Supported Pd-Au Membrane Reactor for Hydrogen Production: Membrane Preparation, Characterization and Testing

A supported Pd-Au (Au 7wt%) membrane was produced by electroless plating deposition. Permeation tests were performed with pure gas (H2, H2, N2, CO2, CH4) for long time operation. After around 400 h under testing, the composite Pd-Au membrane achieved steady state condition, with an H2/N2 ideal selectivity of around 500 at 420 °C and 50 kPa as transmembrane pressure, remaining stable up to 1100 h under operation. Afterwards, the membrane was allocated in a membrane reactor module for methane steam reforming reaction tests. As a preliminary application, at 420 °C, 300 kPa of reaction pressure, space velocity of 4100 h−1, 40% methane conversion and 35% hydrogen recovery were reached using a commercial Ni/Al2O3 catalyst. Unfortunately, a severe coke deposition affected irreversibly the composite membrane, determining the loss of the hydrogen permeation characteristics of the supported Pd-Au membrane.


Introduction
The current development in energy use is oriented towards reducing carbon consumption due to its environmental pollution. Hydrogen as a clean and sustainable energy carrier has gained more attention during the past decades. When hydrogen reacts with oxygen in fuel cells and internal combustion engines, a large amount of energy is released explosively in heat engines and quietly in fuel, releasing water as the product.
The present source of hydrogen comes mainly from synthesis gas, which is a mixture of H 2 , CO and CO 2 , and it is produced by breaking the strong C-H bonds (439 kJ/mol) of hydrocarbons in reforming reactions. Afterwards, hydrogen is purified and separated by different energy intensive steps. A membrane reactor (MR) technology can represent an energetically efficient option to the conventional processes, with the practical advantages of a smaller footprint and capital cost reduction. In this alternative reformer, the hydrogen is produced and continuously removed from the reaction side for permeation through a hydrogen perm-selective membrane, shifting the reaction towards further product formation. As a consequence, the conversion increases and pure hydrogen is produced at the same time.
Furthermore, at stable value of hydrogen permeance, the composite Pd-Au membrane was allocated in a MR module for carrying out methane steam reforming (MSR) reaction for producing hydrogen and to analyze the effect of coke formation during the course of experiments.

Permeation Tests on the Pd-Au/PSS Membrane
The permeation tests on the supported Pd-Au membrane at the onset of the experimental testing were carried out with pure H 2 and N 2 at 350˝C and transmembrane pressures of 50 and 100 kPa, with the results resumed in Table 1. As expected, a higher transmembrane pressure affects positively the hydrogen permeation driving force (in this case, represented by the hydrogen partial pressure difference between retentate and permeate sides), making higher hydrogen permeating flux achievable. Nevertheless, the results do not show high performance of the membrane in terms of H 2 /N 2 ideal selectivity, which can be the consequence of defects, non-homogeneity of Pd-Au layer or Pd-Au alloy not yet formed. Furthermore, it is worth of noting that, at the beginning of the experimental campaign (Figure 1), the composite membrane was probably not at steady state conditions. This should justify why the H 2 /N 2 ideal selectivity trend increases slightly by increasing the transmembrane pressure from ∆p = 50 kPa to ∆p = 100 kPa. Furthermore, at stable value of hydrogen permeance, the composite Pd-Au membrane was allocated in a MR module for carrying out methane steam reforming (MSR) reaction for producing hydrogen and to analyze the effect of coke formation during the course of experiments.

Permeation Tests on the Pd-Au/PSS Membrane
The permeation tests on the supported Pd-Au membrane at the onset of the experimental testing were carried out with pure H2 and N2 at 350 °C and transmembrane pressures of 50 and 100 kPa, with the results resumed in Table 1. As expected, a higher transmembrane pressure affects positively the hydrogen permeation driving force (in this case, represented by the hydrogen partial pressure difference between retentate and permeate sides), making higher hydrogen permeating flux achievable. Nevertheless, the results do not show high performance of the membrane in terms of H2/N2 ideal selectivity, which can be the consequence of defects, non-homogeneity of Pd-Au layer or Pd-Au alloy not yet formed. Furthermore, it is worth of noting that, at the beginning of the experimental campaign (Figure 1), the composite membrane was probably not at steady state conditions. This should justify why the H2/N2 ideal selectivity trend increases slightly by increasing the transmembrane pressure from Δp = 50 kPa to Δp = 100 kPa.  Then, in order to evaluate the permeation characteristic of the membrane at steady state conditions, further permeation tests with pure N 2 , He and H 2 were successively performed and, after each cycle of pure gas permeation test, the composite membrane was flowed under hydrogen at ∆p = 50 kPa also over night. The results of these tests in terms of H 2 , N 2 and He permeating fluxes are also resumed in Figure 1. As shown, in the range 0-100 h, the H 2 permeating flux increased from~2.0ˆ10´3 mol/m 2¨s to 7.5ˆ10´3 mol/m 2¨s , due to the temperature increase from 300˝C to 420˝C. Afterwards, at the set temperature (420˝C) and with the composite membrane exposed to pure hydrogen permeation (a part from the daily realization of pure gas permeation tests), after 400 h, the hydrogen permeating flux reached the steady state condition with an average value of~2.0ˆ10´2 mol/m 2¨s , remaining constant up to 1100 h. This trend confirms the optimum annealing conditions for the formation of Pd-Au alloy over this time [32]. As shown in Figure 2, H 2 /N 2 and H 2 /He ideal selectivities increased in the range 0-100 h due to the higher operating temperature, consequently acting towards higher hydrogen permeating flux. After 300 h at 420˝C, they reached average values of around 500 and 220, respectively, remaining constant up to 1100 h under operation. Then, in order to evaluate the permeation characteristic of the membrane at steady state conditions, further permeation tests with pure N2, He and H2 were successively performed and, after each cycle of pure gas permeation test, the composite membrane was flowed under hydrogen at Δp = 50 kPa also over night. The results of these tests in terms of H2, N2 and He permeating fluxes are also resumed in Figure 1. As shown, in the range 0-100 h, the H2 permeating flux increased from ~2.0 × 10 −3 mol/m 2 •s to 7.5 × 10 −3 mol/m 2 •s, due to the temperature increase from 300 °C to 420 °C. Afterwards, at the set temperature (420 °C) and with the composite membrane exposed to pure hydrogen permeation (a part from the daily realization of pure gas permeation tests), after 400 h, the hydrogen permeating flux reached the steady state condition with an average value of ~2.0 × 10 −2 mol/m 2 •s, remaining constant up to 1100 h. This trend confirms the optimum annealing conditions for the formation of Pd-Au alloy over this time [32]. As shown in Figure 2, H2/N2 and H2/He ideal selectivities increased in the range 0-100 h due to the higher operating temperature, consequently acting towards higher hydrogen permeating flux. After 300 h at 420 °C, they reached average values of around 500 and 220, respectively, remaining constant up to 1100 h under operation. In the range 400-1000 h, when the membrane showed constant properties in terms of hydrogen permeating flux and ideal selectivities, in order to define the correct value of n, a series of hydrogen permeation tests was performed at different transmembrane pressures to calculate the linear regression factor (R 2 ) at different "n". As illustrated in Figure 3, the highest value of R 2 was reached at n = 1, meaning that the permeation characteristics of the composite membrane are far from the Sieverts-Fick law [4], and, meanwhile, confirming a considerable presence of defects as pin-holes in the Pd-Au layer.
Furthermore, with the lifetime of the membrane around 1100 h under continuous operation, CH4 and CO2 permeation tests were also performed besides H2 and the results are summarized in Table  2. As it was mentioned earlier, hydrogen permeates through the Pd-Au layer by a solution-diffusion mechanism. On the contrary, the other gases permeate through the defects of the Pd-Au surface and the pores of the support with a Knudsen diffusion mechanism [36].
The H2/CO2 ideal selectivity of a bit less than 500 represents an interesting result for the separation of hydrogen and carbon dioxide mixtures. However, only a few results are available in In the range 400-1000 h, when the membrane showed constant properties in terms of hydrogen permeating flux and ideal selectivities, in order to define the correct value of n, a series of hydrogen permeation tests was performed at different transmembrane pressures to calculate the linear regression factor (R 2 ) at different "n". As illustrated in Figure 3, the highest value of R 2 was reached at n = 1, meaning that the permeation characteristics of the composite membrane are far from the Sieverts-Fick law [4], and, meanwhile, confirming a considerable presence of defects as pin-holes in the Pd-Au layer.
Furthermore, with the lifetime of the membrane around 1100 h under continuous operation, CH 4 and CO 2 permeation tests were also performed besides H 2 and the results are summarized in Table 2.
As it was mentioned earlier, hydrogen permeates through the Pd-Au layer by a solution-diffusion mechanism. On the contrary, the other gases permeate through the defects of the Pd-Au surface and the pores of the support with a Knudsen diffusion mechanism [36]. reaction testing, particularly evaluating the membrane aging. Then, a qualitative summary of the long-term characteristics of Pd-Au membranes from literature data are displayed in Table 3. Preparation method, thickness of the Pd-Au layer, operating conditions and H2 permeance as well as αH2/He and reference are reported. As shown, the ideal selectivity αH2/He of this work is the lowest compared to the other reported data, but it was calculated after almost 1000 h under operation, while the other data refer to the onset of the experimental tests or for a shorter range time under continuous operation.     The H 2 /CO 2 ideal selectivity of a bit less than 500 represents an interesting result for the separation of hydrogen and carbon dioxide mixtures. However, only a few results are available in the open literature about Pd-Au composite membranes applied in the field of gas separation and reaction testing, particularly evaluating the membrane aging. Then, a qualitative summary of the long-term characteristics of Pd-Au membranes from literature data are displayed in Table 3. Preparation method, thickness of the Pd-Au layer, operating conditions and H 2 permeance as well as α H2/He and reference are reported. As shown, the ideal selectivity α H2/He of this work is the lowest compared to the other reported data, but it was calculated after almost 1000 h under operation, while the other data refer to the onset of the experimental tests or for a shorter range time under continuous operation.

Methane Steam Reforming Reaction in the MR Housing the Pd-Au/PSS
With the aim of examining the performance of the Pd-Au/PSS membrane in an MR, the MSR reaction was carried out (as reported by the square indicator in Figure 1) at 420˝C and 300 kPa of reaction pressure by using also a sweep-gas in the permeate stream in counter-current mode with respect to the feed.
The endothermic reaction of MSR reaction can be expressed according to Equation (1): In the meanwhile, the water-gas shift (WGS) reaction takes place according to Equation (2): Table 4 resumes the operating conditions of MSR reaction, and Figure 4 sketches the reaction results in terms of methane conversion, hydrogen recovery and yield.

Methane Steam Reforming Reaction in the MR Housing the Pd-Au/PSS
With the aim of examining the performance of the Pd-Au/PSS membrane in an MR, the MSR reaction was carried out (as reported by the square indicator in Figure 1) at 420 °C and 300 kPa of reaction pressure by using also a sweep-gas in the permeate stream in counter-current mode with respect to the feed.
The endothermic reaction of MSR reaction can be expressed according to Equation (1): In the meanwhile, the water-gas shift (WGS) reaction takes place according to Equation (2): Table 4 resumes the operating conditions of MSR reaction, and Figure 4 sketches the reaction results in terms of methane conversion, hydrogen recovery and yield.   The first reaction test was carried out at 420˝C, GHSV = 4100 h´1 and after 500 h of the membrane under pure gas permeation tests and with a constant hydrogen permeance. At these conditions, methane conversion is equal to 40% and the hydrogen recovery around 35%, while the yield is around 15%. The low conversion of methane is probably due to the low reaction temperature used during the tests.
Furthermore, owing to a low value of hydrogen perm-selectivity the effect of the conversion shifting is not so much effective compared to dense Pd-based membranes, full hydrogen perm-selective. As a consequence, the hydrogen yield is also low as well as the hydrogen recovered in the permeated stream. Furthermore, as shown in Figure 5, the total hydrogen produced during the reaction was 0.84 mL/min, subdivided in the hydrogen in the permeate and retentate streams. The first reaction test was carried out at 420 °C, GHSV = 4100 h −1 and after 500 h of the membrane under pure gas permeation tests and with a constant hydrogen permeance. At these conditions, methane conversion is equal to 40% and the hydrogen recovery around 35%, while the yield is around 15%. The low conversion of methane is probably due to the low reaction temperature used during the tests.
Furthermore, owing to a low value of hydrogen perm-selectivity the effect of the conversion shifting is not so much effective compared to dense Pd-based membranes, full hydrogen permselective. As a consequence, the hydrogen yield is also low as well as the hydrogen recovered in the permeated stream. Furthermore, as shown in Figure 5, the total hydrogen produced during the reaction was 0.84 mL/min, subdivided in the hydrogen in the permeate and retentate streams. All the results reported in Figures 4 and 5 show a maximum error bar lower than 2%. Hence, in order to improve the methane conversion, the space velocity was decreased to favor higher residence time of the reactants in the reaction zone with a consequent longer contact time between the gas mixture and the catalyst. Nevertheless, at GHSV = 1100 h −1 and 420 °C, the hydrogen recovery increased up to 65%, but the conversion dropped to around 10%. This reverse phenomena can be a sequence of catalyst deactivation due to coke formation.
As a consequence, the experimental tests at 1100 h −1 were repeated, and, once again, coke formation was verified. Then, after the reaction test, pure hydrogen was flowed into the reaction side (~3.0 × 10 −3 mol/min) at 420 °C and for around 2 h. Figure 6 shows the feed and retentate molar flow rates of hydrogen. The results show that, particularly, in the first hour of operation, a substantial part of hydrogen fed to the MR was consumed in the catalytic bed, forming methane and indirectly confirming the coke deposition during the experimental tests. Indeed, during the hydrogen feeding procedure, methane was detected by analyzing the retentate stream to the GC. Hence, after almost 1 h under operation, the flow rate of methane formed during this procedure decreased gradually till 2.25 h, in which the inlet hydrogen stream was equal to the outlet stream. Successively, the reaction test was repeated, but the conversion remained still low and, then, the reaction tests were stopped. To verify the hydrogen permeation characteristic of the Pd-Au/PSS membrane after the reaction tests, permeation tests with pure H2, N2 and He were further performed. Unfortunately, both αH2/N2 and αH2/He ideal selectivities dropped dramatically and, consequently, the MR was cooled down at ambient temperature, removing the membrane from the MR module. Figure 7 shows the status of the membrane before and after the All the results reported in Figures 4 and 5 show a maximum error bar lower than 2%. Hence, in order to improve the methane conversion, the space velocity was decreased to favor higher residence time of the reactants in the reaction zone with a consequent longer contact time between the gas mixture and the catalyst. Nevertheless, at GHSV = 1100 h´1 and 420˝C, the hydrogen recovery increased up to 65%, but the conversion dropped to around 10%. This reverse phenomena can be a sequence of catalyst deactivation due to coke formation.
As a consequence, the experimental tests at 1100 h´1 were repeated, and, once again, coke formation was verified. Then, after the reaction test, pure hydrogen was flowed into the reaction side (~3.0ˆ10´3 mol/min) at 420˝C and for around 2 h. Figure 6 shows the feed and retentate molar flow rates of hydrogen. The results show that, particularly, in the first hour of operation, a substantial part of hydrogen fed to the MR was consumed in the catalytic bed, forming methane and indirectly confirming the coke deposition during the experimental tests. Indeed, during the hydrogen feeding procedure, methane was detected by analyzing the retentate stream to the GC. Hence, after almost 1 h under operation, the flow rate of methane formed during this procedure decreased gradually till 2.25 h, in which the inlet hydrogen stream was equal to the outlet stream. Successively, the reaction test was repeated, but the conversion remained still low and, then, the reaction tests were stopped. To verify the hydrogen permeation characteristic of the Pd-Au/PSS membrane after the reaction tests, permeation tests with pure H 2 , N 2 and He were further performed. Unfortunately, both α H2/N2 and α H2/He ideal selectivities dropped dramatically and, consequently, the MR was cooled down at ambient temperature, removing the membrane from the MR module. Figure 7 shows the status of the membrane before and after the experimental campaign, highlighting the color difference of the membrane surface, moved from the initial gold to gray.
Gade et al. [37] supposed that the presence of H 2 S was responsible of the grain boundary attack with consequent loss of gold. However, the aforementioned authors stated that, in their opinion, the H 2 S attack was not the sole mechanism as a cause of gold depletion as in the case of this work. Therefore, in a next study, we will investigate in details what could be the mechanism for the change in color of the membrane surface from gold to gray during such a reaction as MSR in absence of H 2 S in the feed. experimental campaign, highlighting the color difference of the membrane surface, moved from the initial gold to gray. Gade et al. [37] supposed that the presence of H2S was responsible of the grain boundary attack with consequent loss of gold. However, the aforementioned authors stated that, in their opinion, the H2S attack was not the sole mechanism as a cause of gold depletion as in the case of this work. Therefore, in a next study, we will investigate in details what could be the mechanism for the change in color of the membrane surface from gold to gray during such a reaction as MSR in absence of H2S in the feed.
The support was supplied by Pall AccuSep (New York, NY, USA) having 1.0 cm O.D. (AISI 316L porous tube) with an active length around 4 cm. The porous support was welded to two stainless steel tubes, and one of them is closed for facilitating the membrane housing in the MR module. Then, the total length of the membrane tube was 20 cm. The support was oxidized for 12 h at 500 °C. It was Gade et al. [37] supposed that the presence of H2S was responsible of the grain boundary attack with consequent loss of gold. However, the aforementioned authors stated that, in their opinion, the H2S attack was not the sole mechanism as a cause of gold depletion as in the case of this work. Therefore, in a next study, we will investigate in details what could be the mechanism for the change in color of the membrane surface from gold to gray during such a reaction as MSR in absence of H2S in the feed.
The support was supplied by Pall AccuSep (New York, NY, USA) having 1.0 cm O.D. (AISI 316L porous tube) with an active length around 4 cm. The porous support was welded to two stainless steel tubes, and one of them is closed for facilitating the membrane housing in the MR module. Then, the total length of the membrane tube was 20 cm. The support was oxidized for 12 h at 500 °C. It was
The support was supplied by Pall AccuSep (New York, NY, USA) having 1.0 cm O.D. (AISI 316L porous tube) with an active length around 4 cm. The porous support was welded to two stainless steel tubes, and one of them is closed for facilitating the membrane housing in the MR module. Then, the total length of the membrane tube was 20 cm. The support was oxidized for 12 h at 500˝C. It was then graded with preactivated Al 2 O 3 particles and cemented with Pd in order to decrease the surface pore size and narrow the surface pore size distribution and create the intermediate layer for avoiding intermetallic diffusion.
The surface activation was then carried out using the standard SnCl 2 -PdCl 2 activation procedure and a thin layer of pure Pd was deposited by electroless plating technique. Successively, the Au deposition was performed by the method described in detail in Chen and Ma [29]. After each step of membrane preparation, the He permeance at room temperature was also measured (Figure 8a,b). pore size and narrow the surface pore size distribution and create the intermediate layer for avoiding intermetallic diffusion.
The surface activation was then carried out using the standard SnCl2-PdCl2 activation procedure and a thin layer of pure Pd was deposited by electroless plating technique. Successively, the Au deposition was performed by the method described in detail in Chen and Ma [29]. After each step of membrane preparation, the He permeance at room temperature was also measured (Figure 8a,b). Hence, the Pd-Au/PSS membrane was allocated in the module and fixed by means of graphite gaskets and 3.0 g of Ni/Al2O3 commercial catalyst was packed in the annulus of the MR. To prevent the movement of catalyst particles, glass spheres were placed at each side of the catalytic bed.
The MR was flowed with pure N2 at ambient temperature to check the presence of leakages. A thermocouple was used to measure the temperature of the MR module, which was heated up by using two electrical tapes up to achieving the required temperatures (300-420 °C). The retentate pressure was adjusted by utilizing a back-pressure regulator placed at the outlet side of this stream, while the permeate pressure was always kept at 100 kPa.
A scheme of the experimental setup used in this work is illustrated in Figure 9. In the case of permeation tests, pure gases (H2, N2, He, CH4 or CO2) were flowed into the MR module and the permeating flow rate was measured by means of bubble flow meters. Hence, the Pd-Au/PSS membrane was allocated in the module and fixed by means of graphite gaskets and 3.0 g of Ni/Al 2 O 3 commercial catalyst was packed in the annulus of the MR. To prevent the movement of catalyst particles, glass spheres were placed at each side of the catalytic bed.
The MR was flowed with pure N 2 at ambient temperature to check the presence of leakages. A thermocouple was used to measure the temperature of the MR module, which was heated up by using two electrical tapes up to achieving the required temperatures (300-420˝C). The retentate pressure was adjusted by utilizing a back-pressure regulator placed at the outlet side of this stream, while the permeate pressure was always kept at 100 kPa.
A scheme of the experimental setup used in this work is illustrated in Figure 9. In the case of permeation tests, pure gases (H 2 , N 2 , He, CH 4 or CO 2 ) were flowed into the MR module and the permeating flow rate was measured by means of bubble flow meters. During the reaction tests of MSR, methane was mixed with deionized water (steam/methane = 3.5/1), which was pumped by a P680 HPLC pump (Dreieich, Germany), in a preheated chamber and, then, injected to the reaction zone. N2 was also used as an internal standard gas and as a sweep gas with a flow rate of ~25 mL/min, flowed in counter-current mode with respect to the feed. To remove the unreacted water from the retentate and, eventually, from the permeate streams, they were passed through a condenser. Then, the retentate and permeate dry streams were analyzed by means of a temperature programmed HP 6890 gas chromatograph (GC) (Foster City, CA, USA). Each experimental point reported in this work was constituted of, at least, ten reaction cycles at the same operating conditions in order to ensure the reproducibility of the results. During the reaction tests of MSR, methane was mixed with deionized water (steam/ methane = 3.5/1), which was pumped by a P680 HPLC pump (Dreieich, Germany), in a preheated chamber and, then, injected to the reaction zone. N 2 was also used as an internal standard gas and as a sweep gas with a flow rate of~25 mL/min, flowed in counter-current mode with respect to the feed. To remove the unreacted water from the retentate and, eventually, from the permeate streams, they were passed through a condenser. Then, the retentate and permeate dry streams were analyzed by means of a temperature programmed HP 6890 gas chromatograph (GC) (Foster City, CA, USA). Each experimental point reported in this work was constituted of, at least, ten reaction cycles at the same operating conditions in order to ensure the reproducibility of the results.

Permeation Tests
Pure gases such as H 2 , N 2 , He, CH 4 or CO 2 were flowed into the membrane reactor module before reaction tests for studying the permeation characteristics of the supported Pd-Au membrane.
The pure gases were provided by cylinders.
To analyze the hydrogen permeation characteristics of the Pd-Au/PSS membrane, Equation (3) was used as: J H2 " P H2 pp n H2,retentate´p n H2,permeate q where J H2 is the hydrogen flux permeating through the membrane, P H2 the hydrogen permeance, p H2,reaction and p H2,permeate the hydrogen partial pressures in the reaction and permeate sides, respectively. n is variable in the range 0.5-1 depending on the rate limiting step of hydrogen diffusion. More details about the deviation of Sieverts-Fick law (n = 0.5) were explained elsewhere [44]. The hydrogen over other pure gas ideal selectivity (α H2/i ) was expressed as the ratio of the H 2 permeating flux over the permeating flux of another pure gas of interest at the same transmembrane pressure, as reported in Equation (4): α H2/i " J H2 {J i , i " pHe, N 2 , CO 2 , CH 4 q (4) Methane conversion was described by using Equation (5) during MSR reaction: Conversion p%q " rpCO out`C O 2out q{CH 4 in sˆ100 (5) where CO out and CO 2out represent the outlet molar flow rates of CO and CO 2 , while CH 4,in represents the molar flow rate of methane in the reaction side. Hydrogen recovery was defined by Equation (6): Hydrogen Recovery p%q " rH 2-perm {p H 2-perm`H2-ret qsˆ100 where H 2-perm and H 2-ret are the molar flow rates of hydrogen in the permeate and retentate streams, respectively.

Conclusions
We investigated the performance of a supported Pd-Au membrane in terms of hydrogen permeance and ideal selectivity in pure gas permeation tests under long time continuous operation. The hydrogen permeance increased till achieving steady state conditions after around 400 h. Hence, at 50 kPa of transmembrane pressure and 420˝C, the composite membrane reached H 2 /N 2 ideal selectivity of around 500 with an hydrogen permeance of 2.3ˆ10´2 mol/m 2¨s , remaining stable up to 1100 h under operation.
During MSR reaction tests, 40% of methane conversion and 35% of hydrogen recovery were reached in the MR. However, coke was formed during the reaction tests, and it was probably responsible of the irreversible loss of hydrogen permeation characteristics of the membrane. The MR was cooled at ambient temperature, and we observed that the color of the membrane surface moved from gold to gray. In the future, we will better investigate this effect, making more stable the alloy during the reaction tests, meanwhile, improving the hydrogen permeation selectivity.