Au composite membrane

Journal of Membrane Science 362 (2010) 535–544

Contents lists available at ScienceDirect

Journal of Membrane Science journal homepage: www.elsevier.com/locate/memsci

The effect of H2 S on the performance of Pd and Pd/Au composite membrane Chao-Huang Chen, Yi Hua Ma ∗ Center for Inorganic Membrane Studies, Department of Chemical Engineering, Worcester Polytechnic Institute, 100 Institute Rd., Worcester, MA 01609, United States

a r t i c l e

i n f o

Article history: Received 15 March 2010 Received in revised form 30 June 2010 Accepted 3 July 2010 Available online 31 July 2010 Keywords: Palladium gold alloy Hydrogen sulfide Hydrogen separation Composite membrane Electroless plating

a b s t r a c t The performance of the Pd and the Pd/Au alloy (8 wt% Au) composite membrane prepared by the electroless and galvanic displacement deposition in a 54.8 ppm H2 S/H2 mixture was investigated. Upon testing, the Pd membrane showed the permeance decline in two stages resulted from the surface site blocking and bulk sulfidation. The Pd membrane also failed to separate H2 from other gases after the H2 S exposure due to the deterioration of the structure caused by the possible formation of bulk Pd4 S. On the other hand, the Pd/Au alloy membrane exhibited resistance to bulk sulfidation and the permeance decline during the H2 S exposure was mainly caused by the surface site blocking by the dissociative adsorption of H2 S, which was essentially reversible as evidenced by the near full restoration of the H2 permeance in pure H2 at 500 ◦ C. No significant change in He leak after the H2 S tests substantiated that there was no significant structural changes of the Pd/Au alloy membrane due to the bulk sulfide formation. Higher temperatures resulted in less permeance loss during poisoning and more permeance restoration during recovery, which was due to the exothermic nature of the dissociative adsorption of H2 S on metals. © 2010 Elsevier B.V. All rights reserved.

1. Introduction Extracting H2 from relatively abundant fossil fuels is one of the most practical and economical processes for large scale production of H2 today. It will continue to be so until the maturity of the clean and sustainable H2 production techniques such as the electrolysis of water has been achieved. Coal gasification, in particular, is recently considered as one of the most preferable processes since coal is the most abundant and widely dispersed fossil fuel on the globe. In addition, the H2 concentration in the syngas produced from the coal gasification process can be further increased by a water–gas shift (WGS) reaction along with CO2 as the main by-product. Pd and Pd alloy membranes have been extensively studied in the above-mentioned processes due to their high H2 -selective permeability [1–4]. Higher efficiency and cost effectiveness of the H2 production can be achieved by applying Pd-based membranes in such processes to extract high purity H2 during the reaction. As H2 is being extracted by Pd-based membranes, concentrated CO2 under high pressure in the retentate side reduces the cost of the subsequent carbon sequestration process [5]. Unfortunately, a small quantity of H2 S in the syngas, as a byproduct of the coal gasification, deteriorates the performance of Pd-based membranes. McKinley [6] reported a 70% permeance loss of an 1 mm thick pure Pd foil at 350 ◦ C in the presence of 4.5 ppm

∗ Corresponding author. Tel.: +1 5088315398; fax: +1 5088315853. E-mail address: [email protected] (Y.H. Ma). 0376-7388/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.memsci.2010.07.002

H2 S/H2 over a period of 4 days. The Pd foil lost its luster and appeared dull after the H2 S exposure. When exposed to higher H2 S concentrations, not only the performance but also the structure of Pd membranes has been degenerated. Mundschau et al. [7] observed not only the decline of H2 permeance of the Pd foil upon feeding a 20 ppm H2 S balanced with 60% H2 –He mixture at 320 ◦ C for over 100 h, but also the formation of numerous pinholes on the membrane after the exposure. Kulprathipanja et al. [8] reported the failure of a 4 ␮m Pd/ceramic composite membrane with the loss of selectivity upon exposing to a 115 ppm H2 S in a 50% H2 –N2 mixture within 2 h at 450 ◦ C with the formation of numerous pinholes. Edlund and Pledger [9] observed a rapid failure (within seconds) of the Pd/SiO2 /V composite membrane by a catastrophic rupture with the formation of pinholes and large cracks upon exposure to pure H2 S at 115 psia and 700 ◦ C. Kajiwara et al. [10] also observed the rupture and cracks formation of the Pd/porous alumina composite membranes after exposure to a 6200 ppm H2 S/H2 mixture at 400 ◦ C within 1 h, resulting in the membrane losing its ability to separate H2 from other gases. The loss of permeance mainly resulted from the decrease of the H2 adsorption caused by the adsorbed sulfur on the Pd surface. The adsorbed sulfur blocked the adsorption sites on the Pd surface and generated an energy barrier where each adsorbed sulfur effectively blocked 4–13 sites for the H2 adsorption [11–13]. In addition, the adsorbed sulfur could incorporate with Pd to form bulk Pd4 S depending on the H2 S exposure conditions. When bulk Pd4 S formed, the Pd4 S layer acted as a mass transfer barrier with a lower H2 permeability [8,14]. Furthermore, the formation of bulk Pd4 S generated large stress in the Pd lattices due to the large dif-

536

C.-H. Chen, Y.H. Ma / Journal of Membrane Science 362 (2010) 535–544

ference in the lattice constants between pure Pd and Pd4 S, thereby causing the rupture of the Pd membranes [10]. Several approaches, such as varying the crystal structure (nanostructure or amorphous) [15], have been investigated recently to enhance the resistance to sulfur poisoning of Pd membranes [16]. However, the most promising approach to enhance the sulfur resistance was to alloy Pd with other elements, thereby altering its electronic, chemical properties and reactivity [17]. Although most efforts have been placed in Pd/Cu alloy membranes for higher sulfur resistance [8,18–20], Pd/Au alloy membranes have received increasing attention recently due to their twofold beneficial characteristics. Pd/Au alloy membranes showed higher H2 permeability than pure Pd membranes in the Au content range below 20 wt%, which resulted from the higher solubility of H2 in Pd/Au alloys [21–24]. Pd/Au alloys have also shown higher sulfur resistance compared to many other Pd alloys. McKinley [6] observed a flux decline of 10% on a Pd60 Au40 alloy foil in the presence of 4.5 ppm H2 S at 350 ◦ C, while a decline of 95% was observed on a Pd60 Cu40 alloy foil under the same conditions. Way et al. [25] also observed a 38% permeance decline on a Pd85 Au15 alloy membrane in the presence of 5 ppm H2 S at 400 ◦ C, while a 71% permeance drop was observed on a Pd94 Cu6 FCC phase membrane. In the literature, limited research has been done regarding the H2 S poisoning of Pd/Au alloy membranes, especially in the Au content range below 20 wt%. No systematic data were available regarding the effect of temperature and exposure time on the H2 permeation characteristics of Pd/Au alloy membranes in the presence of H2 S. In addition, the characterization of the important phenomenon of permeance recovery of Pd/Au alloy membranes after H2 S poisoning was rarely reported. One simple and efficient way of Au deposition is the galvanic displacement method, which does not require external sources of current, reducing agent and complexing agent [26–29]. The plating technique only requires simple plating equipment and is environmental friendly. Forming Pd/Au alloy on the top layer of the membrane by the displacement plating technique (and annealing) allows the membrane to have sulfur tolerance, while overall low Au content minimizes the use of the high cost Au. The objective of the present study was to investigate the performance of the pure Pd membrane and the Pd membrane with a thin top Pd/Au alloy layer prepared by the electroless and galvanic displacement plating techniques in a 54.8 ppm H2 S/H2 mixture. The recovery phenomena of both the Pd and Pd/Au alloy membranes in pure H2 after the H2 S exposure were studied. In addition, the effects of the exposure temperature and time on the H2 S poisoning and on the permeance recovery of the Pd/Au alloy membrane were also investigated.

Table 1 Composition and condition of plating bath. Composition/condition

Pd bath

Pd(NH3 )4 Cl2 ·H2 O (g/l) AgNO3 (g/l) Na2 EDTA·2H2 O (g/l) NH4 OH (28%) (ml/l) H2 NNH2 (1 M) (ml/l) Na·Au(Cl)4 ·2H2 O (mM) pH Temperature (◦ C)

4 40.1 198 5.6 – 10–11 60

Ag bath

Au bath –

0.519 40.1 198 5.6 10–11 60

– – – 2–5 2–4 R.T.–60

coupons were prepared by the galvanic displacement plating of Au on the Pd-plated PSS coupons (7–10 ␮m thick Pd). The Pd/Au bi-layer coupons were estimated gravimetrically to have about 8–11 wt% Au (∼0.5 ␮m thick Au). For the pure Pd membrane preparation, the porous Inconel supports were oxidized (700 ◦ C/12 h/air) after cleaning in the alkaline solution. The oxidized supports were then graded with the Al2 O3 slurry to reduce the pore size and to narrow the pore size distribution, followed by the successive Pd and Ag deposition. Both the oxide and the Pd/Ag layers acted as the intermetallic diffusion barriers [31,32]. After the Pd/Ag layers were deposited on the supports, Pd was deposited by the electroless plating until gas-tight (dense). During the Pd plating, vacuum was applied inside of the supports (tube side) after the supports were impermeable to liquid to increase the homogeneity of the deposit layer and to reduce the thickness required for obtaining the gas-tight deposit layer. For the Pd/Au alloy membrane preparation, the galvanic displacement plating of Au was conducted on the gas-tight Pd membranes followed by the heat treatment at 500 ◦ C in H2 atmosphere for more than 48 h. The bath composition and the plating condition of Pd, Ag, and Au deposition were summarized in Table 1. 2.2. Sample characterization The surface characterizations were performed by using an Amray 1610 Turbo scanning electron microscope (SEM) equipped with a Princeton Gamma-Tech Avalon energy dispersive X-ray (EDS) light-element detector and a RBA-1610 5MC type Robinson retractable backscattered electron detector for the qualitative and quantitative analysis. The spacial resolution for the SEM system was in the range of 0.8–1.2 ␮m, and the penetration depth of the X-ray from the EDS system was ∼1 ␮m for the sample investigated (consisted of primarily Pd). The crystal phase identification analysis was conducted using a Rigaku Geigerflex X-ray diffractometer (XRD) equipped with a CuK␣ radiation source ( = 1.54 Å), and a curved crystal monochromator.

2. Experimental

2.3. Membrane permeation-test

2.1. Sample preparation

Fig. 1 shows the permeation-test apparatus used in the study for testing the performance (i.e. H2 permeance and selectivity) of the membranes in pure H2 or H2 S/H2 mixtures. The permeationtest module (in the dotted line rectangle), which was a stainless steel assembly with Swagelok fittings, and the permeation-test procedure were designed and performed based on the work of Mardilovich et al. [30]. The feed gas of pure H2 , He or 54.8 ppm (±1 ppm) H2 S/H2 mixtures flowed through the outside of the membrane (shell side) and a digital mass flow meter was used to measure the permeate flow rate at the inside of the membrane (tube side). The feed gas pressure was monitored by a pressure transducer, and controlled by either a pressure transducer (for H2 and He feed) or a digital mass flow controller (for H2 S/H2 mixture feed). For the characterization in pure H2 and He, a ballast volume of 3.8 l was installed before the module and a He sweep (purge) gas at 1 atm was used

Two types of samples, Pd and Pd/Au deposited coupons and membranes, were prepared. Porous 316 stainless steel (PSS) plates (dimensions: 1 cm × 1 cm × 0.1 cm) were used for the coupon preparation, while porous Inconel tubes (O.D. = 1.27 cm and length = 6 cm) were used for the membrane preparation. Both were 0.1 ␮m-media grade and purchased from Mott Metallurgical Corp. The electroless Pd plating procedure and chemical recipe (including activation process) reported by Mardilovich et al. [30] were adopted in this study for the Pd deposition. For the Au deposition, the galvanic displacement plating technique was used with 2–5 mM sodium tetrachloroaurate dihydrate solutions. Pure Pd coupons were prepared by the electroless plating on the pre-oxidized (700 ◦ C/12 h/air) PSS plates, and the Pd/Au bi-layer

C.-H. Chen, Y.H. Ma / Journal of Membrane Science 362 (2010) 535–544

537

Fig. 1. Schematic of the permeation-testing apparatus.

in the tube side to ease the transition between H2 and He. A He sweep gas was also used to prevent the membrane from oxidizing during the start-up/shut down operation while heating/cooling the testing module between room temperature and elevated temperatures, but not used during the permeation-tests. The module was placed in a Watlow ceramic tube furnace, and the temperature of the module was monitored by Omega type-k thermocouples and controlled by a PID controller. The permeate H2 flux of the membrane at the desired temperature and pressure with or without H2 S was then measured. The He leak of the membrane was also measured, and the ideal selectivity was calculated according to the ratio of H2 to He flux through the membrane at a pressure difference of 1 atm (Pfeed = 2 atm, Ppermeate = 1 atm). The sulfur poisoning-tests were performed by passing a 54.8 ppm H2 S/H2 mixture to the membrane after the membrane reached a steady state permeance in pure H2 at a fixed temperature. After the H2 S exposure, pure H2 was reintroduced to the membrane for the permeance recovery. The He leak and ideal selectivity were also measured and calculated after each poisoning/recovery test.

into the Pd layer forming Pd/Au alloy. However, the existence of the apparent Pd peaks suggested that the formed Pd/Au alloy layer was not as thick as the X-ray penetration depth of ∼5 ␮m. The XRD pattern of the coupon annealed at 450 ◦ C for 96 h was very similar to the one annealed at 400 ◦ C except that the intensity of the alloy peaks annealed at 450 ◦ C were larger and the positions of the alloy peaks shifted towards more closely to the Pd peaks, indicating more complete Pd/Au inter-diffusion and the formation of a thicker Pd/Au alloy layer. The existence of the apparent Pd peaks, however, still suggested the formation of a relatively thin Pd/Au alloy layer. The XRD pattern of the coupon annealed at 500 ◦ C for 48 h (Fig. 2) showed the single Pd/Au alloy peaks without the presence of pure Pd and Au peaks, indicating that the formed Pd/Au alloy layer was at least as thick as the XRD penetration depth. The positions of the Pd/Au alloy peaks were very close to that of pure Pd peaks, which was due to the small Au content in the Pd/Au bi-layer samples. However, the un-symmetrical peak shapes in all planes (especially in the higher planes, i.e. plane (2 0 0), (2 2 0)), still suggested the existence of an Au gradient on the surface. By comparing the XRD patterns annealed at 500 ◦ C for 48 h with those for 96 h as shown

3. Results and discussion 3.1. Pd/Au alloy formation In order to determine the annealing temperature and time required to form FCC Pd/Au alloy on the top surface of the Pd/Au bi-layers, several Pd/Au bi-layer coupons with similar composition (∼8.5 wt% Au with the Pd layer of ∼10 ␮m and the Au layer of ∼0.5 ␮m) were annealed in H2 at different temperatures and for different length of time. Fig. 2 shows the XRD patterns of the Pd/Au bi-layer coupons before and after annealing under different conditions. The XRD pattern of the Pd/Au bi-layer coupon before annealing (labeled as as-deposited) showed distinct peaks for pure Pd and Au phase (at 2 of 38.45◦ , 44.75◦ , 67.1◦ for Au and at 2 of 40.35◦ , 47.05◦ , 68.55◦ for Pd from plane (1 1 1) to (2 2 0)), verifying the existence of the Pd and Au bi-layer. Following the annealing, the XRD pattern of the coupon annealed at 400 ◦ C for 96 h showed the disappearance of pure Au peaks along with the appearance of peaks in between the pure Pd and the Au peak position which corresponded to a FCC Pd/Au alloy, indicating the entire Au layer diffused

Fig. 2. XRD patterns of the Pd/Au bi-layers before/after annealing in H2 under different conditions (the XRD pattern before annealing labeled as as-deposited).

538

C.-H. Chen, Y.H. Ma / Journal of Membrane Science 362 (2010) 535–544

Fig. 3. (a) Secondary-electron imaging (SEI) cross-sectional micrograph; (b) corresponding energy dispersive X-ray spectroscopy (EDS) line scan; (c) SEI cross-sectional micrograph; and (d) corresponding energy dispersive X-ray spectroscopy (EDS) line scan of the Pd/Au bi-layers annealed at 450 and 500 ◦ C for 96 h in H2 , respectively.

in Fig. 2, increasing the annealing time to 96 h at the same temperature of 500 ◦ C did not significantly increase the degree of Pd/Au inter-diffusion and Pd/Au alloy formation. Annealing at the higher temperature of 550 ◦ C for less time (24 h) resulted in a similar Pd/Au alloy formation as the ones annealed at 500 ◦ C for time longer than 48 h as evidenced by the XRD pattern shown in Fig. 2. This indicated that a similar degree of Pd/Au inter-diffusion and alloy formation could be achieved at a higher temperature for a shorter length of annealing time. The result suggested that the annealing temperature had stronger influence on the Pd/Au inter-diffusion and the formation of Pd/Au alloy than the annealing time. The cross-sectional analyses with the SEM and EDS were performed on the annealed Pd/Au bi-layer samples to examine the surface Au content and to verify the existence of the Au gradient. Fig. 3(a) is the cross-sectional micrograph of the Pd/Au bi-layer supported on the oxidized PSS support after annealing at 450 ◦ C for 96 h and Fig. 3(b) is the corresponding EDS line scan. In Fig. 3(a), the 0.5–0.7 ␮m dark layer between the Pd/Au layer and PSS support was the metal oxide intermetallic diffusion barrier layer, and the arrow indicates the direction of the EDS line scan. The thickness of this Pd/Au bi-layer sample was roughly 9 ␮m. The dotted lines in Fig. 3(b) represented the interfaces of the Pd/Au layer between the support and mounting resin. An apparent Au gradient was seen from the surface of the Pd/Au bi-layer with the highest Au concentration of ∼30 wt% at the surface of the Pd/Au bi-layer (the interface between the Pd/Au layer and the mounting resin). The

cross-sectional micrograph and the corresponding EDS line scan of the Pd/Au bi-layer after annealing at 500 ◦ C for 96 h are shown in Fig. 3(c) and (d), respectively. In this sample, the thickness of the Pd/Au bi-layer was roughly 11 ␮m, and an Au gradient at the surface of the Pd/Au layer could also be seen. However, the Au concentration at the surface of the Pd/Au bi-layer sample was lower, roughly 18 wt%. Since the Pd/Au bi-layer samples had very similar compositions before the annealing, the lower Au content at the surface observed after the annealing at 500 ◦ C indicated that the higher annealing temperature resulted in more complete Pd/Au inter-diffusion which was in agreement with the XRD results. The results also indicated that a homogenous alloy had not formed throughout the entire Pd/Au layer even after annealing at 500 ◦ C for as long as 96 h. Yet, annealing the Pd/Au bi-layers at 500 ◦ C for a reasonable amount of time was sufficient to form a Pd/Au alloy top layer with a sufficient thickness to obtain potentially higher H2 permeance (Au content within 20 wt%) [21] and sulfur resistance. 3.2. H2 permeance Table 2 summarized the synthesis conditions of the membranes tested in this work. Membranes M1 and M2 (the repaired membrane of M1) were pure Pd membranes, while membrane M3 was the Pd/Au alloy membrane. Membranes M1 and M2 were characterized in pure H2 and He for more than 300 h and 115 h, respectively, in the temperature range of 250–500 ◦ C, which was within the

Table 2 Membranes investigated. Membrane M1 M2a M3 a

Intermetallic diffusion barrier ◦

700 C/12h/Air 700 ◦ C/12h/Air 700 ◦ C/12h/Air

Grading

Thickness of Pd/Ag barrier (␮m)

Thickness of palladium layer barrier (␮m)

Au content (wt%)

Al2 O3 Al2 O3 Al2 O3

3.6 3.6 2.1

7 10.3 16.0

0 0 8

After the characterization, membrane M1 was repaired with additional plating of 3.3 ␮m Pd. The repaired Pd membrane was referred to as membrane M2.

C.-H. Chen, Y.H. Ma / Journal of Membrane Science 362 (2010) 535–544

Fig. 4. Sieverts’ law regression of the Pd membrane, M1 and the Pd/Au alloy membrane, M3 at 500 ◦ C.

suitable operation temperature range of the water–gas shift reaction. Membrane M3 was firstly prepared on a green (new) support and tested in pure H2 and He at temperatures up to 550 ◦ C for ∼1100 h. However, the accidental loss of H2 supply at the elapsed testing time of 950 h at 550 ◦ C for ∼18 h caused the oxidation of the membrane with the appearance of the membrane turning from shinning silver to dull darkish. It has been shown in the literature that the bulk oxidation of Pd occurred above 200 ◦ C in the presence of oxygen [33,34]. Therefore, membrane M3 has been polished with silicon carbide (grit 800) sand papers by hand to remove the oxide layer and re-deposited with Pd and Au for repairing, and tested again at elevated temperatures up to 550 ◦ C in pure H2 and He for another 800 h. After having developed a significant He leak, membrane M3 was repaired for the second time by the same polishing method and re-plated with Pd and Au. Membrane M3 was then annealed at 500 ◦ C for more than 48 h, followed by the characterization in pure H2 and He for ∼800 h in the temperature range of 250–500 ◦ C. Fig. 4 displays that both the Pd membrane, M1 and the Pd/Au alloy membrane, M3 followed Sieverts’ law in the testing temperature and pressure range, verifying that the rate-limiting step of H2 permeation through the composite membrane was one dimensional diffusion of atomic hydrogen through either pure Pd or Pd/Au alloy H2 -selective layer. In Fig. 4, the H2 permeance of membrane M1 was 53 m3 /(m2 · h · atm0.5 ) at 500 ◦ C, which was equivalent to a free-standing Pd foil with a thickness of 10.5 ␮m [35]. The gravimetrically measured total thickness of the H2 -selective layer of membrane M1 was roughly 11 ␮m including the Pd/Ag barrier

539

layer. The result indicated that the H2 permeance of membrane M1 was very close to that of a free-standing Pd foil of roughly the same thickness. Membrane M2 was prepared by an additional plating of 3.3 ␮m Pd on membrane M1, resulting in the total thickness of the selective layer to be ∼14 ␮m. Membrane M2 exhibited a H2 permeance of 24 m3 /(m2 · h · atm0.5 ) at 400 ◦ C using the H2 flux from a single measurement of pressure difference of 1 atm (Pfeed = 2 atm, Ppermeate = 1 atm) by assuming Sieverts’ law. A 16-␮m thick freestanding Pd foil would produce a similar H2 permeance at 400 ◦ C [35]. Fig. 4 also shows that the H2 permeance of the 18.1 ␮m thick membrane M3 was 6.7 m3 /(m2 · h · atm0.5 ) at 500 ◦ C, which was equivalent to an 80-␮m thick free-standing Pd foil. The result suggested that the membrane M3 showed an abnormally low H2 permeance, especially one would expect an increase of H2 permeance from a membrane alloyed with Au within 20 wt% [21]. The unusually low H2 permeance was most likely caused by the intermetallic diffusion of the support (Inconel) elements (i.e. Ni, Cr) into the Pd–Au layer as evidenced by the cross-sectional SEM morphologies and the EDS line scan of membrane M3 after the characterization (including the H2 S tests) as shown in Fig. 5(a) and (b). The EDS line scan (the insert in Fig. 5(b)) confirmed the diffusion of Ni into the Pd/Au layer within ∼5 ␮m from the support (up to the position at the distance of ∼8 ␮m) with the highest Ni composition of ∼14 wt%. The EDS line scan also indicated the diffusion of Cr into the selective layer within ∼1 ␮m from the support (up to the position at the distance of ∼4 ␮m) with the highest Cr composition of ∼8 wt%. The intermetallic diffusion of support metal (e.g. Fe, Ni, Cr) into the H2 -selective layer has been well-documented for causing the low permeance of composite Pd-based/porous-metal-support membranes [32,36]. The long term test (cumulatively ∼450 h) of membrane M3 at 550 ◦ C during the first two tests was believed to cause the intermetallic diffusion of Ni and Cr into the Pd/Au layer. In the entire duration of the final characterization, membrane M3 showed stable H2 permeance up to the highest testing temperature of 500 ◦ C, indicating that there was no further significant structural or compositional change. The cross-sectional SEM morphologies and the EDS line scan (Fig. 5(a) and (b)) also showed that the thickness of the dense Pd/Au layer was 7–9 ␮m. An Au gradient existed at the top 2–3 ␮m thick layer with the highest Au content of ∼13 wt%. The oxygen and aluminum observed in the support region (below 3 ␮m mark) of the EDS line scan (Fig. 5(b)) was the oxidized layer and the Al2 O3 grading layer. The randomly scattered oxygen in the entire Pd/Au layer (3 –12 ␮m) was from the background noise and the surface contaminations of the SEM sample during the sample preparation.

Fig. 5. (a) Secondary-electron imaging (SEI) cross-sectional micrograph, and (b) corresponding energy dispersive X-ray spectroscopy (EDS) line scan (the insert is the Ni and Cr composition) of membrane M3 after the testing.

540

C.-H. Chen, Y.H. Ma / Journal of Membrane Science 362 (2010) 535–544

Fig. 6. Arrhenius plot for H2 permeance of membranes M1 and M3. The estimated activation energies for H2 permeation were also listed in the plot.

The ideal selectivity (JH2 /JHe , p = 1atm) before the sulfur poisoning-tests was theoretically infinite for membrane M2 (after 115 h) and 88 for membrane M3 (after 800 h) at 400 ◦ C. Since the He leak of membrane M3 was not particularly large (∼0.6 sccm at 400 ◦ C after 800 h testing), the unusually low selectivity of membrane M3 was a result of the abnormally low H2 permeance caused by the intermetallic diffusion discussed in the previous paragraph. As a result, it is believed that there were no abnormally large pinholes/defects in membrane M3 and such a small He leak should not have any significant impact on H2 permeance and n-value [37]. Fig. 6 shows the Arrhenius dependence of the H2 permeance on temperature for membranes M1 and M3. The permeance was calculated with the linear regression of Sieverts’ plot (Fig. 4) at 300, 400, and 500 ◦ C for membrane M1 and 250, 350, 450, and 500 ◦ C for membrane M3. The activation energy for the H2 permeation of membrane M1 was 17.2 kJ/mol, which was close to the value of 16.4 reported by Mardilovich et al. [30], while the slightly lower value of 13.1 kJ/mol for membrane M3 was due to the formation of Pd/Au alloy. The lower activation energy could also be caused by a large mass transfer resistance in the porous supports as reported in the literature [37–39]. Since the supports used for membranes M1 and M3 were similar, the lower activation energy observed in membrane M3 could not be caused by the mass transfer resistance in the porous support. In addition, the n-values estimated by using non-liner regression of the experimental flux data for membranes M1 and M3 were very similar in the testing temperature range (0.64–0.67 for membrane M1 in the temperature of 300–500 ◦ C and 0.57–0.67 for memrbane M3 in the emperature of 250–500 ◦ C). This substantiated the similar mass transfer resistance in the porous supports of membranes M1 and M3. 3.3. Effect of H2 S exposure on membrane performance 3.3.1. Pure Pd membrane Fig. 7 shows the relative H2 permeance of the Pd membrane, M2 and the Pd/Au alloy membrane, M3 as a function of time at 400 ◦ C upon exposure to a 54.8 ppm H2 S/H2 mixture and subsequent recovery in H2 . The concentration of H2 S was chosen in accordance with the U.S. Department of Energy (DOE) guideline [40]. The relative H2 permeance (FH2 /FoH2 ) shown in Fig. 7 was the ratio of permeance (FH2 ) to the original permeance (FoH2 ) before the H2 S exposure at the testing temperature. As shown in Fig. 7, after the introduction of the 54.8 ppm H2 S/H2 mixture, a sharp ∼75% decrease in H2 permeance of membrane M2 occurred instantaneously, and the permeance continued to decline over the 24 h period before it reached the steady state value of ∼93% decline. This indicated that the sulfur poisoning of pure Pd mem-

Fig. 7. Poisoning and recovery of the Pd membrane, M2 and the Pd/Au alloy membrane, M3 at 400 ◦ C in a 54.8 ppm H2 S/H2 mixture.

brane was more than just surface site blocking by the adsorption of sulfur. While the initial rapid decline in H2 permeance was caused by the surface site blocking by sulfur adsorption, the continuous slow decline in permeance suggested the possibility of the bulk sulfidation of Pd and the formation of Pd sulfide (Pd4 S) with time. Since Pd4 S has a lower H2 permeability than that of Pd (2 order of magnitude less) [14], the continuous decline in the H2 permeance occurred as the thickness of Pd4 S layer increased as a function of time. Mundschau et al. [7] also observed a similar H2 permeance decline of a pure Pd foil upon testing with a 20 ppm H2 S in a 60 mol% H2 –He mixture at 320 ◦ C for ∼120 h. Fig. 7 also shows that upon reintroducing pure H2 to membrane M2, no permeance increase was observed for as long as 20 h, indicating that no permeance recovery was possible due to the irreversible formation of the bulk Pd4 S. In contrast, McKinley [6] reported that 100% recovery of a Pd foil in pure H2 was possible after exposing to a 4.5 ppm H2 S/H2 mixture at 350 ◦ C and suggested that only reversible surface site blocking with no irreversible bulk sulfidation occurred under their testing conditions. The difference in permeance recoverability of the Pd membranes between the current study and McKinley’s work [6] might be due, either to the difference in the testing conditions (i.e. H2 S concentration and temperature) or the existence of a threshold thermodynamic condition for the formation of the irreversible bulk palladium sulfide. According to the thermodynamic calculation by Mundschau et al. [7], the threshold H2 S concentration in H2 for Pd to form bulk Pd4 S at 350 ◦ C was 1 ppm by volume with the Gibbs’ free energy (G0 ) of −71.6 kJ/mol for the Pd4 S formation. McKinley’s work [6] of 4.5 ppm appeared to be inconsistent with the thermodynamic computation of Mundschau et al. [7]. The He leak measured at the end of the recovery was ∼10 ml/min, indicating the failure of membrane M2. The large increase of the He leak suggested that the structure of the Pd layer was deteriorated with the formation of pinholes caused by bulk Pd4 S. The formation of pinholes (pores) or even cracks on the Pd layers after H2 S exposure have been observed by several researchers [7,8,10]. The stress generated due to the difference in the lattice constant between Pd and Pd4 S is believed to cause the formation of pinholes or cracks. 3.3.2. Pd/Au membrane In comparison to the Pd membrane, M2, the Pd/Au alloy membrane, M3 showed a different H2 permeation characteristic in the presence of H2 S under the same testing conditions. Fig. 7 shows that membrane M3 also experienced a sharp decline in H2 permeance by ∼85% after the H2 S exposure. However, the permeance reached a steady state value immediately and remained unchanged for the

C.-H. Chen, Y.H. Ma / Journal of Membrane Science 362 (2010) 535–544

541

Fig. 8. SEI micrographs of the Pd/Au coupon samples that were (a) as-prepared; (b) exposed to a 54.8 ppm H2 S/H2 mixture for 24 h at 400 ◦ C; and (c) XRD patterns of the Pd/Au coupon samples before/after the exposure to a 54.8 ppm H2 S/H2 mixture for 24 h at 400 ◦ C.

duration of 4 h. This suggested that the permeance decline of membrane M3 was solely caused by the surface site blocking without bulk sulfidation. It should be noted that the site blocking effect by adsorbed sulfur refers not only to the simple geometric blocking of adsorption sites but also the electronic blocking effect, due to the adsorbed sulfur generating an energy barrier for H2 dissociative adsorption and diffusion. As a result, each adsorbed sulfur effectively blocks 4–13 sites for H2 adsorption [11–13]. Based on the percentage of the observed permeance loss, the sulfur coverage on membrane M3 was approximately between 0.06 and 0.21 ML (monolayer). In fact, 0.3–0.5 ML sulfur coverage was sufficient to block all the H2 adsorption sites [11,12,41]. The results also suggested that, the time required to reach the adsorption equilibrium between H2 S and Pd/Au alloy membrane was very short, within only a few minutes. The ideal H2 /He selectivity of membrane M3 after the H2 S testing showed no significant change as compared to the values before the H2 S exposure (the ideal selectivity at 500 ◦ C was 142 before and 139 after the H2 S testing at 400 ◦ C). No significant change in selectivity indicated that the change in the structure of membrane M3 caused by the H2 S exposure was minimum, substantiating the fact that no bulk sulfide was formed. In fact, the He leak and the ideal selectivities of membrane M3 during the entire H2 S poisoning/recovery test (∼3000 h) were very similar with the He leak at 500 ◦ C oscillating between 0.8 and 1.0 sccm. The sulfur poisoning experiments on the Pd/Au alloy coupons under similar testing conditions were carried out to substantiate the fact that the sulfur poisoning effect on the Pd/Au alloy was essentially a surface phenomenon. Fig. 8(a)–(c) show the surface morphologies, sulfur compositions (analyzed by EDS), and X-ray diffraction patterns of the Pd/Au alloy coupon (∼10 ␮m, 8 wt% Au, annealed at 550 ◦ C/24 h/H2 ) before/after the exposure to a 54.8 ppm H2 S/H2 mixture at 400 ◦ C for 24 h. The results showed that the change in the morphology and crystal phase was negligible. No sul-

fur and bulk sulfide phase was detected by EDS and XRD, indicating that the sulfur poisoning on the Pd/Au alloy was indeed the surface phenomenon. Fig. 7 shows that the permeance of the Pd/Au alloy membrane, M3 after the H2 S exposure could be recovered. A steady state value of ∼65% of the original permeance could be recovered in pure H2 at the poisoning temperature of 400 ◦ C. This indicated that the bonding between the adsorbed sulfur and the Pd/Au alloy surface was partially reversible since the H2 permeance recovery is believed to be associated with the desorption of the adsorbed sulfur. Upon the pure H2 introduction, the gas–solid equilibrium of the H2 S dissociative adsorption shifted due to the decrease of the H2 S concentration in the gas phase, resulting in the desorption of adsorbed sulfur and the recovery of the H2 permeance. The variation of the permeance recovery rate (slope) as shown in Fig. 7 suggested that the desorption rate of the adsorbed sulfur varied during the recovery process. After an instantaneous increase of ∼4% in permeance, a linear recovery to ∼52% of the original permeance within 24 h took place followed by an a much slower rate approaching asymptotically to the steady state value of ∼65% of the original permeance. The possible reason for the variation of the desorption rate of the adsorbed sulfur was that the adsorbed sulfur was bonded to the Pd/Au alloy surface with different binding energies. It has been shown by theoretical calculations that the binding energy of sulfur to Pd decreased as the sulfur coverage was increased (from 5.07 ev/S atom at 0.11 ML to 3.29 ev/S atom at 1 ML) due to the increase of the repulsive force between sulfur atoms with increasing sulfur coverage [42]. As the sulfur coverage decreased during the recovery, the binding energy between sulfur and Pd increased, resulting in the lower desorption and permeance recovery rate. A similar phenomenon of stage-wise permeance recovery has also been observed in Pd/Cu alloy membranes [18]. In addition, the linear permeance recovery rate was found to be temperature dependent. Since the permeance recovery was primarily associ-

542

C.-H. Chen, Y.H. Ma / Journal of Membrane Science 362 (2010) 535–544

Fig. 9. Recovery of H2 permeance of membrane M3 in pure H2 at 400 and 500 ◦ C (membrane M3 was poisoned at 400 ◦ C).

ated with the desorption of sulfur from the membrane surface, the activation energy for desorption of the adsorbed sulfur was estimated to be 94 kJ/mol based on the linear permeance recovery rate in the temperature range of 350–500 ◦ C. The value was within the chemisorption regime and consistent with the dissociative (chemical) adsorption mechanism of H2 S. In order to verify if the unrecovered permeance was the result of the irreversible sulfur on the surface of membrane M3, further recovery in H2 was performed on membrane M3 at the higher temperature of 500 ◦ C and the result is shown in relative permeance in Fig. 9. As seen in Fig. 9, after increasing the temperature to 500 ◦ C for ∼24 h (including ∼5 h in He for the leak measurement), the permeance of membrane M3 was nearly fully recovered at ∼97% of its original 400 ◦ C permeance. Since the dissociative adsorption of H2 S on metals was exothermic, desorption of adsorbed sulfur became more favorable due to the shifting of equilibrium toward gas phase at higher temperatures resulting in more permeance recovery. In addition, more energy was provided to the membrane surface at higher temperatures to desorb the adsorbed sulfur with higher binding energies. A very small portion of the permeance, ∼3%, was still not recoverable after 24 h at 500 ◦ C, indicating the existence of a small portion of adsorbed sulfur with even higher binding energies. Nevertheless, the results suggested that the poisoning of the Pd/Au alloy membrane was caused by only the surface site blocking poisoning mechanism without bulk sulfidation in a 54.8 ppm H2 S/H2 mixture, and ∼100% recovery was possible. A full permeance restoration of the Pd/Au alloy (Pd60 Au40 ) foil has also been reported in the literature after exposing to 4–6600 ppm H2 S/H2 [6], where the permeance loss during the H2 S exposure was also considered to be due to the surface site blocking by the adsorption of H2 S rather than the bulk sulfide formation. 3.3.3. Effect of temperature and time of H2 S exposure on Pd/Au membrane performance The same poisoning experiments including subsequent recoveries in pure H2 were performed on membrane M3 at other temperatures in order to investigate the effect of temperature on H2 S poisoning. At each testing temperature, membrane M3 was exposed to a 54.8 ppm H2 S/H2 mixture for 4 h followed by a recovery period in pure H2 at the same temperature. After a steady state permeance recovery was reached, the temperature was then increased to 500 ◦ C to fully recover the permeance before the next poisoning/recovery test. Fig. 10 displays the relative permeance of membrane M3 upon exposure to a 54.8 ppm H2 S/H2 mixture and subsequent recovery in H2 (before the full recovery at 500 ◦ C) as a function of time in the temperature range of 350–500 ◦ C. The instantaneous decline in

Fig. 10. Poisoning and recovery of membrane M3 in a 54.8 ppm H2 S/H2 mixture as a function of temperature. The temperature next to the curves indicates the poisoning and recovery temperature.

permeance to a steady value after the H2 S introduction during the 4 h duration of exposure was observed at all testing temperatures with the only difference being in the percentage of the permeance decline. This indicated that the surface site blocking effect by the dissociative adsorption of H2 S took place and accounted for the permeance loss in the entire testing temperature range. In addition, the similar permeance recovery phenomenon with the recovery rate changing in stages during the recovery process was also observed at all testing temperatures. However, the recovery rate and degree of recovery were different at different temperatures, substantiating that the permeance recovery was associated with the sulfur desorption from the membrane surface. The temperature dependence of the permeance recovery is shown in Fig. 11. Not only did a higher percentage of permeance remain during the H2 S exposure, but also a higher permeance recovery was observed at higher temperatures due to the exothermic nature of the dissociative adsorption of H2 S on metals. The equilibrium of gas–solid adsorption of sulfur shifted toward gas phase at higher temperatures resulting in less permeance decline and more permeance recovery. The Pd/Au alloy membrane, M3 showed a greater permeance decline in the presence of H2 S when compared to previous work reported in the literature. Upon exposure to a 20 ppm H2 S/H2 mixture at 350 ◦ C, McKinley [6] reported a 56% decline in H2 flux of the Pd/Au alloy foil with 40 wt% Au. Way et al. [25] reported that a 38% decline in permeance was observed at 400 ◦ C when a Pd85 Au15 alloy

Fig. 11. H2 permeance after poisoning (in a 54.8 ppm H2 S/H2 mixture) and after recovery (at the poisoning temperature) of membrane M3 as a function of temperature.

C.-H. Chen, Y.H. Ma / Journal of Membrane Science 362 (2010) 535–544

543

the H2 S exposure mainly resulted from the surface site blocking by the dissociative adsorption of H2 S, which was essentially reversible. The permeance loss during the H2 S exposure as well as the permeance recovery were temperature dependent with less permeance loss and more permeance recovery occurring at higher temperatures due to the exothermic nature of the dissociative adsorption of H2 S on metals. Finally, the exposure duration of H2 S showed no significant effects on the degree of permeance decline at a fixed temperature indicating that the permeance loss caused by the site blocking was the result of equilibrium adsorption of H2 S on the membrane surface. Acknowledgements

Fig. 12. Poisoning and recovery of membrane M3 in a 54.8 ppm H2 S/H2 mixture at 400 ◦ C as a function of H2 S exposure time. The time next to the curves indicates the exposure duration.

membrane was exposed to a 5 ppm H2 S/H2 mixture. The higher H2 S concentrations used in the current study accounted for the higher permeance loss due to the thermodynamic equilibrium of the H2 S adsorption on the membrane surface. In addition, less Au content in membrane M3 might also result in more permeance decline due to the stronger bonding between sulfur and Pd/Au alloy with less Au content. It has been shown by theoretical calculations that the binding energy of sulfur to Pd/Au alloy decreased with increasing Au content in the alloy [25,42]. Fig. 12 summarized the relative permeance of membrane M3 upon exposures to a 54.8 ppm H2 S/H2 mixture with the duration of 4, 8, 12, and 24 h and subsequent recoveries in H2 (before the full recovery at 500 ◦ C) as a function of time at 400 ◦ C. The similar instantaneous decline of 85–90% in permeance upon the H2 S introduction was observed for all four cases. This indicated that the percentage of surface adsorption sites occupied by the adsorbed sulfur was not affected by the exposure time due to the fact that the thermodynamic equilibrium of gas–solid adsorption of sulfur on the Pd/Au alloy membrane surface was fast and time independent. Fig. 12 also shows that longer exposure time increased the time required to recover the permeance to the steady state value. The slightly slower recovery rate appeared to imply that an additional small amount of sulfur was bonded with Pd/Au alloy with higher binding energies as the exposure time was increased. Previous theoretical calculations indicated that beyond a certain sulfur coverage (0.75 ML), the incorporation of sulfur into the subsurface of Pd became more energetically favorable than continuous adsorption on the surface [43]. The longer exposure time potentially increased the subsurface bonded sulfur with a higher binding energy without increasing the surface sulfur coverage. As a result, the permeance recovery after a longer H2 S exposure time occurred with a slower rate. 4. Conclusions The exposure of the pure Pd membrane to a 54.8 ppm H2 S/H2 mixture at 400 ◦ C resulted in two stages of permeance decline due to (i) surface site blocking by the dissociative adsorption of H2 S, and (ii) bulk sulfidation of Pd with the formation of Pd4 S. The latter caused the irreversible permeance loss and led to the failure of the membrane due to the formation of pinholes. The Pd/Au alloy membrane with 8 wt% Au (∼13 wt% Au at the top surface of the membrane) showed resistance to bulk sulfidation upon exposure to a 54.8 ppm H2 S/H2 mixture in the temperature range of 350–500 ◦ C, and underwent no significant structural changes. The permeance decline of the Pd/Au alloy membrane upon

We gratefully thank the financial support provided by the University Coal Research Program, U.S. Department of Energy, Award No. DE-FG26-04NT42194 and Hydrogen Pathway, Award No. DEFC26-07NT43058. References [1] G. Barbieri, A. Brunetti, T. Granato, P. Bernardo, E. Drioli, Engineering evaluations of a catalytic membrane reactor for the water gas shift reaction, Ind. Eng. Chem. Res. 44 (2005) 7676. [2] A. Basile, A. Criscuoli, F. Santella, E. Drioli, Membrane reactor for water gas shift reaction, Gas Sep. Purif. 10 (1996) 243. [3] A. Basile, E. Drioli, F. Santella, V. Violante, G. Capannelli, G. Vitulli, A study on catalytic membrane reactors for water gas shift reaction, Gas Sep. Purif. 10 (1996) 53. [4] A. Criscuoli, A. Basile, E. Drioli, An analysis of the performance of membrane reactors for the water–gas shift reaction using gas feed mixtures, Catal. Today 56 (2000) 53. [5] R. Bredesen, K. Jordal, O. Bolland, High-temperature membranes in power generation with CO2 capture, Chem. Eng. Process. 43 (2004) 1129. [6] D.L. McKinley, Metal Alloy for Hydrogen Separation and Purification, US Patent, 3,350,845, 1967. [7] M.V. Mundschau, X. Xie, C.R. Evenson, A.F. Sammells, Dense inorganic membranes for production of hydrogen from methane and coal with carbon dioxide sequestration, Catal. Today 118 (2006) 12. [8] A. Kulprathipanja, G.O. Alptekin, J.L. Falconer, J.D. Way, Pd and Pd-Cu membranes: inhibition of H2 permeation by H2 S, J. Membr. Sci. 254 (2005) 49. [9] D.J. Edlund, W.A. Pledger, Thermolysis of hydrogen sulfide in a metalmembrane reactor, J. Membr. Sci. 77 (1993) 255. [10] M. Kajiwara, S. Uemiya, T. Kojima, Stability and hydrogen permeation behavior of supported platinum membranes in presence of hydrogen sulfide, Int. J. Hydrogen Energy 24 (1999) 839. [11] M.L. Burke, R.J. Madix, Hydrogen on palladium(1 0 0)-sulfur: the effect of sulfur on precursor mediated adsorption and desorption, Surf. Sci. 237 (1990) 1. [12] P.A. Gravil, H. Toulhoat, Hydrogen, sulphur and chlorine coadsorption on Pd(1 1 1): a theoretical study of poisoning and promotion, Surf. Sci. 430 (1999) 176. [13] N.D. Lang, S. Holloway, J.K. Noerskov, Electrostatic adsorbate–adsorbate interactions: the poisoning and promotion of the molecular adsorption reaction, Surf. Sci. 150 (1985) 24. [14] B.D. Morreale, B.H. Howard, O. Iyoha, R.M. Enick, C. Ling, D.S. Sholl, Experimental and computational prediction of the hydrogen transport properties of Pd4 S, Ind. Eng. Chem. Res. 46 (2007) 6313. [15] K.J. Bryden, J.Y. Ying, Nanostructured palladium–iron membranes for hydrogen separation and membrane hydrogenation reactions, J. Membr. Sci. 203 (2002) 29. [16] H. Gao, Y.S. Lin, Y. Li, B. Zhang, Chemical stability and its improvement of palladium-based metallic membranes, Ind. Eng. Chem. Res. 43 (2004) 6920. [17] J.A. Rodriguez, Physical and chemical properties of bimetallic surfaces, Surf. Sci. Rep. 24 (1996) 223. [18] N. Pomerantz, Y.H. Ma, Effect of H2 S on the performance and long-term stability of Pd/Cu membranes, Ind. Eng. Chem. Res. 48 (2009) 4030. [19] B.D. Morreale, M.V. Ciocco, B.H. Howard, R.P. Killmeyer, A.V. Cugini, R.M. Enick, Effect of hydrogen-sulfide on the hydrogen permeance of palladium–copper alloys at elevated temperatures, J. Membr. Sci. 241 (2004) 219. [20] P. Kamakoti, B.D. Morreale, M.V. Ciocco, B.H. Howard, R.P. Killmeyer, A.V. Cugini, Prediction of hydrogen flux through sulfur-tolerant binary alloy membranes, Science 307 (2005) 569 (Washington, DC, U.S.). [21] V. Gryaznov, Metal-containing membranes for the production of ultrapure hydrogen and the recovery of hydrogen isotopes, Sep. Purif. Methods 29 (2000) 171. [22] A. Sieverts, E. Jurisch, A. Metz, Solubility of hydrogen in the solid alloys of palladium with gold, silver and platinum, Z. Anorg. Allg. Chem. 92 (1915) 322. [23] C.G. Sonwane, J. Wilcox, Y.H. Ma, Achieving optimum hydrogen permeability in PdAg and PdAu alloys, J. Chem. Phys. 125 (2006) 184714/1.

544

C.-H. Chen, Y.H. Ma / Journal of Membrane Science 362 (2010) 535–544

[24] C.G. Sonwane, J. Wilcox, Y.H. Ma, Solubility of hydrogen in PdAg and PdAu binary alloys using density functional theory, J. Phys. Chem. B 110 (2006) 24549. [25] J.D. Way, M. Lusk, P. Thoen, Sulfur-Resistant Composite Metal Membranes, US Patent 20,080,038,567, 2008. [26] Y. Okinaka, Electroless plating of gold and gold alloys, in: G.O. Mallory, J.B. Hajdu (Eds.), Eletroless Plating: Fundamentals and Applications, American Eletroplaters and Surface Finishers Society, Orlando, FL, 1990, pp. 401–420. [27] E.A. Parker, Immersion solutions, in: W. Reid, W. Goldie (Eds.), Gold Plating Technology, Eletrochemical Publications Limited, Ayr, Scotland, 1974, pp. 73–81. [28] M.F. El-Shazly, K. Baker, High build electroless gold processes, in: Proceeding of 1st AES Electroless Plat. Symp., St Louis, Missouri, 1982, Paper 20, 23. [29] M.D. Coutts, A.G. Revesz, Decoration of semiconductor surfaces for electron microscopy by displacement deposition of gold, J. Appl. Phys. 37 (1966) 3280. [30] P.P. Mardilovich, Y. She, Y.H. Ma, M. Rei, Defect-free palladium membranes on porous stainless-steel support, AICHE J. 44 (1998) 310. [31] Y.H. Ma, B.C. Akis, M.E. Ayturk, F. Guazzone, E.E. Engwall, I.P. Mardilovich, Characterization of intermetallic diffusion barrier and alloy formation for Pd/Cu and Pd/Ag porous stainless steel composite membranes, Ind. Eng. Chem. Res. 43 (2004) 2936. [32] M.E. Ayturk, I.P. Mardilovich, E.E. Engwall, Y.H. Ma, Synthesis of composite Pdporous stainless steel (PSS) membranes with a Pd/Ag intermetallic diffusion barrier, J. Membr. Sci. 285 (2006) 385. [33] S.C. Su, J.N. Carstens, A.T. Bell, A study of the dynamics of Pd oxidation and PdO reduction by H2 and CH4 , J. Catal. 176 (1998) 125. [34] Y.L. Lam, M. Boudart, Oxidation of small palladium particles, J. Catal. 47 (1977) 393.

[35] M.E. Ayturk, Synthesis, annealing strategies and in-situ characterization of thermally stable composite thin Pd/Ag alloy membranes for hydrogen separation, Ph.D. Thesis, Worcester Polytechnic Institute, Worcester, MA, 2007. [36] V.M. Gryaznov, O.S. Serebryannikova, Y.M. Serov, M.M. Ermilova, A.N. Karavanov, A.P. Mischenko, Preparation and catalysis over palladium composite membranes, Appl. Catal. A 96 (1993) 15. [37] F. Guazzone, E.E. Engwall, Y.H. Ma, Effects of surface activity, defects and mass transfer on hydrogen permeance and n-value in composite palladium–porous stainless steel membranes, Catal. Today 118 (2006) 24. [38] O. Schramm, A. Seidel-Morgenstern, Comparing porous and dense membranes for the application in membrane reactors, Chem. Eng. Sci. 54 (1999) 1447. [39] S. Thomas, R. Schafer, J. Caro, A. Seidel-Morgenstern, Investigation of mass transfer through inorganic membranes with several layers, Catal. Today 67 (2001) 205. [40] Development, and Demonstration Plan for the Period 2009 Through 2016, U.S. Department of Energy Hydrogen from Coal Program Research, 2009, http://fossil.energy.gov/programs/fuels/publications/programplans/2009 Draft H2fromCoal Sept30 web.pdf. [41] S. Wilke, M. Scheffler, Poisoning of Pd(1 0 0) for the dissociation of H2 : a theoretical study of co-adsorption of hydrogen and sulfur, Surf. Sci. 329 (1995) L605. [42] D.R. Alfonso, First-principles study of sulfur overlayers on Pd(1 1 1) surface, Surf. Sci. 596 (2005) 229. [43] D.R. Alfonso, Initial incorporation of sulfur into the Pd(1 1 1) surface: a theoretical study, Surf. Sci. 600 (2006) 4508.