Microsyst Technol (2008) 14:535–541
DOI 10.1007/s00542-007-0451-9
TECHNICAL PAPER
Fabrication and characterization of a passive silicon-based direct
methanol fuel cell
J. P. Esquivel Æ N. Sabaté Æ J. Santander Æ
N. Torres Æ C. Cané
Received: 18 May 2007 / Accepted: 13 October 2007 / Published online: 30 October 2007
Ó Springer-Verlag 2007
Abstract This paper reports on the fabrication and
characterization of a passive silicon microfabricated direct
methanol fuel cell (lDMFC). The main characteristics of
the device are its capability to work without complex
pumping systems, only by capillary pressure, and the fact
that its performance is not affected by the device orienta-
tion. A simple fabrication process based in deep reactive
ion etching (DRIE), allows obtaining a reliable and low-
cost final device. The device consists of two silicon
microfabricated plates mounted together with a commer-
cial membrane electrode assembly (MEA). The impact of
current collector design on microfuel cell performance is
explored and current–voltage (I–V) and current–power (I–
P) curves of the device at different methanol concentration
and orientation are presented. Optimal performance was
obtained for methanol concentrations between 3 and 5 M,
achieving a maximum power density of 12 mW/cm2. The
results obtained in this work demonstrate the feasibility of
the device and give a guideline for design and conditions
optimization.
1 Introduction
The increasing presence of microelectromechanical sys-
tems (MEMS) in a wide range of applications has
originated an intensive research towards the obtaining of
small power sources that satisfy the requirements of high
J. P. Esquivel (&)
N. Sabaté
J. Santander
N. Torres

C. Cané
Centro Nacional de Microelectrónica-IMB (CSIC),
Campus de la UAB s/n, 08193 Bellaterra, Barcelona, Spain
e-mail: juanpablo.esquivel@cnm.es
power density, long operation lifetime and low cost. In
recent years, many solutions have been proposed to over-
come these challenges. Regarding their integration in
MEMS technology, most of the works have focused on the
development of energy scavenging microsystems (Menin-
ger et al. 2001), microreactors (Jacobson and Epstein 2003)
and microfuel cells (Kamarudin et al. 2007). Among them,
microfuel cell power supplies appear as the most promising
option due to their higher energy density. Particularly, fuel
cells based on polymer electrolytes (PEMFC) are the most
appropriate due to their capability of operating at temper-
atures close to ambient. Although the most effective fuel
used in this kind of cells is hydrogen, its difficulty of
storage and handling makes it hard to incorporate it into
portable applications. In this sense, the use of alternative
fuels as methanol has proved to overcome successfully this
problem, mainly due to its higher energy density and easier
handling as it can be used in liquid state (Larminie and
Dicks 2003). Moreover, in order to optimize the fuel cell
supplied power and facilitate integration, auxiliary devices
such as fuel pumps and air blowers can be eliminated from
the design. This kind of cell system has been denoted as
passive fuel cell (Chen and Zhao 2007). In this way, the
overall DMFC system becomes simpler, more compact and
easier to integrate with other silicon-based devices.
Therefore, in contrast with active systems where the fuel
is supplied with a constant flux the same concentration and
pressure conditions are maintained at every moment, in a
passive system fuel distribution relies on diffusion, so
design and optimization of the architecture of the micro-
fluidic structures is a critical issue.
In this paper, we report on silicon-based microfabricated
direct methanol fuel cell with passive methanol delivery.
The fuel cell is tested at different varying operation
conditions such as methanol concentration and device
123
536 Microsyst Technol (2008) 14:535–541

orientation. Moreover, the effect of the current collector

design both at the anode and the cathode on the fuel cell
response is explored.

2 Design of microfuel cell

The basic structure of a DMFC can be described as a

membrane electrode assembly (MEA) placed between two
conducting plates. The MEA consists of a polymer elec-
trolyte membrane (typically Nafion1) covered on both Fig. 2 Diagram of fuel cell with silicon plates showing the fuel
sides by porous electrodes that contain the catalyst loads reservoir, the array of 80 lm channels and separations between
channels
that enhance the chemical reaction. The MEA is pressed
between the plates, which besides giving support to the
has been defined at the silicon wafer backside and aligned
structure, act as current collectors and distribute the fuel
with the channel array and it is used as a fuel reservoir.
towards the membrane. A schematic diagram with main
Figure 2 shows a scheme of the basic silicon-based mi-
parts of a direct methanol fuel cell can be seen in Fig. 1.
crofuel cell structure.
The approach presented in this work consists on the
development of a silicon-based passive fuel cell with a
compact and simple design. The microfuel cell is based on
3 Fabrication process and assembly
a hybrid device that consists on a commercial MEA
sandwiched between two microfabricated silicon plates. In
The process starts with 100 mm double-side polished Si
addition of giving the mechanical support to the fuel cell
wafers 500 lm thick. A first photolithography is done on
structure, the silicon plates must be structures that facilitate
the front side to define an array of squared windows with
mass transport in the system, ensuring that the fuel reaches
80 lm size distributed over a squared surface of
the membrane and CO2 is correctly degassed. In addition,
5.0 9 5.0 mm. Subsequently, a second photolithography is
due to their role as current collectors, they also have to be
performed on the back side to define the cavity for the fuel
provided with high area and good electrical conductivity to
container. Then, a deep reactive ion etching (DRIE) is
reduce ohmic losses when driving the electrons.
realized first on the front side to obtain 300 lm-deep
Therefore, the plates have been provided with an array
channels, and secondly at the back until the wafer is
of vertical squared channels 300-lm long covering an area
completely perforated. Figure 3 shows a SEM image of the
of 5 9 5 mm. Channels dimensions have been set to
channels array. In order to provide the current collectors
80 9 80 lm in order to ensure the prevalence of the cap-
with an appropriate electrical conductivity, a 150 nm Ti/Ni
illary force versus gravity regardless of device orientation
(Aravamudhan et al. 2005). In order to test the influence of
the exposed area of the MEA to the fuel, three different
designs were fabricated varying the distance between
channels with separations of 20, 40 and 80 lm. A cavity

Fig. 1 Schematic diagram showing the parts and operation of a Fig. 3 SEM image of the front side of the silicon wafer showing the
typical DMFC array of 80 lm channels

123
Microsyst Technol (2008) 14:535–541
Table 1 Main steps of the microfabrication process
Main steps of the microfabrication process
Step Description
Double side
polished wafer
Photolithography
front side
Photolithography
back side
DRIE front side
DRIE back side
Sputter deposition
Ti/Ni
Electrodeposition
Ni/Au
Silicon Photoresist Ti/Ni Sputter Ni/Au
sputter layer is deposited covering the front side of the
wafer and the channels walls. This last conductive layer is
used as a seed layer for the 4 lm thick Ni layer that is
electrodeposited afterwards. This layer, that allows reduc-
ing significantly the resistivity of the current collector, is
then covered by a thin Au layer to prevent oxidation.
Finally, the wafer is cut into 10 9 14 mm chips. The main
steps of the microfabrication process of the silicon current
collectors have been summarized in Table 1.
The final assembly of the microfuel cell is achieved by
placing a MEA (E-TEK ELAT1) consisting of a Nafion1
117 membrane with 4.0 mg/cm2 Pt-Ru catalyst load on the
anode side and 4.0 mg/cm2 Pt on the cathode, between two
microfabricated plates. During the assembling process,
alignment of the arrays of channels on both chips is
ensured, yielding a total active area of 0.28 cm2. In order to
provide the mechanical support to the structure and a larger
fuel reservoir while testing, two micromilled methacrylate
pieces are positioned onto the plates and all the compo-
nents are tightened together with four bolts. An image of
the final assembly can be seen in Fig. 4 and details about
fabrication can be found elsewhere (Santander et al. 2006).
4 Characterization
I–V characteristic curves of the microdevices were
obtained using a Keithley 2400 Sourcemeter as an
537
Fig. 4 Photograph of microfuel cell on the setup structure for testing
electronic load and interfacing with a computer in order to
apply different loads and record the output voltages. The
influence of different parameters such as the silicon current
collectors design, changes in device orientation, variation
of methanol concentration and the operation lifetime with a
limited volume of fuel has been studied. In this section the
obtained measurements are presented and discussed.
Before characterization of the device, a MEA’s pre-
treatment process had to be done to activate the membrane.
The process consisted on wetting the anode side with a
concentration of 0.25 M methanol for several hours and
afterwards submitting the fuel cell to working conditions at
a constant current for a short period of time (Kho et al.
2004).
4.1 Open circuit voltage evolution at different
concentrations
In a direct methanol fuel cell, open circuit voltage (OCV) is
strongly dependent on the methanol concentration. This is
due to the crossing over of methanol from anode to cathode
through the Nafion1 membrane that occurs because of the
combined effect of molecular diffusion and electro-osmotic
drag. This results not only in a fuel loss, but also a decrease
in the overall cell voltage due to the mixed potential on the
cathode. Methanol crossover depends on different factors,
being the most important ones the membrane permeability
and thickness (Liu et al. 2006), the concentration of
methanol (Kim et al. 2006), the operating temperature
(Kho et al. 2005), and the catalyst load in the electrodes
(Bae et al. 2006). In order to evaluate the influence of
methanol concentration on the OCV of cell at room tem-
perature, the evolution of the OCV when filling the fuel
reservoir with 100 ll of methanol at different
123
538 Microsyst Technol (2008) 14:535–541

Fig. 5 Open circuit voltage evolution at different methanol

concentrations

concentrations has been depicted in Fig. 5. This figure

shows the changes in the OCV with time from the moment
of fuel injection of a particular methanol solution into the
fuel reservoir until it reaches a steady value. As expected,
increasing methanol concentrations lead to lower OCV
values. As seen in the figure, when the reservoir is filled,
the OCV experiences an abrupt jump when the methanol
reaches the anode and then drops slowly to a steady value,
as part of the fuel permeates the membrane and reaches the
cathode. In order to compensate a decrease in methanol
concentration associated to fuel losses originated by the
methanol crossover, the fuel reservoir was refilled after a
few minutes. It can be observed that the initial voltage
jump becomes smaller when refilling the reservoir a second
time. This can be due to the accumulation of both methanol
and water in the cathode, which limit access of oxygen to
the catalytic sites and leads to a decrease in the methanol
oxidation rate both in the anode and cathode sides (Kho
et al. 2005). This effect was also visible when the reservoir
was refilled with higher concentrations of methanol.
4.2 Fuel cell performance with different methanol
concentrations
The intermediate current collector design with 80 lm
separation between channels was chosen to evaluate the
behaviour of the microfuel cell working at different
methanol concentrations. The system was tested at con-
centrations of 0.5, 1, 2, 3 and 5 M and the measured I–V
curves have been depicted in Fig. 6a. Measurements were
taken after the OCV stabilization sequence shown in
Fig. 6 a Performance of device working at different methanol
concentrations. b Values of maximum power density, maximum
current density and current density at maximum power of the fuel cell
depending on methanol concentrations
previous section. As already discussed, a reduction in OCV
due to the effect of methanol crossover through the mem-
brane was observed.
The methanol concentration for the maximum power in
a passive cell is much higher than in an active cell, which
has an appropriate concentration of around 1 M (Liu et al.
2004). This is due to the need to compensate the slower
mass transport rate of methanol, which in passive cells is
performed by diffusion. However, increasing concentration
leads to higher methanol crossover and consequent dete-
rioration of cell performance. In practice, a compromise
between transport benefits and crossover negative effects is
found at concentrations between 4 and 5 M (Kamarudin
et al. 2007). Figure 6a, b shows that the maximum power
density delivered by our device increases with the con-
centration and reaches a maximum value of 3.3 mW at
5 M which corresponds to a power density of 11.4 mW/

123
Microsyst Technol (2008) 14:535–541 539

Fig. 7 Cell performance curves with current collectors of different

channel densities

cm2. This power density is higher than any of the values

reported before in silicon-based methanol fuel cells without
external fuel pumping (Kamarudin et al. 2007) and close to
the values reported for stainless-steel fabricated passive
micro cells (Chen and Yang 2003, Liu et al. 2005). In
addition, it was also observed that the maximum current
density of 120 mA/cm2 is achieved at concentration of 3 M
whereas at 5 M the I–V curve drops revealing the existence
of transport limitations beyond the maximum power peak.
This drop might be caused by the additional accumulation
of water that is produced by oxidation of crossover
methanol.

4.3 Silicon plates design comparison

The effect of the channel density in the silicon current
collectors (i.e. separation between vertical channels) was
evaluated comparing the characteristic curves of the three
different geometric designs working with a 2 M methanol
concentration. As it can be seen in Fig. 7, the OCV and the
maximum current density are dependent on the separation
between channels, while the maximum power does not
change significantly. The slight difference observed in the
recorded OCV values is related to the fact that a narrower
separation between fuel channels mean more exposed area
of methanol to the membrane. This enhances methanol
crossover that leads to a reduction of the cell potential.
As expected, all the fuel cell I–V curves show the same
slope at the ohmic region, meaning that transport of elec-
trons is performed with the same efficiency in all designs.
On the contrary, channel density has a significant effect on
fuel cell concentration losses. Current collectors with a
separation of 160 lm between channels suffer from the
Fig. 8 Methanol consumption of microfuel cell. a Operating with
two methanol concentrations delivering the same power. b Fuel cell
with 3 M methanol concentration working at two different current
densities
highest mass transport limitation and show an abrupt drop
before reaching 70 mA/cm2. A separation of 80 lm allows
shifting mass transport limitations up to 90 mA/cm2
whereas the current collectors with a separation of 40 lm
between channels present almost no drop by concentration
losses and reach a current density of more than 115 mA/
cm2.
4.4 Methanol consumption
In order to test the power output performance of our silicon
passive DMFC, different continuous operation conditions
have been tested. For this purpose, the intermediate current

123
540
Fig. 9 Performance at different cell orientations. Fuel cell operating
with methanol reservoir facing up, facing down and sideways
collector design with 80 lm separation between channels
was chosen and the fuel reservoir was filled with 100 ll of
fuel. Methanol consumption tests were performed at con-
stant current operation. Figure 8a shows the transient
discharge of the microfuel cells at methanol concentrations
of 2 and 3 M at an operation point of 9 mW/cm2, that was
achieved by setting cell current values at 35 and 27 mA/
cm2 respectively. A stable cell performance was observed
in both cases until a significant decrease of methanol
concentration in the reservoir caused a sudden drop in the
output cell power. As expected, the increase of methanol
concentration from 2 to 3 M extends operation time of the
fuel cell significantly.
Moreover, the impact of the fuel cell operation point on
its transient discharge has been explored. Figure 8b shows
the power evolution of a microfuel cell fed with 3 M
methanol at two different power loads: 9 and 13.5 mW/
cm2, corresponding to different operation points in the
ohmic region of the I–V fuel curve. At the first point, set to
27 mA/cm2, the fuel cell delivered a stable output power
during a long time. On the contrary, when the cell was
operated at the second point, set to a current of 70 mA/cm2
and corresponding to the maximum power point of the cell
P–I curve, the output stability period was shorted below
5 min. The observed behaviour is related to the passive
feeding of fuel.
At the operating point corresponding to the maximum
cell power, the current is so high that it is difficult to
maintain the fuel concentration at the anode as the overall
methanol concentration in the fuel reservoir decreases, so
the operating point migrates towards the I–V curve region
regulated by mass transport limitations. These results show
123
Microsyst Technol (2008) 14:535–541
that in order to optimize the efficiency of a passive fuel cell
the operation point must be carefully chosen.
4.5 Performance at different device orientation
Dimensions of the channels etched in the silicon current
collectors are chosen small enough to create a capillary
force high enough to overcome gravity. As the channel
surface is covered by a slightly hydrophilic metal, the
capillary force keeps the liquid inside the cavities, pressing
the methanol to the membrane regardless the fuel cell
position. The device capability to work under random
orientation conditions was tested. The system was char-
acterized with a concentration of methanol 3 M in three
different orientations, with the methanol reservoir facing
up, facing down and placed sideways. The resulting I–V
plots are depicted in Fig. 9. It can be observed that the
curves start with the same OCV but finish at different
current density points. The maximum current density was
recorded for the fuel cell working upwards. In this case,
capillary force and gravity work in the same direction. The
worst case was recorded when the cell was placed on its
side, with a 20 mA/cm2 decrease of the maximal current
density, which means a 15% change at the highest intensity
point. However, the variation in the maximum power
density measured at different orientations supposes only a
13% decrease in the performance when the cell is place
sideways.
5 Conclusion
A passive direct methanol microfuel cell has been fabri-
cated and characterized under different conditions. First of
all, it has been shown that the channel density in the current
collectors has an important influence in fuel cell perfor-
mance, mainly affecting fuel cell concentration losses.
Moreover, the I–V characteristic curves of the microfuel
cell have been measured under various methanol concen-
trations showing a maximal power density around 12 mW/
cm2 at a concentration of 5 M despite the mass transport
limitations observed at high current densities.
The evolution of the output power of the device at dif-
ferent current densities and methanol concentrations has
been analysed. These tests have shown that when filled
with 100 ll of methanol at 3 M, the device is able to
provide a power density of 9 mW/cm2 for about 50 min.
However, when the device is operated at its maximum
power, lower energy efficiency is obtained, revealing the
importance of choosing the appropriate operation point in a
passive fuel cell. Finally, as the device has been designed
to be able to work regardless of its orientation, it has been
Microsyst Technol (2008) 14:535–541
shown that a maximal difference of 13% is observed at the
maximum power density when the device is placed side-
ways making this device suitable as a portable power
source.
Acknowledgments The work presented in this paper was supported
by EU Integrated Project SENSATION (FP6-IST-507231) and
Spanish CICYT Project CROMINA (TEC2004-06854-C03-02).
References
Aravamudhan S, Rahman ARA, Bhansali S (2005) Porous silicon
based orientation independent, self-priming micro direct ethanol
fuel cell. Sens Actuators A 123–124:497–504
Bae B, Kho BK, Lim TH, Oh IH, Hong SA, Ha HY (2006)
Performance evaluation of passive DMFC single cells. J Power
Sources 158:1256–1261
Chen CY, Yang P (2003) Performance of an air-breathing direct
methanol fuel cell. J Power Sources 123:37–42
Chen R, Zhao TS (2007) Porous current collectors for passive direct
methanol fuel cells. Electrochimica Acta 52:4317–4324
Jacobson SA, Epstein AH (2003) An informal survey of power
MEMS. In: The international symposium on micro-mechanical
engineering, December 1–3, 2003 ISMME2003-K18
Kamarudin SK, Daud WRW, Ho SL, Hasran UA (2007) Overview on
the challenges and developments of micro-direct methanol fuel
cells (DMFC). J Power Sources 163:743–754
541
Kho BK, Oh IH, Hong SA, Ha HY (2004) The effect of pretreatment
methods on the performance f passive DMFCs. Electrochimica
Acta 50:781–785
Kho BK, Bae B, Scibioh MA, Lee J, Ha HY (2005) On the
consequences of methanol crossover in passive air-breathing
direct methanol fuel cells. J Power Sources 142:50–55
Kim YJ, Bae B, Scibioh MA, Cho EA, Ha HY (2006) Behavioral
pattern of a monopolar passive direct methanol fuel cell stack. J
Power Sources 157:253–259
Larminie J, Dicks A (2003) Fuel cell explained, 2nd edn. Wiley,
England
Liu JG, Zhao TS, Chen R, Wong CW (2005) Effect of membrane
thickness on the performance and efficiency of passive direct
methanol fuel cells. Electrochem Commun 7:288–294
Liu JG, Zhao TS, Liang ZX, Chen R (2006) Effect of membrane
thickness on the performance and efficiency of passive direct
methanol fuel cells. J Power Sources 153:61–67
Liu J, Sun G, Zhao F, Wang G, Zhao G, Chen L, Yi B, Xin Q (2004)
Study of sintered stainless steel fiber felt as gas diffusion backing
in air-breathing DMFC. J Power Sources 133:175–180
Meninger S, Mur-Miranda J-O, Amirtharajah R, Chandrakasan AP,
Lang J (2001) Vibration-to-electric conversion. IEEE Trans
VLSI Syst 9:64–76
Santander J, Sabaté N, Esquivel JP, Torres N, Del Campo FJ, Cané C
(2006) Fabrication and characterization of a passive and
orientation independent micro-DMFC. In: Proceeding of Pow-
erMEMS 2006, San Francisco, USA, pp 77–80
123