+ All Categories
Home > Documents > j.memsci.2011.03.057

j.memsci.2011.03.057

Date post: 03-Jun-2018
Category:
Upload: landa
View: 217 times
Download: 0 times
Share this document with a friend

of 18

Transcript
  • 8/12/2019 j.memsci.2011.03.057

    1/18

    Journal of Membrane Science 375 (2011) 2845

    Contents lists available atScienceDirect

    Journal of Membrane Science

    j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / m e m s c i

    Review

    Correlations in palladium membranes for hydrogen separation: A review

    Samhun Yuna, S. Ted Oyama a,b,

    a Environmental Catalysis and Nanomaterials Laboratory, Department of Chemical Engineering, Virginia Polytechnic Institute & State University, Blacksburg,

    VA 24061-0211, United Statesb Department of Chemical Systems Engineering, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656, Japan

    a r t i c l e i n f o

    Article history:

    Received 17 September 2010

    Received in revised form 27 March 2011Accepted 29 March 2011Available online 5 April 2011

    Keywords:

    Hydrogen separationPalladium membranesPalladium-alloyPalladium composite membranesElectroless platingElectro platingChemical vapor deposition

    a b s t r a c t

    This review describes palladium and palladium alloy membranes for hydrogen separation prepared bydifferent fabrication methods and using different membrane supports. Several correlations of structureandfunction for thosemembranes are provided basedon mechanistic considerations of permeancealongwith structural properties and membrane morphologies. Particular attraction is placed in analysis ofthe hydrogen permeance and selectivity of membranes reported in recent papers. Composite palladiummembranes prepared by the electroless plating technique deposited on alumina substrates are foundto be the most promising for practical applications. It is concluded that the prospects for the use ofpalladium membranes in industrial applications are improving due to extensive research addressingcurrent problems such as durability, hydrogen embrittlement, fouling by hydrocarbons or hydrosulfidecompounds, and the high cost of palladium.

    2011 Elsevier B.V. All rights reserved.

    Contents

    1. Introduction and background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291.1. Palladium membranes for hydrogen separation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291.2. General properties of palladium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291.3. Hydrogen permeation through palladium membranes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

    2. Fabrication methods of palladium membranes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 322.1. Electroless plating (ELP) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 322.2. Chemical vapor deposition (CVD) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 332.3. Physical vapor deposition (PVD) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 332.4. Electroplating deposition (EPD) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

    3. Palladium-based membranes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 343.1. Unsupported membranes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 343.2. Supported membranes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

    3.2.1. Porous Vycor glass supports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 343.2.2. Porous ceramic supports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353.2.3. Porous metal supports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

    4. Performance analysis of palladium membranes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 405. Comparison of Pd membranes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 416. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

    Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42Appendix A. Hydrogen permeation equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42R e f e r e n c e s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

    Corresponding author at: Department of Chemical Systems Engineering, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656, Japan.Tel.: +81 3 5841 0712; fax: +81 3 5841 0712.

    E-mailaddresses: [email protected],ted [email protected](S. Ted Oyama).

    0376-7388/$ see front matter 2011 Elsevier B.V. All rights reserved.

    doi:10.1016/j.memsci.2011.03.057

    http://localhost/var/www/apps/conversion/tmp/scratch_2/dx.doi.org/10.1016/j.memsci.2011.03.057http://localhost/var/www/apps/conversion/tmp/scratch_2/dx.doi.org/10.1016/j.memsci.2011.03.057http://www.sciencedirect.com/science/journal/03767388http://www.elsevier.com/locate/memscimailto:[email protected]:[email protected]://localhost/var/www/apps/conversion/tmp/scratch_2/dx.doi.org/10.1016/j.memsci.2011.03.057http://localhost/var/www/apps/conversion/tmp/scratch_2/dx.doi.org/10.1016/j.memsci.2011.03.057mailto:[email protected]:[email protected]://www.elsevier.com/locate/memscihttp://www.sciencedirect.com/science/journal/03767388http://localhost/var/www/apps/conversion/tmp/scratch_2/dx.doi.org/10.1016/j.memsci.2011.03.057
  • 8/12/2019 j.memsci.2011.03.057

    2/18

    S. Yun, S. Ted Oyama / Journal of Membrane Science375 (2011) 2845 29

    1. Introduction and background

    1.1. Palladium membranes for hydrogen separation

    The subject of palladium membranes for hydrogen separationandmembrane reactors for hydrogenproduction has been coveredextensively since the early work of Gryaznov[1], and has beencovered in a number of reviews [26]. This paper covers recentdevelopments with a concentration on the presentation of struc-tural and performance correlations.

    Hydrogenhas beena valuablematerialsince firstartificially pro-duced by Hohenheim via mixing of metals and strong acids [7]and is consumed on the order of billions of cubic meters per dayin various industrial fields[810].Hydrogen is widely used in thepetroleum industry especially in hydrodealkylation, hydrodesulfu-rization and hydrocracking [4] and has recently attracted attentionas a possible alternative energy carrier to relieve environmentalproblems derived from fossil fuel use[2].Particularly, hydrogenuse in proton exchange membrane (PEM) fuel cells is attractivebecause of the efficiency of the energy conversion and the lack ofrelease of greenhouse gases[1113].Currently, hydrogen is pro-duced in several differentways, such as electrolysisof water,steamreformingof methane, gasification of coal or partial oxidation of oil

    or natural gas, and the total production has increased annually at aconsiderable pace (Fig. 1)[14].

    Common technologies employed for hydrogen separationinclude solvent adsorption, pressure swing adsorption, cryogenicrecovery and membrane separation. Compared with other meth-ods, membrane separation technologies have economic potentialin reducing operating costs, minimizing unit operations and low-ering energy consumption[4].For these reasons and because ofthe increasing demand for high purity hydrogen, the developmentfor effective hydrogen membranes has engendered considerableinterest in academia and industry.

    Membranes made of nickel, palladium, and platinum whichbelong to group 10, and some metallic elements in groups 35 ofthe periodic table have the ability to dissociate anddissolve hydro-

    gen, but only palladium membranes show an outstanding abilityto transport hydrogen through the metal due to a much highersolubility of hydrogen in its bulk over a wide temperature range(Fig. 2)[15].This property has given rise to numerous studies ofpalladium based membranes for the separation and purificationof hydrogen. In addition, there have been many applications ofpalladium in catalytic membrane reactors for reactions involving

    20001980196019401920

    0

    500

    1000

    1500

    2000

    Productionamount(TWh)

    Year

    Total

    Methane steam reforming

    Partial oxidation

    Coal g asification

    Water electro lysis

    Fig. 1. Annual global amount of hydrogen produced.

    16001400120010008006004002000

    0.01

    0.1

    1

    10

    Ni

    Pd

    Log(Solubility)

    Temp /oC

    Fe

    Cu

    Pt

    Fig. 2. Comparison of hydrogen solubility in several metals at a pressure of 1 atm.Solubility is given units of standard cm3 of H2per 100 g of metal.

    hydrogen, such as hydrogenation and dehydrogenation [16,17],methoxymethane reforming[18],methane steam reforming[19],watergas shift [20,21], hydroxylation of benzene [22], and hydro-gen peroxide synthesis[23].General results in membrane reactorscience have alsobeen describedrecently.Examples are correlationof conversion enhancement versus permeance[24], identificationof the regions for the use of 1-D versus 2-D models [25],and thedistinction between primary and secondary products[26].

    Requirements for commercial applications of palladium-basedmembranes include a reasonable membrane cost, high hydro-gen permeance, high hydrogen selectivity versus other gases, andsteady, predictable performance over a long period of time underharsh conditions[27,28], resistance to poisoning by hydrogen sul-fide, chlorine, carbon monoxide, and hydrocarbons and thermal

    stabilityunder thermalcycling[29]. Muchefforthasbeenexpendedto improve these important aspects, and various palladium basedmembranes supportedon ceramic and metallic materialswith highhydrogenpermeanceand hydrogenselectivity have been reported.Concerning durability, several Pd or Pd-alloy membranes werereported to be stable for several months under H2 flow in thetemperature range of 623773K [2931]. Pd-based metallic orcomposite membranes are currently known to be stable for up to10 months according to a National Energy Technology Laboratory(NETL) report[32].

    However, it is still a challenge to prepare palladium basedmembranes with high permeability and hydrogen selectivity withthin defect-free palladium layers which have long term thermaldurability and chemical stability. This chapter reviews the latest

    development in the field with an emphasis on describing generalcorrelations.

    1.2. General properties of palladium

    Despite palladiums unique ability to permeate hydrogen, themetal suffers from several limitations. A first problem is that theabsorption of hydrogen below its critical point of 571 K (298 C)and 2MPa produces two different phases (and )(Fig. 3)[15],both of which retain the pure palladium face-centered cubic (fcc)lattice but with the crystal unit cell lattice parameter increasingfrom 0.3890nm for hydrogen-free palladium to 0.3895nm for the-phase and up to 0.410nm for the -phase at room tempera-ture [15,33] The-phase is obtained at low H/Pd atomic ratios and

    becomes the dominant phase at high temperature. The -phase is

  • 8/12/2019 j.memsci.2011.03.057

    3/18

    30 S. Yun, S. Ted Oyama / Journal of Membrane Science375 (2011) 2845

    0.70.60.50.40.30.20.10.0

    0

    4

    8

    12

    16

    20

    24

    28

    180oC

    200oC

    230oC

    250oC

    270oC

    290oC

    +H

    2vaporpressure/atm

    H/Pd (atomic ratio)

    310oC

    Fig. 3. PCTphase diagram of the palladiumhydrogen system adapted from [15].

    formed at high H/Pd atomic ratios and coexists with the-phase atlow temperature (Fig. 3).The hydrogen vapor pressure is constantin the region of phase coexistence and is bounded by an enve-lope defining max and min, the compositions of maximum andminimum H/Pd ratio for pure palladium. The change in volumeaccompanying the phase transformation can give rise to strain andrecrystallization which lead to bulk and grain boundary defects.

    A second problem is that metals can lose ductility from expo-sure to hydrogen, a process called hydrogen embrittlement, whichcauses cracking of the metal [34]. Finally, the metallic nature ofpalladium gives rise to interactions with carbon containing specieswhich deactivates the surface [35], especially by exposure to unsat-urated hydrocarbons, sulfur, or carbon monoxide[4].

    In order to prevent phase transition and alleviate embrittle-

    ment and poisoning problems,palladium canbe alloyed with othermetallic elements such as Ag, Cu, Fe, Ni, Pt and Y[3641].In gen-eral, the critical temperature for the/-phase transformation canbe lowered considerably by allowing palladium with those met-als. For example, the critical temperature of PdAg (Ag 23%) andPdPt (Pt 19%) were reported to be lowered from 571K (298 C)to around room temperature[42,43].The difference in lattice sizebetween the- and-phases also decreases in palladium-alloys sothat less distortion occurs during hydrogen absorptiondesorptioncycles (Fig. 4)[33,4447].In several studies, the hydrogen perme-ance of palladium-alloys was also reported to be higherthanthat ofpure palladiumwhen yttrium, cerium, silver, copper and gold wereused as the alloying metal. This interesting behavior was observedin some tertiary palladium alloys such as PdRuIn, PdAgRu and

    PdAgRh (Table 1)[48,49].The use of nanometer-sized palladium grains was reported as

    an alternative method for minimizing the lattice distortion fromthe/-phase transition. This occurs because the concentrationof hydrogen on the grain surface and subsurface significantlyincreases compared with that in the interior sites for the case ofnanometer particles[50].

    In thecase of binarypalladium-alloys,it is observed that H2per-meance is generally proportional to the average bond distance ofthe alloys (Fig. 5).This result is reasonable because hydrogen per-meation is controlled by the diffusion of atomic hydrogen throughthe metal lattice[51]and the larger atomic distance facilitates thisprocess. In the case of PdRu careful studies by the group of Wayshow that permeance is not improved over pure Pd[52],and this

    canbe understood from theshorter bond distance in Ru (0.265 nm)

    0 5 10 15 20

    0.004

    0.006

    0.008

    0.010

    0.012

    0.014

    Pd100-x

    Agx/2

    Nix/2

    Pd100-x

    Nix

    min-m

    ax

    /nm

    X / atomic %

    Pd100-x

    Agx

    Fig. 4. Correlation between the lattice parameter differences of palladium and pal-ladium alloys and composition (room temperature)[33,4447].

    than in Pd (0.275nm). However, the PdRu material shows 80%higher hardness, which is an improvement. On the other hand forPdAg permeance is enhanced [5355], andthis fits with thelargerbond distance in Ag (0.289nm).

    Permeability is given by the product of diffusivity and solubility.Investigationof diffusivity andH2 solubilityin Pd-alloymembranesis therefore of interest for the understanding and improvementof H2 permeance. In the literature, the product of diffusivity andsolubility is proportionalto the H2permeance in Pd-alloys[5658].

    Rare-earth palladiumalloys such as PdY or PdCecan achieve ahigher H2solubility because the rare-earthelements are 30% largerthan Pd in atomic size, thus increasing the hydrogen permeationrate, even though the diffusion coefficients in the alloys are rela-tively smaller than that of pure Pd [56]. In the case of PdAg alloys,

    the H2solubility increases with Ag content up to the 2040wt% ofAg while the H2 diffusivity decreases with increasing Ag content.The simultaneous changes in solubility and diffusivity result in a1.7 times higher H2permeability than pure Pd at 23 wt% of Ag andat 623K[57].The estimated solubility for PdAg (30% Ag) and forPdAu (20% Au) are 10 times and 12 times higher, respectively, thanthat of pure Pd at 456 K, while the solubility of PdCu (20% Cu) is 5times smaller than that of pure Pd[58],which corresponds to thepermeability behavior.

    Sulfur poisoningis a significant problem in some feedstocks, forexamples those derived from coal. An advantage of alloys is that

    Table 1

    Higher hydrogen permeance of various palladium-alloys compared with pure pal-

    ladium[48,49].

    Pd-alloy Wt% of alloy metal Average bonddistancea,b/nm

    Permeance ratioPd-alloy/Pd

    Pd 0 0.275 1.0PdY 6.6 0.281 3.5PdY 10 0.284 3.8PdAg 23 0.278 1.7PdCe 7.7 0.280 1.6PdCu 10 0.272 0.48PdAu 5 0.275 1.1PdRuIn 0.5, 6 0.278 2.8PdAgRu 30, 2 0.279 2PdAgRu 19,1 0.278 2.6

    a Bond distance of each metal: Pd (0.275), Y (0.355), Ag (0.289), Ce (0.365), Cu(0.256), Au (0.288), Ru (0.265), In (0.325).

    b Average bond distance:ibonddistanceofMiXi ,Mi: metal,Xi: mole fraction.

  • 8/12/2019 j.memsci.2011.03.057

    4/18

    S. Yun, S. Ted Oyama / Journal of Membrane Science375 (2011) 2845 31

    0.2850.2800.2750.270

    0

    1

    2

    3

    4

    H2

    Permean

    ceratio

    Average bond distance / nm

    Fig. 5. H2permeance ratio as a function of average bond distance.

    they are sometimes more resistant to poisoning by sulfur[59].Thesubject has been reviewed[60].Considerable work has been donewith PdCu membranes because they do not show embrittlementeven at low temperatures[61,62].Studies by Morreale et al. haveshown that the face-centered cubic (fcc) phase is more resistant tosulfur than the body-centered cubic (bcc) phase[63].In transientexperiments the fcc PdCu composition showeda decline of 010%when exposed to 1000ppm of sulfur, while a bcc PdCu composi-tionhadadeclineof99%.StudiesbyPomerantzandMa[64] confirmthese results for PdCu compositions of 8, 18, and 19 wt% Cu, withpermeancelossesof 80% at500 C (773 K). Theyfurthershowed thatthe loss was partially reversible by hydrogen treatment. A recentstudy by Howard and coworkers [65]shows that Pd exposed to1000ppm sulfur at 350 C (623 K) corrodes over a period of hours

    to form a thick (6.6m) PdS4 layer, probably by an autocatalyticprocess. In contrast a Pd47Cu53alloy forms a thin (3 nm) PdCuSlayer. Although this layer cannot dissociate hydrogen or is imper-meable to hydrogen, it does protect the bulk from sulfidation, andcould be removed by a hydrogen treatment.

    PdAu alloy composite membranes are of interest because thepresence of goldreducestheembrittlement problem resultingfromthe hydride phase transition and improves resistance to catalyticpoisoning and corrosive degradation by sulfur compounds whilegiving rise to higher hydrogen permeability than pure Pd up to15% Au content[6,98].In early work, PdAu alloys were preparedby metallurgical processes with a thickness range of 25100m[49,98].This method requires substantial capital investment, butcan produce homogeneous films with precisely controlled compo-

    sition[66].Recent production of thin PdAu alloy membranes byelectroless plating or electroplating has been reported [66,67].Thegroup of Way [66] prepared self-supporting PdAu alloy films withthickness of 513m by electroless plating on mirror-finished 304stainless steel sheeting. They reported that heterogeneous PdAufilms can be homogeneous after heat treatment at greater than1023K. A recent patent reported by Way et al.[68]described thefabrication of Pd or Pd-alloy membranes by sequential electrolessplating on a metal or ceramic substrate and demonstrated thatPdAu membranes had an enhanced resistance to poisoning bysulfur compounds. Their PdAu alloy membranes with a thicknessof 57m plated on -alumina tubes showed about 10% reduc-tion (from 0.2 to 0.18 molm2) in H2flux during a permeation testfor 20 days under a WGS (water gas shift) mixture with 1ppm

    H2S at 673K. Shi et al. [67] prepared 35m PdAu layers via

    Fig. 6. Solution diffusion mechanism of hydrogen permeation through a palladiummembrane.

    sequential electroless plating of Pd and Au onto porous ceramicsurfaces followed by heat treatment at 823 K for 300h under H2flowto obtain a homogeneous alloy structure. Their bestmembraneshowed a hydrogen permeance of 1.7105 molm2 s1 Pa1 at673K, which is the highest value disclosed so far, but is of ques-tionable validity because hydrogen selectivity against other gases

    was not reported. The group of Ma [69]prepared a PdAu alloy(8 wt% Au) membrane by a sequential electroless plating methodon an Inconel tube followed by H2 flow at 773K for 48h. TheirPdAu alloy membrane showed a 40% reduction of H2permeanceduring exposure to a mixture of 54.8 ppm H2S in H2 at 673K and100% recovery of H2 permeance with pure H2 treatment at 773 Kfor 24 h, which indicated that sulfur was reversibly adsorbed onlyon the surface in thecase of this PdAu alloy layer.However,the H2permeance of their alloy membrane (18 m layer) was unexpect-edly threetimeslowerthantheirpure Pd membrane (14m layer),which was attributed to the intermetallic diffusion of the supportelements into the selective layer.

    In summary PdCu (particularly Pd60Cu40) and PdAu are themost promising compositions in terms of permeance and sulfur

    resistance[60].The gold alloys however will be limited by the costof the metals. The challenges are the finding of tertiary alloys withimproved properties and the synthesis of thin membranes.

    1.3. Hydrogen permeation through palladium membranes

    The mechanism of hydrogen permeation through palladiummembranes has been studied extensively and it is well knownthat it generallyfollows a solutiondiffusion mechanism. The stepsinvolved in hydrogen transport from a high to a low pressure gasregionarethefollowing( Fig.6): (a)diffusionof molecularhydrogento the surface of the palladium membrane, (b) reversible dissocia-tive adsorption on the palladium surface, (c) dissolution of atomichydrogen into the bulk metal, (d) diffusion of atomic hydrogen

    through the bulk metal, (e) association of hydrogen atom on thepalladium surface, (f) desorption of molecular hydrogen from thesurface, (g) diffusion of molecular hydrogen away from the surface[29].

    Generally, hydrogen permeation is described by the followingequation:

    J=P(pn

    hpn

    l)

    L (1)

    In this equation, Jis the hydrogen flux, P is the permeability,L is the thickness of the palladium layer, ph and pl are the par-tial pressures of hydrogen on the high pressure (feed) side and thelow pressure (permeate) side, respectively, and n is the pressureexponent. The latter generally ranges from 0.5 to 1 depending on

    what step (ag) is the rate-determining step. According to Siev-

  • 8/12/2019 j.memsci.2011.03.057

    5/18

    32 S. Yun, S. Ted Oyama / Journal of Membrane Science375 (2011) 2845

    Fig. 7. Usual range ofnas a function of palladium thickness at the temperature of623773K.

    erts law[7072],when the rate controlling step is bulk diffusionthrough the palladium layer which is step (c), the value of n is0.5 because the diffusion rate is proportional to the concentra-tion of hydrogen atoms on opposite sides of the metal surface andthis hydrogen concentration is proportional to the square root ofthe hydrogen pressure. When mass transport to or from the sur-face (a, g) or dissociative adsorption (b) or associative desorption(e) become rate determining, the expected value of n is 1 sincethere processes depend linearly on the concentration of molec-ular hydrogen. A detailed description is given in Appendix A.Anexponent of unity suggests that permeation through the palladiumis very fast and usually indicates that the palladium layer is thin,less than 5m(Fig. 7).However, n-values of 0.50.8 have beenreported for thin (5m) Pd layers,n-values of greater than 0.5 can beattributed to defects or pin-holes through which a substantial por-tion of the hydrogenpermeates[11,75]. Thiscan beby the Knudsenor Pouiselle flow mechanisms, and gives rise to exponents higherthan 0.5 (Appendix A). Guazzone et al. [75] described the contribu-tion of different transport mechanisms such as solution-diffusion(giving rise to Sieverts law), Knudsen diffusion or viscous flow tothe total flux. Measurement of the helium flux was used to calcu-late theportion of Knudsen and viscous flows and thetotalflux wasassumed to be the sum of these flows and the solutiondiffusioncontribution.

    A recent detailed analysis by Barbieri and coworkers[76]con-siders cases where there is no single rate-determining step, andit is found that pressure differences and temperature can affect

    the exponent, particularly when adsorption and desorption areimportant. For the particular case of low temperatures and thinmembranes the exponent can be lower than 0.5. When membranetransportis fast as with thin membranesit is expected that externaltransport limitations will become important, giving rise to con-centration gradients in a narrow region (external film) around themembrane. The presence of these gradients is known as concen-tration polarization, and maps have been developed that indicatewhen they will be important [77]. However, even with thick mem-branes where the expected exponent is n = 0.5 deviations can befound[78].As pointed out by Barbieri and coworkers[76]exper-iments to measure n should use materials where the selectivityis infinite to discount the influence of mechanisms such as Knud-sen diffusion on the exponent. In addition, they point out that the

    activation energycan give an indication of the limiting process. Dif-

    Support

    Support

    Intermediate layer

    Pd nuclei

    Support

    Intermediate layer

    Pd layer

    Surface smoothing

    by dipcoating with boehmite sol

    Pd seeding

    by sensitization and activation

    Pd plating

    Intermediate layer

    Fig. 8. Conventional electroless plating procedure.

    fusion through bulk Pd has a low activation energy (

  • 8/12/2019 j.memsci.2011.03.057

    6/18

    S. Yun, S. Ted Oyama / Journal of Membrane Science375 (2011) 2845 33

    Fig. 9. Apparatus for preparing tubular Pd composite membranes by means offorced-flow chemical vapor deposition adapted from[89].

    include ease of coating on materials having any shape, low cost,and use of very simple equipment[86].The drawback of ELP is thecomplicated and time-consuming nature of the method due to therequirement of a number of pre-treatment steps such as activationand sensitization before final plating of the desired metal can becarried out. A particular challenge in the formation of Pd alloys istheunequal reductionpotential of themetallicions which results in

    uneven deposition. This was observed for the case of PdAg alloyswhere dendritic growth occurred[87].This could be overcome byadjusting reactant concentrations (metal/hydrazine ratios) so as tolower the reduction overpotential of Ag to make it closer to Pd.However, considerable effort is needed to find the right conditionsfor preparation.

    2.2. Chemical vapor deposition (CVD)

    Chemical vapor deposition (CVD) is a method for obtaining thinfilms on a substrate by the thermal decomposition of one or sev-eral volatile precursors on the near surface of the substrate. Thismethod is an attractive technique for obtaining Pd layers of con-trolled thicknesses[85].It was first reported as a method for the

    formation of a Pd composite membranes using PdCl2as the metalsourceby Yeet al. [88] whoused an-Al2O3disk as a support. Itohetal. [89] suggested an apparatus for preparingtubular Pd compos-ite membranes by means of forced-flow CVD (Fig. 9).The obtainedPd membrane had a selective layer of 24m of Pd with a H2/N2selectivity of 5000.

    Compared with ELP, the CVD technique often results in betterfilm quality control so that very thin (

  • 8/12/2019 j.memsci.2011.03.057

    7/18

    34 S. Yun, S. Ted Oyama / Journal of Membrane Science375 (2011) 2845

    Table 2

    Bulk Pd or Pd-alloy membranes for H2separation.

    Membrane type L(m) T(K) P(kPa) n H2permeance107 molm2 s1 Pa1

    H2permeabilityBarrera

    H2/N2 Ref.

    Pd disk by ELP 7.2 673 120 0.5 15.6 33,500 40,000 [11]Pd disk by ELP 12.1 673 120 0.5 9.30 33,600 172 [11]Pd59Cu41disk by ELP 16.7 673 120 0.5 10.1 50,600 105 [11]Pd disk 25 623 517 0.5 4.58 34,200 [98]Pd60Cu40disk 25 623 517 0.5 6.69 49,900 [98]

    Pd60Cu40disk 25 673 517 0.5 7.56 56,400 [98]Pd disk from Pd sheet 1000 623 91 0.5 0.76 2.3105 [99]

    a 1 Barrer= 3.351016 molm1 s1 Pa1.

    of conducting support materials such as stainless steel, and cannotbe generally applied.

    3. Palladium-based membranes

    Palladium-based membrane systems reviewed in this chaptercan be broadly divided into two categories: unsupported Pd mem-branes and supported Pd membranes, both of which include purePd and Pd alloy materials. Unsupported Pd membranes are usedin a stand-alone form, usually as foils or tubes without any aux-

    iliary component to give structure or strength for sustaining thetop metal layer. Supported Pd membranes utilize porous materi-als such as Vycor glass, ceramics, or stainless steel to provide astructural scaffold for holding the metallic component.

    3.1. Unsupported membranes

    As reported in a 2002 reviewby Paglieri andWay,earlyresearchin the United States and the former Soviet Union employed Pd orPd alloy membranes with a tubular geometry with thickness in therange of 20100m for structural integrity[4].This type of mem-branes produced ultra high purity H2for use in the semiconductormanufacturing industry or in the recovery of H2isotopes. Unfortu-nately, thehigh cost of thePd material imposed limitson economic

    practicality in comparison with other separation methods [97].Nevertheless, the method remains of importance for small scaleproduction and for research applications to evaluate the intrinsicmaterial properties of pure Pd or Pd alloys. Generally in this typeof membranes, the rate-determining step in H2permeation is bulkdiffusion, so the partial pressure exponent value is 0.5. The per-meance of theses membranes is rather low, with values around1.41089.3107 molm2 s1 Pa1 due to the high membranethickness of at least 20m required for mechanicalstability [28]. Inearly work Mckinley[98]described a diffusion barrier formed by aPd-alloy, which had an improved practical permeability and phys-ical strength in hydrogen diffusion system compared with purePd. He reported that a diffusion layer with 25 m thickness, com-posed of 60wt% of Pd and 40% of Cu showed H2 permeance of

    6.6910

    7 molm

    2 s

    1 Pa

    1 at 623 K, which is 1.5 times higherthan pure Pd.

    A summary of the properties of bulk Pd membranes is pro-vided inTable 2.In this review the permeance and permeabilityvalues from various studies have been converted into units ofmolm2 s1 Pa1 and Barrer, respectively using the thickness ofthe separation layer and partial pressure difference of hydrogen sothat reported values can be compared. Reported values of n in theliterature are indicated inTables 24 and 6.In most studies, thepermeate side (downstream) was at atmospheric pressure and thedeltaPwas 101kPa, but actual reported values were used in thecalculations of permeance.

    Recently, Way and coworkers[11]reported the preparation ofa defect-free unsupported Pd film as thin as 7.2m with H2 per-

    meance of 1.56106

    molm2

    s1

    Pa1

    and H2/N2 selectivity of

    40,000 at 673K. They prepared a Pd disk by an ELP technique ona mirror-finished stainless steel support followed by mechanicalremoval of the membrane. Morreale et al. [99]reported the per-meability of H2 in bulk Pd membranes at conditions of elevatedtemperature in the range of 6231173K and high H2 pressure upto 2.8 MPa. They found that the fitted exponent value of n is 0.62instead of 0.5 at high pressure and proposed that the increase maybebecauseeitherthebulkdiffusion-limitedmechanismisnotoper-ating or because the Sieverts constant and diffusion coefficientchanges with increasing pressure. It is likely that gas-phase mass

    transfer resistance is the cause of the result, as itis expected togrowwith pressure. The evaluated activation energy (Ea) of bulk Pd wasin the range of 1220 kJ/mol and was affected by such factors asthe thickness of the Pd layer, the preparation method in the bulkPd membranes, the grain microstructure of the layers, and exper-imental conditions such as temperature and pressure [29]. Theapparent activation energy is low because it reflects the temper-ature dependence of the entire transport process, including masstransport.

    3.2. Supported membranes

    As mentioned earlier, self-supporting palladium membranesneed to be thick for a sufficient mechanical strength, so such

    membranes have not only a high intrinsic material cost but alsoa low hydrogen flux. Supported membranes can be preparedwith much thinner palladium layers leading to lower expense,so considerable efforts have been expended to develop methodsof preparation using supports. Different supports and depositionmethods have been tried so as to obtain thin palladium filmswith good membrane integrity along with high hydrogen perme-ance and selectivity[100102].Supported palladium membranesreviewed in this chapter can be divided into three broad categoriesbased on the supporting materials used: Vycor glass, ceramics andstainless steel.

    3.2.1. Porous Vycor glass supports

    Vycor glass is a borosilicate glass whose typical starting com-

    position is 63% SiO2, 27% B2O3, 7% Na2O3, and 3.5% Al2O3, fromwhichthe boron has been leached outby a treatmentof a hotdiluteacid solution to leave a regular porous network with pores gener-ally 4300 nm[103]and a final composition of 7580% SiO2, 46%Al2O3and 1012% B2O3[104]. This support material has high tem-perature and thermal shock resistance, and about 28% of porosity,but is mechanically fragile. Porous Vycor glass was one of the firstsupport materials used in the fabrication of composite Pd mem-branes [82,105].A summary of the properties of Pd membranesusing Vycor glass supports is given inTable 3.

    Uemiya et al.[104]first proposed a composite membrane con-sisting of a thin Pd film deposited on a porous glass tube in 1991.The Uemiya group examined thin Pd membranes supported onporous glass tubes using an ELP method and studied the effect of

    copper or silver addition to suppress the problem of H2embrittle-

  • 8/12/2019 j.memsci.2011.03.057

    8/18

    S. Yun, S. Ted Oyama / Journal of Membrane Science375 (2011) 2845 35

    Table 3

    Palladium or palladium-alloy membranes supported on Vycor for hydrogen separation.

    Laye rs on support Method L(m) T(K) P(kPa) n H2 permeance107 molm2 s1 Pa1

    H2permeabilityBarrer

    H2/N2 Ref.

    Pd ELP 15 473 10 1 220 985,000 7 [86]Pd ELP 13 773 194 0.5 9.64 37,400 Higha [104]Pd ELP 27 673 194 0.5 4.44 35,800 High [104]Pd94Cu6 ELP 19 673 194 0.5 2.04 11,600 High [104]Pd93Ag7 ELP 22 673 194 0.5 3.57 23,000 High [104]

    Pd/glass layer withpore-filled Pd ELP 723 100 1 0.49 520 [106]

    a Reported as a high value in the literature.

    ment.TheH2 permeanceof thepurePd membrane,the PdAgalloy,and the PdCu alloy at 673K with a pressure difference of 0.2 MPawere 4.44107 molm2 s1 Pa1, 3.57107 molm2 s1 Pa1,and 2.04107 molm2 s1 Pa1 respectively. Although, the H2permeance of the Pd alloy membranes was lower compared withthe pure Pd membranes, the H2/N2 selectivity remained high at473 K, much lower than the critical temperature of 571 K, whichindicates that the addition of Cu or Ag inhibited the formation ofthe -phase Pd hydride. Kuraoka et al. [106] prepared Pd com-

    posite membranes using Vycor glass as a supporting materialby inserting Pd into the glass pores under vacuum to increasethe adhesion between the support and the top Pd layer. The H 2permeance of their Pd-glass composite membrane at 723 K was0.50107 molm2 s1 Pa1, which was almost the same as thatof the substrate, and the H2/N2selectivity was500. Altinisik et al.[86]prepared Pd composite membranes using porous glass tabletsassubstratesbyELP.TheH2 permeanceof their membranewas veryhigh 2.2105 molm2 s1 Pa1 at473K but the H2/N2selectiv-ity was 7, just slightly higher than the Knudsen selectivity of 3.7,which could be explained by the presence of large pore defects.

    3.2.2. Porous ceramic supports

    Porous ceramic supports are a broad class of non-metallic

    substrates which can be formed into a variety of shapes with con-trollable pore sizes of 5 200nm. Alumina is the most commonceramic materials used for Pd composite membrane preparationbecause it is widely available in different compositions, has goodmechanical and thermal stability, and can be modified through theuse of intermediate layers. Tubular geometry is popular, but disksare often used as well. The simplest form has a symmetric struc-ture and is composed of a single and uniform wall of a materialwith nano-sized pores. To obtain sufficient mechanical strength,thesingle-walledsymmetric structureusuallyneeds a considerablethickness which reduces gas permeation. More advanced aluminasupports have an asymmetric structure[107],composed of a thin

    selectivelayerwithasmallandnarrowporesizedistributionplacedon the surface of a bulk layer with larger particles. These asymmet-ric supports are good substrates for Pd membranes due to theirsmooth outer surface and low gas flow resistance, but the multi-step fabricating process makes them relatively expensive[4].

    Defects and roughness in the support usually produce defectsor pinholes in the deposited top layer[107].Consequently, propersupports should have homogeneous surface characteristics, a nar-rowporesize distribution, and a particle size range smaller than the

    top layer thickness. To suppress pinholes and to obtain a narrowpore size distribution on the supports while preventing a decreaseof permeance, Oyama et al. proposed the use of very thin multiplegraded layers of alumina deposited on various porous substrates(Fig. 12)[108].

    On the other hand, Abate et al. [109] reported that surfaceroughness of the underlying alumina support is required for goodmechanical stability of the Pd alloy membrane and prevent thepeeling or formation of cracks during use. They prepared PdAgalloy membranes by an ELP method directly on an alumina sub-strate with average pore size of 70nm and surface roughness ofseveral microns. The H2 permeance of their membrane at 623Kwas 1.2106 molm2 s1 Pa1 with a H2/N2 selectivity of 450.The gas separation properties of recently published Pd andPd alloy

    composite membranes supported on -alumina are summarizedinTable 4.Wang et al. [110]prepared PdAgRu composite membranes

    supported on porous-alumina using a simultaneous ELP method.The hydrogen permeance of their ternary alloy membranes was34times higher than that of pure palladiummembranes of similarthickness.

    Although this composition has higher permeance than pure Pd(Table 1), the permeance is too high. Since the hydrogen selectivitywas not reported, the result could be due to defects.

    Xomeritakis and Lin [85] prepared thin palladium/aluminamembranesbyCVDusingpalladiumacetylacetonateandpalladium

    Fig. 12. Demonstration of the effect of graded layers adapted from[108].

  • 8/12/2019 j.memsci.2011.03.057

    9/18

    36 S. Yun, S. Ted Oyama / Journal of Membrane Science375 (2011) 2845

    Table 4

    Palladium or palladium-alloy membranes supported on alu mina for hydrogen separation.

    Layers on s upport Method L(m) T(K) P(kPa) n H2 permeance107 molm2 s1 Pa1

    H2permeabilityBarrer

    H2/N2 Ref.

    Pd/-Al2O3withPd/-Al2O3b

    ELP 6 753 100 1 26.0 46,600 2100a [29]

    Pd CVD 1 723 68 1 20.6 6100 780 [39]PdNi alloy CVD 1 723 68 1 20.6 6100 317 [39]Pd/-Al2O3b CVD 1 573 1 2.0 600 200c [85]

    Pd CVD 2 573 30 1 33.4 19,900 5000 [89]Pd/-Al2O3packedwith YSZ

    ELP 5 723 280 0.5 12.5 18,700 High [102]

    Pd77Ag23 ELP 12 623 50 0.5 12.0 43,000 450 [109]Pd69Ag30Ru1 ELP 15 673 75 0.5 73.3 328,000 [110]Pd70Ag30 ELP 13 673 75 0.5 20.0 74,600 [110]Pd ELP 6 673 75 0.5 6.67 11,900 [110]Pd/polymer ELP 5 773 100 1 33.3 49,700 c [111]Pd/-Al2O3b ELP 5 780 413 0.5 12.6 18,800 600 [112]Pd88Ag12/-Al2O3b ELP 11 823 413 0.5 12.1 39,700 2000 [112]Pd/-Al2O3with Pd ELP 1 723 100 1 106 31,600 23 [113]Pd/-Al2O3packed

    with Pd/-Al2O3bELP 2.6 643 413 0.5 4.84 3800 3000 [114]

    -Al2O3/-Al2O3packed withPd/-Al2O3b

    ELP 5 573 413 0.5 13.6 20,300 1000 [115]

    Pd/Sil-1 ELP 5 773 100 1 18 26,800 1300 [116]Pd/anodized

    aluminadELP 1 673 101 1 0.74 200 200 [117]

    a H2/Ar ratio.b Intermediate layer.c H2/He ratio.d Substrate.

    chloride as Pd precursors. They reported that the H2 permeancewas dependenton the crystallinity andmorphology of thePd metaldeposited. The highest H2permeance andH2/He selectivity of theirmembranes were 2.0107 molm2 s1 Pa1 and 200 at 573K,respectively.

    Zhang et al.[29]fabricated thin and dense palladium compos-ite membranes with defect free selective layers using a vacuumassisted electroless plating (VELP) method. They claimed the vac-

    uum was effective because the nitrogen gas produced during theplating procedure could be removed from the surface of the mem-brane where it could cause defects in the palladium film (Fig. 13).The hydrogen permeanceof their membrane prepared by VELP was

    2.6106 molm2 s1 Pa1 with a high H2/Ar selectivity of over2000 at 753 K.

    Tong et al.[111]produced dense Pd membranes without inter-mediate layers by pre- depositing Pd by ELP on a polymer layerwhich was dip-coated on an alumina substrate and then removingthe organic layer with a high temperature (8731273K) treatmentunder a stagnant air atmosphere followed by ELP to obtain a selec-tive layer (Fig. 14). They proposed that Pd nuclei could be thus

    uniformly distributed on the non-porous polymer layer, which isrequired for forming a defect-free Pd layer. After removal of theorganic layer, the presence of adequate space between the Pd layerand the support was reported as a key factor for durability of the

    Fig. 13. SEM photos of top surface of Pd membranes prepared: a) by ELP, b) by VELP[29]. (Reprinted with permission.)

  • 8/12/2019 j.memsci.2011.03.057

    10/18

    S. Yun, S. Ted Oyama / Journal of Membrane Science375 (2011) 2845 37

    Fig. 14. Schematic of a polymer-inorganic process for preparing a Pd membrane on a porous substrate. Cross sections: a) substrate, b) polymer layer + substrate, c) Pdlayer+ polymer layer+ substrate, d) Pd layer + space+ substrate, and e) Pd separation layer+ space+ substrate adapted from[111].

    Pd layer because it provide room for the Pd to expand during ther-mal cycles without stress. The H2permeance of their membrane at773K was 3.3106 molm2 s1 Pa1 and it was impermeable toN2.

    Nair et al. [112] prepared Pd or PdAg alloy membranesof controlled thickness by monitoring the amount of depositedmetal by a quartz crystal microbalance with a resolution of0.01g/cm2 metal loading (Fig. 15). Their apparatus enabled themto simultaneously obtain several Pd or Pd alloy membrane sam-

    ples, which were expected to have uniform properties. The H2permeance of their PdAg membranes of thickness 11m was1.2106 molm2 s1 Pa1 with a good H2/N2selectivity of 2000.

    To address the problem of mechanical failure of compositemembranes due to differences in thermal expansion coefficientsbetween supporting materials (alumina, 6.7106 K1) and thePd layer (11.8106 K1), several techniques were developed. Forexample, Zhao et al.[113]proposed a modified electroless platingmethod involving the use of an activated Pd(II)-modified boehmitesol prepared by a solgel process. They reported that this improvedthe adherence between the substrate layer and the selective layer.Their membrane had around 1m thickness and showed high H2permeance of1106 molm2 s1 Pa1 but only a H2/N2 selec-tivity of 23 at 723 K. The group of Harold prepared encapsulated Pd

    membranes by dip-coating-alumina layers on a thin Pd layer andthen depositing Pdby anELP technique toobtain a Pdselectivelayerwhich was formed inthe pores of-alumina [114]. The Pd particlespackedintheporesofthe -aluminalayer appeared to suppress thedegradation of theselective layer during temperature cycling. Theirmembrane had good thermal resistance with a high H2/N2selectiv-ityof3000butalowH2permeance of 4.8410

    7 molm2 s1 Pa1

    at 643K due to blockage of pores or reduction of the pore size inthe intermediate layers.

    The group of Suzuki fabricated composite membranes packedwith Pd nanoparticles employing an ELP technique under vacuumon-alumina layersbeneath Pdseed layerswithuse of Pd(OAc) a sa

    Fig. 15. Schematic diagram for conducting electroless plating to obtain multiple

    palladium membranes with a uniform quality adapted from[112].

    Pd precursor and hydrazine as a reducing agent (Fig. 16) [115]. Thedispersion of fine Pd particles in the-alumina layer could preventsmall defects from growing into largecracksandcouldsuppressthestress associated with the phase transition. Their membraneshowed a high H2permeance of 1.36106 molm2 s1 Pa1 anda selectivity of 1000 at 573 K.

    Zhang and coworkers prepared stable Pd membranes by grow-ing a Sil-1 zeolite intermediate layer on an -alumina supportand then depositing a 5m layer of Pd by ELP [116]. The best

    membrane had permeance of 1.8106 molm2 s1 Pa1 and atop H2/N2 selectivity of 1300 at 773K with a Pof 100kPa andn = 1, Under long term gas cycling conditions the selectivity was615 at 623 K and 506 at 473 K, but the membrane was stable forthe 240 h duration of the measurements. Seshimo et al. preparedPd membranes by ELP on an anodized alumina substrate usingsupercritical conditions with CO2 and an emulsifier [117]. Theyreport a membrane thickness of 1m, but the permeance wasonly 7.4108 m2 s1 Pa1 andaH2/N2selectivity of 200 at 673Kwith a Pof 101kPa. The low permeance is probably due to thesubstrate, as a membrane prepared by conventional ELP showedsimilar hydrogen fluxes.

    Alumina tubes arealso availableas hollowfibers, andthese havebeenusedassubstrates.Thehollowfibersareproducedbyanextru-

    sion technique in which a polymer-inorganic composite is firstformed and subsequently fired. The resulting fibers are typicallyof 0.72 mm OD and an average pore size of 0.10.5m.

    A recurring problem with Pd membranes is the occurrence ofpinholes, which have the effect of reducing selectivity. A methodof healing pinholes with ELPis the feeding of the palladiumspeciesandthe reductantfrom opposite sides of the membrane [118]. Thiscausesthe deposition of metal precisely where thepinhole is found.However, the absence of reductant on the palladium side couldresult in overly thin membrane layers in some regions.

    3.2.3. Porous metal supports

    Porous metal supports are conductive substrates that can beformed into various shapes with average pore size of 0.2100m.

    Theyare conventionally produced by a powder sinteringor electro-chemical deposition andare good candidates for support materialsbecause their thermal expansion coefficient is similar to that ofPd compared to other materials (Table 5)[119,120]. In addition

    Table 5

    Thermal expansion coefficients of support materials and palladium at 293 K.

    Material Linear thermal expansioncoefficient (106 K1)

    Alumina 5.46.7Borosilicate glass 3.3Steel 1113Stainless steel 1116Palladium 11.8

  • 8/12/2019 j.memsci.2011.03.057

    11/18

    38 S. Yun, S. Ted Oyama / Journal of Membrane Science375 (2011) 2845

    Fig. 16. Preparation of composite Pd membrane by packed Pd into pore of-alumina layer adapted from[115].

    it is possible to obtain easy and sturdy sealing in industrial assem-bliesthan withmore fragile ceramic supports [121,122]. Among theporous metals, stainless steel is the most frequently used materialdue to its ease of fabrication, chemical resistance, and low cost.

    The smallest pore size of commercially available stainless steeldisks or tubes is 0.2m[4].Rough surfaces from large pore sizesor non-uniform pore size distributions often result in defects orpinholes on the surface of thin Pd or Pd alloy membranes[11].Forthis reason, porous metal supports require much thicker Pd filmscompared to ceramic supports to obtain defect-free selective lay-ers. Shu et al. [123] reported that at least 15m of metal depositedby anELP method is required to obtaina dense and defect-free filmwhen using 0.2m porous stainless steel supports. Mardilovich

    et al.[124]showed that the Pd layer needs to be approximatelythree times as thick as the diameter of the largest pore to obtain aHe-tight membrane.

    The roughness and the defects from the macroporous metalsupports can be minimized by shrinking the pore size by employ-ing various methods such as mechanically altering the surface,abrading the surface, sintering at high temperature or depositinginexpensive metals such as Ni or Cu instead of Pd [96,125127].

    Jemaa et al.[126]prepared Pd membranes supported on porousstainless steel disks by an ELP method. They modified the metalsubstratesurface with56mporesizeopeningsbybombardmentof iron shots (shot peening treatment) with an average diame-ter of less than 125m under a release pressure of 5.1 105 Pafollowed by a purge with air to reduce the pore openings on the

    metal surface. They claimed that the original pore openings werereduced to around 1m after a 300s treatment. The H2 perme-ance of their membranes was 5.8 107 molm2 s1 Pa1, whichindicated that gas permeance could be significantly affected by themodificationprocess.NamandLee [96] fabricatedPdNi alloycom-posite membranes supported on a stainless steel disk by vacuumelectrodeposition. They pretreated a 316L stainless steel substratewith average pore size of 500nm by dispersing a submicron Nipowder on the substrate followed by sintering to make the sub-strate surface smooth and enhance adhesion between the topPdNi alloy layer and the substrate. They claimed that defects orpin-holeswere notobservedon the surface of thepretreated mem-brane after deposition of 1m PdNi, while macrospores on thesurface of the unpretreated membranes still existed after deposi-tion of 10m PdNi. The H

    2permeance of their membranes was

    6.76106 molm2 s1 Pa1 witha highH2/N2selectivity of 3000at 723 K. Membrane performances of recently published Pd or Pdalloy composite membranes supported on porous stainless steelwith different preparation methods are summarized (Table 6).

    Tong et al. [128] prepared Pd membranes with thickness of23m by electroless plating of Pd or Pd/Ag layers with a thindiffusion barrier of silver by electro plating on a stainless steelsubstrate filled with aluminum hydroxide gel. They reported thehydroxide gel in the pores did not hinder hydrogen diffusiondue to the significant shrinkage of the gels under thermal treat-ment up at 773K. Their best membrane showed H2permeance of1.7106 molm2 s1 Pa1 with no N2permeance at 673 K.

    The group of Huang produced Pd membranes employing a

    porous 310L stainless steel tube with a 0.5m pore size asa substrate [129]. They coated a porous yttria-stabilized zir-conia (YSZ) layer by atmospheric plasma spray (APS) on theoutside of the metal surface, conducted a MOCVD activationstep using Pd(II)-hexafluoroacetylacetonate as a Pd precursor,and then made a selective Pd layer by an ELP method. Theyclaimed that their MOCVD activation technique was effective forobtaining nanometer-sized Pd particles evenly distributed on thesupport, which resulted in an impervious metal layer in the plat-ing step. However, the H2 permeance of their membrane was7.5107 molm2 s1 Pa1 at 673K with a H2/N2 selectivity of700, which showed that defects or areas uncovered by Pd couldbe present due to the rough porous metal surface. The group ofNam prepared PdCu alloy composite membranes by a vacuum

    assisted EPD using 316L stainless steel disks as supports [121].They used a thin intermediate layer of silica prepared by a solgelmethod to smooth the surface of the mesoporous stainless steelsupports and to improve the structural stability of the Pd alloycompositemembranes. TheH2permeanceof theirmembranes was4.9106 molm2 s1 Pa1 with a high H2/N2 selectivity of over10,000 at 623K.

    Various studies on factors such as preparation time, concen-trations, and temperature of plating baths that can affect thecompactness or grain andcrystallite size of the Pd layers on porousmetals have been conducted. For example, Shi and Szpunar [130]reported the effect of concentration of PdCl2 in a plating bath onPd nanoparticles deposited on the surface of a 316L stainless steelplate and proposed a route to prepare Pd membranes with a verythin selectivelayerof around400nm thickness.The Pd particle size

  • 8/12/2019 j.memsci.2011.03.057

    12/18

    S. Yun, S. Ted Oyama / Journal of Membrane Science375 (2011) 2845 39

    Table 6

    Palladium or palladium-alloy membranes supported on porous metal for hydrogen separation.

    Layers on support Method L(m) T(K) P(kPa) n H2permeance107 molm2 s1 Pa1

    H2permeabilityBarrer

    H2/N2 Ref.

    PdNi0.20.3/Ni powder CVD 1 723 68 1 22.1 6600 >400 [39]PdNi alloy/Cu/Ni

    powderEPD 2 723 55 1 67.6 40,400 3000 [96]

    PdCu alloy/Silica/Nipowder

    EPD 2 623 53 1 48.8 29,100 10,000 [121]

    Pd ELP 20 623 100 0.5 5.0 29,900 5000 [122]Pd ELP 6 673 100 0.5 5.8 10,400 [126]Pd/filled with

    aluminum hydroxidegel/Ag

    ELP 3 673 100 1 17.0 15,200 a [128]

    PdAg/3m Pd/filledwith aluminumhydroxide gel/Ag

    ELP 2 673 100 1 15.0 13,400 a [128]

    Pd/YSZb by APSc ELP 6 673 40 0.5 17.5 31,300 100 [129]Pd with MOCVD

    activation/YSZ byAPS

    ELP 7 673 40 0.5 7.5 15,700 700 [129]

    Pd ELP 0.4 773 276 1 6.5 800 [130]Pd/Cr2O3 ELP 32 723 300 0.5 3.5 38,600 106d [134]Pd/YSZ ELP 11 773 82 0.5 9.7 31,900 [135]Pd/WO3 ELP 12 723 0.5 2.1e 10,000 [136]Pd/NaA ELP 19 723 50 0.5 15.8 89,600 608 [137]

    a Reported as a defect-free Pd or Pd-alloy layer in the literature.b Yttria-stablized zirconia.c Atmospheric plasma spray.d H2/He.e 108 molm2 s1 Pa0.5.

    depositedon thesurface wassmaller withincreasing concentrationof PdCl2(Fig. 17)and the number of layers of the obtained Pd layerincreased as to the crystallite size of Pd decreased. Theirmembranewhich was prepared using 1.82g/l of PdCl2concentration showed

    a H2permeance of 6.5107 molm2 s1 Pa1 at773Kbutdidnotmention the H2selectivity.

    Intermetallic diffusion can be a problem with Pd membraneson metallic substrates, and intermediate layers have been used

    Fig.17. SEMimages showing palladium nanopaticlesdeposited on thestainlesssteel substrateat differentconcentrations of PdCl2a) 4.2g/l,b) 3g/l,c) 2.4g/l,andd) 1.82g/l

    [130]. (Reprinted with permission.)

  • 8/12/2019 j.memsci.2011.03.057

    13/18

    40 S. Yun, S. Ted Oyama / Journal of Membrane Science375 (2011) 2845

    b

    2520151050

    101

    102

    103

    104

    H2

    /N2

    Selectivity

    Top layer thickness / m

    a

    3025201510501E-7

    1E-6

    1E-5

    1E-4

    1

    2

    H2

    Permeanc

    e/molm

    -2Pa

    -1s

    -1

    Top layer thickness / m

    Unsupported

    Vycor

    Alumina

    Metal

    1 low selectivity2 no selectivity

    reported

    Fig. 18. Performances of Pd membranes supported on different materials. a) H2permeance versus top layer thickness, b) H2/N2selectivity versus top layer thickness.

    as diffusion barriers [131133]. The group of Ma reports thatoxidation of a stainless steel (316) substrate produces a nativelayer of chromium oxide that serves as an effective diffusion

    barrier [134]. Their best membrane showed H2 permeance of3.5107 molm2 s1 Pa1 with a very high H2/He selectivity at723K. Zhang et al. found that a layer of solgel derived yttria-stabilized zirconia (YSZ) is a more effectivebarrier than such nativeoxide layers at temperatures above 600 C (873K) [135]. Theyreported that theminimumthickness of a Pd layer required forgas-tightnessis11monanYSZ/PSSsupport,and25monanoxidizedPSS support. Their Pd membrane with the YSZ layer as a diffu-sion barrier showed a permeance of 9.8107 molm2 s1 Pa1

    at 773 K with thermal stability to 923 K. Zahedi et al. report thatWO3forms a good intermediary layer on stainless steel to producea Pd membrane of thickness 12m by electroless plating [136].The membrane has permeanceof 2.1108 molm2 s1 Pa0.5 withan exponent n =0.5 and a N2/H2 selectivity of 10,000 at 723K.

    Cornagliaand coworkers show that a NaAzeolite is an effectivedif-fusionbarrier to produce a Pd membrane of thickness 19m [137].The membrane has permeanceof 1.6106 molm2 s1 Pa1 withan exponentn = 0.5 and a N2/H2selectivity of 608 at 723K.

    4. Performance analysis of palladium membranes

    Overall results are presented in this section to illustrate howthe gas separation properties of these membranes depend on thefabrication methods and supporting materials. To determine theeffect of the supporting materials used in the preparation of the Pdmembranes, the H2 permeance and H2/N2 selectivity was plottedas a function of separation layer thickness (Fig. 18).

    As can be seen inFig. 18a,the H2permeance tends to decrease

    with increase in the selective top layer thickness for each supportclass, which favors the preparation of membranes with as thin aPd layer as possible to obtain high H2permeance and low cost. Thetwo systems which displayed unusually high H2 permeance hadeither a very low H2/N2selectivity or no selectivity was reported.The dashed line indicates a rough upper boundary to the perme-ance data, which highlights the general observation of decreasingpermeance with increasing Pd thickness. Membranes with a topPd layer thickness of less than 5 m displayed a broad range ofH2permeance from 1.110

    7 to 1.0105 molm2 s1 Pa1. Thisindicatesthatthe H2permeancedoes notdepend only on thethick-ness of the Pd layer but also on the structure and thickness of boththe Pd layer and the support. It is also noted that most unsup-portedmembraneshadamuchhigherthicknessthanthesupported

    membranes which resulted in lower H2permeance for the former.

    Fig. 18bshows that the H2/N2 selectivity dependence on thethickness of the separation layer is highly scattered. In Pd mem-branes, the selectivity can be infinite if the Pd layer does not have

    defects or pin-holes. The probability of obtaining these faultlessmembranes generally increases with the layer thickness becausepinholes or other defects will be covered [111]. The selectivity wastherefore expected to be proportional to the top layer thickness,but only a weak dependence was observed. The selectivity shouldbe low in cases where the top Pd layer does not have structuralintegrity regardless of the thickness of the selective layer, becausemolecules will permeate through openings or other defects byKnudsen diffusion or viscous flow (Appendix A).

    The effect of fabrication method is evaluated by plotting H2permeance and H2/N2 selectivity of Pd composite membranes pre-pared by different methods (Fig. 19).Regardless of the preparationmethod, the H2 permeance is inversely proportional to the thick-ness of the Pd layer as obtained for the effect of support (Fig. 18).

    Palladium membranes prepared by EPD or CVD result in lower Pdthickness when compared to those prepared by ELP. Compositemembranes prepared by EPD give superior H2 separation perfor-mance when compared with the membranes prepared by othermethods. However, the EPD method requires conductive supportsand the CVD technique requires volatile and thermally stable Pdprecursors, which limits the usefulness of these methods. For thatreason, extensive efforts have been exerted on the developmentof composite membranes by ELP than by the other methods. Asummary of the results of selectivity versus permeance for variousmethods of Pd membranesare presentedin Fig. 20. RegionA showshigh selectivity and high permeance. Region B shows intermedi-ate values running in a diagonal. Region C shows low selectivityand low permeance. Some of these systems demonstrate H 2 per-

    meance of over 1106

    molm2

    s1

    Pa1

    with H2/N2 selectivityabove1000 (shaded area in Fig. 20), which could represent promis-ing membranes for various commercial applications.

    For membranes in region (B) the H2selectivity is inversely pro-portional to the H2permeance. The H2permeance decreases withthe top metal layer thickness when the structure and morpholog-ical properties of the selective layer and the supports are similar.The H2 selectivity gradually increases as a function of the sepa-ration layer thickness until a non-porous Pd layer is developed onthesubstratesurface. Therefore, H2permeancegenerally decreaseswith H2selectivity increased up to the point where an imperviouslayer is obtained. Interestingly, two regions which show unusualbehavior can be observed.

    Membranes in region (A) with high H2selectivity and high per-meance have a top layer thickness of less than 2 m and thin

  • 8/12/2019 j.memsci.2011.03.057

    14/18

    S. Yun, S. Ted Oyama / Journal of Membrane Science375 (2011) 2845 41

    0 5 10 15 20 25 301E-7

    1E-6

    1E-5

    1E-4

    H2

    Permeance/molm

    -2Pa

    -1s

    -1

    Top layer thickness / m

    CVD

    ELP

    EPD

    a b

    0 5 10 15 20

    101

    102

    103

    104

    H2

    /N2S

    electivity

    Top layer thickness / m

    1

    2

    1 low selectivity2 no selectivity

    reported

    Fig.19. Performance of Pd membranesprepared by differentfabrication methods.a) H2permeanceversus top layerthickness, b) H2/N2selectivity versus top layerthickness.

    intermediatelayers of around 2m with anaveragesub nanometerpore size. The thin layers result in a low resistance to gas per-meation, which leads to high H2 permeance. On the other hand,

    membranes in region (C) with low H2 selectivity and low perme-ance have a thick support layer of around 2mm with an averagepore size of 46 nm or thick intermediate layers of around 10m.Thick supports and intermediate layers create a high resistance topermeation, which results in low H2permeance.

    It is of interest to compare the performance of Pd membraneswith that of polymeric membranes. A comparison of the perme-ation properties of Pd membranes surveyed in this study andpolymeric membranes referred in the literature is given inFig. 21.It is found that regardless of the fabrication method or supportmaterial used for thePd membranes, themajority had performancewellabove thewell-known trade-offline for polymericmembranesreported by Robeson[138]marking the upper bound between theH2/N2selectivity and H2permeation for these materials.

    Polymeric membrane systems generally show low H2 per-meances (below 1108 molm2 s1 Pa1) while Pd membranesshow high permeances (above 1107 molm2 s1 Pa1). Poly-meric membranes are generally non-porous and gas permeationoccurs by the solutiondiffusion mechanism[139].Gas moleculesdiffuse from the feed side to the surface of polymeric membranes

    1E-7 1E-6 1E-5 1E-4

    101

    102

    103

    104

    H2/N2Selectiv

    ity/

    H2Permeance / molm

    -2s

    -1Pa

    -1

    CVD

    ELP

    EPDB

    AC

    Fig. 20. H2/N2 selectivity and H2 permeance of Pd-based composite membranes

    with different fabrication methods.

    10-1

    100

    101

    102

    103

    104

    105

    106

    10-1

    100

    101

    102

    103

    10

    4

    105

    Palladium membrane

    Polymeric membrane

    Upper bound (Robeson plot)

    H2

    /N2

    Selectivity

    H2Permeation / Barrer

    Fig. 21. H2/N2 Selectivity as a function of H2 permeation of different membranecategories.

    where they dissolve and permeate through free-volume elements(0.2 and 0.5 nm in size between the polymer chains) of the mem-brane by random molecular diffusion, followed by desorption anddiffusion into the permeate side. Thus, polymeric membranesexhibitlowH2 permeancewithmoderateH2/N2 selectivity.Despiteextensive research on gas separation, polymeric membranes havenot been able to significantly advance beyond the upper bound ofthe Robeson plot (Fig. 21).This general behavior demonstrates thelimited performance of polymeric membranes for this gas systemcompared to that of Pd membranes.

    5. Comparison of Pd membranes

    For industrial H2 separations, the Pd or Pd alloy layer thick-ness and H2 permeance can be used for a rough estimate of thematerials cost and production yield, respectively. Substrate mate-rial effects were assessed using the two factors (top layer thicknessand H2permeance) for composite Pd membranes prepared by ELP(Table 7 and Fig. 22). The ELPmethod was considered because of itsease of application for any shape or substrate compared with othermethods.

    Membranes using Vycor glass as a substrate require thicker

    separation layers to overcome the low mechanical stability of the

  • 8/12/2019 j.memsci.2011.03.057

    15/18

    42 S. Yun, S. Ted Oyama / Journal of Membrane Science375 (2011) 2845

    Table 7

    H2 permeance and top layer thickness of Pd composite membranes by ELP.

    S upport material P d or Pd alloylayer (m)

    H2permeance(107 molm2 s1 Pa1)

    Vycor glass 1327 0.59.6-Alumina 115 2.033Stainless steel 620 5.018

    0 5 10 15 20 25 301E-8

    1E-7

    1E-6

    1E-5Membrane cost

    Vycor glass

    -Alumina

    Productionyield

    H2permeance/molm

    -2s

    -1Pa

    -1

    Pd or Pd-alloy layer / m

    Stainless steel

    Fig. 22. H2/N2 Selectivity as a function of H 2 permeation of different membranecategories. Comparison of Pd membranes prepared by ELP.

    glass and the difference in the thermal expansion coefficients ofthe materials. Membranes employing porous stainless steel alsoneed a thick Pd or Pd alloy layer to obtain a continuous separationlayer due to the broad distribution of pore sizes and their large sizein the metal substrate. Thick Pd or Pd alloy metal layers result inhigh membrane preparation costs with low H2production yield. It

    is concluded that the most desirable Pd membranes for practicalapplications should be prepared by the ELP technique on porousalumina substrates. This does not consider the costs or ease offabrication, which could still be substantial.

    6. Conclusions

    This review covers palladium-based membrane systems forH2 separation prepared by the electroless plating (ELP), chemi-cal vapor deposition (CVD) and electroplating (EPD) methods. Thefuture of Pd membranes in the industry looks promising becauseseveral composite membranes surveyed here demonstrated goodthermal and mechanical integrity with superior gas properties.The membranes can be classified into two groups, supported and

    unsupported membranes. The supported membranes were sub-divided into three general categories by the type of substrateemployed Vycor glass, ceramic or metal.

    Unsupported membranes are generally thick to have physicalstrength and for this reason exhibit inferior gas separation prop-erties compared to supported membranes because of the highresistance to gas permeation for thick materials. Nevertheless theyare important for the evaluation of the intrinsic properties of Pdmembranes. Supported membranes prepared by both the CVD andEPDmethods need only thin selective Pd layers with high H2selec-tivity because of the structural strength provided by the support.However, the application of these methods is limited for practicalapplications.TheCVDprocessrequiresvolatileandthermallystablePd precursors and that is limiting because commercially available

    Pd precursors are expensive or have very low vapor pressures. The

    EPD method requires conductive supports and this is problem-atic due to the surface roughness or non-uniform pore distributionof the available metal supports. Compared with other membranepreparation techniques, the ELP method allows the easy prepara-tion of Pd membranes using simple equipment and any type ofsupport.

    Membrane structure has a significant effect on hydrogen per-meance and selectivity of Pd membranes. As expected, the H 2permeance is inversely proportional to the Pd layer thickness.The H2 selectivity, however, showed no direct dependence onthe top layer thickness but was affected by imperfections in thelayer such as defects or pinholes on the surface, which are deter-mined not by the thickness but by the quality of the top metallayer.

    Current problems associated with the Pd membranes are H 2embrittlement, poisoining, and thermal and mechanical stability.These problems can be resolved by using Pd alloy layers and/orforming Pd layers with nano-sized crystalline particles. The dura-bility problems associatedwith sulfuror carbon contamination stillneed to be addressed.

    Acknowledgments

    For support of this work the authors acknowledge the Director,National Science Foundation, Division of Chemical, Bioengineer-ing, Environmental, and Transport Systems (CBET) under GrantCBET-084316, the National Energy Technology Laboratory undertheNETL-RUA program,and theKakenhigrant-in-aidKibankenkyuB 22-360,335 from the Japan Ministry of Education, Culture, Sports,Science and Technology (Mombukagakusho).

    Appendix A. Hydrogen permeation equation

    Thefluxofhydrogen(mol/m 2 s) by diffusionthroughdensemet-als is described by Ficks first law (Eq.(2))

    JH2 = DH2CH2 (2)

    where, DH2 is thediffusion coefficient (m2/s) andCis the gradientin concentration (mol/m4).

    For one dimensional diffusion,

    JH2 = DH2CH2L

    (3)

    where, Lis the thickness of the metal layer.When surface reaction or mass transportis rate controlling step,

    Henrys law (Eq.(4))can be used.

    SH =CH2PH2

    (4)

    where, SHis Henrys lawconstant, whichdepends on the solute, thesolvent and the temperature. When the concentration in Henrys

    law is substituted into Ficks first law the result is:

    JH2 = DH2SHPH2L = P

    pH2,h pH2,l

    L (5)

    where, Pis the permeance andpH2,handpH2,lare the partial pres-sures of hydrogen on the high pressure (feed) side and the lowpressure (permeate) side, respectively.

    When diffusionthroughthe bulk metal is rate determining step,Sieverts law (Eq.(6))can be applied. In this case, H2molecules aredissociated into hydrogen atoms prior to the diffusion through themembrane layer.

    SH =CH2

    P1/2H2

    (6)

    where, SHis the Sieverts law constant or solubility.

  • 8/12/2019 j.memsci.2011.03.057

    16/18

    S. Yun, S. Ted Oyama / Journal of Membrane Science375 (2011) 2845 43

    Whenthe concentrationin Sievertslaw is substituted intoFicksfirst law:

    JH2 = DH2SHP

    1/2H2

    L = P

    p1/2H2,h

    p1/2H2,l

    L (7)

    For the case of Knudsen and HagenPouiselle flow the flux isproportionalto pH =pH2,h pH2,l.TheKnudsenflux[140] isgivenby,

    JKH2=

    dpL

    8

    9MRT

    1/2pH2 (8)

    whereL is the thickness of the membrane, is the porosity of themembrane,dpis the pore diameter,is the tortuosity, Ris the gasconstant andMis molecular weight of the diffusing gas.

    The viscous flow HagenPouiselle equation is

    JHPH2 =

    R2

    8LPoPavepH2 (9)

    When these mechanisms occur in conjunction with bulk per-meance through palladium the overall hydrogen flux becomes,

    JTotalH2 =

    1

    (1/JH2

    )+ (1/JK

    H2)+ (1/JHP

    H2)

    (10)

    Then the overall order in hydrogen can be between 0.5 and 1.

    References

    [1] V. Gryaznov, Membrane catalysis, Catal. Today 51 (1999) 391395.[2] J.D. Holladay, J. Hu, D.L. King, Y. Wang, An overview of hydrogen production

    technologies, Catal. Today 139 (2009) 244260.[3] A. Basile, F. Gallucci, S. Tosti, Synthesis, characterization and applications of

    palladium membranes, in: R. Mallada, M. Menndez (Eds.), Membrane Sci-ence and Technology, Elsevier, 2008, pp. 255323.

    [4] S.N. Paglieri, J.D. Way, Innovations in palladium membrane research, Sep.Purif. Rev. 31 (2002) 1169.

    [5] S. Tennison, Current hurdles in the commercial development of inorganicmembrane reactors, Membr. Technol. 2000 (2000) 49.

    [6] J. Shu, B.P.A. Grandjean, A.V. Neste, S. Kaliaguine, Catalytic Pd-based mem-brane reactors: a review, Can. J. Chem. Eng. 69 (1991) 10361060.

    [7] A.C.Andrews,Oxygen, in:C.A. Hampel (Ed.), TheEncyclopediaof theChemicalElements, Reinhold Book Corporation, New York, 1968.

    [8] R.G.Lemus,J.M. MartnezDuart, Updated hydrogenproductioncostsand par-ities for conventional and renewable technologies, Int. J. of Hydrogen Energy35 (2010) 39293936.

    [9] Z. Al-Hamamre, S. Voss, D. Trimis, Hydrogen production by thermal par-tial oxidation of hydrocarbon fuels in porous media based reformer, Int. J.Hydrogen Energy 34 (2009) 823827.

    [10] C.-J. Winter, Hydrogen energy abundant, efficient, clean: a debate over theenergy-system-of-change, Int. J. Hydrogen Energy 34 (2009) S1S52.

    [11] S.K.Gade, P.M.Thoen, J.D.Way, Unsupported palladium alloyfoil membranesfabricated by electroless plating, J. Membr. Sci. 316 (2008) 112118.

    [12] W.C. Lattin, V.P. Utgikar, Transition to hydrogen energy in the United States:a 2006 status report, Int. J. Hydrogen Energy 32 (2007) 32303237.

    [13] J.B. Pearlman, A. Bhargav, E.B. Shields, G.S. Jackson, P.L. Hearn, Modeling effi-ciencyand water balance in PEMfuel cell systems with liquidfuel processingand hydrogen membranes, J. Power Sources 185 (2008) 10561065.

    [14] K. Schoots, F. Ferioli, G.J. Kramer, B.C.C. van der Zwaan, Learning curves forhydrogenproduction technology: an assessment of observed costreductions,Int. J. Hydrogen Energy 33 (2008) 26302645.

    [15] F.A.Lewis, ThePalladiumHydrogenSystem, Academic Press, New York,1967.[16] R. Dittmeyer, V. Hllein, K. Daub, Membrane reactors for hydrogenation and

    dehydrogenation processes based on supported palladium, J. Mol. Catal. A:Chem. 173 (2001) 135184.

    [17] M.P. Gimeno, Z.T. Wu, J. Soler, J. Herguido, K. Li, M. Menndez, Combinationof a two-zone fluidized bed reactor with a Pd hollow fiber membrane forcatalytic alkane dehydrogenation, Chem. Eng. J. 155 (2009) 298303.

    [18] D. Feng, Y. Wang, D. Wang, J. Wang, PdAgAuNi membrane reactor forethoxymethane steam reforming, J. Chem. Eng. Data 54 (2009) 24442451.

    [19] J. Tong, Y. Matsumura, H. Suda, K. Haraya, Experimental study of steamreformingof methane in a thin (6m) Pd-based membrane reactor, Ind.Eng.Chem. Res. 44 (2005) 14541465.

    [20] Y. Bi, H. Xu, W. Li, A. Goldbach, Watergas shift reaction in a Pd mem-brane reactor over Pt/Ce0.6Zr0.4O2catalyst, Int. J. Hydrogen Energy 34 (2009)29652971.

    [21] A. Brunetti, G. Barbieri, E. Drioli, Upgrading of a syngas mixture for purehydrogenproduction in a PdAgmembranereactor, Chem. Eng.Sci. 64 (2009)

    34483454.

    [22] Y. Guo, X. Zhang, H. Zou, H. Liu, J. Wang, K.L. Yeung, Pdsilicalite-1 compos-ite membrane for direct hydroxylation of benzene, Chem. Commun. (2009)58985900.

    [23] L. Shi, A. Goldbach, G. Zeng, H. Xu, Direct H2O2synthesis over Pd membranesat elevated temperatures, J. Membr. Sci. 348 (2010) 160166.

    [24] S.T. Oyama, H. Lim, An operability level coefficient (OLC) as a useful tool forcorrelating theperformanceof membranereactors,Chem.Eng. Sci.151 (2009)351358.

    [25] S.T. Oyama, P. Hacarlioglu, The boundary between simple and complexdescriptions of membranereactors: thetransition between 1-Dand 2-D anal-ysis, J. Membr. Sci. 337 (2009) 188199.

    [26] H. Lim, Y. Gu, S.T. Oyama, Reaction of primary and secondary products in amembranereactor:studies of ethanol steam reforming witha silicaaluminacomposite membrane, J. Membr. Sci. 351 (2010) 149159.

    [27] K. Aasberg-Petersen, C.S. Nielsen, S.L. Jrgensen, Membrane reforming forhydrogen, Catal. Today 46 (1998) 193201.

    [28] A.J.DeRosset, Diffusionof hydrogen through palladium membranes,Ind. Eng.Chem. 52 (1960) 525528.

    [29] X. Zhang,G. Xiong,W. Yang, A modified electroless plating techniquefor thindense palladium composite membranes with enhanced stability, J. Membr.Sci. 314 (2008) 226237.

    [30] R.S. Souleimanova, A.S. Mukasyan, A. Varma, Effects of osmosis onmicrostructure of Pd-composite membranes synthesized by electroless plat-ing technique, J. Membr. Sci. 166 (2000) 249257.

    [31] H. Chen, C. Chu, T. Huang, Comprehensive characterization and permeationanalysisof thinPd/Al2O3composite membranesprepared by suction-assistedelectroless deposition, Sep. Sci. Technol. 39 (2004) 1 4611483.

    [32] Hydrogen from Coal Program, RD&D Plan, External Draft, U.S. Department ofEnergy, Office of Fossil Energy, NETL, September 2010.

    [33] K. Ohira, Y. Sakamoto, T.B. Flanagan, Thermodynamic properties for solution

    of hydrogen in PdAgNi ternary alloys, J. Alloys Compd. 236 (1996) 4249.[34] H.C. Rogers, Hydrogen embrittlement of metals: atomic hydrogen from a

    variety of sources reduces the ductility of many metals, Science 159 (1968)10571064.

    [35] J.P. Collins, R.W. Schwartz, R. Sehgal, T.L. Ward, C.J. Brinker, G.P. Hagen, C.A.Udovich, Catalytic dehydrogenation of propane in hydrogen permselectivemembrane reactors, Ind. Eng. Chem. Res. 35 (1996) 43984405.

    [36] Y. Cheng, K. Yeung, Palladiumsilver composite membranes by electrolessplating technique, J. Membr. Sci. 158 (1999) 127141.

    [37] K.J. Bryden, J.Y. Ying, Nanostructured palladiumiron membranes for hydro-gen separation and membrane hydrogenation reactions, J. Membr. Sci. 203(2002) 2942.

    [38] D. Fort, J. Farr, I. Harris, A comparison of palladiumsilver andpalladiumyttrium alloys as hydrogenseparation membranes,J. Less-Comm.Met. 39 (1975) 293308.

    [39] C.-S. Jun, K.-H. Lee, Palladium and palladium alloy composite membranesprepared by metalorganic chemical vapor deposition method (cold-wall),

    J. Membr. Sci. 176 (2000) 121130.

    [40] S. Uemiya, T. Endo, R. Yoshiie, W. Katoh, T. Kojima, Fabrication of thinpalladiumsilver alloy film by using electroplating technique, Mater. Trans.48 (2007) 11191123.

    [41] M. Katoh, A. Sondoh, T. Horikawa, T. Tomida, Characterization of palladiumand palladiumsilveralloy layers on stainlesssteel support, Intl.J. Mod.Phys.B 20 (2006) 38663871.

    [42] N.I.Timofeev,F.N. Berseneva, V.I.Gromov, Influence of preliminary hydrogenimpregnation on the mechanical properties of palladium and its alloys withsilver, Mater. Sci. 17 (1982) 417419.

    [43] H. Yoshida, K. Okuno, Y. Naruse, T. Kashiwai, Metallurgical considerationson Pd, Pd-alloy and their metalhydrogen systems, Fusion Technol. 8 (1985)23882394.

    [44] A.L. Cabrera, E. Morales, J.N. Armor, Kinetics of hydrogen desorption frompalladium and rutheniumpalladium foils, J. Mater. Res. 10 (1995) 779785.

    [45] Y. Sakamoto, K. Ohira, N. Ishimaru, F.L. Chen, M. Kokubu, T.B. Flanagan,Absorptionof hydrogenby palladiumnickelrhodiumternaryalloys,J.AlloysCompd. 217 (1995) 226234.

    [46] Y. Sakamoto,K. Yuwasa, K. Hirayama, X-ray investigation of theabsorption ofhydrogenby several palladium and nickel solidsolutionalloys,J. Less-Comm.

    Met. 88 (1992) 115124.[47] Y. Sakamoto, F.L. Chen, M. Furukawa, K. Mine, The ( +) hydrogen misci-

    bility gaps in hydrogenated palladium-rich PdY(Gd)Ag ternary alloys, J.Less-Comm. Met. 166 (1990) 4556.

    [48] A.G. Knapton, Palladium alloys for hydrogen diffusion membranes, PlatinumMet. Rev. 21 (1977) 4450.

    [49] V. Gryaznov, Metal containing membranes for the production of ultrapurehydrogen and the recovery of hydrogen isotopes, Sep. Purif. Methods 29(2000) 171187.

    [50] A. Pundt, C. Sachs, M. Winter, M.T. Reetz, D. Fritsch, R. Kirchheim, Hydrogensorption in elastically soft stabilized Pd-clusters, J. Alloys Compd. 293295(1999) 480483.

    [51] G.L. Holleck, Diffusion and solubility of hydrogen in palladium andpalladiumsilver alloys, J. Phys. Chem. 74 (1970) 503511.

    [52] S.K. Gade, M.K. Keeling, A.P. Davidson, O. Hatlevik, J .D. Way,Palladiumruthenium membranes for hydrogen separation fabricatedby electroless co-deposition, Int. J. Hydrogen Energy 34 (2009) 64846491.

    [53] T.A. Peters, M. Stange, H. Klette, R. Bredesen, High pressure performanceof thin Pd23% Ag/stainless steel composite membranes in water gas shift

  • 8/12/2019 j.memsci.2011.03.057

    17/18

    44 S. Yun, S. Ted Oyama / Journal of Membrane Science375 (2011) 2845

    gas mixtures; influence of dilution, mass transfer and surface effects on thehydrogen flux, J. Membr. Sci. 316 (2008) 119127.

    [54] A.L. Mejdell, H. Klette, A. Ramachandran, A. Borg, R. Bredesen, Hydrogen per-meation of thin, free-standing Pd/Ag 23% membranes before and after heattreatment in air, J. Membr. Sci. 307 (2008) 96104.

    [55] G. Barbieri, F. Scura, F. Lentini, G. De Luca, E. Drioli, A novel model equationfor the permeation of hydrogen in mixture with carbon monoxide throughPdAg membranes, Sep. Purif. Technol. 61 (2008) 217224.

    [56] M.L. Doyle, I.R. Harris, Palladium-rare earth alloys: their orderdisordertransformations and behavior with hydrogen, Platinum Met. Rev. 32 (1988)130140.

    [57] Y.H. Ma, I.P. Mardilovich, E.E. Engwall, Thin composite palladium and pal-ladium/alloy membranes for hydrogen separation, Ann. N. Y. Acad. Sci. 984(2003) 346360.

    [58] C.G. Sonwane, J. Wilcox, Y.H. Ma, Solubility of hydrogen in PdAg and PdAubinary alloys using density functional theory, J. Phys. Chem. B 110 (2006)2454924558.

    [59] V. Gryaznov, Hydrogen permeable palladium membrane catalysts, PlatinumMet. Rev. 30 (1986) 6879.

    [60] J.F. Gabitto, C. Tsouris, Sulfur poisoning of metal membranes for hydrogenseparation, Int. J. Chem. Eng. 1 (2009) 394411.

    [61] A. Kulprathipanja, G.O. Alptekin, J.L. Falconer, J.D. Way, Pd and PdCu mem-branes: inhibition of H2 permeation by H2S, J. Membr. Sci. 254 (2005)4962.

    [62] J.B. Miller, B.D. Morreale, A.J. Gellman, The effect of adsorbed sulfur on sur-face segregation in a polycrystalline Pd70Cu30 alloy, Surf. Sci. 602 (2008)18191825.

    [63] 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 palladiumcopper alloys at elevated temperatures, J. Membr. Sci. 241 (2004) 219224.

    [64] N. Pomerantz, Y.H. Ma, Effect of H2S on the performance and long-term sta-bility of Pd/Cu membranes, Ind. Eng. Chem. Res. 48 (2009) 40304039.

    [65] C.P. OBrien, B.H.Howard, J.B.Miller, B.D.Morreale,A.J. Gellman, Inhibitionofhydrogen transport through Pd and Pd47Cu53membranes by H2S at 350 C, J.Membr. Sci. 349 (2010) 380384.

    [66] S.K. Gade, E.A. Payzant, H.J. Park, P.M. Thoen, J.D. Way, The effects of fabri-cation and annealing on the structure and hydrogen permeation of PdAubinary alloy membranes, J. Membr. Sci. 340 (2009) 227233.

    [67] L. Shi, A. Goldbach, G. Zeng, H. Xu, Preparation and performance of thin-layerPdAu/ceramiccompositemembranes,Int. J. HydrogenEnergy 35 (2010)42014208.

    [68] J.D. Way, M. Lusk, P. Thoen, Sulfur-resistant composite metal membranes,USPatent 2008/0038567, Feb. 14, 2008.

    [69] C.-H. Chen, Y.H. Ma, The effect of H2S on the performance of Pd and Pd/Aucomposite membrane, J. Membr. Sci. 362 (2010) 535544.

    [70] A. Sieverts, W. Krumbhaar, ber die lslichkeit von Gasen in Metallen undLegierungen, Ber. Deut. Chem. Ges. 43 (1910) 893.

    [71] R.C.Hurlbert,J.O. Konecny,Diffusion of hydrogen through palladium,J. Chem.

    Phys. 34 (1961) 655658.[72] A. Caravella, G. Barbieri, E. Drioli, Modelling and simulation of hydrogen per-meation through supported Pd-based membranes with a multicomponentapproach, Chem. Eng. Sci. 63 (2008) 21492160.

    [73] R.S. Souleimanova, A.S. Mukastan, A. Varma, Pd membranes formed by elec-troless plating with osmosis: H2 permeation studies, AIChE J. 48 (2002)262268.

    [74] X. Li, T.M. Liu, D. Huang, Y.Q. Fan, N.P. Xu, Preparation and characteri-zation of ultrathin palladium membranes, Ind. Eng. Chem. Res. 48 (2009)20612065.

    [75] F. Guazzone, E.E. Engwall, Y.H. Ma, Effects of surface activity, defects andmasstransferon hydrogen permeance and n-values in compositepalladium-porous, Catal. Today 118 (2006) 2431.

    [76] A. Caravella, F. Scura, G. Barbieri,E. Drioli, Sievertslaw empiricalexponentforPd-based membranes: critical analysis in pure H2permeation, J. Phys. Chem.B 114 (2010) 60336047.

    [77] A. Caravella, G. Barbieri, E. Drioli, Concentration polarization analysis in self-supported Pd-based membranes, Sep. Purif. Technol. 66 (2009) 613624.

    [78] S. Hara, M. Ishitsuka, H. Suda, M. Mukaida, K. Haraya, Pressure depen-

    dent hydrogen permeability extended for metal membranes not obeying thesquare-root law, J. Phys. Chem. B 117 (2009) 97959801.

    [79] M. Coroneo,G. Montante,J. Catalano,A. Paglianti,Modelling theeffectof oper-ating conditions on hydrodynamics and mass transfer in a PdAg membranemodule for H2purification, J. Membr. Sci. 343 (2009) 3441.

    [80] A. Bhargav, G.S. Jackson, Thermokinetic modeling and parameter estimationfor hydrogen permeation through Pd0.77 Ag0.23 membranes, Int. J. HydrogenEnergy 34 (2009), 61645173.

    [81] W.-H. Chen, P