Chemically stable ceramic-metal composite membrane for hydrogen separation

Abstract

A hydrogen permeation membrane is provided that can include a metal and a ceramic material mixed together. The metal can be Ni, Zr, Nb, Ta, Y, Pd, Fe, Cr, Co, V, or combinations thereof, and the ceramic material can have the formula: BaZr.sub.1-x-yY.sub.xT.sub.yO.sub.3- where 0x0.5, 0y0.5, (x+y)>0; 00.5, and T is Sc, Ti, Nb, Ta, Mo, Mn, Fe, Co, Ni, Cu, Zn, Ga, In, Sn, or combinations thereof. A method of forming such a membrane is also provided. A method is also provided for extracting hydrogen from a feed stream.

Claims

1. A hydrogen permeation membrane, comprising: a metal and a ceramic material mixed together, wherein the metal comprises Ni, and wherein the ceramic material has the formula:
BaZr.sub.1-x-yY.sub.xT.sub.yO.sub.3- where 0<x0.5, 0y0.5, 0<(x+y)<1; 00.5, and T is Sc, Ti, Nb, Ta, Mo, Mn, Fe, Co, Ni, Cu, Zn, Ga, In, Sn, or combinations thereof, wherein at least a portion of the ceramic material is prepared with EDTA, citric acid, or a combination thereof prior to mixing with the metal.

2. The hydrogen permeation membrane of claim 1, comprising the ceramic material in about 40% to about 80% by volume.

3. The hydrogen permeation membrane of claim 1, wherein the hydrogen permeation membrane has a thickness of about 0.01 mm to about 10 mm.

4. The hydrogen permeation membrane of claim 1, where 0<y0.5.

5. The hydrogen permeation membrane of claim 4, where 0<0.5.

6. A method of forming a membrane, comprising: mixing a metal and a ceramic powder to form a metal-ceramic mixture, wherein the metal comprises Ni; pressing the metal-ceramic mixture to form a composite membrane; and sintering the metal-ceramic mixture at a temperature of about 1100 C. to about 1700 C., wherein the ceramic powder comprises a ceramic material having the formula:
BaZr.sub.1-x-yY.sub.xT.sub.yO.sub.3- where 0<x0.5, 0y0.5, 0<(x+y)<1; 00.5, and T is Sc, Ti, Nb, Ta, Mo, Mn, Fe, Co, Ni, Cu, Zn, Ga, In, Sn, or combinations thereof, wherein at least a portion of the ceramic powder is prepared with EDTA, citric acid, or a combination thereof.

7. The method of claim 6, wherein the metal-ceramic mixture is sintered in a reducing atmosphere.

8. The method of claim 7, wherein the reducing atmosphere comprises H.sub.2.

9. The method of claim 6, further comprising: heating the metal-ceramic mixture is in an inert atmosphere prior to sintering in the reducing atmosphere.

10. The method of claim 9, wherein the inert atmosphere comprises N.sub.2.

11. The method of claim 9, wherein the inert atmosphere comprises Ar.

12. The method of claim 6, where 0<y0.5.

13. The method of claim 12, where 0<0.5.

14. The method of claim 6, wherein the metal and the ceramic powder are mixed such that the metal-ceramic mixture comprises the ceramic material in about 40% to about 80% by volume.

15. A method of extracting hydrogen from a feed stream, comprising: exposing the feed stream to a first side of a membrane at a temperature of about 600 C. to about 1000 C., wherein the feed stream comprises hydrogen; and collecting pure hydrogen gas from a second side of the membrane opposite of the first side, wherein the membrane comprises a metal and a ceramic material, wherein the metal comprises Ni, and wherein the ceramic material has the formula:
BaZr.sub.1-x-yY.sub.xT.sub.yO.sub.3- where 0<x0.5, 0y0.5, 0<(x+y)<1; 00.5, and T is Sc, Ti, Nb, Ta, Mo, Mn, Fe, Co, Ni, Cu, Zn, Ga, In, Sn, or combinations thereof, wherein at least a portion of the ceramic material is prepared with EDTA, citric acid, or a combination thereof prior to mixing with the metal.

16. The hydrogen permeation membrane of claim 1, wherein the grain size of the ceramic material is less than or equal to 1.17 m.

17. The hydrogen permeation membrane of claim 1, wherein a portion of the ceramic material is not sinter-active.

18. The hydrogen permeation membrane of claim 1, wherein the ceramic material comprises BaZr.sub.0.8Y.sub.0.2O.sub.3-.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) A full and enabling disclosure of the present invention, including the best mode thereof to one skilled in the art, is set forth more particularly in the remainder of the specification, which includes reference to the accompanying figures.

(2) FIG. 1 shows the cross-section SEM image of a sintered NiBZY membrane, according to the Examples. Ni particles (210 m) were embedded in the dense BZY ceramic matrix. The size of BZY grains is about 1 m. The membrane is very dense and free of pores.

(3) FIG. 2 shows the time dependence of hydrogen flux of NiBZY membranes prepared by different methods measured in wet 20% H.sub.2 at 900 C. according to the Examples. All membrane thickness is 0.40 mm. The NiBZY4 membrane shows very stable performance during the whole measurement process (about 160 h).

(4) FIG. 3 shows time dependence of the hydrogen permeation flux of a 0.40-mm-thick NiBZY4 membrane in wet 40% H.sub.2 with 0, 30, 50% CO.sub.2 at 900 C., according to the Examples. The membrane shows very stable and even enhanced performance during the measurement in CO.sub.2 (about 160 h).

(5) FIG. 4 shows the XRD pattern of NiBZY4 membrane surface after permeation test in wet H.sub.2 and CO.sub.2 at 900 C., according to the Examples. The XRD patterns only shows XRD peaks corresponding to BZY or Ni phases, suggesting that Ni and BZY phases remained stable in wet 50% CO.sub.2 at 900 C.

(6) FIG. 5a shows a schematic for the synthesis and chemical stability test processes of BZYNiO2 and NiBZYNiO2 (denoted as NiBZY1 in Table 1).

(7) FIG. 5b shows a schematic for the synthesis and chemical stability test processes of BaY.sub.2NiO.sub.5.

(8) FIG. 6 shows XRD patterns obtained according to the Examples from surfaces of NiBZY1 after boiling for 20 h (a), NiBZY1 (b), NiBZY2 (c), NiBZY3 (d), and NiBZY4 (e) membranes after annealing in wet 17% H.sub.2 and 80% CO.sub.2 at 900 C. for 50 h, where b: BZY (JCPDS 06-0399), n: Ni (JCPDS 04-0850), c: BaCO.sub.3 (JCPDS 05-0378), and y: Y.sub.2O.sub.3 (JCPDS 41-1105).

(9) FIG. 7 shows XRD patterns of fresh BaY.sub.2NiO.sub.5 powder (a), BaY.sub.2NiO.sub.5 powder after annealing in CO.sub.2 at 900 C. for 10 h (b), BaY.sub.2NiO.sub.5 powder after annealing in dry H.sub.2 at 900 C. for 20 h (c), BaY.sub.2NiO.sub.5 powder after annealing in wet H.sub.2 at 900 C. for 20 h (d), where b: BaY.sub.2NiO.sub.5 (JCPDS 41-0463), c: BaCO.sub.3, y: Y.sub.2O.sub.3, n: NiO (JCPDS 78-0643), B: BaY.sub.2O.sub.4 (JCPDS 82-2319), N: Ni.

(10) FIG. 8a shows Rietveld refinement for XRD patterns obtained from surfaces of sintered BZYNiO2 pellets before treatment in wet CO.sub.2 at 700 C. for 100 h. The unit cells and lattice constants are summarized in Table 2.

(11) FIG. 8b shows Rietveld refinement for XRD patterns obtained from surfaces of sintered BZYNiO2 pellets after treatment in wet CO.sub.2 at 700 C. for 100 h. The unit cells and lattice constants are summarized in Table 2.

(12) FIG. 9a shows surface SEM images of BZYNiO2 before treatment in wet CO.sub.2 at 700 C. for 100 h.

(13) FIG. 9b shows surface SEM images of BZYNiO.sub.2 after treatment in wet CO.sub.2 at 700 C. for 100 h.

(14) FIG. 10 shows EDX spectra obtained from spots 1 (a), 2 (b), and 3 (c) in FIG. 9.

DEFINITIONS

(15) Chemical elements are discussed in the present disclosure using their common chemical abbreviation, such as commonly found on a periodic table of elements. For example, hydrogen is represented by its common chemical abbreviation H; helium is represented by its common chemical abbreviation He; and so forth.

(16) As used herein, the prefix nano refers to the nanometer scale up to about 100 nm. For example, particles having an average diameter on the nanometer scale (e.g., from about 0.1 nm to about 100 nm) are referred to as nanoparticles.

DETAILED DESCRIPTION

(17) Reference now will be made to the embodiments of the invention, one or more examples of which are set forth below. Each example is provided by way of an explanation of the invention, not as a limitation of the invention. In fact, it will be apparent to those skilled in the art that various modifications and variations can be made in the invention without departing from the scope or spirit of the invention. For instance, features illustrated or described as one embodiment can be used on another embodiment to yield still a further embodiment. Thus, it is intended that the present invention cover such modifications and variations as come within the scope of the appended claims and their equivalents. It is to be understood by one of ordinary skill in the art that the present discussion is a description of exemplary embodiments only, and is not intended as limiting the broader aspects of the present invention, which broader aspects are embodied exemplary constructions.

(18) Composite metal-BZY-based membranes are expected to possess both high hydrogen permeation flux, mechanical strength, and chemical stability, which are the key factors for successful adoption of hydrogen permeation membrane for practical applications. The metal here can be Nickel, Zirconium, Niobium, Tantalum, Yttrium, Palladium, Iron, Chromium, Cobalt, Vanadium, etc, or the binary alloy of these metals.

(19) Metal-ceramic composite membranes are generally provided, along with their methods of preparation. In one embodiment, the metal-ceramic composite membranes include
M-BaZr.sub.1-x-yY.sub.xT.sub.y).sub.3-
where 0x0.5, 0y0.5; M is Ni, Zr, Nb, Ta, Y, Pd, Fe, Cr, Co, V, or a combinations thereof; and T is Sc, Ti, Nb, Ta, Mo, Mn, Fe, Co, Ni, Cu, Zn, Ga, In, Sn, or a combinations thereof. This structure is referred to herein as M-BZYT. In one embodiment, the volume ratio of BZYT is between about 40% and about 80% (e.g., about 40% to about 70% by volume).

(20) The M-BZYT membranes show excellent high hydrogen permeation flux and chemical stability in H.sub.2O, CO.sub.2, H.sub.2S and other contaminants. Commercial metal powder is used as source of metal phase. Sintering-active BZY powders and metal powders are mixed, pressed, and sintered to obtain dense composite membranes. The membranes shows stable performance in the presence of concentrated CO.sub.2, H.sub.2S.

(21) The invention provides composite membranes based on metal and perovskite oxide BZY for hydrogen permeation. Hydrogen can diffuse through the membrane in the form of atoms through metal phase or protons through BZYT phase. The critical properties of the membranes include permeation flux, chemical stability in H.sub.2O, CO.sub.2, and H.sub.2S-containing atmosphere.

(22) The invented membranes are useful for extracting hydrogen from any feed stream containing hydrogen with a pressure between 1 and 1000 psi at a temperature between 600 and 1000 C. Theoretically, 100% pure hydrogen is obtained because the membranes are dense and allow no other gas to pass through. The flux of the membrane can be affected by many parameters, including phase composition of BZYT phase, volume ratio of metal and BZY, membrane thickness, temperature, and humidity content in feed gas.

(23) The method of forming such membranes can include mixing a metal and a BZYT powder, pressing, and sintering at temperature between about 1100 C. and about 1700 C. The sintering atmosphere can be reducing atmosphere (e.g., 5% H.sub.2/N.sub.2). The atmosphere can also be first in inert gas (N.sub.2, Ar, etc) and then in reducing atmosphere containing hydrogen (e.g., 5% H.sub.2/N.sub.2).

EXAMPLES

(24) Hydrogen separation membranes based on high temperature proton conductors have been pursued for a long time because of its potential to greatly reduce the energy and capital cost of large-scale hydrogen production from steam methane reforming (SMR). A key to their successful applications is the development of a membrane with high performance, chemical and mechanical stability. Yttrium-doped barium cerate (BCY) possesses high proton conductivity but poor chemical stability in H.sub.2O and CO.sub.2. Numerous efforts have been devoted to the improvement of its chemical stability, mainly through the partial replacement of Ce by cations such as In, Sn, Ti, Zr, Nb, Ta, etc. The performance of NiBaCe.sub.0.8Y.sub.0.2O.sub.3- (Ni-BCY) and NiBaZr.sub.0.1Ce.sub.0.7Y.sub.0.2O.sub.3- (NiBZCY) degraded 100% and 43%, respectively, in wet 40% H.sub.2 and 30% CO.sub.2 at 900 C.

(25) Another strategy is to develop hydrogen separation membranes with chemically stable proton conductors such as Ln.sub.6WO.sub.12- (Ln refers to lanthanides), Ca-doped LaNbO.sub.4, and NiLa.sub.0.4875Ca.sub.0.0125Ce.sub.0.5O.sub.2-, but their performances are still much inferior to that of NiBZCYYb, mostly due to their lower proton and/or electronic conductivities. Y-doped BaZrO.sub.3 (BZY) has recently been explored as a proton conductor for electrolyte of solid oxide fuel cells and hydrogen separation membranes because of its excellent bulk proton conductivity and chemical stability. Since BaZr.sub.0.8Y.sub.0.15Mn.sub.0.05O.sub.3- (BZYM) shows very poor hydrogen flux due to the lack of electronic conductivity, dense composite NiBZY membranes have been fabricated and demonstrated much higher hydrogen flux than that of BZYM. During the fabrication of BZY and NiBZY membranes, the main obstacles are BZY's highly refractory nature, poor grain boundary proton conductivity, and high number of grain boundaries due to limited grain growth during sintering. One viable solution to these problems is the adoption of sintering aids (e.g., NiO, ZnO, CuO, LiNO.sub.3) which can significantly lower the sintering temperature of BZY and promote the grain growth. NiO has been reported as the most effective sintering aid in promoting the grain growth, which is crucial to reduce the large grain boundary resistance. Dense BZY ceramic with grains as large as 5 m can be achieved using BZY powders prepared by solid state reactive sintering method with 1 wt. % NiO after sintering at 1500 C. for 24 h, accompanied by the formation of a secondary phase, BaY.sub.2NiO.sub.5, which has a low melting point and promotes the sintering of BZY. In our previous study, dense NiBZY membrane with large BZY grains (1-2 m) was successfully fabricated using BZY powder prepared by solid state reaction method with 2 wt. % NiO (denoted as BZYNiO.sub.2) containing BaY.sub.2NiO.sub.5. In contrast, the BZY grains are very small (0.25 m) in the membrane prepared using NiBZY powder obtained from the gel combustion method and subsequent reduction, which contains no BaY.sub.2NiO.sub.5. Apparently, BaY.sub.2NiO.sub.5 plays an important role in the fabrication of dense NiBZY membrane with large BZY grains. These studies have mainly focused on the sintering behavior, microstructure, and electrical/permeation properties. No work has been reported on the chemical stability of BZY and NiBZY in H.sub.2, H.sub.2O, and CO.sub.2 after the introduction of the sintering aids, which is crucial for their successful applications. Although the weight ratio of NiO sintering aid seems to be low, the weight ratio of BaY.sub.2NiO.sub.5 is high due to the large difference in molecular weight between BaY.sub.2NiO.sub.5 (453.83 g/mol) and NiO (74.69 g/mol). If 2 wt. % NiO (based on the total weight of BaCO.sub.3, ZrO.sub.2, and Y.sub.2O.sub.3) is completely converted to BaY.sub.2NiO.sub.5, there should be 13.8 wt. % BaY.sub.2NiO.sub.5 in the BZYNiO.sub.2 ceramic. Unlike BZY, BaY.sub.2NiO.sub.5 may be unstable in H.sub.2, H.sub.2O, or CO.sub.2-containing atmospheres at elevated temperatures, potentially leading to the chemical instability of BZYNiO.sub.2 ceramic and NiBZY membrane, and thus insulating phases will be formed on the surfaces and the performance will degrade after exposure to H.sub.2O and CO.sub.2. Fortunately, this problem can be mitigated by reducing the amount of BaY.sub.2NiO.sub.5 through tailoring the BZY powders used in the fabrication process. In the NiBZY membrane prepared with BZYNiO.sub.2 (denoted as NiBZY1), 2 wt. % NiO was directly mixed with BaCO.sub.3, Y.sub.2O.sub.3, and ZrO.sub.2 in the preparation of BZYNiO.sub.2. A lot of BaY.sub.2NiO.sub.5 is formed in the following calcination process. NiBZY2 membrane is fabricated using BaZr.sub.0.8Y.sub.0.2O.sub.3- prepared by solid state reaction method without NiO (denoted as BZY20-SSR). However, BaY.sub.2NiO.sub.5 is still formed by the reaction among NiO (formed by partial oxidation of Ni during the sintering in N.sub.2 containing a little oxygen), residual BaCO.sub.3, and Y.sub.2O.sub.3 in BZY20-SSR. Because the amount of BaCO.sub.3 and Y.sub.2O.sub.3 is much less than that in uncalcined BZYNiO2, the amount of BaY.sub.2NiO.sub.5 in NiBZY2 should be less than that of NiBZY1. In order to further reduce the amount of BaY.sub.2NiO.sub.5, we employee BZY20 powder prepared by combined EDTA-citric (CEC) method (BZY20-CEC). CEC is a wet-chemical method and the distribution of particles is much more homogeneous than that of solid state reaction. Therefore, BZY20-CEC is free of residual BaCO.sub.3 and Y.sub.2O.sub.3 after calcination, and will not contribute to the formation of BaY.sub.2NiO.sub.5, which allows us to adjust the amount of BaY.sub.2NiO.sub.5 by partially replacing BZY20-SSR with BZY-CEC. Therefore, another two NiBZY membranes were prepared by further replacing 60% and 80% BZY20-SSR powder in NiBZY2 by BZY20-CEC (denoted as NiBZY3 and NiBZY4, respectively). The amount of BaY.sub.2NiO.sub.5 is expected to follow this order: NiBZY1>NiBZY2>NiBZY3>NiBZY4.

(26) In this study, the performance stability of these NiBZY membranes was evaluated in wet H.sub.2 with and without CO.sub.2 at 900 C. The chemical stability of BaY.sub.2NiO.sub.5, BZYNiO2, and NiBZY was investigated in H.sub.2, H.sub.2O, and CO.sub.2-containing atmospheres. A stable NiBZY membrane was developed, demonstrating excellent chemical and performance stability in the presence of H.sub.2, H.sub.2O, and CO.sub.2.

Experimental

(27) FIGS. 5a and 5b show the schematic of the material synthesis and chemical stability test processes employed in this work. All chemicals were purchased from Alfa Aesar with purity >99.5%. BZY powders were prepared through the solid state reaction method using 0 and 2 wt. % NiO as sintering aid (denoted as BZY20 and BZYNiO2, respectively). For simplicity, only BZYNiO2 is shown in FIGS. 5a and 5b. Stoichiometric amounts of BaCO.sub.3, ZrO.sub.2, and Y.sub.2O.sub.3 were ball-milled in ethanol with ZrO.sub.2 balls for 24 h. 0 and 2 wt. % NiO based on the total weight of BaCO.sub.3, ZrO.sub.2, and Y.sub.2O.sub.3 were added to the powder mixture prior to the ball-milling to obtain BZY20 and BZYNiO2 powders, respectively. The ball-milled powders were dried and then pressed into pellets followed by calcination at 1300 C. for 10 h. The calcined pellets were milled to obtain BZY20 and BZYNiO2 powders, respectively. To obtain dense BZYNiO2 ceramic, calcined and milled BZYNiO2 powders were pressed into pellets and then sintered at 1500 C. for 10 h in air. Some BZY20 powder was also prepared by the combined EDTA-citric acid (CEC) method and calcined at 1100 C. for 10 h (denoted as BZY20-CEC).

(28) TABLE-US-00001 TABLE 1 List of abbreviations for NiBZY (volume ratio 40:60) membranes prepared with different types and amounts of BZY powders. Abbreviation BZY powders NiBZY1 BZYNiO2 NiBZY2 BZY20-SSR NiBZY3 40% BZY20-SSR, 60% BZY20-CEC NiBZY4 20% BZY20-SSR, 80% BZY20-CEC

(29) To study the effect of phase composition on the chemical stability of NiBZY membranes, different BZY powders with different weight contents were used to fabricate 4 different NiBZY membranes (as listed in Table 1). Calculated amounts of BZYNiO2, BZY20, and BZY20-CEC powders were mixed with Ni powders in volume ratio of 60:40 (ceramic vs Ni). These powders were ball-milled, dried, milled, and pressed into pellets with a 20-mm stainless steel die under the pressure of 100 MPa. These pellets were sintered at 1440 C. for 20 h in N.sub.2 and another 20 h in 5% H.sub.2/N.sub.2. During the sintering process in N.sub.2 (contains 0.02 vol % O.sub.2, measured by gas chromatography), partial Ni was oxidized to NiO which reacted with residual BaCO.sub.3 and Y.sub.2O.sub.3, forming BaY.sub.2NiO.sub.5 and promoting the densification and grain growth of BZY phase. During the sintering process in 5% H.sub.2/N.sub.2, residual NiO was reduced to Ni and BaY.sub.2NiO.sub.5 decomposed forming Ni and a secondary phase. The reduction was confirmed by the increase of water content from 0.3% to 2.4% in the gas after switching N.sub.2 to 5% H.sub.2/N.sub.2, monitored by a humidity sensor (Vaisala HMT338). The extended sintering time in 5% H.sub.2/N.sub.2 allowed the elimination of pores generated by the decomposition of BaY.sub.2NiO.sub.5.

(30) The sintered BZYNiO2, NiBZY pellets were polished on a Buehler polisher with SiC sandpapers (320, 600, 1200 grits) and then diamond dispersion solution (9, 3, 1 m) to reveal the surface microstructure. Polished BZYNiO2 pellet was annealed in wet CO.sub.2 (3% H.sub.2O) at 700 C. for 100 h to study its chemical stability in CO.sub.2. A polished NiBZY1 sample was thermally etched at 1300 C. for 30 min in 5% H.sub.2/N.sub.2 for surface microstructure study. To test the chemical stability in water, the polished NiBZY pellets (1 g in weight) were boiled in 30 mL deionized water for 20 h in Teflonlined stainless steel autoclaves. The polished NiBZY samples were also annealed in a gas mixture containing 3 vol % H.sub.2O, 17 vol % H.sub.2 and 80 vol % CO.sub.2 at 900 C. for 50 h to test the chemical stability in wet CO.sub.2.

(31) BaY.sub.2NiO.sub.5 was synthesized by the solid state reaction (SSR) method. Stoichiometric amounts of BaCO.sub.3, Y.sub.2O.sub.3, and NiO powders were ball-milled for 24 h in ethanol with ZrO.sub.2 balls for 24 h. The powders were dried, pressed into pellets, and calcined at 1300 C. for 10 h. The pellets were milled into powders and then annealed in either dry/wet H.sub.2 at 900 C. for 20 h, or wet CO.sub.2 at 900 C. for 10 h. X-ray diffraction (XRD, Rigaku D/Max 2100, with Cu Ka radiation) analysis was used to identify the phases present in the powders and pellets. Rietveld refinements were carried out with GSAS package. Field emission scanning electronic microscopy (FESEM, Zeiss ultra plus) equipped with Energy-dispersive X-ray spectroscopy (EDX, Oxford) was used to study the microstructure and composition of the BZY and NiBZY membranes.

(32) Results and Discussion:

(33) Performance stability in wet H.sub.2

(34) FIG. 2 shows the time dependence of hydrogen flux of various NiBZY membranes at 900 C. in wet 20% H.sub.2 (containing 3 vol % H.sub.2O). The initial performance of the membranes follows such an order: NiBZY1>NiBZY3>NiBZY4 (3.6, 2.8, and 2.0*108 mol/cm.sup.2s, respectively). In order to find the reason for the different initial flux, we investigated the microstructure of the membranes.

(35) All membranes are composed of large Ni grains and small BZY grains. The grain sizes of BZY are estimated to be 1.44, 1.17, and 0.78 m for NiBZY1, NiBZY3, and NiBZY4, respectively. Similar trend is observed for the permeation flux and grain size of BZY in NiBZY membranes. It is well-known that BZY possesses high bulk proton conductivity but suffers from poor grain boundary proton conductivity. Therefore, BZY ceramic with larger grain size and fewer grain boundaries also shows larger total proton conductivity than that with small grain size. Since the NiBZY membranes show the same behavior with BZY ceramics, the higher initial flux of NiBZY membranes with larger BZY grains is attributed to their fewer grain boundaries and higher total proton conductivities. The difference in grain sizes is due to the difference in the amount of BaY.sub.2NiO.sub.5 formed during the sintering process in N.sub.2, and BaY.sub.2NiO.sub.5 promotes the grain growth of BZY phase.

(36) The flux of NiBZY1 first degrades quickly and then slowly, with a total flux loss of 21.5% in 180 h. The flux of NiBZY3 degrades 15.0% in 200 h. However, the flux of NiBZY4 keeps stable during the whole test process (160 h). It seems that there is a trade-off between the initial flux and performance stability of these membranes. Since both Ni and BZY are thermodynamically stable in H.sub.2 and H.sub.2O, it's hard to explain the degradation behavior. Therefore, the phase composition and microstructure was investigated of NiBZY1 membrane after the permeation test. Unlike the fresh membrane which only consists of Ni and BZY, the tested feed and sweep side surfaces contain much less BZY. Both BaCO.sub.3 and YOOH are found on the feed and sweep side surfaces. The SEM images showed that the feed side surface is completely covered by new phases with plate-like structure, while the sweep side surface is only partially covered. EDX results showed that the plate-like phases contain Ba, C, Y, and O, indicating that they are BaCO.sub.3 and YOOH, as revealed by the XRD results. Since the feed gas is wet H.sub.2 without CO.sub.2, BaCO.sub.3 is formed by reaction between Ba(OH).sub.2 and CO.sub.2 in air after the permeation test. These insulating phases (Ba(OH).sub.2 and YOOH) can block the pathways for hydrogen permeation, which explains the degradation behavior, but the sources of these insulating phases are still unclear. To obtain more insight in the sources of Ba(OH).sub.2 and YOOH, we investigated the chemical stability of NiBZY membranes by treating them in boiling water for 20 h or wet 17 vol % H.sub.2 and 80 vol % CO.sub.2 at 900 C. for 50 h.

(37) Results and Discussion:

(38) Chemical Stability in Boiling Water and Wet CO.sub.2

(39) The SEM images obtained from polished surface of fresh NiBZY1, 2, 3, and 4 membranes showed many small (3 m) bumps were found on the surface of NiBZY1. SEM-EDX analysis of polished NiBZY1 membrane shows that these bumps contain Ba, C, and Y, but no Ni, which can be BaCO.sub.3 and Y.sub.2O.sub.3. EDX mapping results show that the porous clusters are rich in Y but are depleted of Ba, indicating that these clusters are Y.sub.2O.sub.3. This can be explained by the decomposition of BaCO.sub.3 at 1300 C. forming volatile BaO, which evaporates to the atmosphere and leaves Y.sub.2O.sub.3 behind. These results suggest that the secondary phases formed in the fabrication process are ready to react with H.sub.2O/CO.sub.2 in air even at room temperature, forming BaCO.sub.3 and Y.sub.2O.sub.3, which grow out of the polished surface. Unlike the small bumps found in NiBZY1, a few large (20 m) bumps with cracks were found on the surface of NiBZY2. There were much fewer bumps on the surface of NiBZY3, and it's difficult to find the bumps on the surface of NiBZY4. Therefore, the number of bumps and the amount of secondary phases follow such an order: NiBZY1>NiBZY2>NiBZY3>NiBZY4.

(40) FIG. 6 shows the XRD pattern obtained from surface of NiBZY1 (a) after treatment in boiling water for 20 h (Similar results are also obtained for NiBZY2, 3, and 4). Only peaks of Ni and BZY are present in both patterns, confirming that the Ni and BZY phases are stable in boiling water. However, the pH value of the deionized water used in the boiling test for NiBZY1 and NiBZY2 increased from 7 to 12. Precipitate was observed after H.sub.2SO.sub.4 was added to the water, indicating formation of BaSO.sub.4 (the sulphates of Ni, Y, Zr were all soluble in water). These results suggested that Ba(OH).sub.2 was formed during the boiling of the membranes. Because Ba(OH).sub.2 was soluble in water and the polished surface was very smooth, its amount on the membrane surface was too low to be detected by XRD. On the other hand, the pH value of water for boiling NiBZY3 and NiBZY4 kept at 8, suggesting very small amount of Ba(OH).sub.2 was formed during the boiling process. After boiling test, significant microstructure damage was observed on the surfaces of NiBZY1 and NiBZY2 membranes. The characteristics of the surface microstructure damage agree very well with the sizes and distribution of secondary phases, such as small dotted bumps in boiled NiBZY1 and small dotted bumps in polished NiBZY1, large cracks in boiled NiBZY2 and large bumps in polished NiBZY2. A close look at the small bumps on boiled surface of NiBZY1 showed that they consist of BZY grains in the bottom and Y.sub.2O.sub.3 on the top. In contrast, NiBZY3 and NiBZY4 showed no cracks or bumps. These results suggest the secondary phases in NiBZY1 reacted with H.sub.2O forming a lot of Ba(OH).sub.2 and Y.sub.2O.sub.3, which can explain the formation of insulating phases on the membrane surfaces and thus the performance degradation in wet H.sub.2.

(41) XRD patterns in FIG. 6 show that a significant amount of BaCO.sub.3 was detected on NiBZY1 and less was found on NiBZY2 after annealing in wet 80% CO.sub.2. Surprisingly, no BaCO.sub.3 was found on NiBZY3 and NiBZY4. This was further verified by SEM images. Most of the surfaces of NiBZY1 and NiBZY2 membranes were covered by the new phases. EDX spectrum showed that the new phases on NiBZY1 is composed of Ba, C, and O, which also suggests that the new phases is BaCO.sub.3. However, there are only a little BaCO.sub.3 on NiBZY3, which is below the detection limit of XRD. For NiBZY4, no BaCO.sub.3 is observed. The amount of BaCO.sub.3 follows such order: NiBZY1>NiBZY2>NiBZY3>NiBZY4. It is noticeable that NiBZY4 shows excellent chemical stability in both boiling water and wet CO.sub.2.

(42) Results and Discussion:

(43) Performance Degradation and Chemical Instability

(44) Results suggested that the performance degradation of NiBZY1 in wet H.sub.2 was caused by the blocking phases (Ba(OH).sub.2 and YOOH) formed on the membrane surfaces. The chemical stability study showed that Ba(OH).sub.2 and Y.sub.2O.sub.3 was formed by the reaction between the secondary phases and water. Therefore, the flux degradation in wet H.sub.2 was caused by Ba(OH).sub.2 and YOOH generated from the reaction between the secondary phase and H.sub.2O. The more secondary phases there are in NiBZY membranes, the more Ba(OH).sub.2+Y.sub.2O.sub.3 is formed when exposed to water, and the more the performance degrades. The question is what the secondary phases are and where they come from. In the previous study, we observed BaY.sub.2NiO.sub.5 after NiBZY membranes were sintered in N.sub.2 only. After the second sintering process in 5% H.sub.2/N.sub.2, BaY.sub.2NiO.sub.5 disappeared but no new phase was found by XRD. It is difficult to identify the new phases directly because they are mixed with large amount of Ni and BZY and are ready to react with H.sub.2O/CO.sub.2 in air. Therefore, we prepared BaY.sub.2NiO.sub.5 separately and investigated its chemical stability under various conditions.

(45) FIG. 7 shows the XRD patterns of BaY.sub.2NiO.sub.5 before and after annealing in different atmospheres at 900 C. After calcination at 1300 C. for 10 h, Immm structured BaY.sub.2NiO.sub.5 with lattice parameters of a=3.757 , b=5.754 , and c=11.315 is obtained. After annealing in CO.sub.2 at 900 C. for 10 h (FIG. 7, b), BaY.sub.2NiO.sub.5 powder decomposes into a mixture of BaCO.sub.3, Y.sub.2O.sub.3, and NiO (BaY.sub.2NiO.sub.5+CO.sub.2=BaCO.sub.3+Y.sub.2O.sub.3+NiO). After annealing in dry H.sub.2 (FIG. 7, c), BaY.sub.2NiO.sub.5 decomposes into BaY.sub.2O.sub.4 and Ni (BaY.sub.2NiO.sub.5+H.sub.2=BaY.sub.2O.sub.4+Ni+H.sub.2O). After annealing in wet H.sub.2 (FIG. 7, d), aside from BaY.sub.2O.sub.4 and Ni, Y.sub.2O.sub.3 and BaCO.sub.3 are found, suggesting that BaY.sub.2O.sub.4 partially reacts with H.sub.2O forming Y.sub.2O.sub.3 and Ba(OH).sub.2 which is converted to BaCO.sub.3 when exposed to air. These results indicate that BaY.sub.2NiO.sub.5 is unstable in H.sub.2, H.sub.2O, and CO.sub.2-containing atmospheres at 900 C., leading to the formation of insulating BaY.sub.2O.sub.4, Ba(OH).sub.2, and BaCO.sub.3, respectively. Coors investigated reduced BaCe.sub.0.2Zr.sub.0.6Y.sub.0.2O.sub.3- (BCZY262, sintered with 1 wt. % NiO by solid state reactive sintering method) by field emission SEM and high resolution transmission electron microscopy (HRTEM). Only Ni metallic nanoprecipitates and some amorphous grain boundaries were observed. Therefore, BaY.sub.2NiO.sub.5 was assumed to decompose into Ni, BaO, and Y.sub.2O.sub.3 in reducing atmosphere. Then BaO and Y.sub.2O.sub.3 were dissolved back to perovskite lattice. In that case, BZYNiO2 should be chemically stable, which cannot explain the catastrophical failure of sintered BCZY262 ceramic during cooling below 400 C. in wet H.sub.2. Instead, the failure can be attributed to the stress originated from volume increase accompanied with the reaction between BaY.sub.2O.sub.4 and H.sub.2O forming Ba(OH).sub.2 and Y.sub.2O.sub.3. BaY.sub.2O.sub.4 can be prepared by calcination of BaCO.sub.3 and Y.sub.2O.sub.3 at 1000 C. for 10 h, suggesting that BaY.sub.2O.sub.4 is more stable than BaO (from the composition of BaCO.sub.3) and Y.sub.2O.sub.3 at elevated temperature in H.sub.2O/CO.sub.2-free condition. Therefore, after sintering in N.sub.2 and then 5% H.sub.2, BaY.sub.2NiO.sub.5 in NiBZY decomposes into BaY.sub.2O.sub.4 and Ni, rather than BaO, Y.sub.2O.sub.3, and Ni. Unfortunately, both BaY.sub.2O.sub.4 and BaY.sub.2NiO.sub.5 easily react with H.sub.2O and CO.sub.2, forming insulating phases.

(46) These results can provide a clear explanation on the performance and chemical instability found in NiBZY1, 2, and 3 membranes. In NiBZY1, BaY.sub.2NiO.sub.5 is formed during the fabrication of BZYNiO2 powder. In NiBZY2, 3, and 4, residual BaCO.sub.3 and Y.sub.2O.sub.3 in BZY20 prepared by the SSR method reacts with NiO (from the oxidation of partial metal Ni in N.sub.2 containing 0.02% O.sub.2) forming BaY.sub.2NiO.sub.5. BaY.sub.2NiO.sub.5 is reduced to BaY.sub.2O.sub.4 and Ni during sintering in 5% H.sub.2/N.sub.2. BaY.sub.2O.sub.4 is distributed as isolated islands and react with H.sub.2O during permeation test in H.sub.2, forming insulating barium and yttrium hydroxides on the membrane surfaces and causing the performance degradation. BaY.sub.2O.sub.4 reacts with H.sub.2O and CO.sub.2 forming Ba(OH).sub.2+Y.sub.2O.sub.3 and BaCO.sub.3+Y.sub.2O.sub.3 during the chemical stability test in boiling water and wet CO.sub.2, respectively. Less BaY.sub.2O.sub.4 will lead to less insulating phases on the membrane surfaces and less performance degradation when exposed to H.sub.2O and CO.sub.2, as observed in the performance and chemical stability study. In NiBZY3 and NiBZY4, the BZY20-CEC grains also act as a covering layer for BaY.sub.2O.sub.4, and keeps BaY.sub.2O.sub.4 away from attack by CO.sub.2 and H.sub.2O, leading to higher chemical stability. Similar strategy has been employed in a BaCe.sub.0.8Sm.sub.0.2O.sub.3-Ce.sub.0.8Sm.sub.0.2O.sub.2- composite electrolyte and successfully avoided the chemical instability issue of BaCeO.sub.3-. Apparently, the amount of BaY.sub.2O.sub.4 is too large to be fully covered by BZY20-CEC in NiBZY3, but is small enough to be fully covered by BZY20-CEC in NiBZY4. We also tried to prepare NiBZY membrane only with BZY20-CEC powder, however, the obtained membrane possessed significant porosity due to lack of BaY.sub.2NiO.sub.5 as sintering aid. The NiBZY4 membrane demonstrated both excellent chemical stability and sinteractivity, and its performance stability was further tested in wet CO.sub.2.

(47) Results and Discussion:

(48) Performance and Microstructural Stability of NiBZY4 Membrane in Wet CO.sub.2

(49) Hydrogen separation membranes are expected to be applied in wet H.sub.2 with considerable amount of CO.sub.2. Therefore, its performance stability in CO.sub.2 is critical to its application. FIG. 3 shows that the flux of NiBZY4 membrane in 40% H.sub.2 increases from 2.5*10.sup.8 to 2.8*10.sup.8 and 2.9*10.sup.8 mol/cm.sup.2s after introduction of 30% and 50% CO.sub.2, respectively. Except for the initial flux increase, the flux keeps stable during the test in wet CO.sub.2 for 225 h. The flux increase is ascribed to the reverse water gas shift (RWGS, H.sub.2+CO.sub.2.fwdarw.H.sub.2O+CO) reaction which leads to the increase of moisture content in feed gas and thus proton conductivity in BZY phase, which is previously observed in NiBZCYYb membrane. The RWGS reaction consumes equal amount of H.sub.2 and CO.sub.2, generating same amount of H.sub.2O and CO. The measured CO content is 1.9% after the introduction of 30% CO.sub.2, indicating that the moisture content increases from 3% (from the water bubbler) to 4.9%, and the hydrogen content decreases from 40% to 38.1%. The former is beneficial for the increase of the proton conductivity of BZY and hydrogen flux, while the latter is deleterious to hydrogen flux by reducing hydrogen partial pressure gradient across the membrane. Nevertheless, the flux increases, suggesting the former is dominant in the case of NiBZY4 membrane. In a very recent report, Zhu et al. developed a dense NiBaZr.sub.0.7Pr.sub.0.1Y.sub.0.2O.sub.3- (NiBZPY) membrane utilizing the benefit of Pr-doping on the sinter-activity of BZY. Because Pr.sup.4+ can easily be reduced to Pr.sup.3+ and even Pr.sup.2+, the doping of Pr in BZY introduces additional oxygen vacancies, which can promote the RWGS reaction, as observed in Ce-doped BZY. The hydrogen content in feed gas decreased dramatically from 40% to 28% after introduction of 30% CO.sub.2 at 850 C. Therefore, the flux decreased from 7.3*10.sup.9 to 6.1*10.sup.9 mol/cm.sup.2s. Besides, the flux of a 0.40-mm-thick NiBZPY membrane in wet 40% H.sub.2 at 900 C. was only 9.2*10.sup.9 mol/cm.sup.2s, much lower than that of NiBZY4 membrane (2.5*10.sup.8 mol/cm.sup.2s in the same condition). Therefore, NiBZY4 is superior to NiBZPY membrane in both performance and performance stability.

(50) After permeation test in wet CO.sub.2, the microstructure of NiBZY4 membrane was investigated by SEM, which showed that the whole membrane is still compact and no obvious porosity is observed. There is no coarsening of Ni particles close to the feed side surface. Only peaks of Ni and BZY can be found in the XRD pattern obtained from the feed side surface. The feed side surface consists of large Ni particles and small BZY grains. In comparison, the feed side surface of NiBZY1 membrane are completely covered by plate-like phases including BaCO.sub.3 and YOOH after test in wet H.sub.2, which is much milder than wet 50% CO.sub.2. Although the surface looks porous, the cross-sectional view shows the porous layer only reaches a depth of several microns. Similar microstructure is observed on the sweep side of the membrane.

(51) After permeation test in wet CO.sub.2, the microstructure of NiBZY4 membrane was investigated by SEM, which shows that the whole membrane is still compact and no obvious porosity is observed. There is no coarsening of Ni particles close to the feed side surface. Only peaks of Ni and BZY can be found in the XRD pattern obtained from the feed side surface (FIG. 4). The feed side surface consists of large Ni particles and small BZY grains. In comparison, the feed side surface of NiBZY1 membrane are completely covered by plate-like phases including BaCO.sub.3 and YOOH after test in wet H.sub.2, which is much milder than wet 50% CO.sub.2. Although the surface looks porous, the cross-sectional view shows the porous layer only reaches a depth of several microns. Similar microstructure is observed on the sweep side of the membrane.

(52) Results and Discussion:

(53) Implication on the Feasibility of Using BZYNiO2 as Electrolyte for SOFCs

(54) An important implication of this study is that BaY.sub.2NiO.sub.5 can react with CO.sub.2 forming BaCO.sub.3 and Y.sub.2O.sub.3, this may be a serious problem for BZYNiO2, which is proposed as a promising electrolyte material with great sinter-activity and high electrical conductivity. Until now, no study has been reported on its chemical stability, which is crucial for its successful application. Rietveld refinements of the XRD profiles in FIG. 8a show that fresh BZYNiO2 consists of BZY and BaY.sub.2NiO.sub.5. After annealing in wet CO.sub.2 at 700 C. for 100 h, BaY.sub.2NiO.sub.5 disappears but BaCO.sub.3 and Y.sub.2O.sub.3 are found (FIG. 8b). SEM image in FIG. 9 showed that the polished surface of BZYNiO2 before exposure to CO.sub.2 is very dense. It is noticeable that some dark spots are randomly distributed in the ceramic matrix. EDX spectra in FIGS. 10a and b, show that the dark phase consists of Ba, Y, Ni, and O, while the light phase consists of Ba, Zr, Y, and O. Considering the XRD results, the dark and light phases are BaY.sub.2NiO.sub.5 and BZY, respectively. After annealing in wet CO.sub.2, the surface of BZYNiO2 is covered by many crystal-like particles with a size ranging from hundreds of nanometers to a few microns. The EDX spectrum in FIG. 10c shows that the particles mainly contain Ba, C, Y, and O, suggesting that they are BaCO.sub.3 and Y.sub.2O.sub.3. It can be seen from the coverage that the amount of BaCO.sub.3 and Y.sub.2O.sub.3 is much more than that of BaY.sub.2NiO.sub.5 on the fresh membrane surface, probably due to diffusion of Ba.sup.2+ from BaY.sub.2NiO.sub.5 inside the BZYNiO2 ceramic. One may ask whether Ba-excessive BZY contribute to the formation of BaCO.sub.3. This is unlikely because there should be a significant change in the lattice parameter if Ba-excessive BZY reacts with CO.sub.2, however, there is almost no change in the lattice parameter of BZY phase (obtained from Rietveld refinement of XRD profiles of BZYNiO2 before and after the annealing in wet CO.sub.2, 4.208 and 4.206 , respectively, FIGS. 8a and b). These results clearly suggest that BaY.sub.2NiO.sub.5 in BZYNiO2 makes it chemically instable in H.sub.2O/CO.sub.2, which needs to be resolved before it is used as electrolyte for solid oxide fuel cells. This problem may be solved by a similar strategy as that in NiBZY4. BZYNiO2 prepared by solid state reaction method can cause the formation of BaY.sub.2NiO.sub.5 which is an efficient sintering aid but also cause chemical instability, and BZY prepared by CEC method is chemically stable but not sinter-active. The combination of these powders can lead to a chemically stable yet still sinter-active proton conductor.

(55) TABLE-US-00002 TABLE 2 Lattice constants of phases in fresh and annealed BZYNiO2 determined by Rietveld refinement of XRD patterns in FIG. 8. Sample Phase Unit cells Lattice constants () Fresh BZYNiO2 BZY Cubic (Pm3m) 4.208 BaY.sub.2NiO.sub.5 Orthorhombic a = 3.757, b = 5.755, (lmmm) c = 11.320 Annealed BZY Cubic (Pm3m) 4.206 BZYNiO2 BaCO.sub.3 Orthorhombic a = 5.306, b = 8.891, (Pmcn) c = 6.471 Y.sub.2O.sub.3 Cubic (la3) 10.595

CONCLUSIONS

(56) Although BZY possesses excellent chemical stability in H.sub.2O and CO.sub.2, BZY ceramic and NiBZY membranes sintered using NiO sintering aid are chemically instable because of the formation of BaY.sub.2NiO.sub.5. BaY.sub.2NiO.sub.5 reacts with H.sub.2, H.sub.2O, and CO.sub.2, forming insulating BaY.sub.2O.sub.4, Ba(OH).sub.2, and BaCO.sub.3 phases, respectively. Previous report suggested that BaY.sub.2NiO.sub.5 was reduced to BaO and Y.sub.2O.sub.3, which were dissolved back to BZY lattice. However, we find that NiBZY membranes contain BaY.sub.2O.sub.4 islands after reduction. Both BaY.sub.2NiO.sub.5 and BaY.sub.2O.sub.4 easily react with H.sub.2O and CO.sub.2 forming Ba(OH).sub.2 and BaCO.sub.3, respectively. Therefore, both BZYNiO2 and NiBZY are chemically instable in wet CO.sub.2, and the hydrogen flux of NiBZY membranes degraded significantly in wet H.sub.2 at 900 C. A comparative study shows that the more BaY.sub.2O.sub.4 there is in NiBZY membranes, the less stable they are. The chemical instability greatly restricts the applications of BZYNiO2 and NiBZYNiO2 as electrolytes for solid oxide fuel cells and hydrogen separation membranes, respectively. Fortunately, the chemical stability of NiBZY membranes can be improved by reducing the amount of BaY.sub.2O.sub.4 formed in the fabrication process. This is achieved by replacing a portion of BZY20 powder in the starting material with BZY20-CEC powder. The BZY20-CEC grains also act as a covering layer for BaY.sub.2O.sub.4 and helps avoid the reaction between BaY.sub.2O.sub.4 and H.sub.2O/CO.sub.2. The new NiBZY4 membrane fabricated with 20% BZY20-SSR and 80% BZY20-CEC powders demonstrated very stable and improved permeation flux in wet 50% CO.sub.2 at 900 C. The phase composition and membrane microstructure were also intact after the test, indicating its excellent chemical stability, surpassing previous NiBZCYYb membrane. It also possesses much higher hydrogen flux than NiBZPY. These facts demonstrate that NiBZY4 is very promising for hydrogen permeation applications.

(57) These and other modifications and variations to the present invention may be practiced by those of ordinary skill in the art, without departing from the spirit and scope of the present invention, which is more particularly set forth in the appended claims. In addition, it should be understood the aspects of the various embodiments may be interchanged both in whole or in part. Furthermore, those of ordinary skill in the art will appreciate that the foregoing description is by way of example only, and is not intended to limit the invention so further described in the appended claims.