Method and system for fabrication of hydrogen-permeable membranes

Abstract

A method for fabrication of an hydrogen-permeable membrane, comprising forming an alloy of a target composition and structure from powders by mechanically alloying; and forming a membrane from the alloy of the target composition and structure.

Claims

1. A method for fabrication of an hydrogen-permeable membrane, comprising: a) forming an alloy of a target composition and structure from powders; and b) forming a membrane from the alloy of the target composition and structure.

2. The method of claim 1, wherein said step a) comprises mechanically alloying the powders.

3. The method of claim 1, wherein said step a) comprises mechanically alloying the powders and annealing the alloy.

4. The method of claim 1, said step a) comprises ball milling the powders.

5. The method of claim 1, said step a) comprises ball milling the powders and heat treatment.

6. The method of claim 1, wherein said step b) comprises forming the membrane by cold rolling.

7. The method of claim 1, wherein said step b) comprises forming the membrane by one of: cold spraying, colloidal spraying and paste painting the alloy of the target composition and structure on a substrate.

8. The method of claim 1, for fabrication of a Pd alloy hydrogen-permeable membrane, said step a) comprising mechanically alloying at Pd with at least one of Cu, Au and Ag powders.

9. The method of claim 1, wherein said step a) comprises mechanically alloying Pd, Cu, and Au powders into a fcc PdCuAu alloy and annealing into a bcc PdCuAu alloy.

10. The method of claim 1, for fabrication of a PdCuAu alloy hydrogen-permeable membrane, said step a) comprising mechanically alloying at least Pd and Cu, and Au powders into a fcc PdCuAu ternary alloy and annealing into a bcc PdCuAu ternary alloy, said step b) comprising cold rolling pellets of the bcc PdCuAu ternary alloy.

11. The method of claim 1, further comprising at least one of annealing the membrane and polishing the membrane.

12. The method of claim 1, further comprising cold rolling, annealing and polishing the membrane.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) In the appended drawings:

(2) FIG. 1 schematically shows separation of hydrogen by a membrane, as known in the art;

(3) FIG. 2 is a flow chart of a method according to an embodiment of an aspect of the present invention;

(4) FIG. 3A is a SEM photograph of BM6 70-?m aggregates;

(5) FIG. 3B is a SEM photograph of the BM6 70-?m aggregates of FIG. 3A at a larger magnification;

(6) FIG. 4A shows an EDX mapping of BM6 powder after ball milling: Na K?1;

(7) FIG. 4B shows an EDX mapping of BM6 powder after ball milling: Pd L?1;

(8) FIG. 5A shows a XRD graph of a powder with composition Pd41.0Cu56.1Au2.9 after ball milling;

(9) FIG. 5B shows a XRD graph of an alloy with composition Pd41.0Cu56.1Au2.9 after first annealing;

(10) FIG. 5C shows a XRD graph of a powder with composition Pd39.4Cu50.1Au10.1 after first ball milling;

(11) FIG. 5D shows a XRD graph of an alloy with composition Pd39.4Cu50.1Au10.1 after first annealing;

(12) FIG. 6A is a SEM picture of BM1 pellets after final annealing;

(13) FIG. 6B is a SEM picture of BM2 pellets after final annealing;

(14) FIG. 6C is a SEM picture of BM3 pellets after final annealing;

(15) FIG. 7A shows BM2 after final annealing;

(16) FIG. 7B shows EDX mapping for BM2 of FIG. 7A: PdL;

(17) FIG. 7C shows EDX mapping for BM2 of FIG. 7A: CuK;

(18) FIG. 7D shows EDX mapping for BM2 of FIG. 7A: AuL;

(19) FIG. 7E shows BM4 after final annealing;

(20) FIG. 7F shows EDX mapping for BM4 of FIG. 7E: PdL;

(21) FIG. 7G shows EDX mapping for BM4 of FIG. 7E: CuK;

(22) FIG. 7H shows EDX mapping BM4 of FIG. 7E: AuL;

(23) FIG. 8A is a SEM micrograph of a BM3 membrane before polishing;

(24) FIG. 8B is a SEM micrograph after polishing at a first location on the surface of the membrane of FIG. 8A;

(25) FIG. 8C is a SEM micrograph after polishing at a second location on the surface of the membrane of FIG. 8A, showing that some imperfections have remained after polishing;

(26) FIG. 9 show a schematic diagram of a permeation chamber for H.sub.2 permeability measurement;

(27) FIG. 10A shows variation of H.sub.2 flow with time at 30, 45 and 60 psig H2 single-gas feed pressure and 464? C. (pressure values on the y axis on the right side), the permeate side not submitted to any pressure control and no sweeping gas being used, measurements performed with a 277 ?m membrane with composition Pd.sub.40.1Cu.sub.53.1Au.sub.6.8 prepared according to the present invention;

(28) FIG. 10B shows evolution pressure on both the retentate and permeate sides of a the membrane of FIG. 10A in function of time;

(29) FIG. 10C shows results of an helium leak test performed on the membrane of FIG. 10A;

(30) FIG. 11 shows a summary of single-gas permeability measurement results;

(31) FIG. 12A shows gas mixture experiments (H.sub.2He 3%-941 ppm H.sub.2S) for Pd 250 ?m Alfa Aesar foil;

(32) FIG. 12B shows gas mixture experiments (H.sub.2He 3%-941 ppm H.sub.2S) for Pd pellets from commercial Pd powder;

(33) FIG. 13A shows results of gas mixture experiments (H.sub.2He 3%-941 ppm H.sub.2S) for BM1-4-:Pd.sub.40.1Cu.sub.59.9Au.sub.0;

(34) FIG. 13B shows results of gas mixture experiments (H.sub.2He 3%-941 ppm H.sub.2S) for BM15-4: Pd.sub.38.8Cu.sub.54.5Au.sub.6.6;

(35) FIG. 14A shows results of gas mixture experiments (H.sub.2He 3%-941 ppm H.sub.2S) for BM4-2: Pd.sub.39.8Cu.sub.49.2Au.sub.11;

(36) FIG. 14B shows results of gas mixture experiments (H.sub.2He 3%-941 ppm H.sub.2S) for BM4-5: Pd.sub.39.8Cu.sub.49.2Au.sub.11 (FIG. 14B);

(37) FIG. 15A and FIG. 15B shows EDX mapping of cross sections of the Pd 250 ?m Alfa Aesar foil as commercially available membrane;

(38) FIG. 15C, FIGS. 15D and 15E shows EDX mapping of cross sections of a BM3-1-Pd.sub.39.2Cu.sub.53.4Au.sub.7.4 membrane prepared according to the present invention respectively.

DESCRIPTION OF EMBODIMENTS OF THE INVENTION

(39) FIG. 2 is a flow chart of a method according to an embodiment of an aspect of the present invention.

(40) First PdCuAu ternary alloys (fcc) are formed by mechanical alloying from Pd, Cu, Au and NaCl (2 wt %) powders (step 12). A phase transition from fcc to bcc is promoted by annealing the PdCuAu alloy at 400? C. for 5 h under a mixture of Ar/5% H.sub.2 (step 14). Then pellets are formed by pressing 300 mg of bcc PdCuAu alloy into a 1.1 cm disk with a thickness of approximately 450 ?m using a load of 20 metric tons for 10 min in a hydraulic press (step 16). The resulting pellets are then sintered at 900? C. for 1 h under a mixture of Ar/5% H.sub.2 (step 18), and then the surface of the pellets are homogenized by cold-rolling with a distance between rolls starting at 500 ?m and decreased by steps of 25 ?m (step 20). A final annealing is performed at 400? C. for 5 h under a mixture of Ar/5% H.sub.2 (step 22), before polishing the pellets (step 24) to obtain an almost-mirror finish like surface.

(41) Pd pellets are heat treated under pure Ar only (steps 14, 18 and 22) to avoid the formation of cracks during cool down.

(42) Membranes thus obtained were tested under single gas conditions, first with He and then with H.sub.2; a final He leak test was further performed. Membranes were considered leak free, meaning that no gas was permeating through defects or pinholes, when no flow was registered in the helium tests, i.e. flow below 0.005 ml/min, which was the resolution limit.

(43) In step 12, Pd, Cu and Au powders were ball milled for at least 10 hours, i.e. for example 18h under Ar. NaCl was used as a process control agent. One stainless steel ball of 2 g and 2 stainless steel balls of 1 g were used. All solids were weighted and sealed under Ar. Ball milling was performed with no temperature control. Amounts of Pd, Cu, Au and NaCl used in 5 different compositions BM.sub.1-5 are listed in Table 1 below:

(44) TABLE-US-00001 TABLE 1 Pd Cu Au NaCl Total solids Synthesis (g) (g) (g) (g) (g) BM1 1.034 0.926 0.000 0.04 2 BM2 0.985 0.838 0.137 0.04 2 BM3-BM5 0.927 0.733 0.300 0.04 2 BM4 0.875 0.640 0.445 0.04 2

(45) As can be seen by X-ray diffraction (XRD), ball milling promotes the formation of one single fcc PdCuAu alloy; in some cases, unalloyed metals can still be identified in the diffractograms (see Table 2 below). However, subsequent heat treatments guarantee complete alloying at all compositions (see Table 3 below).

(46) TABLE-US-00002 TABLE 2 Pd Cu Au (at %) (at %) (at %) Impurities* XRD BM1 39.2 60.8 0.0 NaCl/Fe, Ni fcc BM2 41.0 56.1 2.9 NaCl/Fe fcc + Pd + Cu BM3 39.8 52.7 7.5 NaCl fcc BM4 39.4 50.5 10.1 NaCl/Fe fcc + Pd + Cu + (Au) BM6 95 5 at % Na Pd

(47) As can be seen in the electron microscopy (SEM) photographs of FIG. 3 representative of the aggregates of BM1-BM6, aggregates 70 ?m and below are obtained. The composition of the alloys does not seem to have an effect on the size or shape of the aggregates.

(48) Element distribution images (EDX) of BM6 powder after ball milling shown in FIG. 4 suggest that NaCl is dispersed on the surface of palladium forming granules 2 ?m long.

(49) In step 14, heat treatment is performed at 400? C. under Ar 5% H2 for 5 hours. Complete transition from fcc to bcc phase may occur depending on the alloy composition. Unalloyed metals are no longer observed in XRD (see FIG. 5 [svp pr?cisez l?gende de chaque figure, merci] and Table 3 below).

(50) TABLE-US-00003 TABLE 3 Pd Cu Au XRD after (at %) (at %) (at %) 1.sup.st HT BM1 39.2 60.8 0.0 bcc BM2 41.0 56.1 2.9 bcc BM3 39.8 52.7 7.5 bcc BM4 39.4 50.5 10.1 fcc + bcc

(51) In step 16, 300 mg of bcc PdCuAu alloy was pressed into a 1.1 cm disk and about 450 ?m thick using 20 metric tons for 10 min.

(52) In step 18, the resulting pellets were sintered at 900? C. for 1 h under a mixture of Ar/5% H.sub.2.

(53) In step 20, the thickness of the pellets was decreased and homogenized by cold rolling, with a distance between rolls starting at 500 ?m and decreased by steps of 25 ?m.

(54) Table 4 below shows measurements taken on different days before and after cold rolling. Only the results of membranes that showed 0 ml/min flow under He are shown.

(55) TABLE-US-00004 TABLE 4 Thickness (um) Membrane Before CR After CR* BM1-2 477 362 BM2-2 449 354; 359# BM4-2 421 307; 319# BM5-2 548 440

(56) In step 22, the membranes were annealed at 400? C. for 5 h under Ar/5% H.sub.2. SEM, EDX composition and XRD analysis were performed. Fe content was found to be below 1 at %; there was no clear evidence for the presence of NaCl (see Table 5 below and FIG. 6).

(57) TABLE-US-00005 TABLE 5 Pd Cu Au XRD after (at %) (at %) (at %) 2nd annealing BM1 40 60 0.0 Bcc BM2 40 57 3 Bcc BM3 39 53 7 bcc(+fcc??) BM4 40 49 11 bcc + fcc BM5 39 54 7 bcc

(58) FIG. 7 show EDX mapping after final annealing. EDX mapping shows homogenous distribution of all 3 elements Pd, Cu and Au throughout the membrane (Images taken at 2500 times magnification).

(59) In step 24, all membranes were polished using SiC sanding paper (P800-P1500-P2500). Polishing was finished with Al.sub.2O.sub.3 1 ?m paste on a microcloth. Both sides of the membranes were polished. It was found that polishing decreases defects and pinholes on the surface of the membranes, making the membranes more selective. For example, the flow under 60 psig He may be reduced from 0.80 ml/min to less than 0.005 ml/min (detection limit of He mass flow meter) by polishing (see FIG. 8). Diamond paste (6 ?m and 1 ?m) was used instead of SiC sanding paper P2500 and Al.sub.2O.sub.3, when polishing pure Pd pellets.

(60) Table 6 below shows the thickness in ?m of the membranes obtained after polishing:

(61) TABLE-US-00006 TABLE 6 After polishing Membrane # (um) BM1-2 327 BM2-2 300 BM4-2 304 BM5-2 277

(62) Then H.sub.2 permeability was measured, using the configuration of a permeation chamber as shown in FIG. 9, operated in dynamic mode, i.e. with a constant flow of gas in the feed side (excess is vented to exit), at a temperature of about 350-450? C. (from furnace display), a pressure on feed side below 100 psig (7.8 atm), 3 MF (H.sub.2 0-100 ml/min; H.sub.2 0-20 ml/min; He 0-5 ml/min; detection limit of He MF: 0.005 ml/min); 2 pressure gauges (feed/permeate).

(63) FIG. 10 show single gas experiments for BM5-2, of a composition of Pd.sub.40.1Cu.sub.53.1Au.sub.6.8. This membrane failed the initial He leak test, but was successfully repaired by cold rolling, annealing and polishing. Final thickness was 277 ?m. The membrane was assembled in the permeation chamber between two stainless steel gaskets for support. The chamber was subsequently degassed by alternating three times 5 min vacuum with 30 psig Ar. Afterwards the temperature was raised to 50? C. while keeping Ar inside the furnace at atmospheric pressure. Finally, 10 psig H.sub.2 was introduced and the temperature was raised to 450? C.

(64) Ideal selectivity is used when operating with single gases as:

(65) ${?}_{H2/\mathrm{He}}^{*}=\frac{{\mathrm{Permeability}}_{H2}}{{\mathrm{Permeability}}_{\mathrm{He}}}$

(66) The separation factor for gases in mixtures, also called the selectivity of the membrane, ?.sub.H.sub.2,j, is given by the mole fractions of the components in the feed (x) and in the permeate (y):

(67) ${?}_{H_{2,j}}=\frac{y_{H_2}/y_j}{x_{H_2}/x_j}$

(68) TABLE-US-00007 TABLE 7 G1 G2 Flow Permeability (psig) (psig) (ml/min) (mol .Math. m.sup.?1 .Math. s.sup.?1 .Math. Pa.sup.?0.5) ?* He 57.7 ?1.17 <0.005 <5.0 ? 10.sup.?11 (?P = ?0.3) 29.32 ?0.33 0.89 2.3 ? 10.sup.?8 >469 43.16 ?0.34 1.39 (R.sup.2 = 0.9980) 59.06 ?0.33 1.82 He 58.48 ?1.08 <0.005 4.9 ? 10.sup.?11 (?P = ?0.28)

(69) Table 7 gives the data of each point of FIG. 10.

(70) FIG. 11 shows a summary of single-gas permeability measurement results (circles), with a pressure on retentate of 30-60 psig and permeate at atmospheric pressure. Results by McKinley (U.S. Pat. No. 3,439,474-1969) are also shown (squares), for measurements performed on 25.4 ?m foils supported on porous stainless steel substrates, and permeate at atmospheric pressure.

(71) FIG. 12 show results of gas mixture experiments (H.sub.2He 3%-941 ppm H.sub.2S) for a Pd 250 ?m Alfa Aesar foil membrane as commercially available (FIG. 12A) and for Pd pellet from a commercial Pd powder (FIG. 12B). In the first case (FIG. 12A), a 50% decrease in permeability was observed at time 60 min in the presence of H.sub.2S 941 ppm and switching to pure H.sub.2 at time 90 min allowed a recovery of 70% of the initial permeability; no He was detected downstream. In the second case (FIG. 12B), a 47% decrease in permeability was observed in the presence of H.sub.2S 941 ppm at time 30 min; and switching to pure H.sub.2 at time 60 min allowed a complete recovery to the initial permeability with a final selectivity ?=27.6.

(72) FIG. 13 show results of gas mixture experiments (H.sub.2He 3%-941 ppm H.sub.2S) for BM1-4: Pd.sub.40.1Cu.sub.59.9Au.sub.0 (FIG. 13A), and for BM5-4: Pd.sub.38.8Cu.sub.54.5Au.sub.6.6 (FIG. 13B). In the first case (FIG. 13A), a 12% decrease in permeability was observed at time 30 min in the presence of H.sub.2S 941 ppm and the membrane ruptured at the end. In the second case (FIG. 13B), a 74% decrease in permeability was observed in the presence of H.sub.2S 941 ppm at time 30 min and switching to pure H.sub.2 at time 60 min allowed a 58% recovery to the initial permeability with a final selectivity ?=1.

(73) FIG. 14 show results of gas mixture experiments (H.sub.2He 3%-941 ppm H.sub.2S) for BM4-2: Pd.sub.39.8Cu.sub.49.2Au.sub.11 (FIG. 14A) and for BM4-5: Pd.sub.39.8Cu.sub.49.2Au.sub.11 (FIG. 14B). In the first case (FIG. 14A), a 75% decrease in permeability was observed at time 30 min in the presence of H.sub.2S 941 ppm and switching to pure H.sub.2 at time 60 min allowed a 55% recovery to the initial permeability. In the second case (FIG. 14B), a 48% decrease in permeability was observed in the presence of H.sub.2S 941 ppm and switching to pure H.sub.2 allowed a 71% recovery to the initial permeability with a final selectivity ?=1.9.

(74) FIG. 15 show EDX mapping of cross sections of the Pd 250 ?m Alfa Aesar foil membrane as commercially available (FIGS. 15A, 15B) and of a BM3- 1-Pd.sub.39.2Cu53.4Au7.4 membranes prepared according to the present invention respectively (FIGS. 15C-15E), all being heat treated with H.sub.2S for 15 min at 450? C.

(75) In the case of Alfa Aesar foil membrane as commercially available, FIGS. 15A and 15B show an homogeneous distribution of Pd (FIG. 15A) and S (FIG. 15B) within the bulk of the membrane, which may compromise the efficiency of the membrane for separating hydrogen.

(76) In the case of the Pd.sub.39.2Cu.sub.53.4Au.sub.7.4 membrane prepared according to the present invention, FIGS. 15D and 15E respectively show that Cu and S segregate on the surface of the membrane. As such, these surface elements may be removed by cleaning of the membrane; with only less than about 10% of the elements Cu or S being found in the bulk of the membrane, the membrane may still be efficient for separating hydrogen.

(77) There is thus provided a method for the production of H.sub.2 permeable membranes from ball milled powders. The method allows producing membranes of a thickness comprised in a range between 277 ?m and 327 ?m.

(78) Four different compositions of PdCuAu alloys were tested. About 4 compositions in 10 were found to pass initial leak tests (pellets of PdCuAu alloys).

(79) Tests performed using PdCu, PdCuAg and PdCuAu showed that the membranes obtained had a composition very similar to the powder initially used to form the alloy after milling. This allows controlling the composition of the membrane by controlling the composition of the initial powder mixture. Membranes with composition Pd40Cu(60-x)Aux with x=0, 3, 7 and 11 at % were formed.

(80) A last polishing step was found to allow obtaining highly hydrogen selective membranes; it was shown that membranes that developed leaks during testing could be regenerated by polishing. Selectivity of the membranes under pure hydrogen conditions, called ideal selectivity, was higher than 130; after re-polishing a selectivity higher than 469 was obtained. Selectivity in the presence of H.sub.2S is below 28.

(81) The lowest He flow that was possible to measure was 0.005 ml/min. This corresponds to a permeability of He below 5?10.sup.?9 mol.Math.m.sup.?1.Math.s.sup.?1.Math.Pa.sup.?0.5. Consequently, ideal selectivity factors are above 130. Preliminary gas chromatography (GC) measurements showed that 6 ppmv He is present in the permeate side of a membrane prepared from commercial Pd powder after 15 min under 45 psig He. Comparatively, a commercial Pd 250 ?m foil shows under the same conditions a He content below 1 ppmv. Obtained hydrogen permeability, under pure hydrogen gas conditions, varied between 6.9?10.sup.?9 mol m.sup.?1 s.sup.?1 Pa.sup.?0.5 and 2.1?10.sup.?8 mol m.sup.?1 s.sup.?1 Pa.sup.?0.5.

(82) The selectivity considerably decreases in the presence of H.sub.2S; the longer the time of exposure to H.sub.2S gas mixture, the lower the selectivity. This is specific to the PdCuAu alloys and to the experimental conditions (temperature, pressure and H.sub.2S concentration) and is not related to the method used to prepare the membranes.

(83) In the presence of pure H.sub.2S, copper, and possibly gold, segregate to the surface. This creates a layer that binds to sulfur preventing it from fully migrating to the bulk of the membrane, as it happens with pure Pd. The higher the Au content of the membrane, the lower the S content in the bulk. As the S-contaminants remain mostly on the surface, a cleaning method may be designed to recover the initial activity of the membrane.

(84) However, the CuS and possibly Au layer on top of the membrane may prevent H.sub.2 permeation.

(85) The present method comprises formation of an alloy of a target composition and structure directly from mechanically alloying the precursor powders using ball milling technique. More precisely, the method comprises pressing and sintering the obtained powder, before cold-rolling and polishing.

(86) Cold spray, colloidal spray or paste painting may also be contemplated to prepare membranes supported on porous substrates from the mechanically alloyed powders.

(87) The present alloy preparation method, being an out of equilibrium method, allows fabricating membranes with compositions that cannot be reached using conventional methods, such as membranes in metastable alloys, extended solid solution etc . . . , which in turn allows tailoring specific membrane shaving target properties, in terms of H.sub.2 permeability for example.

(88) Large quantities of powder may be prepared simultaneously, depending on the size of the crusher used, which may have a capacity of up to Kg in industrial settings.

(89) There is generally provided a method for preparation of palladium alloys from powders by mechanical alloying. There is provided a method for fabrication of hydrogen-permeable membranes from alloy powders produced by ball milling.

(90) The scope of the claims should not be limited by the embodiments set forth in the examples, but should be given the broadest interpretation consistent with the description as a whole.