Anda di halaman 1dari 13

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 5 ( 2 0 1 0 ) 1 2 5 9 6 e1 2 6 0 8

Available at

journal homepage:

Enhancing the production of hydrogen via wateregas shift

reaction using Pd-based membrane reactors

Diogo Mendes a, Vania Chibante a, Ju-Meng Zheng c, Silvano Tosti b, Fabio Borgognoni b,
Adelio Mendes a, Luis M. Madeira a,*
LEPAE, Chemical Engineering Department, Faculty of Engineering, University of Porto, Rua Dr. Roberto Frias, 4200-465 Porto, Portugal
ENEA, Unita Tecnica Fusione, C.R. ENEA Frascati, Via E. Fermi 45, Frascati (RM) I-00044, Italy
Dutch Separation Technology Institute (DSTI), 3800 AE Amersfoort, The Netherlands

article info abstract

Article history: In this work, it is described an experimental study regarding the performance of a PdeAg
Received 5 January 2010 membrane reactor recently proposed and suitable for the production of ultra-pure
Received in revised form hydrogen. A dense metallic permeator tube was assembled by an innovative annealing and
28 July 2010 diffusion welding technique from a commercial flat sheet membrane of PdeAg. A finger-
Accepted 29 July 2010 like configuration of the self-supported membrane has been designed and used as
a packed-bed membrane reactor (MR) for producing ultra-pure hydrogen via wateregas
shift reaction (WGS).
Keywords: A CuO/ZnO/Al2O3 catalyst, from REB Research & Consulting, was used for packing the WGS
Hydrogen production membrane reactor. The performance of the reactor was evaluated in terms of CO
Wateregas shift reaction conversion and H2 recovery in a wide range of conditions: temperature from 200  C to
PdeAg membrane 300  C, feed pressure from 1.0 bar to 4.0 bar, vacuum and sweep-gas modes and with
Membrane reactor a simulated reformate feed (4.70% CO, 34.78% H2O, 28.70% H2, 10.16% CO2 balanced in N2).
Hydrogen permeation Also, the effect of the reactants feed composition was investigated and discussed.
CO conversions remained in most conditions above the thermodynamic equilibrium based
on feed conditions. In particular, it is worth mentioning that around 100% of
CO conversion and almost complete H2 recovery was achieved when operating the MR at
300  C with a GSHV 1200 LN kg1 1
cat h , Pfeed 4 bar, Pperm 3 bar and using 1000 mLN min
2010 Professor T. Nejat Veziroglu. Published by Elsevier Ltd. All rights reserved.

1. Introduction phones and electronic gadgets to transportation or stationary

power stations [16,24]. With the foreseen implementation of
The fast depletion of fossil fuels reserves and the related fuel cell devices and the use of heavier crude oil coupled with
environmental concerns are reasons for the renewed interest the growing demand for deep desulphurized fuels, production
in alternative energy resources. Among the potential candi- of hydrogen is expected to increase remarkably.
dates, hydrogen has been pointed out as an ideal energy Hydrogen can be produced from widely available primary
carrier. Hydrogen can be used as a fuel in polymer electrolyte energy sources including fossil fuels and renewables (e.g.
membrane fuel cells (PEMFCs) for electrical power generation biomass, wastes, solar, wind, hydro, geothermal or nuclear
from micro to mega-watt applications, i.e. from mobile power), being considered a promising energy vector [2]. Today,

* Corresponding author. Present address: Dutch Separation Technology Institute (DSTI), 3800 AE Amersfoort, The Netherlands. Tel.: 351
22 508 1519; fax: 351 22 508 1449.
E-mail address: (L.M. Madeira).
0360-3199/$ e see front matter 2010 Professor T. Nejat Veziroglu. Published by Elsevier Ltd. All rights reserved.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 5 ( 2 0 1 0 ) 1 2 5 9 6 e1 2 6 0 8 12597

Nomenclature PBR packed-bed reactor

PEM polymer electrolyte membrane
CH hydrogen concentration in the lattice of the
Pfeed total pressure in the feed side, bar or Pa
metallic membrane, mol m3
pH2 ; ret hydrogen partial pressure in the retentate side, Pan
DH diffusion coefficient of hydrogen atoms, m2 s1
pH2 ; permhydrogen partial pressure in the permeate side,
Ea activation energy for hydrogen permeation,
J mol1
Pperm total pressure in the permeate chamber, bar or Pa
FC fuel cell
Qfeed volumetric flow rate of feed gas mixture at normal
FCO; feed molar flow rate of CO at the inlet of the MR, mol h1
conditions, mLN min1
FCO; exit molar flow rate of CO at the exit of the MR, mol h1
Qsweep-gas sweep-gas flow rate passing in the permeate
FH2 ; perm molar flow rate of H2 at the exit of the permeate
chamber at normal conditions, mLN min1
chamber, mol h1
ideal gas constant, 8.314 J mol1 K1
FH2 ; ret molar flow rate of H2 at the exit of the retentate
RH2 hydrogen recovery
chamber, mol h1
S sorption coefficient of hydrogen in the lattice,
GHSV gas hourly space velocity, LN kg1cat h
mol m3 Pan
JH2 flux of hydrogen through the PdeAg membrane,
T temperature, K or  C
mol m2 s1
TEC thermodynamic equilibrium conversion
LH2 hydrogen permeability, mol m m2 s1 Pan
WGS wateregas shift
L0 pre-exponential factor for hydrogen permeation,
x spatial position along the membrane thickness, m
mol m m2 s1 Pan
XCO carbon monoxide conversion
MR membrane reactor
d PdeAg membrane thickness, m
n hydrogen partial pressure exponent constant

hydrogen production technologies fall into three general permeation, the hydrocarbon conversion can be penalized
categories: thermal, electrolytic, and photolytic processes. because of the reverse reaction rate enhancement (as a conse-
Thermal processes (e.g. reforming and gasification) are the quence of the increase in the number of moles in the forward
most used technologies where energy is used in various steam reforming reaction) [17,25]. In this sense, the use of
feedstocks, such as natural gas, coal, or biomass, to release PdeAg membranes for the H2 upgrading of reformate mixtures,
hydrogen of their molecular structure [20]. via the WGS reaction, seems a more adequate solution.
In the future, the reforming of renewable liquid fuels The WGS is a moderately exothermic reaction and char-
(a process very similar to reforming of natural gas e a rela- acterized by no variation in the number of moles (CO
tively mature technology), is expected to be a promising H2 O4CO2 H2 DHo298 K 41:1 kJ,mol1 ). Thus, CO equi-
choice for H2-based energy systems [12,30]. Among the cata- librium conversion is favored by low temperatures and is
lytic reaction technologies to convert them into hydrogen, the independent of the reaction pressure when operating in
autothermal reforming of bio-ethanol seems to be a feasible a conventional packed-bed reactor (PBR). On the contrary, in
method for the on-board production of hydrogen in fuel cell a Pd-based MR a high pressure on the reaction side facilitates
vehicles [10,14]. However, the common problem with hydrogen permeation and, therefore, from the kinetic point of
reforming processes is that, besides the formation of a H2-rich view, the operating temperature can be reduced because the
stream, high levels of CO and small amounts of impurities reaction equilibrium is shifted towards products formation;
(e.g. CH4, C2H4, C2H6, and CH3CHO) are produced. additionally, the catalyst amount necessary for a given
For PEM fuel cell based vehicles, strict hydrogen fuel conversion can be significantly reduced as well [3]. In this way,
specifications are required [22] and therefore ultra-pure the WGS reaction and the hydrogen separation steps required
hydrogen must be delivered. Amongst the technologies for in the conventional power production processes can be
purifying hydrogen, Pd-based membranes have received replaced by a single WGS membrane reactor, moving in the
increased interest due to their unique hydrogen separation logic of the Process Intensification Strategy [34]. Via process
properties [31]. integration, the number of units is decreased and the total
The integration of PdeAg membranes into hydrocarbon required reactor volume can be largely reduced, while higher
reformers or wateregas shift (WGS) reactors provides a new CO conversions and pure hydrogen recovery can be achieved
effective one step solution, because it produces a pure at lower temperatures [4,18].
hydrogen stream and it allows overcoming the equilibrium In recent years, many researchers have focused their
limitations experienced in conventional reactor configurations attention in the use of PdeAg membrane reactors for the
due to in situ removal of H2. However, the use of PdeAg WGS reaction. A review covering the application of Pd-based
membrane reactors (MRs) as autothermal reformers has some MRs was recently published by Mendes et al. [29]. Among
disadvantages: (i) the damage of the metal membrane alloy due the works reported so far, it is worth mentioning the
to the hot-spots [32,35], and (ii) the decrease of the overall Pd-based membranes that are made of self-supporting
hydrocarbon conversion, depending on the operating condi- metal foils with thickness of 25e100 mm due to their full
tions; for instance, even though the yield of hydrogen in the MR permselectivity towards hydrogen. Nevertheless, some
can be favored at higher pressure because of increasing issues related to membrane lifetime remain to be solved.
12598 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 5 ( 2 0 1 0 ) 1 2 5 9 6 e1 2 6 0 8

With this concern in mind, Tosti et al. [36] developed a new

Pd-based permeator tube, in a finger-like configuration, 2. Experimental
which reduces the restrictions between the membrane tube
and the module and avoids the presence of any mechanical 2.1. Membrane tube preparation and module assembly
(cyclic) stresses, assuring a long lifetime and no change in
the H2 separation ability. The thin wall permeator tube was prepared starting from
Most of the experimental and modeling studies concerning a commercial 50-mm PdeAg sheet (with 25 wt.% Ag). To reduce
Pd or Pd-Ag based WGS reactors reported so far devoted their the metal hardness, it was carried out the annealing of the
attention to the reaction conversion enhancement [1,6,21]. In metal alloy foil [37]. The heat treatment was performed in
fact, a detailed experimental study on the effect of all opera- a high-vacuum (<104 mbar) oven, to avoid oxidation, with
tional parameters (temperature, reaction pressure, feed flow a heating ratio of 5  C min1 up to 900  C, staying at this
rate and reactants feed composition) in terms of H2 recovery temperature for 1 h. This thin metal foil has been cut to
has been hardly addressed, with the exception of very few measure, providing however a sufficient side margin to allow
works [8,9]. Moreover, most of the simulation studies consider the edges to be overlapped in the area of welding. Subsequently,
reaction rate models taken from the literature, where the the metal alloy thin foil has been wrapped around an alumina
effect of CO2 inhibition on the reactor performance [26] is not bar in order to obtain permeating tubes of length 50 mm and
properly accounted. internal diameter 10 mm and proceed with the joining of the
Ideal feeds (CO and H2O balanced in an inert gas) and/or limbs by a diffusion bounding welding. This welding technique
very specific operating conditions are typically tested in MRs, consists of a thermal treatment (under the same operating
in particular in the PdeAg finger-like self-supported conditions of the annealing process) where the limbs of the
membranes, conceived specially for the ultra-pure hydrogen metal foil, overlapped and compressed in a special device, are
production. Indeed, only very few works studied this type of joined mainly because of the silver atoms diffusion in the metal
MR for the WGS reaction and the operating conditions repor- alloy. To apply a controlled pressure over the overlapped limbs
ted are limited and very different from those in the present of the PdeAg tube, a thermo-mechanical press consisting of
work. Therefore, the objective of this study was to assess the Inconel plates and Invar screws is used. During this step, a foil
performance (both in terms of conversion enhancement and of mica (flogopite) with thickness 25e40 mm comes arranged
H2 recovery) of such MR employing a CuO/ZnO/Al2O3 catalyst between the press and the limbs of the PdeAg tube to avoid the
using simulated reformate gas mixture. In particular, it is diffusion of the Pd and Ag atoms to the press during the heat
addressed the effect of the reaction temperature (from 200  C treatment. A dynamometric key is finally used to allow
to 300  C), reaction pressure (in the range of 1e4 bar), feed gas a controlled and precise tightening over the welding area.
hourly space velocity (1200e10,800 LN kg1 1 Finally, the PdeAg tube with 50 mm of length has been joined to
cat h ), reactants gas
feed composition, and different MR operation modes (vacuum a stainless steel VCR connection and a steel plug by brazing. In
and sweep-gas modes). For improving the hydrogen perme- this way, the permeator has the stiffness needed to assure
ation and the reaction rate it is desirable to use high feed a tight assembly into the membrane module. Fig. 1 shows
pressures. In this work, a new strategy was employed to a scheme (a) and a picture (b) of the thin wall permeator tube,
operate this MR at (relatively) high feed pressures (up to 4 bar, which has been inserted into a stainless steel module.
clearly above the ranges tested so far with this system). This The assembly of the PdeAg membrane has a finger-like
was made possible feeding the sweep-gas at a higher pressure configuration in order to allow the membrane elongation and
so that the maximum pressure difference that the PdeAg contraction due to the hydrogenation and thermal cycles.
membrane can withstand was assured. Long-term tests (1 year) already demonstrated the complete
The conversions attained in this work were also compared hydrogen selectivity and durability of these permeators,
with the thermodynamic one based on the inlet conditions showing no formation of defects (holes and cracks) [36]. As
(temperature and feed composition) e maximum limit can be seen in Fig. 1, a stainless steel tube (O.D. 1/16 inch) is
attainable in a conventional packed-bed reactor. The results inserted inside the PdeAg membrane in order to feed the gas
were compared, whenever possible, with literature data. stream; the retentate chamber is in the annular section.

Fig. 1 e Scheme (a) and picture (b) of the PdeAg finger-like configuration membrane reactor.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 5 ( 2 0 1 0 ) 1 2 5 9 6 e1 2 6 0 8 12599

2.2. Experimental set-up column Supelco Carboxen 1010 Plot, 30 m  0.32 mm I.D., and
a TCD (Valco thermal conductivity detector); helium was used
Mass flow controllers (Bronkhorst Hi-Tec, model F201) were as carrier. The relative errors of carbon and nitrogen molar
used to feed pure hydrogen or H2-rich gas mixtures to the MR balances were typically lower than 5%. Regarding hydrogen, it
(cf. Fig. 2). For experiments involving humidified gas streams, is difficult to analyze it by GC-TCD since the thermal conduc-
deionized water was metered, vaporized and mixed in tivity of hydrogen is close to that of helium, used as carrier gas.
a Controller Evaporator Mixer (CEM, Bronkhorst) system with Therefore, the H2 composition in the retentate stream was
the other gases before entering the reactor. The MR was calculated from the mass balance (difference from 100%, dry
encased in an electrical oven (Memmert, type UNE200), basis). Further details concerning the analysis method can be
controlled by a programmable temperature controller. found elsewhere [28].
A thermocouple was inserted through the top of the reactor
for reading the membrane surface temperature (permeation 2.2.1. Permeation experiments
side), approximately at the axial center. The pressure in the Pure hydrogen was fed to lumen side (cf. Figs. 1 and 2) with
lumen/feed side was measured using a pressure transducer pressure ranging from 1.1 bar to 2.5 bar, temperatures ranging
(Druck, ref. 4010, 7 bar) and adjusted using a back-pressure from 200  C to 300  C and flow rate ranging from 50 to
regulator (Swagelok); a second pressure transducer (Druck, 190 mLN min1. Permeance was measured using the stainless
ref. 4010, 5 bar) and a back-pressure regulator (Swagelok) were steel permeator sketched in Fig. 1. Upon heating the module,
used to read and regulate the shell side pressure. Different the membrane was kept in a N2 atmosphere until 300  C was
operation modes were used to produce a pressure gradient reached, then H2 was introduced. The H2 feed gas flowed along
between lumen and shell side: (i) permeate chamber was open the inner side of the membrane and the permeating stream
to the atmosphere, (ii) nitrogen was used as sweep-gas, and flow rate was measured on the shell side at atmospheric
(iii) the shell side was continuously evacuated using a dia- pressure with a mass flow meter after steady-state (approxi-
phragm vacuum pump (Thomas Instruments, ref.: 7011-0069). mately 1 h) had been achieved, at the desired temperature.
In this case, the permeate feed was closed using an on/off Pressure differences between retentate and permeate sides
valve e Fig. 1. were obtained by varying the pressure on the upstream side
The reactor outlet stream was passed through a condenser and keeping the downstream pressure constant at 1 bar.
at room temperature to remove the condensable water. The Neither sweep-gas nor vacuum was used. Pressures in the two
dried gaseous stream flow rate was measured using a mass sides of the membrane tube were monitored via pressure
flow meter (Bronkhorst Hi-Tec, model F201) (cf. Fig. 2); the transducers. Frequently, the selectivity of the membrane was
permeating hydrogen stream was also read using a second obtained against nitrogen. When nitrogen was introduced
mass flow meter (Bronkhorst Hi-Tec, model F201). The gas pressurized in the retentate chamber and the unit closed, no
phase products, N2, CO and CO2, were analyzed in a gas chro- decline in the pressure was noticed after 8 h, confirming the
matograph (Dani 1000 GC), equipped with a chromatographic full hydrogen permeation selectivity of the PdeAg membrane.

Fig. 2 e Scheme of the experimental set-up (BPR: back-pressure regulator; CEM: controller-evaporator mixer; GC: gas
chromatograph; MFC: mass flow controller; MFM: mass flow meter; P1 and P2: pressure transducers; T: thermocouple).
12600 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 5 ( 2 0 1 0 ) 1 2 5 9 6 e1 2 6 0 8

2.2.2. WGS reaction tests FCO; exit

XCO 1  (1)
In a previous work [28] a commercial CuO/ZnO/Al2O3 catalyst, FCO; feed
supplied by REB Research & Consulting, was evaluated (among
other promising samples) to catalyze the WGS reaction. The FH2 ; perm
RH2 (2)
copper-based catalyst showed the best relation activity/ FH2 ; ret FH2 ; perm
stability for the entire range of temperatures tested
Variables FCO; exit and FCO; feed represent the molar flow rates of
(150e300  C) using a simulated reformate feed composition. In
CO at the exit and inlet of the MR, respectively; FH2 ; perm and
this work, the WGS reaction was performed in the tempera-
FH2 ; ret are the corresponding permeate and retentate H2 molar
ture range 200e300  C by packing a certain amount of the
flow rates.
same catalyst into the lumen side of the MR (cf. Fig. 1a). Prior
to the reaction runs, the WGS catalyst was activated in situ
with a mixed gas flow of H2/N2. The pre-reduction protocol 3.1. Permeation tests
considered heating the catalyst, in nitrogen, from room
temperature up to 230  C, at 5  C min1; then, a reduction The H2 mass transfer through thick palladium-based
mixture (5 vol.% of H2/N2, total flow rate of 100 mLN min1) membranes at high temperatures (>150  C) is usually rate-
was admitted and the sample maintained at 230  C for 5 h. limited by the diffusion of hydrogen through the metal lattice
After reduction, the catalyst was cooled or heated to the [31] and is described by the Ficks law:
reaction temperature (analyzed in the range 200e300  C) and
the reactor flushed with N2 prior to feeding the reaction dCH
J DH (3)
mixture. Overheating of the catalyst bed and subsequent dx
metal particles sintering during the reduction was avoided where J is the hydrogen diffusion flux through the metal
controlling closely the hydrogen feed flow rate. The CuO/ZnO/ lattice, DH is the effective diffusion coefficient of atomic
Al2O3 catalyst (1.5 g) was diluted with glass spheres of the hydrogen, and CH is the atomic hydrogen concentration at the
same particle size (250e355 mm) to avoid temperature gradi- spatial position x along the membrane thickness.
ents in the bed (checked previously in a packed-bed reactor), CH is related to the hydrogen partial pressure, pH2 , by
caused by the exothermic reaction. During the experimental Sieverts law (equilibrium conditions) [31]:
campaign, in the absence of the reaction mixture, the system
CH S p0:5
H2 (4)
was kept under N2 atmosphere to avoid re-oxidation.
The reaction tests were carried out using the following feed where S is the sorption coefficient of hydrogen in the lattice. If
gas composition: 4.70% CO, 34.78% H2O, 28.70% H2, and 10.16% the sorbed hydrogen is in equilibrium with gas phase at both
CO2 balanced in N2, which simulates a reformate feed. The membrane surfaces, combining equations (3) and (4), and
reaction pressure (that is, the pressure in the lumen side of the integrating for the membrane thickness, d, one obtains:
membrane) and the feed flow rate were varied in the range
1e4 bar and 30e270 mLN min1, respectively. In the first set of JH2 2 p0:5
H2 ; ret  pH2 ; perm
experiments, hydrogen has been recovered in the shell side of
the membrane cell by vacuum pumping (shell side where LH2 is the hydrogen permeability.
pressure w 30 mbar). Then, nitrogen was fed to the shell side Equation (5) is valid for defect free membranes. To account
(sweep-gas), at a flow rate ranging from 100 to 1500 mLN min1, also for the mass transport through defects the previous
counter-currently, to decrease further the hydrogen partial equation is normally written as
pressure. During this procedure, the total pressure gradient LH2  n 
JH2 pH2 ; ret  pnH2 ; perm (6)
between lumen and shell side was maintained at 1 bar (for d
mechanical reasons). Additionally, to simulate the perfor-
where n stands for the power dependency of the hydrogen
mance of a packed-bed reactor (PBR from now on) for posterior
concentration on its partial pressure; n ranges from 0.5 to 1
comparison with the MR operation, the permeate chamber was
and should be obtained experimentally [15].
closed and data taken after steady-state was achieved.
The relationship between the hydrogen permeability and
The influence of the most important operating variables
the temperature can be described by an Arrhenius-type
(i.e. reaction pressure, temperature, steam-to-CO ratio, feed
equation [15], within the temperature range considered:
space velocity and sweeping gas flow rate) on the reaction
conversion and H2 recovery were studied. Ea
LH2 LoH2 exp  (7)

Assuming that the mass transport is controlled by the

diffusion of atomic hydrogen (n = 0.5), combining expressions
3. Results and discussion
(5) and (7) the Richardsons equation is obtained:
The catalytic performance and separation ability of the PdeAg LoH2 exp 

membrane reactor conducting the WGS reaction have been < T  0:5 
JH2 pH2 ; ret  p0:5
H2 ; perm (8)
evaluated in terms of the steady-state carbon monoxide d
conversion (XCO) and hydrogen recovery (RH2 ). Both parame- The permeation tests carried out at different transmembrane
ters were calculated according to equations (1) and (2), pressure differences and temperatures were used for obtain-
respectively: ing the parameters of equation (6) (LH2 and n) and further its
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 5 ( 2 0 1 0 ) 1 2 5 9 6 e1 2 6 0 8 12601

3.5 -18.6
300 C
280 C
3.0 250 C -18.8
230 C
H2 Flux (mL Ncm min )

2.5 200 C -19.0


2.0 -19.2

ln (LH2 )
1.5 -19.4 This work
(Tosti et al., 2006)
1.0 -19.6 (Iulianelli et al., 2008)
(Uemiya et al., 1991)
(Davis, 1954)
0.5 -19.8

0.0 -20.0
0 50 100 150 200 1.6 1.7 1.8 1.9 2.0 2.1 2.2
pH0.5, feed pH0.5, perm ( Pa 0.5 ) -1 -1
1000 x T /K
Fig. 3 e Hydrogen flux through the 50 mm thick PdeAg Fig. 4 e Hydrogen permeability versus temperature for the
membrane as a function of the difference between the Pd-based membrane of this work and others reported in
square root of hydrogen partial pressure in the feed and the literature.
permeate sides.

3.2.1. Influence of the reaction temperature

permeation activation energy. In Fig. 3, the results presented Fig. 5 illustrates the CO conversion in the WGS reaction and
indicate that the permeation rate of H2 is directly proportional the H2 recovery obtained in the PdeAg MR as a function of the
to the difference between the square roots of the H2 partial temperature. In this section, hydrogen was collected in the
pressure in the feed and permeate sides (n in equation (6), permeate side of the membrane module by vacuum pumping
determined by regression e data not shown). This shows that (Pperm w30 mbar, vacuum mode). A feed pressure of 1.1 bar
the mass transport is controlled by the diffusion of atomic and a GHSV (gas hourly space velocity) 3200 LN kg1 1
cat h
hydrogen through the PdeAg membrane e Sieverts law. were used. The dotted curve shows the WGS thermodynamic
The experimental permeability values towards H2 (LH2 ) at equilibrium conversion (TEC) calculated based on the inlet
different temperatures (T ) were then used to estimate the reformate gas composition [28].
activation energy (Ea) for H2 permeation. Table 1 reports At conditions close to that of a packed-bed reactor, the CO
a comparison of the permeation parameters for Pd-based conversion for 200  C is below the equilibrium conversion due
membranes typically found in the literature with the ones to kinetic limitations (cf. Fig. 5). With the increase of the
from this work. temperature the kinetics is favored, and thus the WGS
The calculated values for the permeation activation energy conversion in the PBR approaches the thermodynamic limit,
and pre-exponential factor (LoH2 ) are in good agreement to which is not favored by the exothermic nature of the reaction
other experimental data, as also shown in Fig. 4. (and is thus smaller than at 200  C). Operating with the PdeAg

3.2. WGS reaction in the PdeAg MR 100 60

98 50
In this section the performance of the PdeAg MR using
a copper-based catalyst for the WGS reaction is analyzed. Such
96 40
evaluation (performed in terms of catalyst activity e XCO e and TEC
XCO / %

RH / %

separation ability -RH2 ) takes into account the effect of the most MR
94 RH 30
important operating variables. Each data point in the graphs

results from the average of at least three analyses.

92 20

90 10
Table 1 e Apparent activation energy and pre-
exponential factor for hydrogen permeation through 88 0
dense PdeAg membranes determined in this work and 180 200 220 240 260 280 300 320
taken from the literature. Temperature / C
Ea/kJ mol1 LoH2 /mol$m m2 s1 Pa0.5 Reference
Fig. 5 e Influence of the reaction temperature on the CO
10.72 5.44  108 This work conversion and H2 recovery for the WGS reaction over a Cu-
11.24 6.64  108 [36] based catalyst in packed-bed reactor and in PdeAg MR
17.60 2.06  107 [23] (Pfeed [ 1.1 bar, Pperm e30 mbar (vacuum mode),
18.45 3.23  10-7 [39]
GHSV [ 3200 LN kgL1 cat h
, and feed composition: 4.70%
18.56 3.85  107 [13]
CO, 34.78% H2O, 28.70% H2, 10.16% CO2 balanced in N2).
12602 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 5 ( 2 0 1 0 ) 1 2 5 9 6 e1 2 6 0 8

MR, XCO is enhanced in the entire range of temperatures used. 3.2.2. Influence of the feed space velocity
In particular, the shift on the conversion is more noticeable at The space velocity was varied by changing the feed flow rate
higher temperatures because of the larger H2 extraction from while keeping the amount of catalyst constant. The effect of
the reaction zone despite the lower H2 partial pressure in the feed space velocity on CO conversion and H2 recovery was
retentate side (lower carbon monoxide conversion). Even investigated by performing the WGS in the PdeAg MR oper-
though the H2 recovery and kinetic rate are both improved by ating in vacuum mode (simulated PBR performance is also
temperature (especially up to 280  C), CO conversion in the MR included for comparison). The results are indicated in Fig. 6.
is penalized by thermodynamic constraints of the exothermic The results show a CO conversion enhancement with the
WGS reaction. PdeAg MR as compared to the PBR for the entire range of space
For the MR properties and operating conditions used in velocities. Moreover, CO conversions at 300  C remained
this work, higher CO conversions (98.0e98.5%) can be reached above the thermodynamic equilibrium value (curve not
at lower temperatures (<230  C) where the equilibrium shown for simplicity) for the entire range of feed space
conversion is favored (cf. Fig. 5). This observation highlights velocities tested. The shift on the conversion (MR vs. PBR)
the fact that PdeAg MRs can be advantageously used also in becomes higher at elevated temperatures, where the H2
terms of energy savings when compared with the conven- recovery is also favored, in line with the results shown in the
tional reactors. previous section. From Fig. 6, it is clear that lower levels of CO
The optimum performance of a PdeAg MR conducting the conversions and H2 recoveries are obtained as GHSV increases
WGS reaction should balance a high CO conversion and a high (except for XCO in the PBR at 300  C because conversions close
H2 recovery. Despite the equilibrium conversion being favored to the TEC limit are always reached). In fact, higher feed space
for low temperatures, this is a very limiting condition in terms velocities lead to lower residence times, which are disadvan-
of H2 recovery. As it can be seen in Fig. 5, the lowest recovery tageous in terms of H2 generation as well as in terms of H2
(RH2 w8.5%) was obtained at 200  C. As mentioned previously, permeation through the Pd membrane. Particularly, at 200  C
H2 permeation through dense PdeAg membranes is an acti- XCO decreases from ca. 99.0% to ca. 75.9% e cf. Fig. 6. For higher
vated process; therefore, RH2 is not favored when operating at temperatures such decrease is less noticeable, due to the fact
lower temperatures. In addition, it is well-known the blocking that under such conditions conversion is already close/above
effect of the Pd membrane surface sites caused by CO for the thermodynamic one and the membrane permeation is
T < 350  C [27]; this inhibitory effect decreases with temper- higher (see also Fig. 5).
ature because less CO is adsorbed on the membrane surface. The effects of the GHSV obtained in this work are in line
The increments of the H2 recovery were more evident at with the trends reported in other works (e.g. [8,11,40]). In
temperatures <250  C (specifically, by moving from 200  C to particular, Uemiya et al. [40] studied the WGS reaction in
250  C) where the H2 permeation kinetics are rate limiting. a 20-mm thick palladium composite membrane obtaining
Above 250  C, RH2 did not increase significantly due to the a conversion enhancement from 75 to 94% when the GHSV
compromise between the kinetic rate of the WGS reaction, was decreased from 2150 to 250 LS kg1 cat h
(12 g of a ferro-
hydrogen permeation through the membrane and the ther- chrome catalyst; ideal feed mixture e H2O/CO 1; 400  C; feed
modynamic constraints of the WGS reaction. This fact was at atmospheric pressure and sweeping the permeate with Ar).
also recently observed by other authors [8]. As shown in Fig. 6, the MR operating at 300  C and
In order to improve the performance of the MR described in GHSV 1200 LN kg1 cat h
produces a maximum H2 recovery of
this work, in particular hydrogen recovery, other relevant 44.3%. In this work, the H2 recovery dependence with the
operation parameters were studied as described in the space velocity follows an exponential decay trend, whatever
following sections. the temperature used. Bi et al. [8] observed a quick and linear

100 50
MR: T = 200 C
MR: T = 230 C
95 MR: T = 250 C
MR: T = 280 C
MR: T = 300 C
XCO / %

RH / %

PBR: T = 200 C 20
MR: T = 200 C
PBR: T = 230 C
80 MR: T = 230 C
PBR: T = 300 C 10
MR: T = 300 C
0 2000 4000 6000 8000 10000 12000 0 2000 4000 6000 8000 10000 12000
-1 -1
GHSV / LNkg cat h GHSV / LNkg cat-1h-1

Fig. 6 e CO conversion and H2 recovery as a function of the gas hourly space velocity (GHSV) for the WGS reaction over a Cu-
based catalyst in packed-bed reactor and in PdeAg MR (Pfeed [ 1.1 bar, Pperme30 mbar (vacuum mode) and feed composition:
4.70% CO, 34.78% H2O, 28.70% H2, 10.16% CO2 balanced in N2).
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 5 ( 2 0 1 0 ) 1 2 5 9 6 e1 2 6 0 8 12603

decay on the H2 recovery (from 84.8% to 48.7%) with an 2.0 bar promoted notably the H2 recovery, from 44.4% to 84.2%,
increase of the space velocity (from 4050 to 9100 LN kg1 1
cat h ). driving up CO conversion from 96.6% to 99.1%. The effect of
From Fig. 6, it can be seen that very similar H2 recoveries the feed pressure is ascribed to a higher permeation driving
were obtained for all range of GHSV tested at 280  C and 300  C. force, as mentioned above. However, it has also a positive
This can be explained by the above mentioned compromise effect on the WGS reaction rate.
between the hydrogen permeation and the equilibrium The CO conversion and H2 recovery obtained with the
constraints of the WGS reaction. PdeAg MR operating on sweep-gas mode as a function of the
feed pressure, keeping constant the overall pressure differ-
3.2.3. Influence of the feed pressure ence P Pfeed  Pperm 1 bar, is shown in Fig. 8.
For a successful application of PdeAg MRs in the WGS reaction The use of sweep-gas acts positively on the hydrogen
it is important to achieve high hydrogen permeation rates. permeation, thus allowing a higher H2 recovery and enhanced
This can be achieved by increasing the hydrogen partial CO conversion e cf. Figs. 7 and 8. Comparing the MR operating
pressure difference between retentate and permeate sides modes, higher conversions and H2 recoveries are achieved
and hence maximizing the driving force for hydrogen when the permeated hydrogen is swept by nitrogen.
permeation (see equation (5)). One method to accomplish this From Fig. 8, one can see that the increase of feed pressure
is to increase the feed pressure despite it does not influence from 2.0 to 4.0 bar raised the H2 recovery from 90.8% to 99.2%,
the WGS reaction from a thermodynamic point of view. Other enhancing the CO conversion from 99.6% to 100%, at 300  C.
options are decreasing the permeate chamber pressure by The decay on CO conversion in the temperature range studied
applying vacuum to the permeate side, or by applying is almost unnoticeable when N2 sweeps the permeated H2. On
a sweep-gas. the other hand, the improvement on the H2 recovery is more
In the next sections, the effect of the feed pressure in the evident. It is thus remarkable that under moderate conditions
MR performance was studied for different operation modes of temperature and pressure, almost complete CO conversion
(at a GHSV 1200 LN kg1 1
cat h ): (i) vacuum mode e the and total H2 recovery have been achieved.
permeate chamber was continuously evacuated by vacuum To the best of our knowledge, only a few experimental
pumping (the vacuum pressure was limited to w30 mbar), and studies concerning the evaluation of the feed pressure effect on
(ii) sweep-gas mode e nitrogen was supplied to the permeate the WGS reaction were carried out on Pd-based MRs (e.g. [7,33]).
side in a counter-current manner to sweep the permeated Also, Bi et al. [8] obtained a H2 recovery enhancement, from
hydrogen. From the mechanical point of view, it is worth 40.5% to 89.2%, and a CO conversion improvement, from 86.0%
noting that the difference between the lumen and external to 95.5%, but by increasing the feed pressure from 4 to 12 bar (at
(total) pressure cannot overcome 2 bar. Moreover, the total 350  C and GHSV 4050 LN kg1cat h
with no sweep-gas).
pressure gradient (P Pfeed  Pperm) must be always positive It is important mentioning that, for real applications,
in order to avoid the collapse of the membrane. Having these a balance should be done between the positive effects due to
issues in mind, a new strategy, at least never tested for this a pressure increase on the MR performance (higher CO
kind of membranes, allowed us to achieve (relatively) high conversion and hydrogen recovery) and the consequent
pressures by using a pressurized sweep-gas, as shown below. disadvantages in terms of operating costs linked to the feed
compression. Vacuum versus sweep-gas mode. Fig. 7 illustrates the
influence of feed pressure on CO conversion for the WGS and The effect of sweep-gas flow rate. As mentioned
H2 recovery obtained in the PdeAg MR. For the same permeate above, a sweep-gas stream can be used to maintain a higher
pressure (w30 mbar), the results show that CO conversion and H2 partial pressure gradient across the membrane, promoting
H2 recovery increased substantially with the feed pressure. CO conversion and H2 recovery. The influence of the sweep-
Particularly, at 300  C the increase of feed pressure from 1.1 to gas flow rate on XCO and RH2 was studied and the results are

100 100
Pfeed = 1.1 bar - Pperm ~ 0.030 bar
98 Pfeed = 2.0 bar - Pperm ~ 0.030 bar

RH / %
X CO / %


92 PBR 40
Pfeed = 1.1 bar - Pperm ~ 0.030 bar
90 Pfeed = 2.0 bar - Pperm ~ 0.030 bar

86 0
180 200 220 240 260 280 300 320 180 200 220 240 260 280 300 320
Temperature / C Temperature / C

Fig. 7 e CO conversion and H2 recovery as a function of the feed pressure for the WGS reaction in the PdeAg MR
(GHSV [ 1200 LN kgL1cat h
, feed composition: 4.70% CO, 34.78% H2O, 28.70% H2, 10.16% CO2 balanced in N2).
12604 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 5 ( 2 0 1 0 ) 1 2 5 9 6 e1 2 6 0 8

100 100


XCO / %

RH / %
98 Pfeed = 2.0 bar - Pperm = 1.0 bar
Pfeed = 3.0 bar - Pperm = 2.0 bar 70
Pfeed = 4.0 bar - Pperm = 3.0 bar
Pfeed = 2.0 bar - Pperm = 1.0 bar
Pfeed = 3.0 bar - Pperm = 2.0 bar
Pfeed = 4.0 bar - Pperm = 3.0 bar

96 50
180 200 220 240 260 280 300 320 180 200 220 240 260 280 300 320
Temperature / C Temperature / C

Fig. 8 e CO conversion and H2 recovery as a function of the feed pressure for the WGS reaction in the PdeAg MR operating on
sweep-gas mode (N2 sweep-gas flow rate [ 1000 mLN minL1, GHSV [ 1200 LN kgL1 cat h
, feed composition: 4.70% CO,
34.78% H2O, 28.70% H2, 10.16% CO2 balanced in N2).

indicated in Fig. 9. The reaction was carried out in the Pfeed side decreases, resulting in a higher driving force for the
range of 2.0e4.0 bar at temperatures of 200 and 300  C. hydrogen permeation. So, the H2 recovery is promoted,
From Fig. 9, it can be seen that CO conversion and H2 favoring the CO shift.
recovery are generally favored when the sweep-gas flow rate Due to the higher conversions achieved, improvements are
is increased. The positive effect of the sweep-gas flow rate is more clearly visible when the system operates with sweep-gas
related to equation (5). In other words, by increasing the in the range of low flow rates. Particularly, at 200  C,
sweep-gas flow rate the hydrogen partial pressure in the shell Pfeed 4.0 bar and Pperm 3.0 bar, the H2 recovery increases

100.0 70
T = 200 C

Pfeed = 2.0 bar - Pperm = 1.0 bar 60
XCO / %

RH / %

Pfeed = 3.0 bar - Pperm = 2.0 bar

Pfeed = 4.0 bar - Pperm = 3.0 bar
Pfeed = 2.0 bar - Pperm = 1.0 bar
98.5 Pfeed = 3.0 bar - Pperm = 2.0 bar
T = 200 C 50
Pfeed = 4.0 bar - Pperm = 3.0 bar

98.0 45
0 250 500 750 1000 1250 1500 0 250 500 750 1000 1250 1500
-1 -1
Qsweep-gas / mLNmin Qsweep-gas / mLNmin

100 100
T = 300 C

98 96

XCO / %

RH / %

96 Pfeed = 2.0 bar - Pperm = 1.0 bar

Pfeed = 3.0 bar - Pperm = 2.0 bar
Pfeed = 4.0 bar - Pperm = 3.0 bar 90
88 Pfeed = 2.0 bar - Pperm = 1.0 bar
Pfeed = 3.0 bar - Pperm = 2.0 bar
92 86
T = 300 C Pfeed = 4.0 bar - Pperm = 3.0 bar
0 300 600 900 1200 1500 0 300 600 900 1200 1500
Qsweep-gas / mLNmin-1 Qsweep-gas / mLNmin-1

Fig. 9 e CO conversion and H2 recovery as a function of the sweep-gas flow rate for the WGS reaction in the PdeAg MR
(GHSV [ 1200 LN kgL1cat h
, feed composition: 4.70% CO, 34.78% H2O, 28.70% H2, 10.16% CO2 balanced in N2).
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 5 ( 2 0 1 0 ) 1 2 5 9 6 e1 2 6 0 8 12605

from 57.3 to 66.8% when the sweep flow rate increases from 100 reforming processes and renewable hydrocarbon feedstocks.
to 1000 mLN min1. Operating at 300  C and at the same pres- Therefore, the study of the reactants molar fraction effect on
sure conditions, the recovery increased from 95.1 to 99.2%. the WGS MR performance gains an important relevance.
Slight decreases on CO conversion and H2 recovery are The studies concerning the influence of the CO and H2O
observed when using sweep-gas flow rates higher than content in the feed were performed in vacuum mode for
1000 mLN min1, particularly at 200  C. The use of a very high a better qualitative evaluation of the PdeAg MR performance.
sweep-gas flow rate can lead to a small temperature decrease The WGS reaction was conducted at GHSV 1200 LN kg1 1
cat h ,

on the outer surface of membrane, decreasing the H2 perme- Pfeed 1 bar and in the 200e300 C temperature range.
ability (Arrhenius law e equation (7)).
The positive effects of the sweep-gas flow rate are again in Influence of the CO content in the feed. From the Le
accordance with literature data. In particular, Tosti el al. [38] Chateliers principle, an increase of the CO concentration in the
report CO conversion enhancement from 84% to total conver- gas feed (lower H2O/CO ratios) favors the CO equilibrium
sion when the sweep-gas flow rate was increased from 230 to conversion, as illustrated in Fig. 10 (dotted line). However, and
470 mL min1. The authors used a MR containing a membrane as reported previously, in a conventional packed-bed reactor
selective only to hydrogen (60-mm Pd layer on a ceramic operating at conditions far from equilibrium (kinetic regime)
support) operating at 330  C and using the stoichiometric feed the CO content in the feed has a negative effect over the activity
ratio (H2O/CO 1) containing only CO plus H2O. of the copper-based WGS catalyst [28]. Despite the importance
It is interesting to highlight the fact that the use of a sweep- of studying the performance of the WGS MR for various reac-
gas plays a more important role in the improvement of the CO tants molar ratios, it is also interesting to observe the influence
conversion when the MR utilizes a membrane 100% selective of the CO feed content. The results are indicated in Fig. 10,
to H2. This occurs because otherwise there is permeation of where the CO concentration was varied from 2.0% to 10.0%.
other reactants through the membrane to some extent. Cris- The experimental results obtained for the PdeAg MR show
cuoli et al. [11] performed the WGS reaction on a mesoporous that H2 recovery is favored by the reactants molar ratio, driving
MR at T 322  C, H2O/CO 0.73 with both retentate and up the CO conversion, overcoming the equilibrium restrictions
permeate pressures at 1 bar. A decrease in the CO conversion for all the tested conditions (Fig. 10). Particularly, at 300  C and
was observed for sweep-gas flow rates higher than H2O/CO ratio 17.39 (2% CO in the feed), a shift of 11% beyond
100 mL min1 due to the counter-diffusion effect of N2 (sweep- the equilibrium value of XCO was reached e cf. Fig. 10.
gas) from the shell to the reaction side. A marginal improvement on the conversion (<1%) was obtained
at 200  C and H2O/CO ratio 3.48; also only 8.9% H2 recovery was
3.2.4. Influence of the steam-to-CO ratio obtained. At these conditions H2 permeability is highly affected
Commercial processes involving the WGS reaction are by temperature and simultaneously by the stronger co-
operated in the presence of excess of steam (i.e. high steam- adsorption of other species at the Pd surface, such as carbon
to-carbon monoxide ratios) to favor the thermodynamic monoxide. For each temperature, it is important to outline that
equilibrium shift. However, from an economic point of view, the MR performance in terms of CO conversion follows the
the large amount of water vapor required is disadvanta- opposite trend as the indicated by the thermodynamic equilib-
geous because it increases the energy consumption (in its rium. Consequently, the enhancement of the CO conversion
vaporization). On the other hand, PdeAg MRs must possess (experimental vs. equilibrium one) increases with the reactants
versatility to operate with WGS feeds from different molar ratio (lower CO concentration in the feed gas stream). This

CO feed content / % CO feed content / %

10 . 0 4.7 2.0 10.0 4.7 2.0
100 70

98 T = 300 C
T = 200 C
94 T = 250 C T = 250 C
XCO / %

RH / %


MR: T = 200 C 20 T = 200 C
88 MR: T = 250 C
T = 300 C
MR: T = 300 C
86 10

84 0
2 4 6 8 10 12 14 16 18 2 4 6 8 10 12 14 16 18
H2O/CO feed ratio H2O/CO feed ratio

Fig. 10 e CO conversion and H2 recovery as a function of the CO feed content for the WGS reaction in the PdeAg MR
(GHSV [ 1200 LN kgL1cat h
and feed composition: 2.0e10.0% CO, 34.78% H2O, 28.70% H2, 10.16% CO2 balanced in N2). Dotted
lines represent the CO equilibrium conversions for the specified inlet conditions.
12606 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 5 ( 2 0 1 0 ) 1 2 5 9 6 e1 2 6 0 8

H 2O feed content / % H2O feed content / %

20.0 34.78 40.0 20.0 34.78 40.0
100 60
T = 300 C
T = 200 C
T = 250 C 50
T = 300 C
40 T = 250 C
XCO / %

RH / %

20 T = 200 C
MR: T = 200 C
MR: T = 250 C
MR: T = 300 C 10

75 0
4 5 6 7 8 9 10 4 5 6 7 8 9 10
H2O/CO feed ratio H2O/CO feed ratio

Fig. 11 e CO conversion and H2 recovery as a function of the H2O feed content for the WGS reaction in the PdeAg MR
(GHSV [ 1200 LN KgL1cat h
and feed composition: 4.70 % CO, 20.0-40.0 % H2O, 28.70 % H2, 10.16% CO2 balanced in N2). Dotted
lines represent the CO equilibrium conversions for the specified inlet conditions.

can be explained not only because the catalyst becomes more

active for lower CO contents as a result of the negative partial 4. Conclusions
order with respect to CO [28], but also because the effective
membrane permeability towards hydrogen increases as the CO In this work, PdeAg membranes were successfully prepared
concentration decreases. Indeed, the strong covalent interac- by an innovative annealing and diffusion welding technique
tion of CO with the membrane surface affects H2 permeation. in order to separate and produce ultra-pure hydrogen. The
It is therefore possible to conclude that the PdeAg MR combination of an active low-temperature CuO/ZnO/Al2O3
conducts the WGS reaction on the kinetic regime. Therefore, it catalyst with a high H2 permeable and selective PdeAg self-
is also possible to infer that the non-equilibrium MR conver- supported membrane resulted in a WGS membrane reactor
sions obtained depend not only on the operating conditions that exhibited significant improvements in comparison to
but also on the WGS reaction path, as generally referred by analogous systems reported in literature. The influence of
Barbieri et al. [5]. various parameters, such as temperature, pressure, sweep-
gas mode, reactants feed composition, and feed space Influence of the H2O content in the feed. In principle, velocity, was investigated in a detailed parametric study.
an MR gives the same level of conversion at a lower ratio of For this effective H2 separation/production system, higher
steam-to-carbon monoxide as compared with that achieved CO conversions can generally be achieved at lower tempera-
in a conventional reactor operating at a higher ratio. It is tures where the equilibrium conversion is favored. On the
obvious, therefore, that a membrane reactor serves to reduce other hand, H2 recovery can be improved increasing the
the amount of steam needed to achieve reasonable levels of operating temperature and/or applying a higher H2 partial
conversion. A set of experiments was then performed where pressure difference between the retentate and permeate sides
the H2O concentration was varied from 20.0% to 40.0%. of the dense PdeAg membrane. The performance of the WGS
It is clear from Fig. 11 that, in general, the water vapor MR can also be improved operating the system at lower feed
content enhances the PdeAg MR performance in the space velocities. Such conditions promote a higher residence
temperature range of 200e300  C. In the literature it is repor- time in the catalytic bed as well as permit that more H2
ted that the water content is crucial for the performance of permeates the Pd membrane interface.
a commercial CuO/ZnO catalyst [19]. Operating at conversions The best operating mode to achieve higher MR perfor-
beyond the thermodynamic equilibrium, the MR system mance is using a sweep stream for increasing the hydro-
behaves in accordance with Le Chateliers principle (see gen permeation driving force. It is noteworthy that around
dotted lines in Fig. 11). This fact indicates that the variation of 100% of CO conversion and almost complete H2 recov-
each reactant concentration (CO or H2O) has different effects ery can be achieved operating the MR at 300  C with
on the WGS MR performance. Once again, the experimental a GSHV 1200 LN kg1 cat h
, Pfeed 4 bar, Pperm 3 bar and
results show that H2 recovery is favored by increasing the using 1000 mLN min1 of sweep-gas. Because the Pd-Ag
reactants molar ratio, driving up the CO conversion e cf. membrane prepared cannot withstand high-pressure
Fig. 11. At 300  C, where the H2 recovery is higher, CO differences between retentate and permeate sides (above 2
conversion was promoted from 91.5% to 97.2% by increasing bar), the feed and the sweep-gas pressures were both
the H2O/CO ratio from 4.26 to 9.58. Even operating the MR in increased for improving the hydrogen permeation. This
vacuum mode and at lower temperatures, the results obtained operating mode resulted in improved reaction rate and
follow the same trend as reported elsewhere ([7,40]). hydrogen recovery.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 5 ( 2 0 1 0 ) 1 2 5 9 6 e1 2 6 0 8 12607

Higher purge-gas flow rates could increase the removal of [10] Casanovas A, Saint-Gerons M, Griffon F, Llorca J.
hydrogen, promoting the CO conversion, due to the large Autothermal generation of hydrogen from ethanol in
driving force for hydrogen permeation. However, the CO a microreactor. Int J Hydrogen Energy 2008;33:1827e33.
[11] Criscuoli A, Basile A, Drioli E. An analysis of the performance
conversion was essentially constant for sweep-gas flow rates
of membrane reactors for the water-gas shift reaction using
above 1000 mLN min1, indicating that such flow rate was gas feed mixtures. Catal Today 2000;56:53e64.
large enough for the removal of the hydrogen produced. [12] Das D, Veziroglu TN. Advances in biological hydrogen
It was also found that changing the reactants concentra- production processes. Int J Hydrogen Energy 2008;33:6046e57.
tion in the feed has opposite effects on the MR performance. [13] Davis WD. Diffusion of gases through metals: I. Diffusion of
While carbon monoxide conversion (and also hydrogen gases through palladium. US Atomic Energy Commission
recovery) is improved by a higher water content, following the Report No. KAPL-1227; 1954.
[14] Deluga GA, Salge JR, Schmidt LD, Verykios XE. Renewable
thermodynamic trend, the reverse effects are noticed for
hydrogen from ethanol by autothermal reforming. Science
higher CO concentrations in the feed (following the opposite 2004;303:993e7.
trend as the thermodynamic limit). It was concluded that the [15] Dittmeyer R, Hollein V, Daub K. Membrane reactors for
PdeAg MR operates in kinetic regime, where the non-equi- hydrogenation and dehydrogenation processes based on
librium conversions obtained depend not only on the oper- supported palladium. J Mol Catal A Chem 2001;173:135e84.
ating conditions but also on the WGS reaction path. [16] Dyer CK. Fuel cells for portable applications. J Power Sources
Finally, it is important to remark that the ultra-pure
[17] Gallucci F, De Falco M, Tosti S, Marrelli L, Basile A. Co-current
hydrogen stream produced (permeate stream) is suitable to
and counter-current configurations for ethanol steam
directly feed a PEMFC. On-going studies concerning the reforming in a dense Pd-Ag membrane reactor. Int J
thickness reduction of the membrane (<1 mm) are being Hydrogen Energy 2008;33:6165e71.
carried out in order to reduce the membrane costs, increasing [18] Gallucci F, Paturzo L, Basile A. A simulation study of the
the permeance while keeping the selectivity towards H2. steam reforming of methane in a dense tubular membrane
reactor. Int J Hydrogen Energy 2004;29:611e7.
[19] Gines ML, Amadeo N, Laborde M, Apestegua CR. Activity and
structure-sensitivity of the Water-gas shift reaction over Cu-
Acknowledgments Zn-Al mixed-oxide catalysts. Appl Catal A 1995;131:283e96.
[20] Holladay JD, Hu J, King DL, Wang Y. An overview of hydrogen
production technologies. Catal Today 2009;139:244e60.
Diogo Mendes is grateful to the Portuguese Foundation for
[21] Hwang KR, Ihm SK, Park JS. A catalytic membrane reactor for
Science and Technology (FCT) for his doctoral grant (reference:
water-gas shift reaction. Korean J Chem Eng 2010;27:816e21.
SFRH/BD/22463/2005). The authors also acknowledge financing [22] ISO. Hydrogen fuel e product specification e Part 2: proton
from FCT through the project PTDC/EQU/ERQ/66045/2006. exchange membrane (PEM) fuel cell applications for road
vehicles. ISO/TS 14687e2. Int Organization for
Standardization; 2008.
references [23] Iulianelli A, Longo T, Basile A. Methanol steam reforming
reaction in a PdeAg membrane reactor for CO-free hydrogen
production. Int J Hydrogen Energy 2008;33:5583e8.
[1] Adrover ME, Lopez E, Borio DO, Pedernera MN. Simulation of [24] Larminie J, Dicks A. Fuel cell systems explained. 2nd ed.
a membrane reactor for the WGS reaction: pressure and Chichester: John Wiley & Sons; 2003.
thermal effects. Chem Eng J 2009;154:196e202. [25] Lin WH, Liu YC, Chang HF. Hydrogen production from
[2] Balat H, Krtay E. Hydrogen from biomass e Present scenario oxidative steam reforming of ethanol in a palladium-silver
and future prospects. Int J Hydrogen Energy 2010;35:7416e26. alloy composite membrane reactor. J Chin Inst Chem Eng
[3] Barbieri G, Brunetti A, Granato T, Bernardo P, Drioli E. 2008;39:435e40.
Engineering evaluations of a catalytic membrane reactor for [26] Ma DH, Lund CRF. Assessing high-temperature water-gas
the water gas shift reaction. Ind Eng Chem Res 2005;44: shift membrane reactors. Ind Eng Chem Res 2003;42:711e7.
7676e83. [27] Mejdell AL, Jondahl M, Peters TA, Bredesen R, Venvik HJ.
[4] Barbieri G, Brunetti A, Tricoli G, Drioli E. An innovative Effects of CO and CO2 on hydrogen permeation through
configuration of a Pd-based membrane reactor for the a similar to 3 mm Pd/Ag 23 wt.% membrane employed in
production of pure hydrogen e experimental analysis of a microchannel membrane configuration. Sep Purif Technol
water gas shift. J Power Sources 2008a;182:160e7. 2009;68:178e84.
[5] Barbieri G, Scura F, Brunetti A. Mathematical modeling of [28] Mendes D, Garcia H, Silva VB, Mendes A, Madeira LM.
Pd-alloy membrane reactors. In: Mallada R, Mendendez M, Comparison of nanosized gold-based and copper-based
editors. Membrane science and technology series. catalysts for the low-temperature wateregas shift reaction.
Amsterdam, The Netherlands: Elsevier; 2008b. Ind Eng Chem Res 2009;48:430e9.
[6] Basile A, Chiappetta G, Tosti S, Violante V. Experimental and [29] Mendes D, Mendes A, Madeira LM, Iulianelli A, Sousa JM,
simulation of both Pd and Pd/Ag for a water gas shift Basile A. The water-gas shift reaction: from conventional
membrane reactor. Sep Purif Technol 2001;25:549e71. catalytic systems to Pd-based membrane reactors e a review.
[7] Basile A, Criscuoli A, Santella F, Drioli E. Membrane reactor Asia-Pac. J Chem Eng 2010;5:111e37.
for water gas shift reaction. Gas Sep Purif 1996;10:243e54. [30] Olah GA, Goeppert A, Prakash GKS. Beyond oil and gas: the
[8] Bi YD, Xu HY, Li WZ, Goldbach A. Water-gas shift reaction in methanol economy. Weinheim, Germany: Wiley-VCH; 2006.
a Pd membrane reactor over Pt/Ce0.6Zr0.4O2 catalyst. Int J [31] Paglieri SN, Way JD. Innovations in palladium membrane
Hydrogen Energy 2009;34:2965e71. research. Sep Purif Methods 2002;31:1e169.
[9] Brunetti A, Barbieri G, Drioli E. Upgrading of a syngas mixture [32] Patil CS. Membrane reactor technology for ultrapure
for pure hydrogen production in a Pd-Ag membrane reactor. hydrogen production. Ph.D. thesis, University of Twente, The
Chem Eng Sci 2009;64:3448e54. Netherlands; 2005.
12608 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 5 ( 2 0 1 0 ) 1 2 5 9 6 e1 2 6 0 8

[33] Pinacci P, Broglia M, Valli C. Evaluation of the water gas shift [37] Tosti S, Bettinali L, Violante V. Rolled thin Pd and PdeAg
in a palladium membrane reactor. In: 9th international membranes for hydrogen separation and production. Int J
conference on catalysis in membrane reactors (ICCMR9), Hydrogen Energy 2000a;25:319e25.
Lyon e France; 2009. p. 77e78. [38] Tosti S, Violante V, Basile A, Chiappetta G, Castelli S, De
[34] Stankiewicz A, Moulijn JA. Process intensification. Ind Eng Francesco M, et al. Catalytic membrane reactors for tritium
Chem Res 2002;41:1920e4. recovery from tritiated water in the ITER fuel cycle. Fusion
[35] Tiemersma TP, Patil CS, Annaland MV, Kuipers JAM. Eng Des 2000b;49:953e8.
Modelling of packed bed membrane reactors for autothermal [39] Uemiya S, Matsuda T, Kikuchi E. Hydrogen permeable
production of ultrapure hydrogen. Chem Eng Sci 2006;61: palladium silver alloy membrane supported on porous
1602e16. ceramics. J Membr Sci 1991a;56:315e25.
[36] Tosti S, Basile A, Bettinali L, Borgognoni F, Chiaravalloti F, [40] Uemiya S, Sato N, Ando H, Kikuchi E. The water gas shift
Gallucci F. Long-term tests of PdeAg thin wall permeator reaction assisted by a palladium membrane reactor. Ind Eng
tube. J Membr Sci 2006;284:393e7. Chem Res 1991b;30:585e9.