International Journal of Hydrogen Energy 32 (2007) 5059 – 5066

www.elsevier.com/locate/ijhydene

Purification process for chemical storage of hydrogen for fuel cell

vehicles applications
Yousef Swesi a , Didier Ronze b,c,∗ , Isabelle Pitault a , Roland Dittmeyer d , Fabien Heurtaux e
a Laboratoire de Génie des Procédés Catalytiques, CNRS/CPE, BP 2077, 69616 Villeurbanne, Cedex, France
b Université de Lyon, Lyon F-69003, France
c IUT A, Département Génie Chimique, Villeurbanne F-69622, France
d Karl Winnacker Institute, DECHEMA, Theodor Heuss Allee 25, D-60486 Frankfurt am Main, Germany
e RENAULT, Direction des Technologies Automobiles Avancées, 1 avenue du Golf, 78288 Guyancourt, France

Received 28 April 2007; received in revised form 9 August 2007; accepted 9 August 2007
Available online 24 October 2007

Abstract
An alternative way to provide hydrogen in fuel cell vehicles (FCV) consists of storing hydrogen in liquid hydrocarbons. Supplying hydrogen
is then performed, e.g. by dehydrogenation of cycloalkanes and purification through various methods. Two main separation processes were
investigated: membranes (zeolite and palladium, both on porous ceramic supports) and adsorption. Separation through zeolite membranes is shown
to be inappropriate for FC application since the toluene content in the permeate is still too high (> 2000 ppm). Palladium membranes are more
promising for this application. The permeation for the two membranes were found at 4.73 × 10−4 mol m−2 s−1 and 5.06 × 10−2 mol m−2 s−1 ,
respectively (at P of 50 kPa and toluene inlet mole fraction of 12%, temperature 400 ◦ C), while the separation factor of the two membranes
for the hydrogen over toluene (Sf(H2 /Tol) ) was found infinite and 40, respectively. When toluene is present at high concentration, the diffusion
of hydrogen is remarkably hindered due to a relatively strong adsorption of toluene in the membranes (on the micropores in case of zeolite,
and on the Pd film in case of palladium).
䉷 2007 International Association for Hydrogen Energy. Published by Elsevier Ltd. All rights reserved.

Keywords: Hydrogen storage; Purification; Adsorption; Membrane

1. Introduction of 50–250 ◦ C at atmospheric pressure [3–6] or in three-phase

To use hydrogen in power supply devices (fuel cells), an or-
ganic liquid hydride (cyclo-alkane) is currently considered for
hydrogen storage as an alternative to liquefied or high-pressure
gas hydrogen tanks [1,2]. The whole process will include the
feeding of the vehicle with a cycloalkane, its dehydrogenation
in the vehicle (giving an aromatic compound), the discharge
of the aromatic from the vehicle, and its regeneration in a
hydrogenation plant. The hydrogenation of aromatics was in-
vestigated by several authors. The reaction can be performed
either in gas–solid fixed bed reactors in the temperature range
∗ Corresponding author. IUT A, Département Génie Chimique, Villeur-
banne F-69622, France. Tel.: +33 4 72 43 17 61; fax: +33 4 72 43 16 73.
E-mail addresses: swy@lobivia.cpe.fr (Y. Swesi), dro@lgpc.cpe.fr,
didier.ronze@lgpc.cpe.fr (D. Ronze).
0360-3199/$ - see front matter 䉷 2007 International Association for Hydrogen Energy. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijhydene.2007.08.015
semibatch or batch reactors at pressures of 2.0–4.0 MPa
and at temperatures of 95–140 ◦ C [7]. In this configuration,
cycloalkane is used as a hydrogen carrier. In the vehicle, the
dehydrogenation step itself is split as follows: heating and
evaporation of the cycloalkane, its endothermic dehydrogena-
tion (heat needed for that reaction being supplied by a combus-
tion reaction of toluene and air (cf. Fig. 1)), cooling of outlet
gas (separation of the condensable products in the mixture of
aromatic compounds and hydrogen), and purification of the
produced hydrogen. The purification steps can occur after gas
cooling and/or during the dehydrogenation step.
Main advantages of this process are on one hand the absence
of direct CO2 emissions from the vehicle, and on the other hand
milder reaction conditions, compared to those of hydrocarbon
reforming. To evaluate the feasibility of this process, we chose
the MTH (methylcyclohexane–toluene–hydrogen) cycle. These
5060 Y. Swesi et al. / International Journal of Hydrogen Energy 32 (2007) 5059 – 5066

liquid MCH at ambient

Hydrogenation
temperature and 4bar

Va porisation and
H2 heating 380˚C, 4bar

Pt /alumina
+ 3H2

Option 1 Option 2
75% H 2 + 25%Tol
Heat

Burner Z- membrane
Condensation at
or Pd-membrane at
ambient T
380˚C , 4bar
Air+Toluene

Toluene liquid
H 2 <5% Tol
ambient T
Condenser

Option 1a Heater Option 1b

Toluene

Temperature
Z- membrane Swing
or Pd-membrane at Adsorption
4bar, T

Condenser
Air

Toluene
pure H 2 < 1 ppm Tol

Fuel Cell

Fig. 1. Methylcyclohexane–toluene–hydrogen cycle.

molecules were selected due to their advantageous properties:
important hydrogen storage capacity (6.1 wt%), lower toxicity
(compared to benzene), and good reactivity in the dehydro-
genation over a well-known catalyst: the endothermic reaction
of methylcyclohexane dehydrogenation (reaction heat about
205 kJ mol−1 ) is usually performed on metal/alumina catalysts
at 280–410 ◦ C. Moreover, at high operating pressure, complete
conversion of methylcyclohexane is hard to achieve due to ther-
modynamical equilibrium limitations [8–11].
The whole study of this process has been divided into two
parts (reaction and separation). This paper deals only with the
second one, i.e. the choice of separation methods for toluene
and hydrogen and the possible configurations of coupling the
reaction and separation steps.
Among the various available separation methods, two can
be conveniently used for hydrogen purification in vehicles:
adsorption and diffusion through membranes. Both are inves-
tigated in this study. Out of many possible process configura-
tions, two options were proposed (Fig. 1). The first one (option
1) consists of a condensation unit at the outlet of the reactor,
at ambient temperature to reduce the toluene concentration in
the hydrogen, followed by either membranes (1a) or adsorption
(1b) [12].
The second one (option 2) consists of a separation at
high toluene concentration (25 vol%) and high temperature
(≈ 380 ◦ C) at the outlet of the reactor by diffusion through
membranes, with the need of an additional adsorption unit if
toluene remains in the purified gas. In both cases, the pro-
duced liquid toluene is sent to an ex-situ hydrogenation unit to
produce MCH.
Various membranes can be used to purify hydrogen, for ex-
ample, silica molecular sieves, carbon molecular sieves, palla-
dium, and zeolite membranes. Analysis of the literature leads
to the selection of the two last membranes as promising candi-
dates for this separation.
Palladium membranes are very efficient for separation of hy-
drogen from other gases, because the separation in this type
of membrane depends on dissociative adsorption. A very sig-
nificant review [13] (1400 references) was published in 2002,
summarizing the studies and advances made on this type of
membrane. Different Pd membranes on various supports have
been studied, for instance, Pd on porous glass tubes [14] or on
 Y. Swesi et al. / International Journal of Hydrogen Energy 32 (2007) 5059 – 5066 5061

porous stainless steel [15], and Pd/alumina composites on alu- Permeate

Graphite seals
Membrane
mina supports [16]. Pd membranes produced by Li et al. [17]
were investigated using pure hydrogen at temperatures between
320 and 577 ◦ C; the hydrogen permeation flow reached up to Gas mixture Retentate
6.67×10−3 m3 m−2 s−1 for hydrogen pressures from 1.34×105
to 2.7 × 105 Pa at the high-pressure side while the low-pressure
side was maintained at 105 Pa. Pure Pd membranes are known Enameled zone
to become brittle in the presence of hydrogen at temperatures Sweep gas
below 300 ◦ C, due to a phase change in the Pd–H system [13].
Fig. 2. Zeolite membrane module.
The separation in zeolitic membranes is controlled by
the diffusion in the porous structure of the zeolite and by
adsorption–desorption phenomena. Many types of zeolites
(MFI, ZSM-5, Silicalite, NaX,…) on different supports (porous external diameter 0.006 and 0.01 m, respectively). Whereas
ceramics, Vycor 䉸 , Teflon䉸 ) are available. Several preparation the exact thickness of the zeolite is difficult to measure due
methods were investigated by Dalmon and his team [18,19], to the composite nature of the material, permeation modelling
to prepare composite membranes by hydrothermal synthesis suggested a value in the range of 5.20 m.
of zeolites within the porous network of a support. Dong The module (Fig. 2) consists of a cylindrical membrane
et al. [20] prepared MFI zeolite membranes used for the placed in a stainless steel tube. High-temperature graphite seals
separation of linear hydrocarbons and hydrogen. The zeolite at the membrane ends isolate the internal compartment from
membrane becomes selective for hydrogen in the temperature the external one. Two inlets (gas mixture and sweep gas) and
range between 200 and 500 ◦ C. Recently, Lai and Tsapatsis two outlets (permeate and retentate) are provided. An oven al-
[21] synthesized a siliceous ZSM-5 zeolite membrane for lows setting of the temperature of the module.
the separation of the cyclic hydrocarbons benzene, p-xylene,
o-xylene, and cyclohexane. Giroir-Fendler et al. [22] prepared
2.1.2. Palladium membrane
MFI membrane for the catalytic dehydrogenation of i-butane
The palladium membrane was deposited on the inner side of
on porous ceramics (-Al2 O3 ); at 400 ◦ C, their results revealed
an asymmetric porous ceramic support tube (pore size of the
a permeance of 2 × 10−4 and 2 × 10−6 m3 m−2 s−1 Pa−1
surface layer 0.1 m, supplier Inocermic GmbH, Hermsdorf,
for hydrogen and i-butane, respectively. Arruebo et al. [23]
Germany). The dimensions of the membrane were as follows:
used MFI (silicalite) membranes on different supports for the
length 0.10 m, internal/external diameter 0.006/0.01 m, thick-
separation of hydrocarbons from natural gas at different tem-
ness of the palladium layer 9.10 m, area 2.35 × 10−3 m2 . The
peratures and pressures. Moron et al. [24] synthesized zeolite
module consists of two flanges enclosing the membrane and a
A membranes by secondary growth on the surface of macro-
non-porous alumina tube having a similar thermal expansion
porous -alumina and stainless steel tubular supports; their
coefficient (Fig. 3). The membrane is not enamelled at its two
membrane was used for the separation of H2 /C3 H8 mixtures,
ends, which are closed with graphite seals. The module is heated
and high separation factors of 7.28 were reported at 50 ◦ C.
by a heater cable; the temperature was measured by three ther-
The purpose of this study is to demonstrate the process fea-
mocouples along the membrane. A test with pure nitrogen was
sibility. In the first part, two types of membranes were used for
run on this membrane in order to confirm the absence of defects
H2 /toluene separation: the first membrane (Pd) was prepared at
in the palladium layer; our results (leak flux between 5 × 10−10
the Karl-Winnacker-Institute (DECHEMA e.V., Frankfurt am
and 1.67 × 10−9 m3 s−1 Pa−1 ) show higher leak rates than
Main, Germany) and the second one (zeolite) at the Catalysis
those reported by the supplier (1.19×10−11 m3 s−1 Pa−1 ), but
and Environment Research Institute (IRCELyon Villeurbanne,
remain acceptable in view of the measurement errors and the
France). In the second part, the membrane performances are
ideal permselectivity H2 /N2 defined as the single-gas hydro-
compared to those of some adsorbents (adsorbent performances
gen flux vs. the single-gas nitrogen flux through the membrane
have previously been measured and published [25]).
for the same temperature and driving force (hydrogen partial
pressure difference). In fact, this tells that the expected H2 flux
2. Experimental section
is 1000 times higher than the N2 flux.
2.1. Membranes
2.1.3. Permeation set-up
2.1.1. MFI zeolite membranes An experimental set-up was developed to test the palladium
MFI zeolite composite membranes were prepared inside and zeolite membranes (Fig. 4).
the pores of asymmetric porous alumina tubes (pores of The membrane can be supplied with three different gas
12–0.8–0.2 m successively) provided by Pall-Exekia. These mixtures:
supports were 0.15 m long, with 1 cm diameter, and 1 cm
enamel endings for sealing purposes. A detailed study of • nitrogen used as sweep gas (co-current or counter-
this preparation has been described in Refs. [19,26,27]. The current); the flow rate is controlled by a flow meter (range
membrane area is 2.8 × 10−3 m2 (length 0.15 m, internal and 0.500 cm3 min−1 ),
5062 Y. Swesi et al. / International Journal of Hydrogen Energy 32 (2007) 5059 – 5066
Retentate
Graphite seals
Heater cable
Sweep gas
Fig. 3. Palladium membrane module.
N2 5 1
4 ΔP
H2 5 1
2 Module
6
Fig. 4. Experimental set-up used for gas mixture separation. (1) mass flow rate; (2) saturator; (3) pressure controller; (4) pressure indicator; (5) pressure
controller; (6) toluene liquid; (7) condenser; (8) pneumatic valve.
• different test mixtures of n-butane, toluene, hydrogen, and
nitrogen. The flow rates are controlled by three flow meters
(0.50 cm3 min−1 for nitrogen, 0.100 cm3 min−1 for hydro-
gen, and 0.50 cm3 min−1 for n-butane),
• hydrogen saturated with liquid toluene.
Before the experiments, the flow meters were calibrated us-
ing a volumetric piston flow meter (Model Dry Cal NEXUS
DCL-ML). The flows of permeate and retentate were measured
(at ambient conditions) by the volumetric piston flow meter.
The transmembrane pressure P is defined as the pressure
difference between retentate and permeate (so over the mem-
brane), the latter being fixed by the pressure controller (3) in
Fig. 4; it was measured by a differential pressure sensor and
controlled by a pneumatic valve. The membrane temperature
was controlled via a PID controller; temperature increase and
decrease were executed at a controlled rate (1 ◦ C min−1 ) in or-
der to minimize mechanical stresses due to thermal expansion.
• Analysis system: Gas compositions were measured with a
gas chromatograph equipped with two detectors: a flame
Permeate
Membrane
Porous ceramic
support
Dense ceramic
tube
Gas mixture
+
3 Permeated gas
- 7
8 Retentate gas
4
Membrane
GC
ionization detector (FID) for the analysis of hydrocarbons
(toluene, methylcyclohexane, n-butane,…) and a thermal
conductivity detector (TCD) for the analysis of hydrogen.
The characteristics of the chromatographic analysis are as
follows:
◦ capillary column: CP-PoraBOND U fused silica STUD
25 m × 0.32 mm, df = 7 m,
◦ detectors: FID (T = 250 ◦ C) and TCD (T = 250 ◦ C),
◦ carrier gas: H2 (for FID) and N2 (for TCD).
• Temperature programming: T = 150 ◦ C (5 min), rise of
15 ◦ C min−1 until 250 ◦ C, T = 250 ◦ C (10 min).
2.2. Adsorption
The experimental set-up used for adsorption experiments
has been published recently [25]. It consists of a stainless steel
column (internal and external diameter 0.0041 and 0.006 m,
respectively, length 0.15 m) filled with the adsorbent to be
tested; the external porosity of the bed was 0.38. The toluene
concentration in the carrier gas (helium) at the inlet was set
by two Dewar saturators filled with liquid hydrocarbon in
 Y. Swesi et al. / International Journal of Hydrogen Energy 32 (2007) 5059 – 5066
order to reach the equilibrium. The system includes two
6-way valves which allow to work in two types of operation,
either co-current (adsorption and desorption in the same direc-
tion) or counter-current (adsorption and desorption in opposite
directions).
3. Results
3.1. Zeolite membrane
3.1.1. Permeation measurement
The set-up was first validated by a series of experiments using
n-butane/H2 mixtures in the same conditions as in the literature;
the results were found similar to those published [22]. Pure
hydrogen permeance  was measured at room temperature;
the permeance of 0.19 × 10−6 mol m−2 s−1 Pa−1 obtained is in
agreement with the reference of the membrane supplier ( ≈
0.17 × 10−6 mol m−2 s−1 Pa−1 ).
Hydrogen and nitrogen permeation on a -Al2 O3 and
-Al2 O3 membrane is the sum of the contributions of Knudsen
diffusion and Poiseuille flow [28]:
Fp = FpV Pave + FpK .
The first term FpV represents the viscous contribution to the
permeance total, the second term FpK comes from the Knudsen
diffusion, Pave is the average pressure between retentate and
permeate side. The results presented in Fig. 5 show that hydro-
gen permeance is linear vs. the pressure; the very low value of
the slope (about 5 × 10−6 ), i.e. the value of coefficient FpV ,
indicates that the convective transport can be neglected; there-
fore, the contribution of the Knudsen flow through the zeolite
membrane itself is about 99.82% of the total.
Moreover, these results give a low nitrogen permeation ,
about 0.032 mol m−2 s−1 Pa−1 , which is very small compared
with the hydrogen permeation of 0.19 mol m−2 s−1 Pa−1 .
0.04
0.03
ΠN2 (μmol. m-2. s-1. Pa-1)
0.02
0.01
0 range of the gas phase after MCH conversion [11]) showed
150 200 250 300
Average pressure (kPa)
Fig. 5. Nitrogen permeation through zeolite membrane (P = 3.5 kPa at
450 ◦ C).
5063
0.25
n-C4
0.20
Π (μmol. m-2. s-1. Pa-1)
H2
0.15
0.10
0.05
0.00
0 100 200 300 400
Temperature (˚C)
Fig. 6. Permeance of hydrogen and n-butane mixtures reached with
the MFI–alumina composite membrane (H2 /N2 /n-C4 mole fraction of
14.9/69.9/15.2, sweeping flow rate 8.7 × 10−7 m3 s−1 , outlet pressure
1.22 × 105 Pa).
The other widely used test on the membrane was the
hydrogen/n-butane separation vs. temperature (cf. Fig. 6). The
results show an increasing n-butane flux at temperatures lower
than 100 ◦ C and a decrease of this flux at higher temperatures,
as the hydrogen flux appears. This phenomenon is due to
the n-butane adsorption at low temperature, which blocks the
micropores and hinders the hydrogen permeation.
The above results obtained by two tests of permeation con-
firmed there is no major defect in the membrane.
3.1.2. Hydrogen/toluene permeation
Separation of toluene from H2 /toluene mixtures was done
at different temperatures (between 200 and 450 ◦ C) and trans-
membrane pressures (between 34 and 107 kPa). Results with-
out sweep gas (representing the real operating conditions in the
vehicle) are presented in Fig. 7. The hydrogen permeance, and
therefore the H2 /toluene separation factor, increased with tem-
perature. At room temperature, strong hydrocarbon adsorption
hinders hydrogen diffusion through the membrane. The experi-
ments were performed several times in a random order, and we
noticed that the membrane performance drastically decreased
with time, due to an increase in toluene adsorption. Neverthe-
less, at 400 ◦ C and toluene mole fraction (ytol ) of 13% in hy-
drogen (pressure drop P of 50 kPa), the maximum toluene
flow through the membrane was 2 × 10−4 mol h−1 , whereas
the hydrogen flow was 41 × 10−4 mol h−1 , giving in this case
a H2 /toluene separation factor of 30 (Fig. 7).
Results obtained with different toluene concentrations in the
feed (between 4% and 25%, corresponding to the composition
350 400 a slightly decreasing permeance of hydrogen with increasing
toluene concentration (Fig. 8).
Moreover, the minimum observed toluene concentration in
the permeate is not lower than 0.2%, i.e. 2000 ppm (Fig. 9),
which is too high to feed a fuel cell.
5064 Y. Swesi et al. / International Journal of Hydrogen Energy 32 (2007) 5059 – 5066

Inlet toluene
0.012
mole fraction
14%
Hydrogen 10000
12%

Toluene in permeate side (ppm)

Π μmol. m-2. s-1. Pa-1

10%
0.008 8000
ytol = 12% 8%

ytol = 10%
6000
ytol = 9%
0.004
ytol = 8%
4000

Toluene 2000
0
200 300 400 500
Temperature (˚C) 0
200 300 400 500
50 Temperature (˚C)
ΔP = 34 kPa
ΔP = 50 kPa Fig. 9. Outlet toluene concentration vs. temperature for the MFI–alumina
40 ΔP = 72 kPa composite membrane (hydrogen feed: 6.77×10−6 m3 s−1 , permeate pressure
1.02 × 105 Pa).
ΔP = 94 kPa
Sf (H2/Tol)

30 ΔP = 107 kPa
0.1

20

10
0.08
J (mol. s-1. m-2)

0
200 300 400 500
Temperature (˚C)
H 2/N 2 450˚C
0.04 H 2/Tol 450˚C
Fig. 7. Permeance (a) and separation factor (b) of hydrogen and toluene
mixtures reached with the MFI–alumina composite membrane (H2 flow rate H 2/N 2 375˚C
6.77 × 10−6 m3 s−1 , outlet pressure 1.02 × 105 Pa). H 2/Tol 375˚C
H 2/N 2 350˚C
0
0.012 0 0.4 0.8 1.2 1.6 2
(PR0.5- PP0.5) (kPa )0.5
Π (μmol. m2. s-1. Pa-1)

Hydrogen Fig. 10. Permeation of hydrogen from a mixture with toluene through the
0.008
Pd/alumina composite membrane (hydrogen flow rate 6.77 × 10−6 m3 s−1 ,
toluene concentration between 10% and 14%, inside pressure from 1.25×105
ΔP = 101 kPa to 2 × 105 Pa).
0.004 ΔP = 74 kPa
ΔP = 50 kPa lead to poor membrane performance; under such conditions,
all the hydrogen fed is permeated through the membrane, and
Toluene
0 the transmembrane hydrogen pressure difference (the driving
0 0.05 0.1 0.15 0.2 0.25 force for hydrogen permeation) reaches zero even before the
Toluene mole fraction end of the membrane; the result is a low hydrogen permeance.
Fig. 8. Hydrogen and toluene permeances vs. toluene concentration for the
Toluene/hydrogen separation results for higher flow rates are
MFI–alumina composite membrane (hydrogen feed: 6.77 × 10−6 m3 s−1 , presented in Fig. 10 (according to Sievert’s law, the results are
permeate pressure 1.02 × 105 Pa). given as a function of the difference of the square root of the
hydrogen partial pressure on the two sides of the membrane);
this figure shows that the presence of toluene reduces the per-
3.2. Palladium membrane meated hydrogen flow by 3–10%. This phenomenon could be
attributed to adsorption of toluene on the membrane surface
Preliminary experiments with toluene/hydrogen mixtures which inhibits the adsorption and uptake of hydrogen by the
showed that low hydrogen flow rates (6.77 × 10−6 m3 s−1 ) palladium. The membrane showed almost infinite H2 /toluene
 Y. Swesi et al. / International Journal of Hydrogen Energy 32 (2007) 5059 – 5066
separation factors; no traces of hydrocarbons were detectable
in the permeate.
The hydrogen flow crossing
√ the membrane, at the tem-
perature of 380 ◦ C and P = 1.513 kPa0.5 , was 7.52 ×
10−2 mol m−2 kPa−0.5 (H2 /N2 mixture) in the absence of
toluene and 8.073 × 10−2 mol m−2 kPa−0.5 (H2 /N2 /toluene
mixture) in presence of toluene.
3.3. Adsorption
In this part, the results of the adsorption study are summa-
rized in order to allow a comparison with the membranes. The
performances and the comparison of some adsorbents are pub-
lished elsewhere [25]. The aim of this comparison was to reach
the best balance between several parameters: capacity of ad-
sorption, desorption dynamics, pressure drop, and influence of
humidity on the adsorption. In our process, an adsorbent must
to be able to adsorb a large quantity of toluene in hydrogen but
it also must be able to easily desorb toluene with air from the
fuel cell compressor. Different available adsorbents were tested:
minerals (alumina, silica, zeolites), organics (porous polymers),
and activated carbons.
Activated carbon presents high adsorption capacity and weak
pressure drop compared to the other adsorbents, but desorption
is slow. The zeolite USY is inappropriate due to its very slow
desorption and a significantly higher pressure drop. Among
the porous polymers, the best were the Porapak 䉸 -Q and
Porapak 䉸 -T, which have an optimum balance between toluene
adsorption capacities and desorption dynamics. Moreover,
the study of water adsorption showed that Porapak 䉸 -Q is
suitable for our application, mainly in counter-current mode
that allows to recover a significant quantity of hydrocarbon
adsorbed. No kinetic limitations were found on the selected
adsorbent (Porapak 䉸 -Q), but only thermodynamic control. As
the specifications for the FC feed (toluene content lower than
1 ppm) can be reached (in counter-current desorption mode),
Porapak 䉸 -Q is a convenient candidate for the separation of
toluene from hydrogen.
4. Discussion
In option (1), presented in Fig. 1, a condensation unit right be-
hind the reactor achieves a first separation, as the toluene mole
fraction can be reduced from 25% to 5% vol. (depending on the
condenser temperature), the gas stream being then reheated to
obtain appropriate purification conditions (T > 200 or 300 ◦ C
for membrane separation and T > 50 ◦ C for adsorption). The
experimental results obtained with zeolite membranes showed
that this type of membrane is not permselective enough (cf.
Fig. 7). A residual concentration of toluene in the outlet hy-
drogen flow (at least about 2000 ppm, cf. Fig. 9) was always
observed under our operating conditions and would imply the
use of another separation unit (such as adsorption) after the ze-
olite membrane to reach the FC requirements (< 1 ppm). The
palladium membrane exhibited 100% separation (no detectable
traces of toluene in the purified hydrogen), but an important
drawback of this configuration is the need of heating up the gas
5065
to above 300 ◦ C after having it cooled down for condensation.
This disadvantage is decreased by the adsorption option (1b),
which occurs at lower temperature (< 80 ◦ C).
In option (2) separation is achieved at high temperature
(380 ◦ C) and toluene concentration (25 vol%). The same con-
clusions as before are valid for the zeolite membrane, which
leaves an unacceptably high quantity of toluene in the puri-
fied gas; moreover, the permeance decreased slightly when the
toluene concentration was varied from 2.5% to 25% (cf. Fig. 8).
The palladium membrane gives good separation and reaches a
high hydrogen flux.
In both cases, the presence of toluene reduces slightly (about
10%) the permeance of hydrogen, and the separated toluene
(retentate side) must be condensed. For a typical vehicle ap-
plication, the required membrane area is estimated (with the
FC data given by Brown [29]) at about 7 m2 (T = 374 ◦ C,
P = 0.5 × 105 Pa, and hydrogen flow rate = 1 g s−1 ), that is
not technically unreasonable.
5. Conclusion
The zeolite membrane showed interesting results in terms
of permeance of hydrogen, even if the separation factor still
needs to be improved to reach a lower toluene partial pressure
in the permeate. This separation depends on adsorption and
surface diffusion; the working temperature must be higher than
200 ◦ C, in order to avoid strong adsorption of hydrocarbons
in the pores: at these temperature the permeance was found to
be 4.11 × 10−4 and 4.73 × 10−4 mol m−2 s−1 at 400 ◦ C (with
P of 50 kPa), but the toluene concentration remains too high
for the application considered (FC). Zeolite membranes are not
able to produce 100% pure gas streams (as do thin-film Pd
membranes) which would make a further adsorption step (e.g.
activated carbon) necessary.
On the other hand, hydrogen separation with a palladium
membrane from H2 /toluene mixtures is achieved satisfacto-
rily: under our experimental conditions hydrogen permeance
is 5.02 × 10−2 mol m−2 s−1 at 350 ◦ C and at high temperature
(450 ◦ C) it is 5.79 × 10−2 mol m−2 s−1 (P of 50 kPa, toluene
mole fraction of 12%). For the fuel cell vehicles (FCV) applica-
tion, Pd membranes seem to be more promising than zeolites,
because of their better separation performances. Nevertheless,
their mechanical properties should be improved, for instance,
by using different supports [30] (porous stainless steel) to be
implemented in a future vehicle.
Regarding the purification by adsorption, the Porapak 䉸 -Q
is an appropriate candidate for the separation of toluene from
hydrogen. Presently, the specifications for the FC feed can be
reached in a counter-current desorption mode at relatively low
temperature. This process is relatively complex to control, but
as the membranes today are rather easily breakable, it remains
an interesting way to feed a fuel cell with purified hydrogen.
Acknowledgements
We would like to thank Dr. M. Forissier for his useful and
pertinent advices, the Libyan government for Y. Swesi’s grant,
5066 Y. Swesi et al. / International Journal of Hydrogen Energy 32 (2007) 5059 – 5066
DECHEMA for supplying Pd membranes and a suitable reactor,
and Dr. S. Miachon (IRCELyon CNRS) for supplying zeolite
membranes.
References
[1] Saito Y, Hodoshima S, Shono A. Hydrogen storage in organic hydrides
at low exergy consumption. In: WHEC 16, 13–16 June 2006, Lyon,
France, p. S15–239.
[2] Shudo Y. Demonstrative study on the production of hydrogen and
aromatic compounds originated from biogas in the dairy area. In: WHEC
16, 13–16 June 2006, Lyon, France, p. I–197.
[3] Ali AGA, Ali LI, Aboul-Fotouh SM, Aboul-Gheit AK. Hydrogenation
of aromatics on modified platinum–alumina catalysts. Appl Catal A
1998;170:285–96.
[4] Lindfors LP, Salmi T, Smeds S. Kinetics of toluene hydrogenation on
Ni/A12 O3 catalyst. Chem Eng Sci 1993;48(22):3813–28.
[5] Thybaut JW, Saeys M, Marin GB. Hydrogenation kinetics of toluene on
Pt/ZSM-22. Chem Eng J 2002;90:117–29.
[6] Klvana D, Chaouki J, Kusohrsky D, Chavarie C. Catalytic storage of
hydrogen: hydrogenation of toluene over a nickel/silica aerogel catalyst
in integral flow conditions. Appl Catal 1988;42:121–30.
[7] Toppinen S, Rantakylä T-K, Salmi T, Aittamaa J. The liquid phase
hydrogenation of benzene and substituted alkylbenzenes over a nickel
catalyst in a semi-batch reactor. Catal Today 1997;38:23–30.
[8] Ali JK, Baiker A. Dehydrogenation of methylcyclohexane to toluene in
a pilot-scale membrane reactor. Appl Catal A 1997;155:41–57.
[9] Dittmeyer R, Höllein V, Daub K. Membrane reactors for hydrogenation
and dehydrogenation processes based on supported palladium. J Mol
Catal A 2001;173:135–84.
[10] Zhang LY, Xu GH, An Y, Chen CP, Wang QD. Dehydrogenation of
methyl-cyclohexane under multiphase conditions. Int J Hydrogen Energy
2006;31:2250–5.
[11] Ferreira-Aparicio P, Rodriguez-Ramosa I, Guerrero-Ruiz A. On the
performance of porous Vycor membranes for conversion enhancement
in the dehydrogenation of methylcyclohexane to toluene. J Catal
2002;212(2):182–92.
[12] Heurtaux F, Forissier M, Meille V, Pitault I, Duhamel L. Système
de production d’hydrogène à bord d’un véhicule en utilisant la
déshydrogénation de composés organiques. French Patent FR2860455,
2005.
[13] Paglieri SN, Way JD. Innovation in palladium membrane research. Sep
Purif Methods 2002;31:1–169.
[14] Uemiya S, Sato N, Ando H, Kude Y, Matsuda T, Kikuchi E. Separation
of hydrogen through palladium thin film supported on a porous glass
tube. J Membr Sci 1991;56:303–13.
[15] Lin YM, Rei MH. Separation of hydrogen from the gas mixture out of
catalytic reformer by using supported Palladium membrane. Sep Purif
Technol 2001;25:87–95.
[16] Huang T, Wei M, Chen H. Permeation of hydrogen through
palladium/alumina composite membrane. Sep Sci. Technol 2001;36:
199–222.
[17] Li A, Liang W, Hughes R. Fabrication of dense palladium composite
membrane for hydrogen separation. Catal Today 2000;56:45–51.
[18] Casanave D, Giroir-Fendler A, Sanchez J, Loutaty R, Dalmon JA. Control
of transport properties with a microporous membranes reactor to enhance
yield in dehydrogenation reaction. Catal Today 1995;25:309–14.
[19] Miachon S, Landrivon E, Aouine M, Sun Y, Kumakiri I, Li Y,
Prokopová OP, Guilhaume N, Giroir-Fendler A, Mozzanega H, Dalmon
J-A. Nanocomposite MFI—alumina membranes via pore-plugging
synthesis: preparation and morphological characterisation. J Membr Sci
2006;281(1–2):228–38.
[20] Dong J, Lin YS, Liu W. Multicomponent hydrogen/hydrocarbon
separation by MFI-type zeolites membranes. AIChE J 2000;46:1957–66.
[21] Lai Z, Tsapatsis M. Gas and organic vapor permeation through b-oriented
MFI membranes. Ind Eng Chem Res 2004;43:3000–7.
[22] Giroir-Fendler A, Peureux J, Mozzanega H, Dalmon JA. Characterization
of a zeolite membrane for catalytic reactor application. Stud Surf Sci
Catal 1996;101:127–36.
[23] Arruebo M, Coronas J, Menéndez M, Santamaria J. Separation of
hydrocarbons from natural gas using silicalite membranes. Sep Purif
Technol 2001;25:275–86.
[24] Moron F, Pina MP, Urriolabeitia E, Menéndez M, Santamaria J.
Preparation and characterization of Pd–zeolite composite membranes for
hydrogen separation. Desalination 2002;147:425–31.
[25] Swesi Y, Kerleau P, Pitault I, Heurtaux F, Ronze D. Purification of
hydrogen from hydrocarbons by adsorption for vehicles application. Sep
Purif Technol 2007;56:25–37.
[26] Giroir-Fendler A, Julbe A, Ramsay JDF, Dalmon JA. Matériau
inorganique composite poreux, notamment sous forme de membrane, et
procédé d’obtention d’un tel matériu. French Patent No. 005562, 1994.
[27] Pachtová O, Kumakiri I, Kocirik M, Miachon S, Dalmon JA. Dynamic
desorption of adsorbing species under cross membrane pressure
difference: a new defect characterisation approach in zeolite membranes.
J Membr Sci 2003;226:101–10.
[28] Huang TC, Chen HI. A study on the preparation and gas permeation of
porous alumina supports. Sep Sci Technol 1995;30:2189–209.
[29] Brown LF. A comparative study of fuels for on-board hydrogen
production for fuel-cell-powered automobiles. Int J Hydrogen Energy
2001;26:381–97.
[30] Huang Y, Dittmeyer R. Preparation and characterization of composite
palladium membranes on sinter-metal supports with a ceramic barrier
against intermetallic diffusion. J Membr Sci 2006;282:296–310.