Journal of Power Sources 147 (2005) 116127

A radial microfluidic fuel processor

Ashish V. Pattekar, Mayuresh V. Kothare
Integrated Microchemical Systems Laboratory, Department of Chemical Engineering, Lehigh University, 111 Research Drive, Bethlehem, PA 18015, USA

Received 13 December 2004; accepted 14 January 2005

Available online 14 April 2005

Abstract
A novel radial flow packed-bed microreactor has been developed to enable hydrogen delivery in miniature fuel cells for portable power
generation using the methanol reforming process. Conventional rectangular microchannel based designs typically used for on-chip chemical
processing were found to be inadequate for practical high-throughput micro packed-bed application due to the associated high pressure drops.
The radial design exploits several physical phenomena inherent to the device operation and provides an order of magnitude reduction in the
required driving pressure while maintaining the compact design of the microreactor and providing greater than 98% conversion of methanol
to hydrogen at flow rates sufficient to supply hydrogen to a 20 W fuel cell. The developed microreactor opens the path for fuel cell based
portable energy systems with energy storage capacities up to 10 times those possible with current battery technology.
2005 Elsevier B.V. All rights reserved.

Keywords: Microreactor; Micro fuel cells; Microreformer; Microfluidics; Lab-on-a-chip; Radial flow

1. Introduction and literature survey the successful commercialization of miniature hydrogen pro-
ton exchange membrane (PEM) fuel cells. Also, the stored
Miniaturized lab-on-a-chip and plant-on-a-chip de- energy density of hydrogen in compressed or liquid form is
vices have been successfully developed for several real-world significantly lower in comparison to storage in the form of liq-
applications in recent years [14]. Power-plant-on-chip and uid hydrocarbons, such as methanol, which can be reformed
palm power are buzzwords of great interest currently in the for generation of the hydrogen gas as and when needed. Hy-
micro-power community where battery technology is thought drogen storage in metal hydrides and nanostructured carbon
to be quickly approaching its peak in energy storage capacity [8,9] has been proposed but a number of limitations have been
per unit volume/weight of the power source [5]. As a promis- recognized, including high weight of the hydride storage unit,
ing alternative, hydrogen fuel cells have long been considered lower energy density, the passive nature of the hydrogen re-
a power source of the future due to their high efficiency in the lease mechanism and high cost. Therefore, the development
conversion of fuel to usable energy and low emission of pollu- of a compact and efficient miniaturized fuel reformer for
tants and greenhouse gases during operation [6,7]. In the con- on-demand hydrogen delivery to miniature PEM fuel cells
text of portable and palm power sources, fuel cells have the has been the subject of considerable research activity in re-
potential of providing energy storage densities several times cent years [1019].
those possible using current state-of-the-art lithium-ion bat-
teries [5]. However, the difficulties and hazards involved in
the storage and handling of the hydrogen fuel in either com- 2. The radial microreactor
pressed gas or liquid form have been a major impediment in
In this paper, we present a novel radial flow micro packed
bed reactor-on-a-chip with an integrated micro-vaporizer for
Corresponding author. Tel.: +1 610 758 6654; fax: 1 610 758 5057. hydrogen delivery using methanol reforming:
E-mail addresses: [email protected] (A.V. Pattekar),
[email protected] (M.V. Kothare). CH3 OH + H2 O
CO2 + 3H2 (1)

0378-7753/$ see front matter 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.jpowsour.2005.01.024
 A.V. Pattekar, M.V. Kothare / Journal of Power Sources 147 (2005) 116127 117

CH3 OH
CO + 2H2 (2) been converted to hydrogen. At the same time, the pressure
decreases as the gas moves closer to the outlet due to the
CO + H2 O
CO2 + H2 . (3)
presence of a pressure gradient which drives the flow and
We have previously demonstrated rectangular microchannel this causes a further increase in the gas volumetric flow rate
based packed-bed microreactors for this application [19] as and gas flow velocity [19]. The result of both these effects
part of our ongoing work on this subject. The packing of mi- is that higher velocity in local regions closer to the outlet
crochannels (400 m 1000 m) with catalyst particles of leads to a smaller residence time in these regions, result-
mean diameter 5070 m resulted in highly efficient contact- ing in lower catalyst utilization. Also, since the bulk of the
ing of reactants with the catalyst and excellent conversions pressure drop in the microchannels occurs in the high gas
at relatively low temperatures in the range of 195210 C. velocity regions, such a design leads to high pressure drops
However, the inlet pressure of the methanolwater mixture in the device during operation, as was experimentally ob-
required to drive the flow and generate sufficient hydrogen served.
for an 810 W fuel cell was routinely found to be in the range Vaporization of liquid feed in a microreactor also poses
of 70100 psig in these conventional microchannel reactors. unique challenges due to the sudden expansion of the fluid
Such high inlet pressures are neither achievable nor desir- during boiling which causes flow pulsation and severe pres-
able in portable devices nor can they be controlled using sure drop oscillation [22] in the two-phase flow region. This
available miniature pumping devices and valves [20,21]. For pulsation in the vaporization region leads to unsteady opera-
true high energy density portable operation, the microreactor tion and the possibility of damage to the device unless suffi-
design should, therefore, be able to operate at significantly cient room is provided for the nearly thousandfold expansion
lower pressure drops while maintaining the high throughput of the fluid as it changes from liquid to vapor.
rate for hydrogen production. The unique radial design pre- All of the above problems have been resolved by employ-
sented here achieves this goal by exploiting several key char- ing a radial flow microreactor geometry which compensates
acteristics of the methanol reforming reaction as discussed for the expected gas expansion by providing a variable (in-
below. creasing) flow cross section along the reaction path. As shown
An analysis of the different physical and chemical phe- in Fig. 1(a), the reacting gases flow radially outward from the
nomena occurring inside the microchannel reactor reveals the smaller inner circular boundary to the larger outer circular
inadequacy of simple microchannel based designs for practi- boundary, resulting in significantly reduced pressure drops
cal micro-packed bed applications. The methanol reforming compared to conventional microchannel based designs for a
reaction causes an increase in the total number of moles of given feed flow rate. While radial flow reactors have been
the reaction mixture, leading to an increase in gas mixture used in conventional chemical plants in the past [23], their
volume at a given pressure. This results in an increase in use has been limited due to the more complex manufacturing
the gas flow velocity towards the outlet of the microchan- process involved in implementing these designs at the macro
nels where a significant amount of methanol has already scale. This is not an issue in micro scale radial reactors, since

Fig. 1. Radial flow microreactor: (a) schematic; (b) fabricated device; (c) vaporization region; and (d) catalyst particle filter (patent pending).
118 A.V. Pattekar, M.V. Kothare / Journal of Power Sources 147 (2005) 116127

the microfabrication steps involved are not much different for

the radial and microchannel configurations.
Apart from the lower pressure drops, the radial design
results in a very compact device, since a significantly larger
amount of catalyst can be accommodated in a reactor of given
volume compared to a microchannel based design in which
the reactor walls take up a significant portion of the net reactor
volume [19]. While this may not always be an important issue
in conventional chemical plants which are typically the size of
large buildings, compactness of design is of prime importance
in portable devices where any increase in device size may
severely limit its applicability. Thus, the radial microreactor
exploits the unique advantages of the radial design at micro-
scale, making the device especially well suited for plant-on-a-
chip applications where low pressure drop and compactness
of design are of prime importance.

3. Fabrication procedure

A schematic of the steps followed in fabricating the ra-

dial microreactor are given in Fig. 2. Standard photolithog-
raphy steps followed by deep-reactive ion etching (DRIE)
are used for etching features on a silicon substrate to form
the vaporizer and filters for trapping catalyst particles. This
is followed by anodic bonding to a pyrex wafer having
drilled holes for catalyst introduction (two catalyst intro-
duction ports) and microfluidic interfacing (one inlet and
two symmetrically located outlets as shown in Fig. 1(b)).
The procedure for carrying out the microfluidic interfacing
is described elsewhere [24]. Finally, the catalyst is intro-
duced into the annular reaction chamber by carrying out
fluidized packing of Cu/ZnO/Al2 O3 microparticles (S ud-
Chemie catalyst C18-7) of diameters in the range of 50
70 m. The catalyst powder is placed on the catalyst intro-
duction ports and the inlet/outlet ports of the microreactor
are connected to vacuum. This draws the microparticles into
the reaction chamber where they are trapped in the annular
region between the catalyst particle filters to form the radial
packed bed shown in Fig. 2(e). The fabricated device along Fig. 2. Schematic of the steps followed in the fabrication of the radial mi-
croreactor.
with magnified views of the different features is shown in
Fig. 1(b)(d).
of known flow rate maintained using a mass flow controller
is mixed with the microreactor exhaust gases before feeding
4. Experimental verication into the gas analyzer. This provides for calculation of the ex-
act hydrogen production rate by measuring the concentration
4.1. Experimental set-up of hydrogen relative to argon. A pressure sensor connected
between the syringe pump and the microreactor inlet is used
A schematic of the experimental set-up used for testing the to monitor the pressure of the methanolwater mixture enter-
microreactor is shown in Fig. 3 and the actual set-up is shown ing the microreactor.
in Fig. 4. It consists of a syringe pump for feeding the liq- The mass spectrometer takes in the desired amount of gas
uid methanolwater mixture of required composition to the for analysis and the rest is either vented through a vacuum
microreactor at desired flow rates and a mass spectrometer pump connected to the outlet side (Fig. 3) or collected in a
(Stanford Research Systems QMS-200 gas analyzer) which sample balloon as shown in Fig. 4. The temperature of the
provides an on-line analysis of the composition of the re- microreactor is maintained at desired levels using a hotplate
acted gases from the microreactor exit. An inert gas (argon) on which the microreactor is placed. As demonstrated previ-
 A.V. Pattekar, M.V. Kothare / Journal of Power Sources 147 (2005) 116127 119

Fig. 5. Comparison of pressure drops in the microchannel device with pres-

sure drops in the radial device at three times the throughput rate (feed: 2:1
molar H2 O:CH3 OH).

Fig. 3. Schematic of experimental setup.

liquid feed. Though the presence of the meandering circular
walls causes the feed to move in non-radial directions within
ously [19], it is also possible to use on-chip heaters and tem- the vaporization region, the features are arranged such that
perature sensors for maintaining the operating temperature the fluid exits the vaporization region uniformly in all di-
of the microreactor (Fig. 3) to avoid the use of bulky external rections. This was verified by feeding liquid water at room
hotplates and simplify the experimental set-up. temperature at the inlet and noting that the flow out of the
vaporization zone was uniform in all directions [25].
4.2. Pressure drop experiments The feed vaporization ability of the device was tested sepa-
rately prior to carrying out the reaction runs by feeding liquid
It was experimentally verified that this device could be op- water at various flow rates and temperatures [2628] and not-
erated at less than 16 th the pressure drops and with through- ing the maximum flow rate up to which the entire feed could
puts up to three times those possible using a microchannel be vaporized within the vaporization region. The device was
configuration of the same overall reactor size, as shown in able to successfully vaporize up to 1525 cm3 h1 of liquid
Fig. 5. water feed at operating temperatures in the range of 150
200 C without any liquid droplets appearing in the catalyst
4.3. Vaporization experiments region, as shown in the vaporization runs. Snapshots from
one of the vaporization runs at 200 C [28] and water feed
The radial geometry of the design provides for expan- flow rates of 15 and 20 cm3 h1 are shown in Fig. 6(a) and
sion of the evaporating liquid radially outward from the inlet (b), respectively. Liquid droplets can be seen clearly within
at the center and successfully dampens the pulsation effect the vaporization region at the high 20 cm3 h1 flow rate, but
discussed earlier. Special features were incorporated in the all fluid entering the catalyst zone (marked by the outer cir-
device as shown in Fig. 1(c) to provide sufficient area of con- cular boundary of the vaporizer shown in Fig. 6(a) and (b))
tact between the liquid feed and the heated silicon surface is in gaseous phase (note the absence of liquid droplets in
to ensure proper vaporization. The central radial micro-walls this region), thus, verifying complete vaporization. Since the
in Fig. 1(c) serve as the initial point of contact for the liquid maximum flow rate of the methanolwater liquid feed during
feed entering at the center. Subsequently, the two-phase fluid the subsequent reaction runs needed to produce hydrogen for
mixture meanders around the concentric walls and passes on a 20-W fuel cell was less than 10 cm3 h1 , the vaporization
to the catalyst packed reaction zone through another set of capacity was demonstrated to be sufficient for the range of
parallel walls providing further area for vaporization of the interest of the feed flow rates.

5. Results from reaction runs

Results from experimental reaction runs of the radial re-

actor are shown in Table 1 and in Fig. 7. The temperature
of the microreactor was varied in the range of 150 to about
250 C for different feed flow rates and feed compositions.
The hydrogen production rate for each run rises as increas-
ing the reactor operating temperature results in faster kinetics
to the point when equilibrium conversions are reached and
no further increase in hydrogen production rate is possible.
Since the methanol reforming reaction can proceed to almost
Fig. 4. The experimental setup. complete conversion at equilibrium under the operating con-
120 A.V. Pattekar, M.V. Kothare / Journal of Power Sources 147 (2005) 116127

Fig. 6. Vaporizer operation at 200 C: (a) water feed 15 cm3 h1 ; (b) water feed 20 cm3 h1 .

ditions studied here [29], this represents close to complete 6. Mathematical analysis
conversion of methanol in the radial reactor. The conversion
values were calculated using the data reported in the table and Rigorous mathematical modeling and analysis was carried
were found to be higher than 98% for each of the reaction out to understand the observed improvement in performance
runs. As shown in Table 1, higher water to methanol ratios with the radial configuration in comparison to the microchan-
in the feed lead to an increase in the number of moles of hy- nel configurations developed earlier. The main equations of
drogen produced per mole of methanol in the feed due to the the model are discussed below, followed by results from nu-
suppression of carbon monoxide production via reaction (3) merical integration of the developed models.
above.
The peak hydrogen production rate achieved with the ra- 6.1. Coupled ow and reaction kinetics modeling
dial device was about 145 SCCM (sufficient for a 20 W fuel
cell), as shown in Fig. 7. The pressure drop across the de- A comprehensive model based on coupled reaction kinet-
vice at this peak production rate was found to be 10.2 psi. ics and momentum transport equations is provided below.
In contrast, the microchannel device developed earlier [19] FORTRAN programs were developed to integrate the model
provided a peak hydrogen production rate of 72 SCCM, at a equations numerically using the method of lines. As dis-
pressure drop exceeding 95 psi. Thus, it was experimentally cussed below, the simulation results support the experimen-
verified that the radial device can produce more than twice the tally observed behavior (significantly lower pressure drops
amount of hydrogen compared to the microchannel device, at higher throughputs in the radial device compared to the
at less than 19 th of the pressure drop. microchannel device) and serve as further verification of the
advantage of the radial geometry for high throughput, low
pressure drop applications.
Darcys law for flow through porous media (Eq. (4)) [30]
and the continuity equation for compressible fluids (Eq. (5))
[31] are used to model gas flow through the micro-packed bed,
where u is the fluid flow velocity vector; Kp , the permeability
of the porous medium; P denotes the pressure; denotes the
fluid viscosity and , the fluid density:
Kp
u = ux i + uy j + uz k = P (4)

( u) = 0 (5)

The gas is modeled as a compressible fluid in accordance

with the ideal gas law:
Fig. 7. Hydrogen production rate in radial microreactor at various tempera-
tures and feeds. PV = nRT (6)

Table 1
Results from experimental operation of the radial microreactor
Feed composition Feed flow rate Moles of H2 produced Mol% CO
(molar H2 O:CH3 OH) (liquid, 20 C) (cm3 h1 ) per mole CH3 OH feed in product
1:1 7.0 2.83 3.13
1.5:1 7.5 2.85 2.52
2:1 8.0 2.96 2.18
 A.V. Pattekar, M.V. Kothare / Journal of Power Sources 147 (2005) 116127 121

where V denotes the volume occupied by n moles of the gas denoting the forward rate of the reactions (mol m3 s1 )
at temperature T and R is the universal gas constant. This can given in Eqs. (15)(17) above by r1 , r2 , and r3 , respectively
also be written as the rate of production of each species is given by:
nM MP rCH3 OH = r1 r2
== (7) (18)
V RT
rH2 O = r1 r3 (19)
where M denotes the gas molecular weight. Eqs. (4), (5)
and (7) can be combined into a single governing equation rCO2 = r1 + r3 (20)
describing the flow of the gas mixture through the packed
bed of catalyst particles: rH2 = 3r1 + 2r2 + r3 (21)

 rCO = r2 r3
Kp M (22)
PP = 0 (8)
RT The rates r1 , r2 and r3 are functions of the partial pressures
and since Kp and R are constants and assuming that T is (Pj , N m2 ) of the different species in the reacting mixture,
maintained constant across the reactor geometry by proper which are given by:
temperature control, this becomes: Pj = nj P (23)


M where P is the total pressure (N m2 ) and nj is the mole
PP = 0. (9)
fraction of each species, given by:
Considering variations only in radial direction (axisym- Fj
nj = (24)
metry), we have: FT


1 MP P where j represents any of CH3 OH, H2 O, CO2 , CO and H2
r =0 (10) and F is the flow rate in mol s1 . This is in turn used to
r r r

 calculate the molecular weight M for use in Eq. 14 via the
MP P equation:
r =0 (11)
r r n

M= nj M j . (25)

2 j=1
rMP 2 P
rM P rP P M MP P
+ + +
r 2 r r r r Also, the effect of variation in viscosity of the reaction
rMP P mixture as its composition changes with reaction is included
= 0. (12) in Eq. (14) above. The mixture viscosity (), which is a func-
2 r r
tion of the temperature (T ), pressure (P), and composition
Then using the notation (nj ), is calculated during the numerical solution by using

 correlations and mixing rules available in the literature [32].
P In order to incorporate the effect of chemical reactions
=m (13)
r in the above fluid dynamics model, we write the following
balance for the flow rate Fj (mol s1 ) of each species in the
and removing the partial derivative sign for this one- radial flow device:
dimensional (radial) representation, we have from Eqs. (12)
dFj dFj
and (13) above, = = rj (mol (m3 s1 )) (26)

 dV 2rh dr
dm m2 m m dM m d

= + . (14) dFj
dr P r M dr dr = (2rh)rj (27)
dr
Eqs. (13) and (14) constitute the fluid mechanics model for
where h is the thickness of the radial catalyst packed bed
the radial reactor. In this model, the effect of reaction is cap-
and rj represents the rate of production of each species,
tured through variations in effective molecular weight M
given in Eqs. (18)(22) above.
and viscosity of the reacting mixture. For the methanol
Eqs. (13), (14) and (27) represent the complete coupled
reforming system,
model for fluid flow and reactions in the radial microreactor,
CH3 OH + H2 O
CO2 + 3H2 (15) which can be solved using a numerical integration technique,
such as the method of lines.
CH3 OH
CO + 2H2 (16) A similar analysis for the microchannel design (analyzing
a single reactor microchannel) gives the following model for
CO + H2 O
CO2 + H2 (17) the tubular configuration, with z representing the distance
122 A.V. Pattekar, M.V. Kothare / Journal of Power Sources 147 (2005) 116127

Table 2 size, the radial design provides twice the hydrogen produc-
Parameters used in the simulationconstants tion rate of the microchannel design, at a fraction of the
CST1 = 7.5 106 mol m2 Sc = 102 m2 g1 pressure drop. The radial device fabricated in this work had
CST1a = 1.5 105 mol m2 b = 1300 kg m3 an overall size (net reactor volume) comparable to a previ-
CST2 = 7.5 106 mol m2 R = 8.314 J (mol K)1 ously fabricated microchannel based design [19] with seven
CST2a = 1.5 105 mol m2 T = 245 C = 518.15 K parallel microchannels, each microchannel of dimensions
400 m 1000 m 10.4 cm. Thus, in order to make a fair
Table 3 comparison of the expected pressure drops, the flow rate used
Parameters used in the simulationk and K parameter values calculated at in the simulation of a single microchannel was appropriately
the reforming temperature of 245 C = 518 K scaled to 17 th of the total feed into the microchannel device.
kR = 31813.7 m2 (mol s)1
KHCOO(1) = 0.18902 bar1.5 As seen in Fig. 8(a), for a methanol + water feed flow
kW = 86502.8 m2 (mol s)1 KH(1a) = 0.59799 bar1 rate sufficient to generate enough hydrogen for a 10-W fuel
kD = 2732.26 m2 (mol s)1 KH(2a) = 425.41 bar1 cell, the pressure drop across the microchannel device was

KCH = 0.68135 bar0.5
KCH = 3836.6 bar0.5
3 O(1) 3 O(2) found to be about 7.8 atm. In contrast, the pressure drop
KOH(1) = 0.49241 bar0.5
KOH(2) = 3836.6 bar0.5

across the radial device was found to be about 0.7 atm for a
methanol + water feed flow rate sufficient to generate enough
along the reaction channel and w and t representing the hydrogen for a 20-W fuel cell (see Fig. 9(a)).
microchannel width and thickness, respectively: As seen in Fig. 8(b), the tubular (microchannel) device has

 a steep rise in the flow velocity near the exit due to the increase
dP
=m (28) in the number of moles during the reaction and expansion of
dz the reacting mixture, which leads to non-uniform catalyst

 utilization in this region and high pressure drop during the
dm m2 m dM m d
= + (29) operation. These effects are successfully countered in the ra-
dz P M dz dz
dial geometry and there is a significantly lower variation in


dFj the flow velocity magnitude in the radial device compared to
= (wt)rj . (30) the microchannel device as seen in Figs. 9(b) and 8(b), lead-
dz
ing to a more uniform catalyst utilization and significantly
In order to compute the reaction rates r1 , r2 and r3 , the lower pressure drop in the radial device. Thus, the simulation
comprehensive kinetic model developed by Peppley et al. [33] results support the experimental observations regarding the
was implemented in the simulations. The main equations of higher throughput capacity of the radial device with operation
this model are given in Eqs. (31)(33) below and the various at a fraction of the overall pressure drop, as noted in Section
parameters used in the model are listed in Tables 2 and 3. 5 above.
All temperature-dependent k and K values listed in It must be noted that the main objective of the simulations
Table 3 are calculated at the reforming temperature of 245 C, was to demonstrate the advantage of the radial geometry in
based on formulae available in [33].
  

kR KCH P /P 0.5 1 P 3 P /k P P CST1 CST1a Sc b
O(1) CH 3 OH H2 H 2 CO 2 R CH 3 OH H 2 O
r1 =       
3
(31)
1 + KCH P /P 0.5 + K P P 0.5 + K P /P 0.5 1 + K 0.5 P 0.5
3 O(1) CH 3 OH H2 HCOO(1) CO 2 H2 OH(1) H 2 O H2 H(1a) H2

K
kW T 2
OH(1) (PCO PH2 O /PH2 )(1 PH2 PCO2 /kW PCO PH2 O )(CS1 ) Sc b
0.5
r1 =   (32)

(1 + KCH PCH3 OH /PH0.5
+ KHCOO(1) PCO2 PH0.5
+ KOH(1) (PH2 O /PH0.5 ))2
3 O(1) 2 2 2

 

kD KCH PCH3 OH /PH0.5 (1 PH2 2 PCO /kD PCH3 OH )CST2 CST2a Sc b
3 O(2) 2
r3 =
(33)
(1 + KCH3 O(2)
(PCH3 OH /PH0.5
2
) + KOH(2) (PH2 O /PH0.5
2
))(1 + KH(2a)
0.5 P 0.5 )
H2

where CST1 , CST1a , CST2 and CST2a are the concentrations of ac-
tive sites on the catalyst surface, Sc is specific surface area
and b is catalyst bulk density, respectively [33]. providing a significant reduction in the operating pressure
drops. To that end, any kinetic model which reasonably rep-
6.2. Simulation results resents the reforming process would have been adequate. The
comprehensive kinetic model [33] used in this work is consid-
The results from the numerical integration of the above ered to be the most general form, and was, therefore, chosen
models for the radial and microchannel device are given in for the simulation studies to verify the hypotheses regarding
Figs. 8 and 9. It is clear that for the same overall reactor the advantages of the radial microreactor configuration.
 A.V. Pattekar, M.V. Kothare / Journal of Power Sources 147 (2005) 116127 123

Fig. 8. (a) Pressure profile in microchannel reactorfluid pressure (atm) vs. distance along the microchannel (m); (b) velocity profile in microchannel reactor
fluid velocity (m/s) vs. distance along the microchannel (m) (at a hydrogen production rate sufficient for a 10-W fuel cell).

Fig. 9. (a) Pressure profile in radial reactorfluid pressure (atm) vs. distance from device center (m); (b) velocity profile in radial reactorfluid velocity (m/s)
vs. distance from device center (m) (at a hydrogen production rate sufficient for a 20-W fuel cell).

7. Energy density calculations and comparisons needed for 20 W operation [37]. Thus, a 20 h continuous op-
eration at 20 W would require 8.40 g mol of H2 . This trans-
We will now proceed to calculate the projected energy lates into a net methanol storage requirement of 2.947 g mol,
storage capacities possible using the integrated system con- or 94.32 g methanol, assuming 2.85 mol H2 produced per mol
sisting of the radial microreformer coupled with a micro fuel CH3 OH feed to the reformer as per Table 1, for a feed ratio of
cell and to draw out important conclusions regarding the vi- 1.5:1 molar H2 O:CH3 OH. Now, a 1.5:1 molar H2 O:CH3 OH
ability of this technology in comparison with current state of mixture has a density of 0.9067 g/cm3 at 20 C and contains
the art batteries. 0.49225 g methanol per cc of the mixture [32]. Thus, the net
7.1. The need for alternative portable power sources water-methanol (1.5:1 molar) feed mixture storage require-
ment for a 20 W, 20 h operation would, therefore, be 173.73 g,
Kelley et al. [34] and have provided a comparison of the or 191.61 cm3 .
specific energy density (, Watthour energy storage per kg Based on the microfabricated fuel cell sizes reported
weight of the battery system) possible using current state of by Kelley et al. [34], a typical micro fuel cell for 20 W
the art miniature power sources (summarized in Table 4). application (basis: FC mass 0.031 g, volume 0.012 cm3 ,
Considering the projected improvements in the lithium-ion with an active area of 0.25 cm2 , operating at a power density
battery [35], it can be expected that the energy storage ca- of 80 mW cm2 ) would have a mass of 0.031 20/(0.08
pacity of commercially available rechargeable batteries may 0.25) = 31 g and a volume of 12 cm3 . Also, the reformer
be improved up to about 450 Wh l1 or about 275 Wh kg1 reported in this work, when scaled for the above application
over the next few years [35]. In contrast, based on the DARPA (hydrogen supply for a 20 W fuel cell = 0.42 g mol h1 H2
palm power initiative guidelines [36], it is clear that several production rate) would have a mass of about 35 g including
key applications exist for portable energy storage systems the packaging for thermal insulation, and a volume of about
with capacities in the range of 1000 Wh kg1 (20 W, 3-h mis- 28 cm3 . Thus, the reformerfuel cell combination would have
sion) to over 3000 Wh kg1 (20 W, 10-day mission). Thus, a
new micro-power technology may be the only alternative if
Table 4
these higher energy densities are to be attained for portable
Specific energy density of current state of the art miniature power sources
power.
Battery type Specific energy (Wh kg1 )
7.2. PEM fuel cell specications Ni/Cd battery 60
Ni/metal hydride battery 65
For a typical PEM fuel cell operating at 60% thermal ef- Li/Lix Mn2 O4 battery 130
ficiency, a hydrogen flow rate of 0.42 g mol h1 would be LiC6 /Lix CoO2 battery 90
124 A.V. Pattekar, M.V. Kothare / Journal of Power Sources 147 (2005) 116127

Table 5 tem efficiency and the energy storage density by a significant

Energy requirement for operation of microreformer producing hydrogen for amount. For example, a 72% H2 utilization in the fuel cell
a 20 W fuel cell
would result in useful electrical energy output from the fuel
Reformer process step Energy requirement (W) cell equal to (20 W 20 h 0.72) = 288 Wh, with a hydro-
Sensible heating of water in feed 0.6954 gen flow rate of 0.1176 mol h1 in the fuel cell anode off-gas
Sensible heating of methanol in feed 0.5504 (effluent). If the heating requirement for the reformer oper-
Vaporization requirement for water 2.4483
Vaporization requirement for methanol 1.3942
ation is supplied by combustion of the un-utilized hydrogen
Endothermic heat of reaction 2.3850 in the anode off-gas, this translates into a heating capacity
Total power required for operation 7.4733 of 7.8302 W based on the lower heating value (LHV) of hy-
drogen, which is more than sufficient to provide the reformer
requirement of 7.4733 W given in Table 5. This system would
an intrinsic mass of 66 g and a volume of about 40 cm3 , in
then have an overall energy storage density of 1200 Wh kg1 ,
addition to the storage requirements for methanol and water
or about 1241 Wh l1 .
discussed above. This makes the total mass and volume of the
Recent developments in high temperature PEM fuel cells
power source (system + fuel) equal to about (66 + 174) =
have led to novel membranes that can operate at high temper-
240 g and (40 + 192) = 232 cm3 , respectively.
atures, up to 200 C [38]. High temperature PEM fuel cells
offer several advantages, including simpler water manage-
7.3. Reformer energy requirement ment and high CO tolerance (up to about 2% CO in the fuel),
which significantly reduces the amount of post-reforming
To make a fair comparison between miniature fuel cells fuel clean-up steps required. With proper control of the re-
and state of the art miniature batteries, not only does one forming conditions to maintain levels below 2%, CO clean-
need to consider a basis related to the type of application up can be completely avoided, significantly simplifying the
being evaluated and the volume and weight of the energy overall system. Apart from this, another important advantage
conversion devices involved, but also the overall energy of the high temperature PEM fuel cell is that the waste heat
requirement for operating the power source (e.g., liquid is available for exchange at a higher temperature, resulting in
feed vaporization and endothermic reforming reaction heat the possibility of further heat integration in the overall sys-
requirements, heat loss to the ambient, and overall fuel cell tem. For example, continuing the above analysis, 40% of the
efficiency, etc. in the case of the microreformer-micro fuel thermal energy in the hydrogen used in the fuel cell is wasted
cell combination). A detailed analysis of the the energy as heat (60% thermal efficiency). This translates into net heat
requirement of each of the process steps carried out in the energy of [0.42 mol h1 0.72 (fuel utilization) 0.40
microreformer is provided in Appendix A and summarized 239700.0 J mol1 (LHV of H2 )]/3600(s h1 ) = 8.054 W at
in Table 5. As shown, a reformer providing sufficient 200 C available for further integration into the system and
hydrogen for a 20 W fuel cell application would require also to take care of the heat loss, providing a further improve-
about 7.4733 W power to vaporize the liquid fuel/water ment in the energy storage density. Including a 15% penalty
mixture and carry out the reforming reaction. for the energy requirement related to fluid pumping and
power conditioning losses, this makes the overall energy stor-
7.4. Projected energy storage density age densities of 1200 0.85 = 1020 Wh kg1 , or 1241
0.85 = 1055 Wh l1 easily attainable in a practical system.
Assuming absolutely no thermal integration between the Water produced in the fuel cell may be recycled into
reformer and the fuel cell, if the entire reformer energy re- the system to satisfy the reformer requirement, further
quirement is satisfied by electricity produced in the fuel improving the overall efficiency and energy storage
cell, this translates into a useful energy storage capacity of capacity of the system. For example, in a fuel cell op-
20 h (20 7.4733) W = 250.534 Wh. erating at 80% fuel utilization, the effluent is steam at
Going back to the 20 W, 20 h basis, the total mass and 200 C at a flow rate of (0.42 0.80) = 0.336 g mol h1 .
volume of the power source (fuel + system) are 240 g and Noting that the steam required in the reformer is
232 cm3 , respectively. This translates into an energy storage 1.5 (0.42/2.85) = 0.2211 g mol h1 , the steam require-
capacity of 1044 Wh kg1 and 1080 Wh l1 for this system. ment in the reformer may be completely satisfied through
Of course, this calculation does not consider the energy re- the fuel cell effluent if proper flow separation and control
quirement for pumping the fuel/water mixture at inlet pres- systems are implemented. Such a highly integrated system
sure, the heat loss to the surrounding during the operation will also require significantly lower energy for operating the
and the efficiency of the power conditioning circuits that are reformer, since the heating requirements for the water feed
typically necessary in fuel cell based power supplies, which are satisfied directly through the waste heat in the fuel cell.
would potentially bring down the above values by as much as For the 20 W, 20 h basis, the net volume and weight of the
25%, to about 783 Wh kg1 and about 810 Wh l1 . However, fuel cartridge would then be 94.32 g, or about 121 cm3 (pure
there is also no allowance for heat integration between the fuel methanol). Assuming a slightly larger fuel cell + reformer
cell and the reformer, which could increase the overall sys- system (80 g, 55 cm3 ) for incorporating the integration
 A.V. Pattekar, M.V. Kothare / Journal of Power Sources 147 (2005) 116127 125

mechanism, the net size of fuel + system in this case would nology. As development continues and these problems are
be 174.32 g and 176 cm3 . The net reformer energy require- resolved, the microreformer micro fuel cell combination
ment would be about 4.4268 W (since steam at 200 C is with estimated energy storage densities in the range of up
available from the fuel cell). The hydrogen flow in the anode to 2693 Wh kg1 can definitely be expected to appear in the
off-gas would be 0.42 0.20 = 0.084 mol h1 , representing portable energy market of the future.
a heating capacity of 5.593 W, which is more than enough to
satisfy the reformer energy requirement. The energy storage
Acknowledgements
capacity of this highly integrated system would then be
(20 W 20 h 0.80) = 320 Wh. Again, including a 15%
This work was supported in part by the U.S. National Sci-
penalty for the energy requirement related to fluid pumping
ence Foundation (XYZ-on-a-Chip initiative Grant CTS-
and power conditioning losses, this translates into an energy
9980781) and the Pittsburgh Digital Greenhouse. Fabrica-
storage density of 1560 Wh kg1 or about 1545 Wh l1 , a
tion of the microreactor was performed in part at the Cor-
five- to six-fold improvement compared to the best state of the
nell Nano-Scale Science and Technology Facility (a member
art rechargeable batteries available today. Longer run times
of the National Nanofabrication Users Network) which is
result in higher energy storage capacity, since the ratio of the
supported by the National Science Foundation under Grant
storage volume and weight of the fuel to the system would
ECS-9731293, its users, Cornell University and Industrial
increase. For a three-day run, this translates into a stored en-
Affiliates. The authors would also like to acknowledge the
ergy of (20 W 72 h 0.80) = 1152 Wh, with a net storage
help of J. Zelinski (Lehigh Universitys Physics Department
weight and volume of (339.55 + 80) g and (435.6 + 55) cm3 ,
Machine Shop) in the drilling of the pyrex wafers.
respectively. With the 15% penalty for system losses, this
would make the energy storage density for a 3-day run to be
2334 Wh kg1 or about 1996 Wh l1 . For a 10-day (240 h) Appendix A
run, the energy density would be 2693 Wh kg1 or about
2167 Wh l1 , thus, proving micro fuel cell technology to be A detailed analysis of the energy requirement of each of
a viable alternative for portable power needs of the future. the process steps carried out in the microreformer is provided
in this appendix. The results of these calculations are sum-
marized in Table 5.
8. Conclusion
A.1. Sensible heating of methanol
A completely novel approach to carrying out chemical
reactions in a microreactor was presented in this paper. Con- The vapor pressure of methanol is given by [39]:
ventional microchannel based designs were found to be inad-
sat
equate for microreaction systems which involve increase in log10 (PCH 3 OH
)
the moles due to reaction and corresponding expansion of the
3.2447 103
reaction mixture. The radial device presented here not only = 45.6171 13.988 log10 (T )
solves the problems associated with the microchannel geom- T
etry, but also it allows for rigorous modeling and analysis of + 6.6365 103 T 1.0507 1013 T 2 (A.1)
the operation due to the simple geometry. Rigorous modeling
where PCHsat
3 OH
is measured in mmHg and T is the absolute
and simulation studies of the device operation revealed the
temperature. Thus, at the reformer inlet pressure of 1.7 atm
advantages of the new design and confirmed the reasoning
(see Fig. 9(a)), the boiling point of CH3 OH is 351.85 K, ob-
behind adopting this approach for developing the microreac-
tained from the above equation.
tor. The developed microreactor produces sufficient hydro-
The specific heat capacity of liquid methanol CpCH3 OH,(l)
gen for fuel cells in the range of up to 20 W and thus opens
the path for portable fuel cell based energy systems utilizing (J (mol K)1 ) is given by [39]:
reforming technology for hydrogen delivery. CpCH3 OH,(l) = 40.152 + 3.1046 101 T 1.0291
As the demand for high energy density power sources
grows with the introduction of ever more power hungry mul- 103 T 2 + 1.4598 106 T 3 (A.2)
tifunctional portable electronic devices, miniature fuel cells
where T is the absolute temperature. Therefore, the sensible
are likely to provide an excellent solution to the problems
heat requirement for liquid methanol from a room tempera-
faced by the portable rechargeable batteries of today. Surely,
ture of 298.15 K to its boiling point of 351.85 K is:
several challenges need to be overcome before widespread
introduction of this technology can be successful. Integra-  351.85
tion of the reformer and fuel cell system, balance of plant CpCH3 OH,(l) dT = 4433.458 J mol1 . (A.3)
298.15
issues, efficient power conditioning systems, and reliability
CH3 OH,(g)
of long-term operation are just some of the issues that need Also, the specific heat capacity of methanol vapor Cp
to be resolved in a prototype power source utilizing this tech- (J (mol K)1 ) is given by [39]:
126 A.V. Pattekar, M.V. Kothare / Journal of Power Sources 147 (2005) 116127

CpCH3 OH, (g) = 40.046 3.8287 102 T + 2.4529 Therefore, the sensible heat requirement for steam from its
boiling point of 388.70 K to the reactor operating temperature
104 T 2 2.1679 107 T 3 + 5.99 of 245 C = 518.15 K is:
1011 T 4 . (A.4)  518.15
CpH2 O,(g) dT = 4501.367 J mol1 . (A.12)
Therefore, the sensible heat requirement for methanol vapor 388.70
from its boiling point of 351.85 K to the reactor operating Thus, the total sensible heat load for water is:
temperature of 245 C = 518.15 K is:
 518.15 H2 O,sensible

Htotal = 11322.71 J mol1 . (A.13)
CpCH3 OH,(g) dT = 9009.682 J mol1 . (A.5)
351.85 For a water flow rate of 0.2211 mol h1 (sufficient to produce
Thus, the total sensible heat load for methanol is: (0.2211/1.5) 2.85 = 0.420 mol h1 of H2 ), this translates
into an energy requirement of 11322.71 0.2211/3600 =
= 13443.14 J mol1 .
CH OH,sensible

Htotal3 (A.6) 0.6954 W.

For a methanol flow rate of 0.1474 mol h1 (sufficient to pro-
duce 0.1474 2.85 = 0.420 mol h1 of H2 ), this translates A.4. Water vaporization
into an energy requirement of 13443.14 0.1474/3600 =
0.5504 W. The specific enthalpy of saturated water (liquid) at
388.70 K is 8727.15 J mol1 , and that of saturated steam (va-
A.2. Methanol vaporization por) is 48591.65 J mol1 [32]. Thus, the energy requirement
for vaporization of the feed water at the microreformer inlet
The enthalpy of vaporization of methanol is given by [39]: pressure of 1.7 atm is

0.377
v T (48591.65 8727.15) J mol1
0.2211
mol s1

HCH3 OH = 52.723 1 kJ mol1 . (A.7) 3600

512.58
= 2.4483 W. (A.14)
Thus, the methanol vaporization heating requirement at the
boiling temperature of 351.85 K is


351.85 0.377 0.1474 A.5. Endothermic heat of reaction
52.723 1 kJ mol1 mol s1
512.58 3600
Considering the overall reforming reaction
= 1.3942 W. (A.8)
(g) (g)
CH3 OH(g) + H2 O(g)
CO2 + 3H2 (A.15)
A.3. Sensible heating of water
we have:
At the reactor feed pressure of 1.7 atm, the boiling point
H298.15 K = 48.97 kJ mol1 (A.16)
of water is 388.70 K, obtained from steam tables [32]. For
calculating the liquid water heating requirement, we have the
following correlation for specific heat capacity of water [39]: =
H245 C =
H518.15 K
 518.15
CpH2 O,(l) = 92.053 3.9953 102 T 2.1103 =
H298.15 K +
Cp (T )dT
298.15
4 2 7 3
10 T + 5.3469 10 T J (mol K)1(A.9)
= 58.25 kJ mol1 . (A.17)
Therefore, the sensible heat requirement for liquid water
from a room temperature of 298.15 K to its boiling point where
of 388.70 K is:
 388.70
Cp = 3Cp(g)H + Cp(g)CO Cp(g)H Cp(g)CH (A.18)
2 2 2O 3 OH
CpH2 O,(l) dT = 6821.339 J mol1 . (A.10)
298.15 which is given by [39]:

Also, the specific heat capacity of water vapor Cp 2

H O,(g)

Cp = 29.655 + 0.149555T 4.10398 104 T 2
(J (mol K)1 ) is given by [39]:
+ 3.34252 107 T 3 9.01766 1011 T 4
3
CpH2 O,(g) = 33.933 8.4186 10 T + 2.9906
(A.19)
105 T 2 1.7825 108 T 3
Thus, the heat requirement for carrying out the endothermic
+ 3.6934 1012 T 4 . (A.11) reforming reaction is 58250 0.1474/3600 = 2.3850 W.
 A.V. Pattekar, M.V. Kothare / Journal of Power Sources 147 (2005) 116127 127

Appendix B. Supplementary data [18] A.V. Pattekar, M.V. Kothare, Microreactor and method of use to pro-
duce hydrogen by methanol reforming, US Patent Pub. App. no.
20040179980. Provisional disclosure filed December, 2002 (in final
Supplementary data associated with this article can
form: December 2003).
be found, in the online version, at 10.1016/j.jpowsour. [19] A.V. Pattekar, M.V. Kothare, A microreactor for hydrogen production
2005.01.024. in micro fuel cell applications, J. Microelectromech. Syst. 13 (2004)
718.
[20] M. Richter, A. Drost, J. Kruckow, A high performance silicon microp-
ump for fuel handling in DMFC systems, in: Proceedings of the 2003
References Fuel Cell Seminar, Miami beach, FL, 2003.
[21] M.A. Unger, H. Chou, T. Thorsen, A. Scherer, S.R. Quake, Monolithic
[1] P. McFadden, BIOSENSORS: broadband biodetection: holmes on a microfabricated valves and pumps by multilayer soft lithography, Sci-
chip, Science 297 (2002) 2075. ence 288 (2000) 113116.
[2] V. Percec, Chemistry: catalyst hunt accelerates, Nature 424 (2003) 135. [22] W. Qu, I. Mudawar, Measurement and prediction of pressure drop in
[3] T.R. Boussie, G.M. Diamond, C. Goh, K.A. Hall, A.M. LaPointe, M. two-phase micro-channel heat sinks, Int. J. Heat Mass Transfer 46
Leclerc, C. Lund, V. Murphy, J.A.W. Shoemaker, U. Tracht, H. Turner, (2003) 27372753.
J. Zhang, T. Uno, R.K. Rosen, J.C. Stevens, A fully integrated high- [23] A.A. Savoretti, D.O. Borio, V. Bucala, J.A. Porras, Non-adiabatic
throughput screening methodology for the discovery of new polyolefin radial-flow reactor for styrene production, Chem. Eng. Sci. 54 (1999)
catalysts: discovery of a new class of high temperature single-site group 205213.
(IV) copolymerization catalysts, J. Am. Chem. Soc. 125 (2003) 4306 [24] A.V. Pattekar, M.V. Kothare, Novel microfluidic interconnectors for
4317. high temperature and pressure applications, J. Micromech. Microeng.
[4] M.A. Burns, B.N. Johnson, S.N. Brahmasandra, K. Handique, J.R. 13 (2003) 337345.
Webster, M. Krishnan, T.S. Sammarco, P.M. Man, D. Jones, D. [25] See http://www.lehigh.edu/asp2/mov/movie-1-uni-flow.wmv for a
Heldsinger, C.H. Mastrangelo, D.T. Burke, An integrated nanoliter video of the uniform flow distribution of the fluid exiting the vaporiza-
DNA analysis device, Science 282 (1998) 484487. tion region (also available as part of supplementary online material at
[5] R.F. Service, Shrinking fuel cells promise power in your pocket, Sci- the Journal of Power Sources web-site).
ence 296 (2002) 1222. [26] See http://www.lehigh.edu/asp2/mov/150C-various-flow-rates.wmv
[6] S. Park, J.M. Vohs, R.J. Gorte, Direct oxidation of hydrocarbons in a for a video of vaporizer operation at 150 C and various liquid water
solidoxide fuel cell, Nature 404 (2000) 265267. feed flow rates. (Also available as part of supplementary online material
[7] G.A. Deluga, J.R. Salge, L.D. Schmidt, X.E. Verykios, Renewable hy- at the Journal of Power Sources web-site).
drogen from ethanol by autothermal reforming, Science 303 (2004) [27] See http://www.lehigh.edu/asp2/mov/180C-various-flow-rates.wmv
993997. for a video of vaporizer operation at 180 C and various liquid water
[8] L. Schlapbach, A. Zuttel, Hydrogen-storage materials for mobile ap- feed flow rates. (Also available as part of supplementary online material
plications, Nature 414 (2001) 353. at the Journal of Power Sources web-site).
[9] U. Bunger, W. Zittel, Hydrogen storage in carbon nanostructures [28] See http://www.lehigh.edu/asp2/mov/200C-various-flow-rates.wmv
still a long road from science to commerce? Appl. Phys. A 72 (2001) for a video of vaporizer operation at 200 C and various liquid water
147151. feed flow rates. (Also available as part of supplementary online material
[10] P. Pfeifer, M. Fichtner, K. Schubert, M.A. Liauw, G. Emig, Mi- at the Journal of Power Sources web-site).
crostructured catalysts for methanolsteam reforming, in: Proceed- [29] J.C. Amphlett, M.J. Evans, R.A. Jones, R.F. Mann, R.D. Weir, Hydro-
ings of the 3rd International Conference on Microreaction Technology, gen production by the catalytic steam reforming of methanol. Part 1.
DECHEMA, Frankfurt, Germany, 1999. The thermodynamics, Can. J. Chem. Eng. 59 (1981) 720727.
[11] A.Y. Tonkovich, J.L. Zilka, M.J. LaMont, Y. Wang, R.S. Wegeng, Mi- [30] W.L. McCabe, J.C. Smith, P. Harriot, Unit Operations of Chemical
crochannel reactors for fuel processing applications. I. Water gas shift Engineering, fifth ed., McGraw-Hill, New York, 1993 (Chapter 7).
reaction, Chem. Eng. Sci. 54 (1999) 29472951. [31] R.B. Bird, W.E. Stewart, E.N. Lightfoot, Transport Phenomena, Wiley,
[12] P.J. de Wild, M.J.F.M. Verhaak, Catalytic production of hydrogen from New York, NY, 1960.
methanol, Catal. Today 60 1 (2000) 310. [32] R. Perry, D. Green, Chemical Engineers Handbook, sixth ed.,
[13] S. Fitzgerald, R. Wegeng, A. Tonkovich, Y. Wang, H. Freeman, J. McGraw-Hill, New York, NY, 1984.
Marco, G. Roberts, D. VanderWeil, A compact steam reforming re- [33] B.A. Peppley, J.C. Amphlett, L.M. Kearns, R.F. Mann, Methanol steam
actor for use in an automotive fuel processor, in: R.S.W.W. Ehrfeld, reforming on Cu/ZnO/Al2 O3 catalysts. Part 2. A comprehensive kinetic
I. Rinard (Eds.), Proceedings of the 4th International Conference on model, Appl. Catal. A: Gen. 179 (1999) 3149.
Microreaction Technology, Atlanta, GA, 2000, pp. 358363. [34] S. Kelley, G. Deluga, W. Smyrl, Miniature fuel cells fabricated on
[14] A.V. Pattekar, M.V. Kothare, S.V. Karnik, M.K. Hatalis, A microreac- silicon substrates, AIChE J. 48 (2002) 10711082.
tor for in situ hydrogen production by catalytic methanol reforming, [35] A.V. Pattekar, M.V. Kothare, Miniature fuel processors and fuel cells for
in: Proceedings of the 5th International Conference on Microreaction portable power: industry trends and potential applications, in: Proceed-
Technology (IMRET 5), Strasbourg, France, 2001. ings of the 2003 AIChE Annual Meeting, San Francisco, CA, 2003.
[15] J.D. Holladay, E.O. Jones, M. Phelps, J. Hu, Microfuel processor for [36] See http://www.darpa.mil/dso/thrust/matdev/palmpower/research.html
use in a miniature power supply, J. Power Sources 108 (2002) 21. for a summary of the goals of the DARPA palm power project.
[16] A.V. Pattekar, M.V. Kothare, Design and fabrication of a microreactor [37] Fuel Cell Handbook, sixth ed., EG&G Services Inc., US Department
for hydrogen production by catalytic methanol reforming, in: Proceed- of Energy, Morgantown, WV, November 2002.
ings of the 2002 AIChE Annual Meeting, Indianapolis, IN, 2002. [38] A. Schechter, R.F. Savinell, Imidazole and 1-methyl imidazole in phos-
[17] M.V. Kothare, A.V. Pattekar, Fuel processing microreactors for hydro- phoric acid doped polybenzimidazole, electrolyte for fuel cells, Solid
gen production by methanol reforming, in: Proceedings of the Knowl- State Ionics 147 (2002) 181187.
edge Foundations 5th International Small Fuel Cells 2003: Small Fuel [39] C.L. Yaws (Ed.), Chemical Properties Handbook, McGraw-Hill, New
Cells for Portable Power Applications, New Orleans, LA, 2003. York, NY, 1999.