deposited Pd layer and analyzed the RBS data using the RUMP software package (Rutherford Backscattering Spectroscopy Analysis Package, Genplot, Cortland, OH). We performed X-ray photoelectron spectroscopy (XPS VG ESCALAB 220i) with an Al monochromatic source Kα (12 KV, 65 W, hν ¼1486.6 eV). Binding energy was calibrated with pure Ag with Ag 3d5/2 at 368.2 eV [29] and we used Casa XPS software to analyze the chemical shifts of elements. 2.3.3. Thermal properties We used differential scanning calorimetry (DSC, TA instrument Q20) with a heating rate of 20 K/min and nitrogen flow rate of 50 ml/min to investigate the thermal properties of the amorphous membranes. The TA instrument software calculated the Tg and Tx (Universal Analysis processing software). 2.3.4. Hydrogen permeability test We measured hydrogen permeability in our custom-built crossflow testing system [30,31]. Copper gaskets sealed our selfsupporting membranes into the stainless steel permeation module. Prior to use, we first annealed the copper gaskets in air at 703 K for 15 min, next polished them up to 2400 grit sand paper, and finally ultrasonically cleaned them to optimize their sealing properties. The leakage rate of the copper gaskets did not exceed 5% of the tested hydrogen flow rate at 573 K. The effective permeation area of the membranes sealed into the permeation module was 1.82 cm2. We used a muffle furnace to heat the module to 573 K with a ramp rate of about 16 K/min. K-type thermocouples monitored the temperatures at both the exterior of the module and the interior of the module near the membranes. During the initial ramp to operating temperature we flushed N2 on the feed side of the membrane and Ar on the permeate side. Mass flow controllers regulated the flow rate of all gases. Nitrogen and argon were ultra-high purity grade (99.999%), and a hydrogen generator (VWR International, Model H2PEM-100-L) provided hydrogen with purity of 99.99% þ. After reaching the desired temperature, we shut off the nitrogen feed purge and introduced hydrogen and pressurized the cell. A back pressure regulator controlled the feed pressure to 140 kPa. A pressure gauge displayed the pressure of the permeate side, which we maintained at 1 atm at all times. We used an Agilent 7890 gas chromatograph (GC) to monitor the concentration of gases in the permeate stream. The total testing time was 5–6 h . We calibrated the GC with customized gas standards from Air‐Products. We confirmed the permeate flow rate with a soap bubble flow meter. We calculated permeability using Sieverts' law J¼
P 0:5 P H P 0:5 L t
ð1Þ
In Eq. (1), J [mol s 1 m 2] is the flux of hydrogen, P [mol m 1 s 1 Pa 0.5] is the permeability, t [m] is the thickness
266
T. Lai et al. / Journal of Membrane Science 489 (2015) 264–269
of the membrane; PH [Pa] is hydrogen partial pressure of the feed side, and PL [Pa] is hydrogen partial pressure of the permeate side.
3. Results and discussion 3.1. Structure Fig. 1 shows a photograph of the as-cast membrane. The membranes have a diameter of approximately 20–25 mm with an average thickness of 45 mm 75 mm. Fig. 2 shows the XRD patterns of as-cast membranes and after Pd deposition and annealing. The as-cast XRD spectra only shows a broad peak centered around 381; no sharp (crystalline) peaks exist in the ascast membrane, confirming the amorphous structure. After Pd deposition and annealing, the broad amorphous peak remained and the sharp peaks matching the crystalline XRD pattern of Pd appear. No other crystalline peaks appeared after deposition, indicating that the annealing did not significantly change the amorphous structure. The RBS measurements indicated that the thickness of the Pd layer was about 120 nm on both sides of the membranes. As mentioned before, annealing is necessary to promote the adhesion of Pd with the membranes.
3.2. Thermal properties Table 1 lists the Tg, Tx and ΔT of the three compositions we investigated. The Tgs are comparable to both the theoretical and experimental results [23,28,32,33]. As the concentration of Zr increased, Tg and Tx decreased. The Tgs do not change after Pd deposition and annealing. The minimum size of the SCLR is 49 K, which indicates that these three compositions have good glass forming ability. According to previous research on Cu–Zr systems, the three compositions we chose can be formed into bulk metallic glass [25,26,33]. We chose 573 K to perform the hydrogen permeability test, which is at least 63 K lower than the minimum Tg of the alloys we used. Our previous work has shown that it is possible to anneal an amorphous Zr-based alloy in an inert environment at 50 K below the measured Tg for 8.5 h and then cycle it to the Tg without changing the ultrasonically measured alloy properties [24]. We performed stability tests of Zr54Cu46 with DSC. We annealed samples for two hours at 580, 633 and 653 K under nitrogen. After holding at these temperatures, we performed a regular DSC scan (from the holding temperature to 773 K, with a heating rate of 20 K/s). The samples annealed at 580 and 633 K have the same DSC curves (including Tg, Tx and heat of crystallization) as the as-cast samples. However, we found that after annealing the samples at temperatures 11 K less than the measured Tg of the as-cast alloy (i.e. annealing at 653 K), the heat of crystallization is slightly different from the as-cast samples. 3.3. Hydrogen permeability and thermal stability Fig. 3 shows the permeability of uncoated and coated Cu–Zr membranes at 573 K with a feed of pure hydrogen at a pressure of 140 kPa (we kept the permeate side of the membrane at atmosphere pressure with an Ar sweep). There is no significant difference in the permeabilities of uncoated and palladium coated Cu63Zr37 and Cu46Zr54. However, there is a significant difference in the permeability of the uncoated and palladium coated Cu40Zr60. Table 1 Measured Tg , Tx and ΔT of amorphous metals.
Fig. 1. A picture showing an as-cast Cu40Zr60 amorphous piece.
Fig. 2. XRD pattern of Cu46Zr54 after casting and after Pd-deposition and annealing.
Composition
Tg (K)
Tx (K)
ΔT (K)
Cu63Zr37 Cu46Zr54 Cu40Zr60
732 664 640
788 713 690
56 49 50
Fig. 3. Hydrogen permeability of uncoated and Pd-coated amorphous metallic membranes.
T. Lai et al. / Journal of Membrane Science 489 (2015) 264–269
There appears to be no correlation between the Zr-content of the membranes and the overall hydrogen permeability. Fig. 4 shows the RBS spectra of as-synthesized membranes (before coating with Pd and before permeation testing). The penetration depth of RBS can exceed 10 mm with a resolution limit of 3 nm [34]. The RBS test enables us to collect the depth profiles of the surface region of the tested samples [34]. Our RUMP simulation results show that the oxides layers are present on the surface of the samples. In Fig. 4, a RUMP simulation shows that peaks around 1.65 MeV (Channel 330) indicate the presence of zirconium oxides and a peak around 1.5 MeV (Channel 300) indicates the presence of copper–zirconium oxides. In Fig. 4(a), Cu63Zr37 samples have zirconium oxide and copper–zirconium oxide layers of total thickness about 130 nm. In Fig. 4(b), there is a zirconium oxide layer approximately 8 nm thick and layers of copper–zirconium oxides approximately 10 nm thick on the surface of as-synthesized Cu46Zr54. In Fig. 4(c), there is about 5 nm of zirconium oxides on the surface of Cu40Zr60. We also performed XPS which has a penetration depth of a few nanometers [35]. Our XPS results (not shown here for brevity) indicate that all the three compositions have small amounts of zirconium and copper oxides
267
present on the surface. The presence of surface oxides on the Cu63Zr37 (surface oxides 130 nm thick) and Cu46Zr54 (surface oxides 18 nm thick) amorphous metallic membranes explain the insignificant difference in permeability between uncoated and Pdcoated samples. This also explains why the Pd-coated Cu46Zr54 membranes exhibit a slightly lower average permeability (although not one that is statistically different) than the uncoated Cu46Zr54 membranes. Although Pd effectively dissociates hydrogen, the surface oxides on the amorphous metallic membranes may act as a transport barrier, counteracting the catalytic effect of the Pd. In the case of Cu40Zr60, there is limited, if any, oxide ( 5 nm thick) present on the membrane surfaces, therefore we still see the catalytic effect of the Pd in increasing the hydrogen permeability. Thin layers (10s of nanometers) of surface oxide could readily result in large resistance to the hydrogen transport because hydrogen transport through dense metallic membranes is by a dissociation–solution–diffusion process [36]. It is very difficult to eliminate all oxides in the synthesis of a metallic glass [37,38]. Thermodynamically, it is favorable for oxides to migrate to the surfaces of a metallic glass [39]. Zirconium-oxides have a larger negative heat of formation than
Fig. 4. RBS spectrum of as-synthesized (a) Cu63Zr37, (b) Cu46Zr54 and (c) Cu40Zr60 (before coating and before hydrogen permeation testing). The peaks in the circles indicate the presence of oxides.
Fig. 5. (a) Permeability vs time during the hydrogen permeation test of a Pd-coated Cu63Zr37 sample (b) XRD pattern of an uncoated Cu63Zr37 membrane permeate side after completion of the hydrogen permeation test.
268
T. Lai et al. / Journal of Membrane Science 489 (2015) 264–269
copper-oxides (at 298 K the heats of formation of oxides are CuO: 156.1 KJ/mol, Cu2O: 170.7 KJ/mol, ZrO2: 1097.5 KJ/mol [40]). However, according to the RUMP simulation, we found that the composition with the highest concentration of Zr (Cu40Zr60) has the least oxides on the surface. Sen et al. found that the Cu–Zr alloy with higher Zr concentration has a lower surface oxidation rate both in crystalline and glassy states [41]. This could be attributed to the different growth rate of different oxides which is determined by the Cu/Zr composition ratio as shown in Fig. 4. Hickman et al. found that between 573 K and 773 K only copper oxides can form in Zr20Cu80 [42]. Both reports by Sen et al. [42] and Hickman et al. [41] agree with our observations. Also, it is possible that the surface composition of the as-cast membranes is different than the bulk composition. Because Cu has a lower surface energy (1.79 J/m) than Zr (2.0 J/m) [38], copper has the tendency to migrate to the surface of the alloy. Fig. 5(a) shows the hydrogen permeability of a Pd-coated Zr37Cu63 membrane as a function of time during the permeation test. Fig. 5(b) shows the XRD spectra of the permeate side of Cu63Zr37 after the completion of the hydrogen permeation test. The system permeation error is within 10%. As Fig. 5 shows, after two hours of testing, the permeability starts to decrease and the permeability continues to decrease through the end of the test ( 6 h). This may be the result of the amorphous structure relaxation [22]. The onset of crystallization may happen sometime during the third hour of testing (as indicated by the decrease in permeability seen around 2.5 h of testing). Although we used a testing temperature of 573 K, which is 63 K lower than the Tg of Cu63Zr37, we hypothesize that with the presence of hydrogen, the structure relaxation and formation of hydride may left shift the TTT curve of Cu–Zr alloys (the nose moves to a lower temperature) which accelerated the crystallization of amorphous membranes [21,22]. Crystallization of an amorphous alloy further decreases the hydrogen permeability [10,11]. The crystallization of Cu–Zr may be the result of the formation of hydrides. As shown in Fig. 5(b), there are peaks corresponding to zirconium hydrides and zirconium oxides after permeability test. Zr has a large, negative hydride heat of formation [15]. The strong affinity of Zr for hydrogen facilitates hydrogen absorption of the alloy with higher concentration of Zr [10,14]. But the formation of zirconium hydride promotes crystallization and is detrimental to the thermal stability.
4. Conclusions We synthesized Cu–Zr (Zr at%¼37, 54, 60) free-standing, amorphous metallic membranes by splat quenching and verified their amorphous nature with XRD. Palladium catalytic coatings promoted hydrogen dissociation and recombination. We measured hydrogen permeability of the amorphous free-standing membranes at a feed hydrogen pressure of 140 kPa. The hydrogen permeability was lower than that predicted by the first-principle calculations. The observed permeability was explained by oxide formation on the surface. The variation in oxide thickness may be the result of the formation of different oxides as a function of the Cu/Zr composition ratio and their different growth kinetics. The stability tests and the permeability tests indicate that the presence of hydrogen accelerated crystallization which leads to significant permeability degradation. Overall, the thermal stability of amorphous metallic membranes is very important for hydrogen separation applications.
Acknowledgments We wish to acknowledge the financial support of Ira A. Fulton Schools of Engineering. Acknowledgment is made to the Donors of
the American Chemical Society Petroleum Research Fund for support of this research (award # 52461-DNI10). We gratefully acknowledge the use of facilities with the LeRoy Eyring Center for Solid State Science at Arizona State University. We also would like to acknowledge Dr. William L. Johnson, Pamela Albertson, and George Kaltenboeck at the California Institute of Technology for facilitating the use of their arc-melter and the splat quencher. We would like to acknowledge Dr. Jerry Y.S. Lin and his students for help in designing the hydrogen permeation testing system. We would also like to acknowledge Fred Pena at ASU for his extensive help in building and maintaining the hydrogen permeation testing system. We sincerely thank Dr. David Sholl and postdoc Dr. Zhu in Georgia Institute of Technology for discussions about the permeability results.
References [1] Hydrogen production, (2013). Available from: 〈http://energy.gov/eere/fuel cells/natural-gas-reforming〉. [2] M.D. Dolan, N.C. Dave, A.Y. Ilyushechkin, L.D. Morpeth, K.G. McLennan, Composition and operation of hydrogen-selective amorphous alloy membranes, J. Membr. Sci. 285 (1–2) (2006) 30–55. [3] N.W. Ockwig, T.M. Nenoff., Membranes for hydrogen separation, Chem. Rev. 107 (10) (2007) 4078–4110. [4] S.N. Paglieri, J.D. Way., Innovations in palladium membrane research, Sep. Purif. Methods 31 (1) (2002) 1–169. [5] F.A. Lewis., The palladium hydrogen system – structures near phase-transition and critical-points, Int. J. Hydrog. Energy 20 (7) (1995) 587–592. [6] C. Nishimura, M. Komaki, M. Amano., Hydrogen permeation characteristics of vanadium–nickel alloys, JIM, Mater. Trans. 32 (1991) 501–507. [7] S. Hao, D.S. Sholl, Comparison of first principles calculations and experiments for hydrogen permeation through amorphous ZrNi and ZrNiNb films, J. Membr. Sci. 350 (1-2) (2010) 402–409. [8] 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 (1-2) (2000) 289–294. [9] W.L. Johnson., Bulk glass-forming metallic alloys: science and technology, MRS Bull. 24 (10) (1999) 42–56. [10] S.N. Paglieri, N.K. Pal, M.D. Dolan, S.M. Kim, W.M. Chien, J. Lamb, et al., Hydrogen permeability, thermal stability and hydrogen embrittlement of Ni–Nb–Zr and Ni–Nb–Ta–Zr amorphous alloy membranes, J. Membr. Sci. 378 (1–2) (2011) 42–50. [11] S.N. Paglieri, N.K. Pal, S.M. Kim, M.D. Dolan, W.M. Chien, J. Lamb, et al., Ni–Nb– Zr–(Ta, Co) amorphous alloy membranes for hydrogen separation, in: Abstracts of Papers of the American Chemical Society, 2011, p. 242. [12] M.D. Dolan, S. Hara, N.C. Dave, K. Haraya, M. Ishitsuka, A.Y. Ilyushechkin, et al., Thermal stability, glass-forming ability and hydrogen permeability of amorphous Ni64Zr36 xMx (M ¼Ti, Nb, Mo, Hf, Ta or W) membranes, Sep. Purif. Technol. 65 (3) (2009) 298–304. [13] S. Hara, N. Hatakeyama, N. Itoh, H.M. Kimura, A. Inoue, Hydrogen permeation through palladium-coated amorphous Zr–M–Ni (M ¼Ti, Hf) alloy membranes, Desalination 144 (1–3) (2002) 115–120. [14] S. i. Yamaura, M. Sakurai, M. Hasegawa, K. Wakoh, Y. Shimpo, M. Nishida, et al., Hydrogen permeation and structural features of melt-spun Ni–Nb–Zr amorphous alloys, Acta Mater. 53 (13) (2005) 3703–3711. [15] R. Griessen, A. Driessen., Heat of formation and band structure of binary and ternary metal hydrides, Phys. Rev. B 30 (8) (1984) 4372–4381. [16] A.J. Maeland, L.E. Tanner, G.G. Libowitz., Hydrides of metallic glass alloys, J. Less Common Met. 74 (2) (1980) 279–285. [17] S. Yamaura, Y. Shimpo, H. Okouchi, M. Nishida, O. Kajita, H. Kimura, et al., Hydrogen permeation characteristics of melt-spun Ni–Nb–Zr amorphous alloy membranes, Mater. Trans. 44 (9) (2003) 1885–1890. [18] E.S. Park, J.H. Na, D.H. Kim, Abnormal behavior of supercooled liquid region in bulk-forming metallic glasses, J. Appl. Phys. 108 (5) (2010) 053515-1–053515-9. [19] S.I. Yamaura, S. Nakata, H. Kimura, A. Inoue., Hydrogen permeation of Ni–Nb– Zr metallic glasses in a supercooled liquid state, Mater. Trans. 47 (12) (2006) 2991–2996. [20] K.W. Park, J.P. Ahn, H.K. Seok, Y.C. Kim., Relationship between activation energy for hydrogen permeation and hydrogen permeation properties of amorphous Cu50Zr50 and Cu65Zr35 membranes, Intermetallics 19 (12) (2011) 1887–1890. [21] N. Eliaz, D. Eliezer., An overview of hydrogen interaction with amorphous alloys, Adv. Perform. Mater. 6 (1) (1999) 5–31. [22] K. Dini, R.A. Dunlap., Crystallization and hydrogen absorption in amorphous Cu60Zr40 and Cu50Zr50, J. Phys. F: Met. Phys. 15 (2) (1985) 273–277. [23] L. Ge, X. Hui, E.R. Wang, G.L. Chen, R. Arroyave, Z.K. Liu., Prediction of the glass forming ability in Cu–Zr binary and Cu–Zr–Ti ternary alloys, Intermetallics 16 (1) (2008) 27–33.
T. Lai et al. / Journal of Membrane Science 489 (2015) 264–269
[24] M.L. Lind, G. Duan, W.L. Johnson., Isoconfigurational elastic constants and liquid fragility of a bulk metallic glass forming alloy, Phys. Rev. Lett. 97 (1) (2006) 015501. [25] D.H. Xu, G. Duan, W.L. Johnson., Unusual glass-forming ability of bulk amorphous alloys based on ordinary metal copper, Phys. Rev. Lett. 92 (2004) 24. [26] A. Zhang, D. Chen, Z. Chen., Bulk metallic glass-forming region of Cu–Zr binary and Cu–Zr based multicomponent alloy systems, J. Alloy. Compd. 477 (1–2) (2009) 432–435. [27] S.Q. Hao, D.S. Sholl., Computational prediction of durable amorphous metal membranes for H2 purification, J. Membr. Sci. 381 (1–2) (2011) 192–196. [28] L. Ge, X.D. Hui, G.L. Chen, Z.K. Liu., Prediction of the glass-forming ability of Cu-Zr binary alloys, Acta Phys.-Chim. Sin. 23 (6) (2007) 895–899. [29] R. Nyholm, N. Mårtensson., Experimental core-hole ground state energies for the elements 41Nb to 52Te, Solid State Commun. 40 (3) (1981) 311–314. [30] V. Jayaraman, Y.S. Lin, M. Pakala, R.Y. Lin., Fabrication of ultrathin metallic membranes on ceramic supports by sputter-deposition, J. Membr. Sci. 99 (1) (1995) 89–100. [31] T. Lai, M.L. Lind., Heat treatment driven surface segregation in Pd77Ag23 membranes and the effect on hydrogen permeability, Int. J. Hydrog. Energy 40 (1) (2015) 373–382. [32] J.S.C. Jang, C.C. Koch., The glass transition temperature in amorphous Cu1 xZrx alloys synthesized by mechanical alloying/milling, Scr. Metall. 23 (10) (1989) 1805–1810. [33] D.H. Xu, B. Lohwongwatana, G. Duan, W.L. Johnson, C. Garland., Bulk metallic glass formation in binary Cu-rich alloy series – Cu100 xZrx (x ¼34, 36, 38.2,
[34]
[35] [36] [37]
[38] [39] [40] [41]
[42]
269
40 at%) and mechanical properties of bulk Cu64Zr36 glass, Acta Mater. 52 (9) (2004) 2621–2624. S.H. Sie, Rutherford backscattering spectrometry and nuclear reaction analysis, Surface Analysis Methods in Materials Science, Springer, Verlag Berlin Heidelberg New York, 2003. R. Schulz, K. Samwer, W.L. Johnson., Kinetics of phase separation in Cu50Zr50 metallic glasses, J. Non-Cryst. Solids 61–62 (Part 2) (1984) 997–1002. T.L. Ward, T. Dao., Model of hydrogen permeation behavior in palladium membranes, J. Membr. Sci. 153 (2) (1999) 211–231. A. Gebert, K. Buchholz, A. Leonhard, K. Mummert, J. Eckert, L. Schultz., Investigations on the electrochemical behaviour of Zr-based bulk metallic glasses, Mater. Sci. Eng.: A 267 (2) (1999) 294–300. J. Schroers., Processing of bulk metallic glass, Adv. Mater. 22 (14) (2010) 1566–1597. A. Gebert, P.F. Gostin, L. Schultz., Effect of surface finishing of a Zr-based bulk metallic glass on its corrosion behaviour, Corros. Sci. 52 (5) (2010) 1711–1720. I. Ihsan Barin, Thermochemical Data of Pure Substances, VCH, Weinheim, Germany, 1989. P. Sen, D.D. Sarma, R.C. Budham, K.L. Chopra, C.N.R. Rao., An electron spectroscopic study of the surface oxidation of glassy and crystalline Cu–Zr alloys, J. Phys. F: Metal Phys. 14 (2) (1984) 565. J.W. Hickman, E.A. Gulbransen., Oxide films formed on titanium, zirconium, and their alloys with nickel, copper, and cobalt, Anal. Chem. 20 (2) (1948) 158–165.