Design and Fabrication of Micro Direct Methanol Fuel Cell with New ...

Design and Fabrication of Micro Direct Methanol Fuel Cell
with New Concept Planar Structure
by Means of Micro Electro Mechanical Systems
by
Shinji Motokawa
Thesis submitted to Waseda University
March, 2005

CONTENTS:
Preface 1
Chapter 1: General Introduction 3
1.1 Fuel Cells 5
1.1.1 Polymer Electrolyte Fuel Cell 12
1.1.2 Direct Methanol Fuel Cell 15
1.2 Requirement of Electric Device for Battery 17
1.3 Miniaturized System by Using Micro Fabrication Technology 20
1.3.1 Micro Electro Mechanical Systems (MEMS) 20
1.3.2 Power MEMS 21
1.4 Micro Fuel Cell on Si Wafer by Using MEMS Technology 22
1.4.1 Micro polymer electrolyte fuel cells 22
1.4.2 Micro direct methanol fuel cells 23
1.5 Numerical Models and Computational Simulation of Fuel Cells 24
References 25
Chapter 2: Design and Fabrication of New Concept Micro Direct Methanol
Fuel Cell Utilizing MEMS Technology 27
2.1 Characteristic of the Configuration of the New Concept Micro Direct
Methanol Fuel Cell on Si Wafer 30
2.1.1 Working principle of the operation 35
2.1.2 Cell materials 36
2.1.3 Mass flow control of fuel and oxidant 37
2.1.4 Relative advantage and disadvantage of different configuration of
micro direct methanol fuel cell 37
i

2.2 Design and Fabrication of Test Cell 39
2.2.1 Design of the test cell 39
2.2.2 Consideration of fabrication process of the test cell 40
2.3 Fabrication Process 40
2.3.1 Formation of fluidic channel and feedhole for mass flow by using
lithography and chemical etching
2.3.2 Formation of electrode on the lateral wall in channels by electro
deposition
2.3.3 Micro DMFC assembly 43
References 49
Chapter 3: Electrochemical Evaluation of New Concept Micro Direct
Methanol Fuel Cell
3.1 Performance of the Test Cell 54
3.1.1 Experimental 54
3.1.2 Performance of the test cell 56
3.1.3 Effect of flow velocity on the performance 59
3.1.4 Effect of concentration of methanol on the performance 61
3.1.5 Effect of temperature 62
3.1.6 Durability of the test cell 64
3.1.7 Conclusion 64
3.2 Study of Conditions of Pt and Pt-Ru Catalyst Layer 65
3.2.1 Experimental 65
3.2.2 Effect of conditions of electro deposition on morphology and
catalysis of Pt and Pt-Ru
3.2.3 Conclusion 69
3.3. Comparison of Ni/Au and Ti/Au on SiO2 as Current Collector 70
3.3.1 Experimental 70
3.3.2 Effect of thin metal layer on properties of current collector 71
3.3.3 Conclusion 71
ii
40
42
53
67

3.4 Study of micro flow channel electrode 72
3.4.1 Experimental 72
3.4.2 Effect of scale parameters of channels 76
3.4.3 Conclusion 77
References 78
Chapter 4: Numerical Model for New Concept Micro DMFC 79
4.1 Theories and Model 81
4.1.1 Assumption and basic formula 81
4.1.2 Effect of distance between channel electrodes 85
4.1.3 Effect of Nafion thickness 86
4.1.4 Effect of flow velocity 88
4.1.5 Ratio of flow rate of both channels and methanol permeability 89
4.1.6 Ratio of flow rate of both channels and proton transport 90
4.1.7 Conclusion 90
References 91
Chapter 5: Development of Direct Methanol Fuel Cell System for
Electronic Devices
5.1 Micro DMFC System 95
5.1.1 Design guidelines micro DMFC system 95
5.1.2 Prospect for practical application 96
5.2 Micro DMFC System Integration on Single Si Wafer 97
5.2.1 Integration of 3 cells on single Si wafer 98
5.2.2 Performance of the 3cell stack system 102
5.2.3 Conclusion 102
5.3 Study of separator materials for particle application 103
5.3.1 Experimental 103
5.3.2 Performance of the micro DMFC on flexible resin substrate 107
5.3.3 Conclusion 108
iii
93

5.4 Perspective future work 108
References 109
Chapter 6: General Conclusions 111
List of Achievement 117
Acknowledgement 119
iv

Preface
Recently, concepts of portable, mobile, and wearable for electronic devices spread
out rapidly as the remarkable progress in information society. Therefore, the
requirements for the power sources of an electronic device, such as more compact and
higher power density per weight, would rapidly increased as repletion of the contents
despite efforts of reduction the consumption power. In the future, these requirements
might lead to efforts for the improvement of conventional batteries, such as Lithium ion
and Nickel-Hydrogen. However, the growth might be insufficient toward the
requirements. The direct conversion of chemical into electrical energy via fuel cells has
been at the centre of attention of electrochemical research and technology development.
In particular, a direct methanol fuel cell system is relatively simple and could be easily
miniaturized since it does not need a fuel reformer, complicated humidification, or
thermal management system.
The direct conversion of chemical into electrical energy via fuel cells has been at
the centre of attention of electrochemical research and technology development.
Additionally, a direct methanol fuel cell system is attractive as new power sources for
portable electric devices, such as note PC, mobile phone etc, due to it is unnecessary a
reformer, high power density, performable at low temperature, and easy for handling.
On the other hand, Micro-Electro-Mechanical Systems (MEMS) is the integration
of mechanical elements, sensors, actuators, and electronics on a common silicon
substrate through micro-fabrication technology. While the electronics are fabricated
using integrated circuit (IC) process sequences (e.g., CMOS, Bipolar, or BICMOS
processes), the micromechanical components are fabricated using compatible
"micromachining" processes that selectively etch away parts of the silicon wafer or add
new structural layers to form the mechanical and electromechanical devices. MEMS
promises to revolutionize nearly every product category by bringing together
silicon-based microelectronics with micromachining technology, making possible the
realization of complete systems-on-a-chip.
There are the backgrounds of the study discussed in this thesis. Design and
fabrication of micro fuel cells by using micro fabrication technology on previous work
were investigated.
In this study, a micro direct methanol fuel cell with new concept planar structure
was designed and fabricated by means of micro electro mechanical systems fabrication
technology and electrochemical concerns. The cell operation was demonstrated and the
cell performance was investigated with respect to fabrication process such as channel
1

shape, conditions of catalyst formation, and the catalyst morphology. A computer fluidic
dynamics model has been adapted to investigate the effects of several parameters of the
micro DMFC. Moreover, by designing the micro direct fuel cell array with several cells
as stack and adaptation of a polymer material as base substrate, the possibility of
practical use was discussed.
2

Chapter 1
General Introduction
3

maximum energy available is determined by the amount of chemical reactant stored
within the battery itself. The battery will cease to produce electrical energy when the
chemical reactants are consumed. In a secondary battery, the reactants are regenerated
by recharging, which involves putting energy into the battery from an external source.
The fuel cell, on the other hand, is an energy conversion device that theoretically has the
capability of producing electrical energy for as long as the fuel and oxidant are supplied
to the electrodes. Different fuels can be used, such as hydrogen, ethanol, methanol, or
gaseous fossils fuels like natural gas. Solid or liquid fossil fuels need to be gasified first
before they can be used as fuel. Oxygen or air can be used as oxidant.
The characteristics of fuel cells are summarized as follows; (1) the theoretical
efficiency of the energy conversion is very high, especially at low temperature. (2) The
scale merit is small. (3) No pollutant (especially NOx, SOx), no noise and no vibration.
The environmentally friendly generation system is possible. (4) The fuel cell is an
electrochemical system, which means the reaction at the electrode/electrolyte interface
is most important. The electrochemical reactor is a 2 dimensional reactor and the
volume efficiency is generally not good. (5) The cell voltage of a single cell is normally
less than 1V. In order to get a high power density, the reactants (fuel and oxidant)
should react smoothly at the electrode/electrolyte interface. (6) Total system is very
expensive, compared to a conventional electric generation system or an internal
combustion engine.
Fig.1.2 Four cells of Grove’s battery to drive an electrolytic cell, 1842.
6

The first hydrogen-oxygen, very dilute sulfuric acid Fuel Cell was described by
Grove in a postscript from January 1839 to a latter he wrote in December 1838 to the
Philosophical Magazine. The new cell consisted of two platinum strips surrounded by
closed tubes containing hydrogen and oxygen, respectively, formed by preliminary
electrolysis of the electrolyte as shown in Fig.1.2 Grove showed that he was already
well aware of the fact that the three-phase contact is essential: “As the action could only
be supposed to take place with ordinary platinum foil, as the line or water-mark where
the liquid, gas and platinum met [1].
The thermodynamic aspects of a fuel cell are as follows [2]. The fuel cell is an
electrochemical device which converts the free-energy change of an electro chemical
reaction into electrical energy. One may thus write the expression.
where,
°
°
∆ = −nFEr
G , (1.1)
°
E r is the reversible potential of the
encountered fuel cell reaction is
1
+ O2
= H O . (1.2)
2
H 2
2
cell. The simplest and most commonly
The free-energy change of this reaction under standard conditions of temperature and
pressure (T=25 o C, PH2 = PO2 = 1 atm, H2O in liquid state) is 56.32 kcal mol -1 . The
number of electrons transferred in this reaction is 2. Thus, the reversible potential is
1.229 V. The variations of the reversible potential (Er) with temperature and pressure are
expressed by the equations
⎛ ∂E
⎞
= +
∆S
nF
0
( T − 298)
+ E + ( T −
0
Er Er
⎜ ⎟
r
⎝ ∂T
⎠ p
( ∆n)
0 RT
= E − ln P , (1.4)
nF
Er r
298
)
, (1.3)
where, ∆ n is the change in number of gas molecules during the reaction. The entropy
change, ∆ S for reaction (1.1) is -39 entropy units, while ∆ n is -3/2. Thus, Eqs. (1.3)
and ( 1.4) show that the reversible potential decreases with an increase of temperature,
while
the behavior is opposite with and increase of pressure.
At
temperature above 100 o C, water is produced as a vapor in the cell. The value of
∆ S is considerably less when water is produced in this state than as a liquid. Thus
d is -0.54 mV/ o ∂ E / ∂T
is -0.25 mV/ C in the latter. The effect
o C in the former case an
of pressure on Er is also less when water is produced as a vapor than as a liquid ( ∆ n =
-1/2 in the former and -3/2 in the latter case).
7

Hydrogen is not a primary fuel. Thus, attempts were made in the 1960s to use
primary fuels such as hydrocarbons (CH4 to C10H22) and coal in fuel cells. However,
due to the high degree of irreversibility of the anodic oxidation reactions of the
hydrocarbons, these attempts proved futile. Thus, these fuels were processed to produce
hydrogen for low-temperature fuel cells and H2 and CO for the higher temperature fuel
cells by steam-reforming reactions of the hydrocarbons and by gasification of coal.
Several other types of fuels (methanol, ethanol, ammonia, and hydrazine) were also
researched
in the 1960s. Practically all of these fuel cell reactions have a
thermodynamic reversible potential from 1.0 to 1.2V.
Even if there were no efficiency losses in H2-O2 fuel cells (due to activation, mass
transport, and ohmic overpotentials), heat would still have to be rejected from a fuel cell
because ∆ S is negative for reaction (1.2). Thus, the theoretical efficiency of H2-O2
fuel cells at 25 C, based on the enthalpy change of the reaction (commonly referred to
o
as the higher heating value by mechanical engineers), is 83%. There is only one fuel cell
reaction where ∆ S is positive, namely,
1
C + O → CO.
(1.5)
2 2
For this reaction, the theoretical efficiency is 137% at 150 o C. The reaction
C + O → CO (1.6)
2
has entropy change of 0 e.u. Thus, for
this reaction, the theoretical efficiency is close to
unity. As a rough rule of thumb, if
2
∆ n is positive, the entropy change is positive (due
to increasing disorder), while if ∆ n is negative (increasing order), ∆ S is negative,
and if ∆ n is zero ∆ S is also zero.
The electrode kinetics aspects of a fuel cell are as follows. The vitally important role
of electrode kinetics on the performance of fuel cells (particularly those operating at low
and intermediate temperature, 25-200 o C) is best illustrated by a typical cell potential
versus current density plot as shown in Fig.1.3. Three distinct regions are illustrated in
this plot. The predominant cause of the difficulties in attaining high energy efficiencies
and high power densities in low- to medium-temperature fuel cells is the low
electrocatalytic activity of most electrode materials for the oxygen electrode reaction.
The hydrogen electrode shows a linear relationship of its half-cell potential versus
current density plot from zero to the highest value of current density in the fuel cells
using phosphoric acid (T= 200 o C) , potassium hydroxide (T= 80 o C), or a proton-
conducting electrolyte (Nafion or Dow membrane, T= 85 o C). This is not the case with
the oxygen electrode, where a semi-exponential relation between its half-cell potential
and current density is observed. Thus, at low current densities, the entire loss in the fuel
8

cell potential from the reversible value is due to activation overpotential at the oxygen
electrode.
Fig. 1. 3 Typical plot of cell potential vs. current for fuel cells illustrating regions
of control by various types of overpotentials.
Another problem encountered is that the reversible potential is not attained even at
zero current density. This problem is again due to the oxygen electrode. The exchange
current density for this reaction is so low that competing anodic reactions (for example,
oxidation of the platinum electrocatalyst, corrosion of carbon, oxidation of organic
impurities) play a significant role. The net results is that the open circuit potential
(OCP) is a mixed potential, which is lower than the reversible potential for the H2-O2
fuel cell reaction by about 0.1 to 0.2V. Thus, even at close to zero current densities, the
efficiency of a fuel cell is lower than its theoretical
value by 8-16%. The cell potential
(E) versus current density (i), from a current density 0 to the value at the end of the
linear region,
may be expressed by the relation
where,
The
Cell voltage / V
1.0
0.5
Thermodynamic reversible cell potential (E)
Cell potential losses due to activation
polarization
Resistance losses due to ohmic
polarization
Current density / mA/cm2 Current density / mA/cm2 Current density / mA/cm2 E = E − blogi
− Ri , (1.7)
0
E0 Er
blogi0
+ = , (1.8)
Mass transport losses
due to concentration
polarization
parameters b and i are the Tafel slope and exchange current density for the
0
9

oxygen reduction reaction, and R accounts for the linear variation of overpotential
(predominantly ohmic) with current density, which is observed in the intermediate
range.
The situation is more complex when organic fuels (hydrocarbons and alcohols) are
used directly in fuel cells. The exchange current densities for these reactions are as low
as or even lower than those for the oxygen reduction reaction. Thus, the low open
circuit potentials and the exponential decrease of the
half-cell overpotentials with
current densities at both electrodes account for their relatively poor performance. It is
worth while rationalizing the shape of the E versus I plot (Fig.1.3). by differentiating
Eqs. (1.7):
∂E
b
= − − R . (1.9)
∂i
i
At a low current density, the differential resistance of the cell is high because of the first
term on the right side of Eqs. (1.9); consequently, there is a steep fall of cell potential
with increasing current density. At higher current densities, b

fabrication of fuel cell electrodes with optimized structures; thus, mass transport
limitations are rarely encountered at current densities up to a few A/cm 2 . In
high-temperature fuel cells using molten carbonate (T= 650 o C) and solid oxide (T=
1000
o C) electrolytes, the exchange current densities of the fuel cell reaction are quite
high (>1mA/cm 2 ), and thus the cell potential versus current density plot is linear
throughout the entire current density range (0 to a few hundred mA/cm 2 ). The typical
several cells potential versus current density plot as shown in Fig.1.4
Theoretical and experimental electrode kinetic studies of fuel cell reactions
(1.1)-(1.6) have led to the engineering design, development, and demonstrations of fuel
cell power plants exhibiting high levels of performances (high energy efficiency, high
power density, and reduction in noble metal loading. The major accomplishments are (i)
The design and fabrication of porous gas diffusion electrodes with optimized structures
to enhance (a) diffusion of dissolved gases (H2, O2) to reaction sites, (b)
electrochemically active sites, and (c) ionic transport through porous electrodes; (ii) use
of supported electrocatalysis (Pt crystallites on high-surface-area carbon) to
significantly reduce noble metal loadings, as compared with loadings when unsupported
platinum electrocatalysts were used; (iii) inhibition of CO poisoning in phosphoric acid
fuel cells by elevation
of operating temperature; (iv) utilization of alloys and
heat-treated metal-organic macrocyclics as electrocatalysts which exhibit higher
exchange non-noble metal electrocatalysts for high-temperature fuel
cells with molten
carbonate or solid oxide electrolytes; and (vi) use of thin electrolyte layers to minimize
ohmic over potentials.
The efficiency (ε ) of a fuel cell varies with current density in the same manner as
that of the cell potential with current density (cf. Fig. 1.3) because
nFE
ε = . (1.10)
∆H
The power density of a fuel cell is expressed by the relation
P = Ei . (1.11)
The shape of P versus i or P versus E plot is a parabola if the E versus i relation is linear.
The parabola is distorted for low- and intermediate-temperature fuel cells because of the
semi logarithmic relation between cell potential and current density at low values of i
and sudden drops of cell potential at high current densities.
11

The overall fuel cell reaction is thus as expressed by Eqs (1.2). The oxygen
reduction reaction is the slower reaction (the exchange current for this reaction is at
least three orders of magnitude lower than that of hydrogen oxidation), and thus the
challenge is the enhancement of the electrocatalytic activity of this reaction.
The functioning of a fuel cell relies on the formation of a stable three-phase
boundary in the immediate vicinity of the electrocatalyst site. This boundary, for fuel
cells
with liquid electrode, the liquid electrolyte, which has worked its way into the
pores to the electrocatalyst. Alternatively, a tin film of the electrolyte layer may form on
the electrocatalyst, and the reactant gas can dissolve in the electrolyte interface and then
diffuse through the electrolyte film to the electrocatalyst.
In the SPFC, there is no liquid electrolyte and it is very probable that a thin film
forms on the electrocatalyst. The presence of excess liquid water hampers the easy
access of gas into the porous structure of the electrode, and the fuel cell undergoes a
decrease in performance due to mass transport limitations of the oxidant gas. This
flooding situation is exacerbated by the ingresses of water from the hydrogen side. The
proton formed during hydrogen oxidation is usually strongly hydrated and causes
transport of water from the anode to the cathode. This phenomenon
has further
repercussions
in that the loss of water causes drying out of the membrane and an
increase in ohmic behavior at the anode-membrane interface. It is therefore of
paramount importance that methods for efficient addition of water to the hydrogen
electrode and removal of water from the oxygen electrode be divided.
The SPFC operates at a lower temperature than the PAFC, MCFC, and SOFC,
which are in a more advanced state of development. The limit on the temperature at
which the fuel cell operates is set by the thermal stability and conductivity
characteristics of the polymeric membrane that is used as its electrolyte. With Nafion, it
is best not to exceed an operating temperature of 85 o C. With the new class of ionomers
developed by the Dow Chemical Company, this temperature can be from atmospheric to
8atm. Air pressure up to 8 atm have been used. Operation at higher pressure is necessary
for attaining higher power
densities, particularly with air as the cathodic reactant. The
pressures,
in general, are maintained equal on either side of the membrane. This
minimizes the problem of gas crossover through the membrane. The crossover reduces
the cell potential and also increases the risk of forming an explosive mixture of
hydrogen and oxygen.
The water management in the membrane and electrode assembly of the SPFC is
fairly complex and requires dynamic control to match the varying operating conductions
of the fuel cell. A simple humidification scheme, used at Texas A&M
University and
13

Los Alamos National Laboratory, is to disperse the gas through a ceramic fit immersed
in a column of water. The rate of the gases introduced into the fuel cell. High flow rates
cause entrainment stoichiometric flow requirements . Steam generation and atomization
techniques have also been tried for humidifying the reactant gases.
These polymers have the following desirable
properties: high oxygen solubility,
high
proton conductivity, high chemical stability, low density, and high mechanical
strength. As proton transport occurs from the anode to the cathode, a proton-conducting
solid electrolyte is essential. The polymeric membranes developed by DuPont (Nafion)
used in the first few versions of the SPFC.
The gas-diffusion electrodes that have been used so far contain unsupported (2-10
mg/cm 2 ) or supported platinum (platinum supported on carbon; 0.4 mg/cm 2 , 10% or
20 % Pt on C) electro-catalysts. The latter type of electrode is essentially the same as
that developed for the PAFC and has not yet been optimized for SPFC. The electrodes
have a carbon backing (cloth or paper). A Teflon emulsion is used to bond the platinum
particles (nm diameter) to the carbon layer. The active layer is deposited a few microns
thick on the substrate layer (about 10 µm for unsupported and 50 µm
for supported
electrocatalysts).
At Los Alamos National Laboratory and at Texas A&M University
have shown that the performance of electrodes with low platinum loading (0.4 mg/cm 2 )
is comparable to those with high platinum loading. This was achieved by impregnating
the active layer of the former type of electrode with a proton conductor.
The graphite support structures have gas flow
paths to allow for laminar flow of
gases
across the electrode. Rectangular and circular flow paths in axial and concentric
directions have been attempted at Genera Electrics, United Technologies, and Delco
Remy Division of Genera Motors Corporation. Flow-modeling
studies on these
electrodes
are in progress in several laboratories.
Tsutsumi et al [5] reported the fuel cell using DME has equal performance with
methanol as fuel. Fujiwara et al [6] demonstrated a direct ascorbic acid fuel cell by
using variety novel metal and carbon material as anodic catalyst.
14

Methanol is oxidized a carbon dioxide at the anode of µ-DMFC. However the oxidation
reaction proceeds through the formation of carbon monoxide as an intermediate which
strongly adsorbs on the surface of Pt catalyst [9]. Therefore, a potential, which is much
more anodic than the thermodynamic value, is needed to obtain a reasonable reaction
rate [10].
Furthermore, the µ-DMFC generally needs sub-systems such as pump, a valve and a
gas separator to increase energy and power density. In contrast, the µ-PEFC using
hydrogen as a fuel has higher power density than that of the µ-DMFC; however, the
problem of fuel storage still remains [11]. Currently, chemical hydrides [12] are under
investigation for portable hydrogen sources.
Many studies of micro fuel cells are being performed from view points of catalytic
electrodes, ion conductive polymers, cell structures, systematization, packaging,
fabrication and fuel storage. Recently, new types of nano-materials such as anhydrous
ion conductive polymer with fullerene and a catalytic electrode with carbon nano-horns
have been demonstrated. These material technologies are important for variety sizes of
fuel cells. On the other hand, micro fuel cells need special developments on structures,
systematization, packaging and fabrication unlike technology is a crucial tool for these
developments [13-16].
Micro fuel cell systems can be divided into three main types. One is the system that
directly converts fuel such as methanol into electricity. Second is the system of
reformed micro fuel cell. The hydrogen used in reformed hydrogen system is obtained
from methanol or a hydrocarbon such as methane or propane [17-19]. The reason it is
called reformed hydrogen is that the methanol or hydrocarbon is processed in a device
known as a reformer to produce the hydrogen just upstream of the fuel cell. The other
system of micro fuel cells is borohydrides system. Borohydrides complex compound
(NaBH4 or KBH4) can be used as a source of stable hydrogen supply if dissolved in
dilute solution [20-23].
Among of them, micro fuel cell system combined with a micro-fuel reformer seems
one of promising candidates to power laptop computers, video camcorders etc. The
micro fuel cell/fuel reformer system is expected to become a more compact system
compared to a DMFC system, because a PEFC has approximately one order of
magnitude higher power density than a DMFC. Additionally, it has potential to be
fueled by various hydrocarbons including methanol, ethanol, butane and dimethyl ether.
The micro-fuel reformer basically has a system configuration similar to
conventional fuel reformers, consisting several key components such as a reforming
reactor, a CO separator, a heat source and fluid control peripherals. In the micro-fuel
18

eformer, these components should be as small as possible and of low power
consumption.
These approaches have their own pros and cons. For example, a DMFC system is
simple in structure and consists of a small number of components. However, methanol
needs to be diluted in order to prevent the fuel from going through the electrolyte
membrane to the cathode side (this phenomenon is called “methanol crossover”, which
lowers power generation efficiency. On the other hand, the reformer-based system can
achieve much higher power generation efficiency because it extracts hydrogen out of
the fuel and directly supplies hydrogen to the power generation cell. However, there
former is complex in structure and needs a number of components. All these
technologies are currently under development to apply a high power density and a
reliable operation in the hands of the customer.
19

1.3 Miniaturized System by Using Micro Fabrication Technology
1.3.1 MEMS
Micro-Electro-Mechanical Systems (MEMS) is the integration of mechanical
elements, sensors, actuators, and electronics on a common silicon substrate through
micro-fabrication technology. While the electronics are fabricated using integrated
circuit (IC) process sequences (e.g., CMOS, Bipolar, or BICMOS processes), the
micromechanical components are fabricated using compatible "micromachining"
processes that selectively etch away parts of the silicon wafer or add new structural
layers to form the mechanical and electromechanical devices [24].
MEMS promises to revolutionize nearly every product category by bringing
together silicon-based microelectronics with micromachining technology, making
possible the realization of complete systems-on-a-chip. MEMS is an enabling
technology allowing the development of smart products, augmenting the computational
ability of microelectronics with the perception and control capabilities of micro-sensors
and micro-actuators and expanding the space of possible designs and applications.
Microelectronic integrated circuits can be thought of as the "brains" of a system and
MEMS augments this decision-making capability with "eyes" and "arms", to allow
micro-systems to sense and control the environment. Sensors gather information from
the environment through measuring mechanical, thermal, biological, chemical, optical,
and magnetic phenomena. The electronics then process the information derived from the
sensors and through some decision making capability direct the actuators to respond by
moving, positioning, regulating, pumping, and filtering, thereby controlling the
environment for some desired outcome or purpose. Because MEMS devices are
manufactured using batch fabrication techniques similar to those used for integrated
circuits, unprecedented levels of functionality, reliability, and sophistication can be
placed on a small silicon chip at a relatively low cost.
There are at least seven MEMS fabrication technologies falling into three classes:
bulk micromachining, surface micromachining, and high-aspect ratio micromachining
(HARM). Bulk micromachining is a subtractive process that involves the selective
removal of the wafer substrate material to form the MEMS structure, which can include
cantilevers, holes, grooves, and membranes. Surface micromachining is an additive
process that involves depositing combinations of thin structural and sacrificial layers,
wherein the sacrificial layers are subsequently removed to form raised structures that
can include gears, comb fingers, cantilevers, and membranes. HARM includes deep
ultraviolet or x-ray lithography techniques collectively known as LIGA (from the
20

German Lithographie, Galvanoformung, Abformung, meaning lithography,
electroplating, and molding). LIGA makes it possible to create micro-components out of
polymers, metals, and ceramic materials using micromachined molds. Although the
traditional approach in MEMS has been to develop application specific fabrication
processes, the trend is reversing and the industry is seeing a dominant few process
technologies starting to emerge.
1.3.2 Power MEMS
A micro-scale combustor can achieve power densities (power per unit volume) of
2000 mega-watts per cubic-meters whereas the best lithium battery technology only
delivers 0.4 mega-watts per cubic-meters. As a result, there is substantial interest in
power MEMS as a replacement in applications that would ordinarily be battery powered.
Military interest derives from the need to deploy remote sensors and devices, whereas
commercial interest centers on replacement of batteries in a variety of portable
electronics. One of the most pressing hurdles remaining is that all power MEMS
prototypes to date have been demonstrated using gaseous fuels. However, liquid
fuels-and therefore liquid fuel atomization-are crucial to the future development of
power MEMS. Without the energy density of liquid fuels (500-700 times greater than
gaseous fuels like hydrogen and methane), power MEMS devices will not have long
enough refueling intervals to be of use. A method of liquid fuel delivery and atomization
is being developed at the ATESR Laboratory. Conventional atomization techniques
where liquid fuels are forced at high pressure through pinhole orifices encounter
daunting challenges when shrunk to the MEMS scale. Recently, novel micro energy
source is shift from preexisting batter. The system is called Power MEMS. Power-
MEM is based on concern of chemical energy convert electric or kinetic energy.
21

1.4 Micro Fuel Cell on Si Wafer by Using MEMS Technology
Yamazaki et al [25] suggested, by using the MEMS technology, the possibility of
reduction of the diffusion paths by preparing fine channels in which the materials pass
not by diffusion but by bulk flow. Thereby, the performance of the cell stack might be
considerably increased. Hockaday et al. indicated the promising of miniature direct
methanol fuel cell in his patents [26-28].
Fuel cells based upon a silicon substrate will not match the raw materials costs of
traditional carbon-based fuel cell components, but if packaging integration and on-chip
control become cost drivers for the system, the silicon based fuel cell power source will
offer advantageous over the more traditional carbon-based embodiments.
1.4.1 Micro polymer electrolyte fuel cells
The micro electro mechanical systems (MEMS) technology has been developed at
the various requests of environmental and internal sensors, machining of silicon and
metal derivatives, optical and biomedical systems, and micro-fluidics. The prospective
potential for miniaturization and economical mass production of small fuel cells is the
main reason for the use of MEMS processes to fabricate the micro fuel cells; however,
we can expect more basic effect of the micro fabrication process on the performance of
the fuel cell of various sizes. Most conventional cell stacks of PEMFC consist of
separators with gas channels, carbon paper sheets, carbon particles, and catalysis
supported by carbon particles. In those stacks, the molecules of the fuel must move
from the gas channels to the reaction points on the catalysis particles in the anode. In the
cathode, the molecules of the air must be transported from the channels to the surface of
the catalysts while the water molecules move in the reverse direction. These molecules
move through the pores in the electrodes by the diffusion process. Therefore, the
efficiency of the material transportation is not high. By using the MEMS technology, we
can reduce these diffusion paths by preparing fine channels in which the materials pass
not by diffusion but by bulk flow. This may considerably increase the performance of
the cell stack [25].
The components of a novel miniature fuel cell/fuel reformer system fueled by liquid
gases such as butane and propane were prototyped by MEMS technology and tested. In
this system, fuel, air and water are supplied to the fuel reformer by utilizing the vapor
pressure of the liquid gas for the reduction of power consumption by peripherals and the
simplification of the system [29]. A Micro-Polymer Electrolyte Fuel Cell (µ-PEFC) with
“alternating structure” was demonstrated. The alternating structure has a series of single
22

cells formed in one plane, and the polarization of each single cell is alternately inverted.
This structure has self-assembled cell interconnection on micro-machined silicon
substrates.
1.4.2 Micro direct methanol fuel cells
A direct methanol type micro fuel cell is most attractive for energy supplier to
portable electric devices; however, the power density is still low. The main reasons of
the low power density of a DMFC are the low catalytic activity of electro chemical
methanol oxidation at low temperature and the methanol crossover from the fuel side to
the air side through a PEM. Therefore, the DMFC generally needs sub-system such as a
pump, a valve and a CO2 separator to improve energy and power densities [13, 30, 31].
Two types of DMFC systems are investigated, passive DMFC and active DMFC.
A passive DMFC system has the basic advantages of lower system volume and
weight, simpler system design, simpler mode of operation costs. However, this system
create low power density because methanol permeation through the membrane. This
migration of methanol from the anode to the cathode side leads to fuel losses and
reduced voltage due to the formation of a mixed potential at the cathode.
Active DMFC system consists of fuel delivery, fluid pumping, gas-liquid separation,
water balance, humidity control, and air delivery. It incorporates micro pumps, valves,
methanol concentration sensor and moisture sensor. In order to achieve high power
density and device compactness, pure methanol will be stored tin this system and the
water obtained at cathode will be pumped back to compensate the water loss at anode.
During the operation of the DMFC system, the methanol concentration is monitored
using a methanol concentration sensor and the feedback is provided to the control valve
of the methanol reservoir.
Pure methanol is delivered from the reservoir into the circulating stream to maintain
the set concentration. Carbon dioxide bubbles generated at the DMFC stack are
removed with a gas-liquid separate. The air flow at the cathode, driven by natural
convection and enhanced by the air feed pump, provides the oxygen to the stack. The air
stream decreases the water balance; a micro pump delivers the water from the cathode
back to the anode, to compensate for the lost amount. Due to the relatively high surface
to size ration of this micro system, the micro DMFC will be likely staying close to room
temperature at all times. To conserve power, all the active components, such as micro
valves and pumps, will be electro-statically operated on demand.
23

1.5 Numerical Models and Computational Simulation of Fuel Cells
C. Marr et al. [32] investigated the composition and performance optimization of
cathode catalyst platinum and catalyst layer structure in a proton exchange membrane
fuel cell by including both electrochemical reaction and mass transport process. K. Scott
et al. [33] reported that the performance and modeling of a direct methanol fuel cell
based on a solid polymer electrolyte membrane (SPE). The performance of the fuel cell
is affected by the cross-over of methanol from the anode to the cathode through the
polymer membrane and this behavior is modeled. A direct methanol fuel cell (DMFC)
model has been developed and experimentally verified, with which fundamental
calculations of the DMFC were carried out by H. Dohle et al. [34]. The performance of
the fuel cell is adversely influenced by methanol permeation from the anode to the
cathode. Moreover, the formation of a mixed potential is possible both at the anode and
cathode and has a large negative effect on the energetic performance of the fuel cell.
24

References
[1] W.R.Grove, P. Mag., 21, 417 (1842).
[2] L.J.M.J. Blomen, M.N. Mugerwa, Plenum Press, New York (1993).
[3] N. Mano, F. Mao et al., Chem. Commun., 4, 518 (2003).
[4] A. Heller, Phiys. Chem. Chem. Phys., 6, 209 (2004).
[5] Y. Tsutsumi, T. Satou and A. Yoshizawa, proc. Abstracts of Papers, 12 th ,
International Symposium on Alcohol Fuels, 403 (1998).
[6] N. Fujiwara, K. Yasuda, T. Ioroi, Z. Siroma, Y. Miyazaki, and T. Kobayashi,
Electrochem. Solid-State Lett., 6, A257 (2003).
[7] R.G. Hockaday, M. DeJone, C. Navas, P.S. Turner, H.L. Vaz and L.L. Vazul, A
BETTER POWER SUPPLY FOR PORTABLE ELECTRONICS: MICRO FUEL
CELLS, 791 (2000).
[8] T. Yamaguchi, et al., Advanced Materials, 15, 14 (2003).
[9] R. Parsons, T. VanderNoot, J. Electroanal. Chem. 9, 257 (1988).
[10] M. P. Hogarth, G. A. Hards, Platinum Met. Rev. 150, 40 (1996).
[11] S. R. Narayanan, H. Frank, B. Jeffries-Nakamura, M. Smart, W. Chun, G. Halpert, J.
Josek, and C. Cropley, “Proton conducting membrane fuel cells I”, The
Electrochemical Society Proceedings Series, PV 95-23, Pennington, NJ.
[12] S. C. Amendola, S. L, Sharp-Gold man, M.S. Janjua, M. T. Kelly, P. J. Petillo, and
M. Binder, J. Power Sources, 85, 186 (2000).
[13] S. C. Kelley, G. A. Deluga, and W. H. Smyrl, “ A Miature Methanol/Air Polymer
electrolyte Fuel Cell”, Electrochem. Solid-State Lett., 3, 407 (2000).
[14] R. F. Savinell, J.S. Wainright, L. Dudik, K. Yee, L. Chen, C. C. Liu, Y. Zhang, and
M. Litt, “Microfabricated Fuel Cells for Portable Power”, The 197 th Meeting of
The Electrochemical Society, Abstract, 63 (2000).
[15] J.D. Morse, A. F. Jankowski, J. P. Hayes, and R.T. Graff, “A Novel Thin Film Solid
Oxide Fuel Cell for Microscale Energy Conversion”, SPIE Conference on
Micromachined Devices and Components, 223-226 (1999).
[16] J.D. Morse, A. F. Jankowski, R. T. Graff, and J.P. Hayes, “Novel proton exchange
membrane thin-film fuel cell for micro scale energy conversion”, J. Vac. Scil.
Technol. A, 18, 4, 2003-2005 (2000).
[17] J. D. Holladay et al. “ Micro fuel processor for use in a miniature power supply”, J.
Power Sources, 108, 22-27 (2002).
[18] K.S. Chang et al., “ CONBUSTOR-INTERATED MICRO-FUEL PROCESSOR
WITH SUSPENDED MEMBRANE STRUCTURE”, Digest of Technical papers,
25

Transducers ’03 (The 12 th International Conference on Solid-State Sonsors,
Actuators and Microsystems), Boston, Massachusetts, USA, 635-638 (2003).
[19] L.R. Arana et al., “CONBUSTION-ASSISTED HYDROGEN PRODUCTION IN
A HIGH TEMPERATURE CHEMICAL REACTOR/HEAT EXCHANGER FOR
PORTABLE FUEL CELL APPLICATIONS”, Digest of Technical Papers,
Transducers ’03 (The 12 th International Conference on Solid-State Sensors,
Actuators and Microsystems), Boston, Massachusetts, USA, 1734-1737 (2003).
[20] J. Y. Lee, U. S. Pat. 5,599,640 (1997).
[21] S. Amendola, U. S. Pat. 5,804, 329 (1998).
[22] R. Jasiski, Electrochem. Technol., 3, 40 (1965).
[23] S. M. Lee, J. H. Kim, H. H. Lee, P. S. Lee, and J. Y. Lee, J. Electrochem. Soc., 149,
A603 (2002).
[24] S. Shoji and M. Esashi, J. Micromech. Microeng., 4, 157 (1994).
[25] Y. Yamazaki, Electrochim. Acta, 50, 663 (2004).
[26] R.G. Hockaday, U.S. Patent No.4673624
[27] R.G. Hockaday, U.S. Patent No.5631099
[28] R.G. Hockaday, U.S. Patent No.5759712
[29] K.-B. Min, S. Tanaka and M. Esashi, Electrochemistry, 70, 924 (2002).
[30] EG&G services, Parsons Inc., Science Application International Corporation, Fuel
Cell Hand book, US Department of Energy, Fifth edition, (2000).
[31] W. Y. Sim, G. Y. Kim, and S. S. Yang, “Fabrication of Micro Power Source using a
Micro Direct Methanol Fuel Cell for the Medical Application”, Proc. IEEE Micro
Electro Mechanical Systems Conference, Interlaken, Switzerland, January 21-25,
341 (2001).
[32] C. Marr and X. Li, J. Power Sources, 77, 17(1999).
[33] K. Scott, W. Taama and J. Cruickshank, J. Power Sources, 65, 159 (1997).
[34] H. Dohle, J. Divisek and R. Jung, J. Power Sources, 86, 469 (2000).
26

Chapter 2
Design and Fabrication of New Concept
Micro Direct Methanol Fuel Cell Utilizing MEMS Technology
27

2.1 Characteristic of the Configuration of the New Concept Micro
Direct Methanol Fuel Cell on Si Wafer
Application of MEMS technology for micro-fuel cell was reported by Kelly et al [1,
2]. The bilayer and monolithic design have compared by using a mathematical model to
quantify the effects of the secondary current distribution by Mayers et al [3]. The
Schematic monolithic design is summarized in Fig. 2.1. The comparison of power
densities for bilayer and monolithic designs, based upon volume of unit cell as depicted
in Fig.2.2. The specific volume power is estimated by 40% of the conventional design
from calculation by Mayer. Jingrong Yu et al. [4] presented the fabrication and
performance evaluation of a miniature twin-fuel-cell on silicon wafers. The miniature
twin-fuel-cell was fabricated in series using two membrane-electrode-assemblies
(MEA) sandwiched between two silicon substrates in which electric current, reactant,
and product flow. H.-Y Cha et al. [5] have developed the novel design and the
fabrication processes for micro-direct methanol fuel cell (µ-DMFC). The MEA consists
of two identical polymer chips positioned on both sides of the proton exchange
membrane. Lee et al [6] presented a design configuration for integrated series
connection of polymer electrolyte fuel cells in a planar array. The series path is oriented
in a “flip-flop” configuration, presenting the unique advantage of a fully continuous
electrolyte requiring absolutely no interconnecting bridges across or around the
membrane. Electrical interconnections are made by thin-film metal layers that coat
etched flow channels patterned on an insulating substrate. Lu et al. [7] have developed a
silicon-based µ-DMFC. Anode and cathode flow-fields with channel and rib width of
750 µm and channel depth of 400 µm were fabricated on Si wafers. A MEA was
specially fabricated to mitigate methanol crossover. This MEA features a modified
anode backing structure in which a compact micro-porous layer is added to create an
additional barrier to methanol transport thereby reducing the rate of methanol crossing
over the polymer membrane. Hahn et al. [8] fabricated self-breathing PEM fuel cells
with a size of 1 × 1 cm and 200 µm of thickness were fabricated. J.S. Wainright et al.
[9] demonstrated high fuel utilization by a fuel cell operating with on-board source of
hydrogen from aqueous sodium borohydride, as well from solid metal hydride systems.
Medicine is also a demanding field for this kind of miniature fuel cells as an
implantable micro-power source for medical devices, such as cerebrospinal fluid shunt
pump and a micro-insulin pump [10].
30

Fig.2.1 Schematic diagram of monolithic design [3].
Fig.2.2 Comparison of power densities for bilayer and monolithic designs, based upon
volume of unit cell [3].
31

Fig.2.3 Images of a single µ-fuel cell fabricated using microfabrication technology.
Electropolished tunnels beneath porous silicon layer prepared by electro chemical etch.
Photo from [3].
Fig.2.3 shows the cross-sectional images of a single µ-fuel cell fabricated using
microfabrication technology proposed by Mayers et al. The monolithic design has an
advantage for stepwise pile on microfabrication due to planar alignment. However, on
their research, the configuration and the fabrication process, such as utilization of
porous silicon as fuel and oxygen diffusion layers, would be complicated, relatively.
Therefore, a further miniaturization of a system with these configurations might be
difficult.
The research cited shows that the adaptation of micro-electronic techniques to
micro-electrochemical systems such as DMFC can be successful. We have to emphasize
that while related work exists, the design and testing of a micro-fabricated silicon-base
miniature methanol/oxygen polymer electrolyte fuel cell remains a novel concept. Since
always, different applications imply different system design characteristics, operation
parameters, as well as materials employed in the device. Towards that aim, we
embarked in an effort to develop a µ-DMFC with good compatibility with wafer level
MEMS process for 1-100 mW class application, such as distributed micro-sensors and
wireless MEMS. This communication will focus on the design, fabrication, and
preliminary performance evaluation of a novel concept of µ-DMFC. The novelty of the
structure proposed in this work is that we have fabricated and arranged the anodic and
cathodic micro-channels in plane fabricated onto a single silicon substrate.
32

Our approach has several advantages and differs in many ways from previous designs:
(a) It is of planar structure and essentially an unfolded fuel cell as shown in Fig.2.5.,
which integrates the anode and cathode onto a single Si surface. Whereas, the bilayer
design uses separate Si wafers with channels for the anode and cathode.
(b) Our design does not require a membrane electrode assembly (MEA), i.e., the
membrane is placed on the patterned electrodes. In the bilayer design, the membrane
has to be placed between the two electrodes and the whole is hot-pressed to make the
MEA.
(c) In our design, the fuel and oxidant are supplied to the cell in isolated, separate
microchannels. Both the fuel and oxidant are distributed in microchannels
throughout the wafer, and they both possess an exhaust. The isolation of fuel and
oxidant precludes one from crossing the fuel and oxidants streams.
(d) The characteristic length of the system that is the distance that the protons must
travel from anode to cathode is adjustable by using silicon microfabrication
technique. The shorter distance might be making the system less sensitive to ohmic
impedance effect.
(e) The efficiency of the current collectors is high, because the catalyst layers are
supported on the metal directly. In addition, the current collectors are directly
deposited in the microchannels. The current does not need to be pulled out by
relatively large metal lines.
(f) Catalyst electrodes are directly fabricated in the bottom and sidewalls of the
microchannels. Therefore, this design offers simplicity in the stepwise integration of
substrate, electrodes, and membranes.
H + H
CO2 + H + H
CO2 +
Fig.2.5 Schematic cross-sectional image of a new concept planar µ-DMFC.
H +
PEM H +
PEM
Anode Cathode
CH CH3OH 3OH 3OH
e- ee-
CH 3 OH/H 2 SO 4
solution
Catalyst
layer
Current collector
34
O 2
H + H + H + H +
H H2O 2O
e- ee-
O O O2sat./H 2sat./H 2sat./H 2sat./H 2sat./H 2sat./H 2sat./H 2sat./H2SO 2SO 2SO 2SO 2SO 2SO 2SO 2SO 4
solution

2.1.2 Cell materials
Fuel cells are basically composed of electrolyte, electrodes, diffusion layer, current
collector, and separator. These components required for materials, corrosion resistance,
thermo-stability, high electric conductivity.
Silicon wafer was used as base substrate in order to use MEMS fabrication
technology consists of micro-fabrication technology. It is not necessary electric
conductivity for a silicon wafer so that is not used as a current collector. However, we
have used ion doped silicon wafer because that is available. The 200 µm of thickness
was chosen to reduce the etching time and to obtain the strength when forming the
through holes.
It is necessary high electric conductive as current collector in order to reduce ohmic
resistance. The main problem of planar type fuel cells is the ohmic resistance due to
increasing the length as against the cross section on interconnection. The electrode and
the interconnection were fabricated by sputtering a layer of copper over a layer of gold
on the top of the silicon wafer. Lu et al [10] was used Cr/Cu/Au (with thickness of
0.01/3/0.5 µm) deposited on the front-side of each wafer by electron beam evaporation
in order to minimize contact resistance between the MEA and the Si wafer. A lager
thickness of the metal film leads to reduce the ohmic resistance. In our case, to prevent a
leak of solution caused by the different thickness at the deposited and non-deposited
metal film on silicon oxide, thinner metal film should be used, respectively. Au was
chosen as current collector, however silicon oxide is poor adhesive as against Au,
therefore, and titanium was used as adhesive layer.
For electrode catalyst, it is necessary high activity for methanol oxidation as anodic
catalyst and for oxygen reduction as cathodic catalyst. Among several possibilities, a
Pt-Ru alloy has shown the most promising performance for the hydrogen oxidation
reaction in the presence of CO [16, 17], and a Pt is used as cathodic catalyst continually.
Pt as cathodic catalyst and Pt-Ru as anodic catalyst was used.
For polymer electrolyte membrane (PEM), it is necessary a high ionic conductivity,
high chemical stability, and shape stability. Owing to their high chemical stability,
perfluorinated sulphonated cation exchangers are generally used as the membrane
electrolyte at present. DuPont TM Nafion112 (thickness: 50 µm, equivalent weight 1100
g/ml, ionic conductivity 0.083 Ω -1 cm -1 ) has been used as the polymer electrolyte
membrane of the µ-DMFC.
36

2.1.3 Mass flow control of fuel and oxidant
The fuel is fed into the anode flow field, moves through the diffusion medium, and
reacts electrochemically at the anode catalyst layer. The diffusion medium is typically a
carbon cloth of carbon paper. The oxidant is fed into the cathode flow field, moves
through the diffusion medium, and is reduced at the cathode. The carbon dioxide and
the water produced at the anode and the cathode exits the fuel cell through either the
cathode or the anode flow field. These movements must be accomplished without
hindering the transport of reactants to the catalyst layers in order for the fuel cell to
operate efficiently. An Active type DMFC has disadvantage for configuration of the
system. The specification of the system has to be discussed in response to occupancy.
Passive type allows simplicity on the system.
Fig.2.7 shows the schematic of experimental set-up for measurement. Potentiostat/
Galvanostat HZ-3000 1512µ (Hokuto Denko) was used for measurement of
electrochemical properties. To evaluate the basic properties reproducibility, the system
with pump was used. For mass flow control, fuel and oxidant were supplied in a
push-pull manner using micro syringe pump ESP-64 (EICOM) with samples collected
into 2.5 ml manual simpler injector with separately drivable two channels.
Mass flow control
Micro syringe pump
IN OUT
µ-DMFC test cell
Fig.2.7 Schematic of experimental set-up for measurement.
37
Potentiostat / Galvanostat
Electrochemical properties

2.1.4 Relative advantage and disadvantage of different configuration of micro
direct methanol fuel cell
A primary advantage of the planar structure is the ability to form all of components
on the same structure, analogous to IC or MEMS manufacturing [15].
So-called active DMFC required assistance equipment, such as pump and adjust
system of concentration methanol solution as fuel. However, it is required that to
scavenge generated co-products, such as carbon dioxides in cell. Micro-syringe pump
and actuator have been investigated. In future work, a mass flow controller would be
combined our fuel cell systems on a chip.
38

2.2.2 Consideration of fabrication process of the test cell
Prototype test cells were fabricated by using well-known micromachining
technology, such as photolithography, deep reactive ion etching.
2.3 Fabrication Process
2.3.1 Formation of fluidic channel and feedhole for mass flow by using lithography
and chemical etching
The µ-DMFC was prepared using a series of fabrication steps tailored from MEMS
techniques. This procedure is shown in schematic from in Fig.2.9. Beginning with 20
mm x 25 mm oriented silicon (p-type, 1-10 Ω cm, 200 ± 20 µm thick) polished
on both front and back sides, a 500 nm layer of silicon dioxide was grown thermally
(wet oxidation at 1100 o C for 1 h and 30 min). To make feedholes and channels, the
silicon dioxide on the front and backside of the wafer were patterned by
photolithography. Windows were then opened using buffered hydrogen fluorides etch.
To complete the fabrication process, the channels and feedholes were then etched using
Deep Reactive Ion Etching (D-RIE) process. D-RIE was performed with a STS
Multiplex ICP Deep Reactive Ion Etcher, with an inductively coupled plasma at 13.56
MHz, coil rf power 600 W, and electrode rf power 15 W. The etchant gas was SF6 (130
sccm, 22 mT) and intermittent passivation with C4F8 (85 sccm, 18 mT) enabled vertical
sidewall profiles. Etch rate was typically 2.5-3.0 µm/min. The feedholes are circulars, 1
mm diameter, and with a hole spacing of 8 mm and 10 mm horizontally and vertically,
respectively. The width and the depth of the microchannel are both equal of 100 µm,
while the clearance between channels is 100 µm. Subsequently, a 50 nm layer of silicon
dioxide was grown thermally (dry oxidation at 1100 o C for 12 min) on the entire surface
of the silicon wafer again. This step was to make sure that silicon dioxide layer
produced electrical insulation between silicon and the metal so that short circuit
between electrodes would be prevented enough. Finally a photosensitive dry film for
printed wiring boards was applied to prepare for selective deposition of metal film.
Ti/Au for current collectors was formed by electron-beam deposition and lift-off method.
A 100 nm gold layer was deposited in an ULVAC CRTM-6000 electron beam
evaporation chamber, preceded by 20 nm titanium layer just beneath the gold to
promote adhesion.
40

(a) Thermal oxidation
(b) Lithography ( double side alignment )
(c) Dioxide etching
(d)
Si etching using D-RIE ( channels )
( e) Si etching using D-RIE ( through holes )
(f) Thermal oxidation
SiO 2
Si
Resist
Fig.2.9 Fabrication process of prototype µ-DMFC.
41
(g) Lithography (Dry film)
(h) Ti / Au deposition
(i) Lift off
Pt Pt-Ru
(j) Pt, Pt-Ru electroplating
(k) Assembly
Ti/Au
Dry
film
Ti/Au
Glass
PEM

Etch profile considerations for 3-D electrode channels
On our proposed µ-DMFC, the 3-D electrode channels are the most important for
the performance. The particular profile was determined by etch method. An isotropic
plasma etch (Deep-RIE) provided vertical walls, while anisotropic wet etching in
potassium hydroxide resulted in 54.7 o sloped walls according to silicon crystal planes.
Softened edges ware created by 30 sec immersion of samples in a caustic solution
selected for preferential etching of sharp corners (1:2:1 mixture of 50 % hydrofluoric
acid, nitric acid and 69.5 % acetic acid). Factorial experimentation has been applied to
investigate the adequacy of metal film conduction over etched topology, and results
conclude that film thickness dominates over other design parameters [16]. Fig.2.12
shows cross-sectional SEM images of the channel with different channel profiles.
Fig.2.12. SEM images of the channel with different etch profiles:
a) vertical wall by Deep-RIE, b) smooth edge by Deep-RIE, acid etching, c) sloping
wall by alkali etching. Channel dimensions: 100 µm of width and 50 µm of depth.
44

For 3-D electrode patterning, 3-D photolithography is the most important process.
Poor step-coverage by spin coating for these 3-D structures was a technical issue. The
optimization of distance between electrodes would be drew one’s interest on proposed
µ-DMFC configuration. Thereby, we have prepared different type chromium masks,
which have 24, 50, and 100 µm of distance between electrodes, respectively. On a
prototype test cell, a dry film put on a substrate with channels at ordinary pressure. The
patterns were developed in 8 % sodium bicarbonate. A dry film technique that is capable
of 100 µm 3-D electrode patterning. However, it was not enough to make an electrode
patterning of very precise (i.e. below 50 µm) because of poor adhesion between a dry
film and a silicon substrate. The issues when resist coat is summarized in Fig.2.13.
Several groups study about spray coating for slope wall were reported [20,21]. The
Electronic Vision EV101 spray coater [22] and AZP4620 positive photoresist diluted
with acetone were used. The wafer was heating during spray-coating steps in order to
avoid the spread or aggregation of droplets on the wafer. The resist coated wafer was
then exposed with the electrode photo mask for 60 sec. The patterns were developed in
an AZ-400K developer solution (diluted with water in 1:4 ratio). Fig.2.14 shows
cross-sectional SEM image of resist coated by spray coater on side wall in the channel
formed by KOH anisotropic etching. Resist thickness is 2 µm (convex corner) and 14
µm (concave corner). Consequently, as can be seen in Fig.2.15, the Au/Ti patterning
substrate was obtained.
45

Resist coating
Development
C: 100 µm
C: 50 µm
Dry film laminating
c)
Vertical wall
C: 100 µm
C: 50 µm
C: 24 µm
Sloping wall
Spray coating
Fig.2.13 Schematic cross sectional images of the channel with vertical and sloping walls
at the resist coating and the developing.
a) poor coverage on edge and side wall, b) residue on the bottom, c) dissolution or
abrasion.
46
a)
a)
b)
b)
c)

Fig.2.14 Cross-sectional SEM image of resist coated by spray coater on side wall in the
channel formed by KOH anisotropic etching.
Au
SiO 2
Fig.2.15. SEM image of plane view of the Au/Ti electrodes formed on lateral wall in the
channels and on the flat parts.
47
Au

Concern of etch profile
In a prototype test cell, to make channels, Deep-RIE was used. Etch rate was
typically 2.5-3.0 µm/min. The etch profiles of the holes and channels is vertically. For
most micromachining and active circuit processing, (100) orientation silicon is used, for
which hydroxide etchant produce pyramidal pits with 54.7 o (111) sidewall angles
relative ton the (100) surface. Of all the anisotropic etchant, aqueous potassium
hydroxide (KOH) and tetramethylammonium hydroxide (TMAH) solutions are the most
commonly used. Wet etching is a blanket name that covers the removal of material by
immersing the wafer in a liquid bath of the chemical etchant. Etch rate was typically 1.0
µm/min.
48

References
[1] C. S. Kelly, G. A. Deluga, and W. H. Smyrl, Electrochem. and Solid-State Lett., 3,
407 (2000)
[2] C. S. Kelly, G. A. Deluga, and W. H. Smyrl, AIChE J., 48, 1071 (2002)
[3] J. P. Meyers, H. L. Maynard, J. Power Sources, 109, 76 (2002)
[4] J. Yu, P. Cheng a, Z. Maa, B. Yi, Electrochim. Acta, 48 1537 (2003)
[5] H.-Y Cha, H.-G Choi, J.-D Nam, Y. Lee, S. M. Cho, E.-S Lee, J.-K Lee, C.-H
Chung, Electrochim. Acta, 50 795 (2004)
[6] S. J. Lee, A. Chang-Chien, S. W. Cha, R. O’Hayre, Y. I. Park, Y. Saito, F. B. Prinz, J.
Power Sources, 112, 410 (2002)
[7] G.Q. Lu, C.Y. Wang, T.J. Yen, X. Zhang, Electrochim. Acta, 49, 821 (2004)
[8] R. Hahn, S. Wagner, A. Schmitz, H. Reichl, J. Power Sources, 131, 73 (2004)
[9] J.S. Wainright, R.F. Savinell, C.C. Liu, M. Litt, Electrochim. Acta, 48, 2869 (2003)
[10] G. Q. Lu, C. Y. Wang, T. J. Yen, X. Zhang, Electrochim. Acta, 49, 821 (2004)
[11] J. Yu, P. Cheng, Z. Ma, B. Yi, J. Power Sources, 124, 40 (2003)
[12] J. Yu, P. Cheng, Z. Ma, B. Yi, Electrochim. Acta, 48, 1537 (2003)
[13] K-B. Min, S. Tanaka, and M. Esashi, Proc. IEEE Sixteenth Annual International
Conference on MEMS, 379 (2003)
[14] J. S. Wainright, R. F. Savinell, C. C. Liu, M. Litt, Electrochim. Acta, 48, 2869
(2003)
[15] R. Dillon, S. Srinivasan, A. S. Arico, V. Antonucci, J. Power Sources (2004) in
press.
[16] A. Kabbabi, R. Faure, R. Durand, B. Beden, F. Hahn, J.-M. Leger, C. Lamy, J.
Electroanal. Chem. 444, 41 (1998)
[17] H.A. Gasteger, N.M. Markovic, P.N. Ross, Jr., J. Phys. Chem., 99, 8290 (1995)
[18] C. Catteneo, M.I. Sanchez de Pinto, H. Mishima, B.A. Lopez de Mishima, and D.
Lescano, J. Electroanal. Chem. 461, 32 (1999)
[19] C. Coutanceau, A.F. Rakotondrainibé, A. Lima, E. Garnier, S. Pronier, J-M. Léger
and C. Lamy, J. Appl. Electrochem., 34, 61 (2004)
[20] V.K. Slingh, M. Sasaki, J.H. Song, K. Hane, J. Appl. Phys. Part 1, No. 6B, 42,
4027 (2003)
[21] N.P. Pham, E. Boellaard, J.N. Burghartz, J. of Micro electrochemical Systems,
3,13(2004)
[22] B. Wieder, C. Brubaker, T.Glinsner, P.Kettner, N. Nodes, Pacific Rim Workshop
on Transducers an Micro/Nano Technologies, (2002)
49

Chapter 3
Electrochemical Evaluation of New Concept
Micro Direct Methanol Fuel Cell
51

In this chapter, the Cell performance of the µ-DMFC with new concept
configuration was discussed. Test results were able to confirm that this new concept of
µ-DMFC generates electricity. The performance of the cell was measured at ambient
temperature was of mW/cm 2 class. The laboratory aims further to improve the
performance of this µ-DMFC and establish the assembling technique for practical
applications.
5353

3.1 Performance of the Test Cell
In this section, first, the cell performance for the prototype test cell was
demonstrated. As a result, we were able to confirm the operation of the µ-DMFC as
power source. To increase with the cell performance, the fabrication process of the
electrodes was discussed about by using both wet and dry process, respectively.
Furthermore, the characteristics of the cell performance were discussed based on
electrochemical concern.
3.1.1 Experimental
The patterned silicon substrates (200 µm thick), above-described in chapter 2, were
used. The membrane is pretreated as follows. It is boiled in 1M H2O2 solution for 1 hour
and then rinsed in boiling deionized water for 1 hour to remove any organic compounds.
Subsequently, it is boiled in 1 M H2SO4 for 1 hour to remove metal compound and to
replace Na + with H + in the membrane. Finally, the µ-DMFC was assembled by placing
the pretreated Nafion112 membrane (50 µm thick) between the patterned silicon and a
glass substrate and the whole was clamped mechanically for testing. Fig.3.1 shows the
photograph of the assembled µ-DMFC showing the clamping device and electrical
connections; a) prototype which made of acrylic resin material. It was insufficient to
clamp a µ-DMFC about the strength and the reproducibility at the measurement.
Whereby, we have fabricated second type which made of stainless material with a slight
improvement on the prototype. as shown in Fig.3.1b). Fuel and oxidant were supplied in
a push-pull manner using micro-syringe pump with samples collected into 2.5 ml
manual simple injector with separately drivable two channels.
Owing to the deleterious effect of methanol crossover, dilute methanol solution (i.e. 1
M) was typically used in DMFC. However, under these conditions there will be
insufficient fuel to create power. Thus it would be desirable to operate with methanol
concentration just above the minimum necessary to provide methanol to the anode.
Mayers and Maynard [1] propose an optimal operating concentration of about 2 M
methanol.
The effective surface area of the platinum as cathode catalyst was calculated from
cyclic voltammograms in 0.5 M H2SO4 at 25 o C, Undertaken in the range of -0.2-1.2V
(vs. Ag/AgCl) using a triangular potential sweep at a scan rate of 50 mV s -1 . The CV of
methanol oxidation of the platinum–ruthenium catalyst was in 0.5 M H2SO4 at 25 o C,
the 2nd. curve was plotted in Figures.
54

Potentiostat
/Galvanostat
µ-DMFC
Micro-syringe
a) b)
Fig.3.1 Photographs of experimental set-up. a) prototype which made of acrylic resin
with four bolts, b) second type which made of stainless materials with six bolts.
55

3.1.2 Performance of the test cell
Fig.3.2 shows cell polarization curves operated using 2 M CH3OH/0.5 M H2SO4
solution as fuel and O2 saturated/0.5 M H2SO4 solution as oxidant at room temperature
and atmospheric pressure. The maximum power density was0.44 mW/cm 2 at 3.6
mA/cm 2 for cell. The open circuit voltage was apparent 300 mV. As can be seen from
Fig.3.2, the polarization curves exhibit kinetic and ohmic control. The operation of
µ-DMFC has been confirmed.
Cell voltage / mV
400
300
200
100
0
0
Current density / mA cm-2 1 2 3 4
Current density / mA cm-2 1 2 3 4
0.5
0.4
0.3
0.2
0.1
Fig.3.2 Cell voltage and power density vs. current density plots for prototype µ-DMFC
in 2 M CH3OH/0.5 M H2SO4 as fuel and O2 saturated/0.5 M H2SO4 as oxidant with a 10
µl/min of the flow rate at the anode and cathode, respectively. Channel dimensions; 100
µm of width, 70 µm of depth, and 100 µm of clearance between channels with vertical
walls. The electrode area is 0.014cm 2 . The condition of electro-plating is depicted in
section 2.3.2. Pt was used as anode and cathode catalyst, respectively.
56
0
Power density / mW cm-2 Power density / mW cm-2

Fig.3.3 shows cell polarization curves operated at Pt and Pt-Ru as anode catalyst
using 2 M methanol solution at ambient temperature and atmospheric pressure. A Pt was
used as cathode catalyst, respectively. As can be seen in Fig.3.3, The OCV for a cell
with Pt as anode catalyst was 300 mV while for a cell with Pt-Ru as anode catalyst of
400 mV. The maximum power density is 0.44 mW/cm 2 at 3 mA/cm 2 for a cell with Pt
anode catalyst. While, the maximum power density reached 0.78 mW/cm 2 at 3.6
mA/cm 2 for a cell with Pt-Ru anode catalyst. This is because the kinetics for methanol
oxidation is enhanced at Pt-Ru electrode.
When it is compared with other macro DMFC unit cell, the output voltage of the
fabricated µ-DMFC unit cell is low. The reasons for the low output voltage may be
described as follows:
(1) The flow rate of fuel and oxidant solutions was adapted 10 µl/min, respectively.
The optimization of the flow rates might be improved the cell performance.
(2) The composition of the anode catalyst used in our cell is not optimal. The
composition used in most of the technical works is a platinum/ruthenium of about
50/50 atomic ratio. It is still necessary to optimize the catalyst composition so as
to increase the catalyst utilization.
57

Cell voltage / mV
Power density / mW cm-2 Power density / mW cm-2 Power density / mW cm-2 500
400
300
200
100
0
1.0
0.8
0.6
0.4
0.2
0
0
0
Pt-Ru
Pt
Current density / mA cm-2 2 4 6
Current density / mA cm-2 2 4 6
Current density / mA cm-2 2 4 6
Pt-Ru
Pt
Current density / mA cm-2 2 4
Current density / mA cm-2 2 4
Current density / mA cm-2 2 4
Fig.3.3 Cell voltage and power density vs. current density plots for prototype µ-DMFC
in 2 M CH3OH/0.5 M H2SO4 as fuel and O2 saturated/0.5 M H2SO4 as oxidant with a 10
µl/min of the flow rate at the anode and cathode, respectively. Channel dimensions; 100
µm of width, 100 µm of depth, and 100 µm of clearance between channels with vertical
walls. The electrode area is 0.018cm 2 . The condition of electro-plating is depicted in
section 2.3.2. Pt was used as cathode, and Pt-Ru was used as anode catalyst.
58
6

3.1.3 Effect of flow velocity on the performance
Fig.3.4 shows a polarization curves on test cell 2 using 2 M CH3OH/0.5 M H2SO4
solution at room temperature and atmospheric pressure. The oxygen solution was 30-80
µl/min. These showed task dependent the performance increases as a velocity of oxidant
solution was increased. As expected, an increase in the performance is observed with
increasing fuel solution velocity. This measurement showed cell performance increases
as flow rate of oxidant solution was increased. This result may be due to low dissolved
oxygen in the oxidant solution. The OCV was 530 mV. The maximum power density
was 0.84 mW/cm 2 at 3.7 mA/cm 2 , at 15 µl/min of fuel flow rate and 50 µl/min of
oxidant flow rate.
59

Cell voltage / V
0.6
0.5
0.4
0.3
0.2
0.1
0
0 1 2 3 4 5 6 7
Current density / mA cm -2
Current density / mA cm -2
I-V,30
I-V,40
I-V,50
I-V,80
I-P,30
I-P,40
I-P,50
I-P,80
I-V f:15, o:30
I-V f:15, o:40
I-V f:15, o:50
I-V f:15, o:80
I-P f:15, o:30
I-P f:15, o:40
I-P f:15, o:50
I-P f:15, o:80
Fig.3.4 Cell voltage and power density vs. current density plots for µ-DMFC in 2 M
CH3OH/0.5 M H2SO4 with a flow rate of 10 µl/min. The O2 saturated/0.5 M H2SO4 was
supplied with a flow rate of 30, 40, 50, and 80 µl/min. Channel dimensions; 100 µm of
width, 30 µm of depth, and 50 µm of clearance between channels with sloped walls.
The electrode area is 0.0079 cm 2 . The condition of electro-plating is depicted in Fig.3.8
(d)
60
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
Power density/ mW cm -2
Power density/ mW cm -2

3.1.4 Effect of concentration of methanol on the performance
At using high concentration methanol, the crossover causes not only a gradual drop
but a deterioration of cathode activity [2, 3]. Lu et al [4] has developed a silicon-based
micro direct methanol fuel cell (µ−DMFC). Anode and cathode flow fields with channel
and rib width of 750 µm and channel depth of 400 µm were fabricated on Si wafers. A
membrane-electrode assembly (MEA) was specially fabricated to mitigate methanol
crossover. This MEA features a modified anode backing structure in which a compact
micro porous layer is added to create an additional barrier to methanol transport thereby
reducing the rate of methanol crossing over the polymer membrane. The cell with the
active area of 1.625 cm 2 was assembled by sandwiching the MEA between two
micro-fabricated Si wafers. Extensive cell polarization testing demonstrated a maximum
power density of 50 mW/cm 2 using 2 M methanol feed at 60 o C. When the cell was
operated at room temperature, the maximum power density was shown to be about 16
mW/cm 2 with both 2 and 4 M methanol feed. It was further found that the present
µ−DMFC still produced reasonable performance under 8 M methanol solution at room
temperature.
61

3.1.5 Effect of temperature
Fig.3.5 shows current density versus cell voltage of test cell 3 with different
temperature, i.e, 30, 40, and 60 o C. The apparent open circuit voltage was increased as
the temperature increased. However, the voltage rapidly decreased at high current
density region at the higher temperature. This is because the mass of dissolved oxygen
in oxidant solution reduced at the temperature.
62

3.1.6 Durability of µ-DMFC
Fig.3.6 shows the long life stability of test cell when it was holding 0.62 mA/cm 2 .
Although, the cell voltage slightly decreased after 60 min, the long life stability have
been demonstrated.
Cell voltage/ V
0.6
0.4
0.2
0
0 30
Time/ min
60
Fig.3.6 Long life stability of µ-DMFC at 30 o C, at 0.62 mA/cm 2 of current density. Cell
voltage vs. time plots for µ-DMFC in 2 M CH3OH/0.5 M H2SO4 as fuel, O2 sat./0.5 M
H2SO4 as oxidant, 5 µl/min of a fuel flow rate, 15 µl/min of an oxidant flow rate.
Channel dimensions; 100 µm of width, 30 µm of depth, and 50 µm of clearance
between channels with sloped walls. The electrode area is 0.0079 cm 2 . The condition of
electro-plating is depicted in Fig.3.8. (a).
3.1.7 Conclusion
Test results were able to confirm that this new concept of µ-DMFC generates
electricity. The cell performance of the cell was demonstrated at ambient temperature
was of mW/cm 2 class. The laboratory aims further to improve the performance of this
µ-DMFC and establish the assembling technique for practical applications.
64

Current
t on t off
a) (A) 12C
b) (B) 12C
c) (A) 0.1C+(B) 11.9C
(A) (B)
i p
Time
i DC
t on= 100ms, t off= 300ms, i p=30mA cm -2
i DC=30mA cm -2
(A) Pulse electrodepositon
(B) DC electrodepositon
t on= 100ms, t off= 300ms, i p=50mA cm -2 , i DC=30mA cm -2
d) [(A) 0.1C +(B) 0.9C ] 12cycle t on = 100ms, t off = 500ms, i p =50mA cm -2 ,i DC =30mA cm -2
Fig.3.7 Current diagrams of electroplating of catalysts. a) pulse electroplating, b) direct
current electroplating, c) combination of pulse and direct current electroplating, d)
combination of pulse and direct current electroplating. Additionally, repeat that at 12
times.
The process of uniform coating by electroplating consisted of three steps. The first
step was to nucleation generation uniformly by using pulse electroplating, the second
was to grow particle using direct current (DC), and third was repetition in order to
obtain given loading as shown in Fig.3.7
The conditions of electroplating were: in the first step, current density =50 mA/cm 2 ;
ton = 100 ms, toff = 500 ms; charge density =0.1 C/cm 2 , in the second step, current
density =30 mA/cm 2 , charge density =0.9 C/cm 2 . in the third step, total charge density
was 12 C/cm 2 . The electroplating process was carried out at 25 o C for both electrodes.
66

3.2.2 Effect of condition of electroplating on morphology and catalysis of Pt and
Pt-Ru
First of all, The Ti/Au was torn off on the Si substrate when the Pt and PtRu catalyst
layers deposited onto the Ti/Au by condition (a) (as depicted in Fig.3.7). It is considered
that the effect of internal stress in Pt and PtRu layers appeared sensitivity.
Fig.3.8 shows that the electroplated catalyst layers on the Ti/Au in the channels. It is
not uniform that the catalyst layer from condition (b), i.e., only from direct current
electroplating produced a uniform coating of the surface exposed, sidewalls, and bottom
of the microchannels electrodes.
It is known that pulse electroplating, as compared with direct electroplating, has
many advantages in terms of controlled particle size, stronger adhesion, uniform
electroplating, selectivity of hydrogen, reduction of internal stress, etc.
67

(i)
(ii)
(iii)
Cathode
Cathode Anode
Wall
Bottom
Anode
Cathode Anode
Fig.3.8 SEM images of deposited catalyst layer in lateral wall of the channel. The
condition of electroplating is (i) condition (b), (ii) condition (c), (iii) condition (d), as
depicted in Fig.3.7.
68

3.2.3 Conclusion
In this section, we have studied conditions of fabrication of Pt and Pt-Ru catalyst
layer. The combination of direct current and pulse electroplating was attempted. As a
results, uniform deposition on Ti/Au layer in the channel was achieved.
69

3.3. Comparison of Ni/Au and Ti/Au on SiO2 as Current Collector
A gold as current collector is an important component to keep a high efficiency of
collecting electrons. Dry and wet process have been discussed conventionally for thin
metal layer depositing
3.3.1 Experimental
The specimens were cleaned in SPM (80 mixture of 4 parts of concentrated H2SO4
and 1 part of 30% H2O2) and then in 1%-HF aqueous solution to remove the oxide layer
formed by SPM cleaning. After cleaning, these wafers were rinsed in ultrapure water
with the resistivity of 18.2 MΩ cm.
For both electroplating an electroless deposition, the simple bath C that consists of
only NiSO4 and (NH4)2SO4 listed in Table 2.4 was employed.The bath was adjusted at
pH 9.0 with NH4OH. The bath temperature of electroless deposition was adjusted at 80
o C.
70

3.3.2 Effect of thin metal layer on properties of current collector
B
B’
B B’
Lead layer
Si
Sidewall of channel
Channel
Top of channel
EB depositon (Ti/Au)
Not fine
Electroless deposition (Ni/Au)
After Au deposition
Ununiform Uniform
After Pt electro-plating
Fig. 3.9 Photographs of deposited metal on lateral wall in channel.
Fine
Fig.3.9 shows the comparison of Ti/Au layer by electron beam depositon methode
and Ni/Au layer by combination of electroless plated on silicon oxide in the channel.
Apparently, Ni/Au layer is unifom deposited on lateral wall in channel compared to
Ti/Au layer. Additionally, after Pt electroplating on each thin metals, it is clear that
Ni/Au layer was also superior as a seed layer. However, Ni/Au layer was tore of the
silicon oxide, when Pt loading increased to enhance the cell performance, due to its poor
adhesive on silicon oxide.
3.3.3 Conclusion
By utilizing such deposition behavior, was poor adhesive towards Pt electrodepositon
71

a)
Outlet
DHE
µ-DMFC µ-DMFC
b)
C.E
C.E
R.E
Inlet
Potentiostat /
Galvanostat
R
Inlet
- +
W.E
Outlet
DHE
W.E
Pt
Glass
Nafion
Fig.3.10 Schematic of experimental set-up for electrochemical measurement by using
Dynamic Hydrogen Electrode (DHE).
73
H 2

Fig.3.11 shows cyclic voltamgrams using Pt working electrode in 0.5 M H2SO4. A
typical shape of the voltammograms was obtained. The peaks corresponding to the
adsorption and desorption of hydrogen, formation and reduction of oxide are clearly
distinguished. However, the peak in anodic current was broad, which indicates that the
organic species adosorped, such as co-products of methanol permeated from anode
channel. Things to be emphasized in the voltammograms are that the shape is stable
even after 10 cycles. The results obtained in this evaluation that the dynamic hydrogen
electrode can actually be used in real electrochemical analyzing.
Current density/ mA cm-2 Current density/ mA cm-2 1
0
-1
-2
2-10 cycles
-0.1 0.3 0.7 1.1 1.5
Potential/ V vs. DHE
Fig.3.11 Cyclic voltammograms in 0.5 M H2SO4 at 25 o C obtained using Pt working
microchannel electrode (cathode side) on the µ-DMFC. A dynamic hydrogen electrode
and a Pt counter electrode (anode side) was used. Scan rate: 50 mV/s
74

Fig.3.12 shows the polarization curves of the single electrode for both of anode and
cathode electrodes, respectively. At the measurement, a 0.5 M H2SO4 solution was
supplied into the opposite micro channel electrodes. As expected, the current of oxygen
reduction was low due to its few dissolution of oxygen in the oxidant solution. As can
be seen in anode polarization curves, it is clear that the overpotential is high for
methanol oxidation. This is because that Pt was used as anode catalyst at this
measurement. Although the anode performance might be enhance by using more active
catalyst such as PtRu, at present results, it is considered that the cell performance is
dependent on the cathode characteristics.
Potential/ V vs. DHE
1
0.8
0.6
0.4
0.2
0
0
Cathode
Anode
Current density/ mA cm-2 5 10
Current density/ mA cm
15 20
-2
5 10 15 20
Fig.3.12 I-V characteristics of test cell in 2M CH3OH / 0.5M H2SO4 with a flow rate of
20 µl/min. The O2 sat./0.5M H2SO4 with a flow rate of 20 µl/min. a DHE as reference
electrode was used.
75

3.4.2 Effect of scale parameters of channels
It is shown in Fig. 3.13 that the oxygen electrode experiences a shift of at least 100
mV when operated in a µ-DMFC. This effect is due to the methanol crossover across
the Nafion TM membrane. It is therefore evident that the design of a reference electrode
was needed to find out where to locate the losses of the DMFC.
Clearly, these data showed dependent increases the value of current as the oxidant
flow rate was increased. The effect of increase of oxygen supplied to the cathode
electrode appears clearly.
Current/ mA
0
-0.08
-0.16
0 0.2
Blank
(a)
(b)
(c) (d)
0.4
Potential/ V vs. DHE
Blank5
10
20
40
80
0.6 0.8 1
Fig.3.13 I-V characteristics of µ-DMFC in 2M CH3OH / 0.5M H2SO4 with a flow rate
of 10 ml/min. the oxidant flow rate of; (a) 10, (b) 20, (c) 40, (d) 80 ml/min.
76

3.4.3 Conclusion
Aspects of the design, materials and fabrication of a microfabricated methanol fuel
cell have been presented. Our concept of a novel structure lies in that the anodic and
cathodic micro-channels arranged in one plane, unlike the conventional bipolar design.
The first objective of the experimental trials was simply to verify the feasibility of this
novel structure on basis of MEMS technology. Thus a µ-DMFC on a silicon wafer has
been successfully fabricated using photolithography, deep reactive ion etching, and
electron beam deposition. In addition, quasi-reference electrodes could be built in the
prototype cell. Test results were able to confirm that this new concept of µ-DMFC
generates electricity. The performance of the cell measured at ambient temperature was
of mW/cm 2 class.
A reliable DHE reference electrode was introduced into the planar µ-DMFC, and
was used for the electrode characterization. Improvements should be made in the fields
of channel size and the electrode optimization.
77

References
[1] J.P. Meyers, H.L. Maynard, J. Power Sources, 109, 76 (2002)
[2] R. Jiang and D.Chu, J. Electrochem. Soc., 151(1), A69 (2001)
[3] X.Ren, S. Gottesfeld, J. Electrochem. Soc., 148(1), A87 (2001)
[4] G.Q. Lu, C.Y. Wang, T.J. Yen, X. Zhang, Electrochim. Acta, 49, 821 (2004)
[5] A. Küver, I. Vogel, W. Vielstich, J. Power Sources, 52, 77(1994)
[6] K. Fufukawa, K. Okajima, M. Sudoh. J. Power Sources, (2004) in press
[7] Kyoung Hwan Choi, Han Sung Kim, and Tae Hee Lee, J. Power Sources, 75, 230
(1998)
78

Chapter 4
Numerical Model for New Concept Micro DMFC
7979

In this chapter, a study has conducted on the effects of channel parameters on
transport phenomena of produced proton from anodic channel to cathodic channel and
on the performance of a proton exchange membrane fuel cell. By applying three
dimensional flow analysis models, Computational Fluid Dynamics (CFD), the effects of
clearance between channels, etc..
4.1 Theories and Model
In this section, we made the assumption that a Nafion membrane is porous materials
with numberless clusters which have the role of proton conductive. On that note,
actually, the phase of proton transport is H2O. The assumption of pseudo- porous
material is neglected the pressure. Nafion 112 (50 µm), 115 (125 µm), 117 (175 µm)
was used polymer electrolyte membrane (PEM). The membrane was assumed to pseudo
porous materials. A Nafion membrane has an ion -cluster region, which size is about 4
nm, containing sulphonate groups. The cluster size was adapted as the grain size of
pseudo porous materials.
The starting point for the work presented here is the commercially available
Coventor ware 2003 (Microfluidics Analysis, React Sim). Lee et al [1] to determine the
electromechanical coupled field effect of the PZT actuation, Coventor simulator was
utilized. Numerous approximate one-dimensional and two-dimensional models of
varying degrees of complexity and simplification have appeared in the literature.
Asuccinct review of much of the early work was prepared by Kleinstreuer and Belfort.
[2] The present work is directed towards removing some of the assumptions discussed
above and to validate a general purpose finite volume model for use in the simulation of
membrane systems. As a first step, we performed simulations in two dimensions to
enable validation of the modeling framework but note that this approach is easily
extended to certain three-dimensional flows.
4.1.1 Assumption and basic formula
A 3-D model, shown schematically in Fig.4.1, was set up to represent the model
consists of a membrane sandwiched between 3 face electrodes channels (bipolar type)
and a membrane put on alternative 3 face electrode channels (planar type). The channel
has a vertical wall, respectively. The physical dimensions of the geometry ware varied
to match the conditions in the cited reference.
The computational results showed that fuel consumption can be increased by
decreasing the permeability of the flow distributor, especially when the permeability
decreases to a value below that of the anode. In terms of the reactant, gas utilization
8181

concentrations of the reagents, raised to the power of their stoichiometric coefficients,
and follows equations that look like this:
Table 4.1 Parameters values used base case
D
D
D
D
D
D
D
D
l
d
w
Diffusion coefficiency
Proton in H 2 O(T:300K)
Diffusion coefficiency
Proton in Nafion(T:300K)
Water volume fraction:0.5
Diffusion coefficiency
H 2 O in H 2 O (T:300K)
Diffusion coefficiency
H 2 O in Nafion (T:300K)
Water volume fraction:0.5
Diffusion coefficiency
Methanol in Nafion (T:288K)
Diffusion coefficiency
Methanol in Nafion
(2M, T:293K, N115)
Diffusion coefficiency
Oxygen in H 2 O (T:298K)
Diffusion coefficiency
Oxygen in Nafion (T:300K, N117)
Length of electrode
Depth of channels
Width of channels
83
10,000
3000
2,200
1,000
1260
590
2410
600
25,000
50
100
µm2 µm /s 2 /s
µm2 µm /s 2 /s
µm2 µm /s 2 /s
µm2 µm /s 2 /s
µm2 µm /s 2 /s
µm2 µm /s 2 /s
µm2 µm /s 2 /s
µm2 µm /s 2 /s
µm
µm
µm
[2]
[2]
[2]
[2]
[3]
[4,5]
[3]
[6]

ν
κ
a A b a b
+ ν B →ν
C + ν D
Rate = k(
T + T
ref
)
β
E0
exp( −
T + T
ref
)[ A]
ν
a [ B
]
ν
b
The stoichiometric coefficients, which are entered by the user, are used to calculate
the correct proportions of products and reagents, and are also used in the rate calculation.
The number of reagents and products are controlled using the small arrows directly
above the appropriate table. The reaction can be made reversible by entering a non-zero
Reverse Rate.
If a temperature-dependent reaction rate is desired, enter non-zero values for the
Reference Temperature (Tref), Temperature Exponent (B), and Activation Energy (E0).
All applicable reaction parameters can be varied with SimMan. The units for the
reaction rate constants depend on the stoichiometry, but should always use standard
Coventor Ware chemistry units (µM, s).
84

References
[1] D. Lee, J.S. Ko and Y.T. Kim, Thin Solid Films, 468, 285 (2004)
[2] K.D. Kreuer, Solid State Ionics, 97, 1 (1997)
[3] R.C. Reid, “The properties of gases and liquids” Handbook, (1985)
[4] X. Ren, T.E. Springer, J. Electrochem. Society 147(2), 466 (2000)
[5] Tnaja Kalio, Thesis in Helsinki University of Technology, (2003)
[6] C. Chuy, J. Electrochem. Soc., 147(12), 4453 (2000)
[7] C.S. Kelly, G.A. Deluga, and W.H. Smyrl, AIChE J., 48, 1071 (2002)
[8] Z.H. Wang, C.Y. Wang, J. Electrochem. Soc., 150, A508 (2003)
91

9292

Chapter 5
Development of Direct Methanol Fuel Cell System
for Electronic Devices
93

5.1 Micro DMFC System
5.1.1 Design guidelines for micro DMFC system
By various construction techniques, the current, the voltage and the geometric
design of the fuel cell may be adapted to the requirements of the application. A high
current is realized by enlarging the active area of the membrane electrode assemblies, a
high voltage by series connection of a certain number of cells and the exterior
dimensions are determined by the arrangement of series-connected cells in a stack or in
the plane of a membrane (banded structure membrane). Several different possible
concepts are shown in Fig.5.1. [1] The components of micro DMFC system, such as
pumps, valves, fans, and DC/DC converter, should be constructed for the packaging.
Fig.5.1 Construction principles of membrane fuel cells [1].
95 95

5.1.2 Prospects for practical application
Concern of materials
In that case of using air breathing method on cathode configuration, the products,
which is poison for biogenic such as HCOOH, HCHO species, diffusive through
membrane from anode side membrane. Produced H2O on cathode catalyst compulsorily
remove from active sites in order to hold the activity. Therefore, the controlled flow
management promises safety and stability, whereby, an optimized pump as mass flow
controller is required.
The system must be considered to provide the necessary reactants to the fuel cell,
scavenge the by-products and to regulate the system, while minimizing volume, mass
and if at all possible, system complexity.
In an ordinary stack, a bipolar plate serves as current collector and also contains the
flow field for fuel supply to the electrode area. As the current path follows the surface of
the membrane and as a short circuit between adjacent cells must be excluded, the cell
frame of this type of cell will be fabricated from electrically insulating polymeric
materials. This feature makes cheap and simple mass production for the cell housings
possible.
96

5.2 Micro DMFC System Integration on Single Si Wafer
Technical issues of plane stacking type µ-DMFC are as follows
1. Reduction the ohmic resistance of interconnection between unit cells.
2. Pressure in plane must be maintain for sealing and reduction of ohmic contact)
3. Ionic insulation between unit cells.
4. Electrical insulation between unit cells.
5. Prevent of exhaust of intermediates, such as Formaldehyde and Formic acid.
6. Reduction of cross over of fuel.
7. Purified Oxidant.
The investigations focused on pattering technologies for the fabrication of micro
flow fields, design studies for integrated flow fields, material compatibility for fuel cells,
patterning of membrane electrodes, and serial interconnection of single cells in a planar
arrangement, laminating and assembling processes.
Determining the optimal manifold design and the fraction of the input and the output
stream that bypasses each channel will be a challenge problem, well suited to modeling
by a CFD (computational fluid dynamics) software package Coventor ware as shown in
Fig.5.2. The manifold geometry must be optimized to minimize the pressure drop across
the system. Fig.5.2 shows the pressure drop calculated for the channels with vertical
wall of various geometries. The pressure drop is determined by the smaller of the
channel depth or width, and is generally smaller for larger channels. The channels are
sufficiently small that all flow is laminar.
The planar design is made possible by the small scale lengths used in a miniaturized
device; there is no analogous system in large fuel cells. The characteristic length of the
system is the distance that the protons must travel from anode to cathode, which here is
the clearance between channels. If this distance is too long, then the system will have a
large Ohmic impedance; too short and the methanol crossover rate will be too large.
Additionally, if the channels are too wide, then the catalyst layer will not be utilized
uniformly the section of the catalyst layer closest to the opposite electrode will be most
active [2]. The monolithic design may appear similar to a “strip cell”, but it is quite
different. In a strip cell, the fuel and oxidant are mixed together at the inlet and flow
over a reaction zone that has two regions of catalysts- one operating as the cathode and
one as the anode. It would be unwise top-remix either H2/O2 or CH3OH/O2.
Lee et al [3] presented a design configuration for integrated series connection of
polymer electrolyte fuel cells in a planar array. The series path is oriented in a
97

more appropriate because of no separate mechanically including the possibility of
lateral ionic conduction within the membrane.
a)
b)
Fig.5.3 Design of chrome mask for µ-DMFC array. a) Single cell, b) Ten cells on 4 inch
wafer.
Fig.5.4 shows the fabrication process flow of µ-DMFC array. The starting material
was a 4 inch diameter, oriented silicon (p-type, 1-10 Ω cm, 200 ± 20 µm thick)
double-side polished silicon wafer. 2 µm thick silicon oxide layer was formed by
thermal oxidation, beforehand. To make feedholes, the silicon dioxide on the backside
of the wafer was patterned by photolithography. Custom alignment of multiple electrode
pairs onto a single membrane was coordinated with the supplier (BCS Technology,
Byran, TX). Windows were then opened using buffered hydrogen fluorides (BHF) etch.
The feedholes were then etched using Deep Reactive Ion Etching (D-RIE) (details in
Section 2.1.2). To make channels as electrodes and manifolds from external to each cell,
the silicon dioxide on the top and backside of the wafer were patterned with similar to
the feedholes. Windows were then opened using BHF. The channels were prepared by
anisotropic silicon etching, aqueous solution of hydrogen potassium (KOH, 30 wt%) , at
80 o C. The feedholes from external are circulars, 1 mm diameter, and with a hole spacing
of 8 mm and 10 mm, respectively. The electrode was patterned on the front side of the
wafer. A 500 nm layer of silicon dioxide was deposited on the silicon surface by
tetraethoxysilane chemical vapor deposition (TEOS-CVD) to prevent short-circuit
between each cell. An AZP positive resist was coated by spray coater to enhance the
step-coverage on the lateral walls. Ti/Au for current collectors was formed by
99

Fig.5.5 shows a single diced silicon µ-DMFC array chip. The catalyst layers of
anodic Pt-Ru and cathodic Pt were deposited by using electro-deposition (see. Section
3.2.1). The width and the depth of the micro-channels are both equal of 100 µm, while
the clearance between channels is 100 µm as shown in Fig.5.5.
The feed holes and channels in the silicon wafers were prepared by anisotropic
silicon etching from the back and front of the wafer with silicon dioxide acting as
intrinsic etch stop layer [2].
Fig.5.5 Photograph of fabricated µ-DMFC array with one cent coin.
101

5.2.2 Performance of the 3 cells stack system
The performance of 3 cells µ-DMFC array is shown in Fig.5.6. 3 cells µ-DMFC
array have successfully demonstrated the expected additive performance of the
integrated series, and peak power in a 3-cell silicon assembly with methanol and oxygen
has exceeded 0.28 mW/cm 2 at 0.78 mA/cm 2 .
Cell voltage / mV
800
600
400
200
Current density / mA cm-2 0
0
0 0.4 0.8 1.2
Current density / mA cm-2 0
0
0 0.4 0.8 1.2
Fig.5.6 Polarization curves of µ-DMFC array.
5.2.3 Conclusion
In this chapter, a µ-DMFC array with 3-cells in series on silicon wafer has been
successfully fabricated by micro electronic fabrication techniques including spray
coating, photolithography, dry and wet etching and chemical and physical vapor
deposition. The operation was confirmed, however the cell voltage is lower than that of
three-fold of unit cell.
102
0.4
0.2
Power density / mW cm-2 Power density / mW cm-2

5.3 Study of separator materials for particle application
In order to reduce the overall cost of the fuel cell with a view toward
commercialization, it is necessary to produce this cell on cheaper base substrates. A
number of plastic materials exist which are suitable for replication of surface relief
microstructures e.g. poly (methyl methacrylates) (PMMA) and polycarbonates (PC). In
this study, a commercial, cyclic olefin copolymer (COC) was used do to its excellent
properties. These materials exhibit unique property combinations such as optical clarity,
excellent dielectric strength, moisture barrier, high water vapor resistance and high
temperature resistance compared with typical polymers [5].
K. Shah et al. [6] developed miniature hydrogen-air proton exchange membrane fuel
cells on silicon and polydimethylsiloxane (PDMS) base substrates using conventional
and non-conventional micro-fabrication technologies. Although brittle, glass and silicon
substrates were chosen for initial prototyping to take advantage of well-characterized
processes adopted from bulk micromachining. However, in broader perspective etching
is certainly applicable to polymers, metals and other substrates as well. Furthermore,
complex geometric structures can be pattern-transferred to a variety of alternative
materials using a micro machined master for further design flexibility.
5.3.1 Experimental
First, the Si master with channels as micro relief was fabricated. The original
material was a 200 µm thick p-type silicon wafer with sensitivity from 0.1 to 10
Ωcm. The first step was to deposit silicon dioxide layer on the wafer for Si etch mask.
To make micro-channels, the silicon dioxide on the front side of the wafer was patterned
by photolithography. A buffered hydrofluoric acid (BHF) etching was applied on silicon
dioxide of the front side wafer and then the exposed surface of silicon etched with
potassium hydroxide (KOH) solution at about 80 o C. Subsequently, the silicon oxide of
surface was etched with BHF to ensure that there was no fresh silicon surface exposed
in the channels so that the metal layer would bond well in contact with the silicon wafer.
Finally, two depositing steps were sequent applied to the front side of the silicon wafer:
(1) 30 nm of titanium layer as adhesive, (2) 200 nm of a gold layer as seed layer for
electro-deposition. For replication of the microchannels, a commercial ZEONEX 480
cyclo olefine copolymer (Zeon) was used.
Fig.5.7 shows the essential steps in the fabrication of the nickel master. A
generation nickel master (also called a shim or stumper) is obtained by electroplating.
The plating was done in house. The bath composition and the conditions of
103

electro-plating are summarized in Table 5.1. The plating was stopped when a total
nickel thickness of about 300 µm had been obtained. The nickel master was then
separated from the original silicon master. In the separation procedure the silicon master
was removal by immersing it into tetramethylammonium hydroxide (TMAH) solution.
Table 5.1. Conditions of fabricating Ni mold master
Nickel(II) sulfamate tetrahydrate
Boric acid
Nickel(II) bromide trihydrate
pH
Temperature
Current density
Si
Si
Si
Seed layer
350
30
10
3.0-4.0
50
5-10
g L-1 g L-1 g L-1 g L-1 g L-1 g L-1 o C
A dm-2 A dm-2 a) Formation of seed layer
Ni
Seed layer
Seed layer
Si master
b) Deposition of Ni
c) Removal of Si
Ni master
Fig.5.7 Fabrication process of Ni master for imprinting method.
Fig.5.8 shows SEM image of the nickel master. The channel shape was fine
104

transcription. The dimension of the nickel master was 2×2 cm. The nickel master was
stuck on a resin in order to fix at hot-embossing.
a)
b)
Fig.5.8 SEM image of the fabricated nickel master. a) channel part, b) bend part of
channels.
Sample
Fig.5.9 Schematic cross-sectional view of the hot embossing system.
Hot plate
Si plate
Ni master
Si plate
Hot plate
Fig.5.9 shows hot embossing press used for replication. The Electric EVG520
Semi-Automated Wafer Bonding System was used. A third generation nickel master
was mounted on the hot plate of the temperature controlled, polished plates. The plastic
105

disk was then sandwiched between the nickel master and the silicon smooth plates with
400 µm of thickness. The hot plate was heated to a temperature that was well above the
glass transition temperature, Tg, of the cyclo olefine® copolymer (ZEONEX 480
Tg~130 o C). A pressure of 500 N/cm 2 was then applied for 20 minutes. When the plates
had cooled down to 20 o C below Tg, the pressure was released and the high quality
plastic replica was separated from the master.
A 100 nm gold layer as current collector was deposited in an ULVAC CRTM-6000
electron beam evaporation chamber. The electrode was patterned by photolithography.
Fig.5.10 and Fig.5.11 show the electrode patterned COC based substrate. The figure
indicates that the shape of the diffractive structure is preserved to a high degree
throughout the replication procedure.
100 µm
Fig.5.10 Photographs of the electrode patterned COC substrate.
Fig.5.11 Photograph of COC based µ-DMFC.
106
Patterned
resist
Au/ COC
COC substrate
Metal pattern
Through hole
Micro channel

5.3.2 Performance of the micro DMFC on flexible resin substrate
Fig.5.12 shows the performance of the COC based µ-DMFC utilizing 2 M methanol
as the fuel and 0.5 M H2SO4 as the proton transport medium. The cell was operated at
room temperature with ambient pressure. The flow rate of both fuel and oxidant
solutions was 10 µl/min, respectively. Prototypes using COC material have successfully
demonstrated. The maximum power density was 0.39 mW/cm 2 at 2.73 mA/cm 2 .
As can be seen in Fig.5.12, the performance was lower than that of prototype cell
using Si substrate The acidic methanol analyzes solution does transport protons
generated on the Pt-Ru anode catalyst layer to the membrane when the Pt-Ru alloy is
remote from the Nafion membrane. The protons are transported through the acidic
methanol solution to the membrane, for migration to the cathode.
Cell voltage / mV
300
200
100
0
0
0 1 2 3 4
Current density / mA cm-2 Current density / mA cm-2 Fig.5.12 Polarization curves of COC based µ-DMFC. The flow rate of both fuel and
oxidant solutions was 10 µl/min, respectively
107
0.5
0.4
0.3
0.2
0.1
Power density / mW cm-2 Power density / mW cm-2

5.3.3 Conclusion
A polymer material as base substrate was employed to evaluate the possibility of
adaptation to µ-DMFC system. Evaluation of the cell performance by electrochemical
measurements and adaptation of hot embossing method to fabricate the polymer replica
confirmed that the polymer material can actually be used in simple fabrication process.
If the lifetime of the cell performance is elongated, the polymer µ-DMFCs will be used
for practical use.
5.4 Perspective for Future Work
While the extreme aspect ration for each flow pass in the planar design is an
exaggeration of what on would want to use in practice, it does illustrate an important
point: that the flow channels lie in a single plane precludes one from crossing the fuel
and oxidant streams. In addition, one must put down metal lines to pull the current,
because the silicon substrate must be insulating, to prevent short circuiting one another
out. The planar design, then, offers simplicity only in the stepwise integration of
substrate, electrodes and membranes. This simplicity is paid for, however, both in power
density and in terms of complexity within the substrate.
108

References
[1] A Heinzel, C Hebling, M. Müller, M. Zedda, and C. Müller, J. Power Sources, 105,
410 (2002).
[2] Jingrong Yu, Ping Cheng a, Zhiqi Maa, Baolian Yi, Electrochim. Acta, 48, 1537
(2003).
[3] S.J. Lee, A. Chang-Chien, S.W. Cha, R. O’Hayre, Y.I. Park, Y. Saito, F.B. Prinz, J.
Power Sources, 112, 410 (2002).
[4] R. Hahn, S. Wagner, A. Schmitz and H. Reichl, J. Power Sources, 131, 73 (2004).
[5] W. Kaminskym, Maromal. Chem. Phys., 197, 3907 (1996).
[6] K. Shah, W.C. Shin, R.S. Besser, J. Power Sources, 123, 172 (2003).
109

110

Chapter 6
General Conclusions
111

112

We have proposed the novel concept which consists of novel design and
technological solutions for fabrication of micro direct methanol fuel cell by using
MEMS and electrochemical technology. In future, further miniaturization of micro fuel
cells to generate 1-100 mW class output are also becoming important for distributed
micro-sensors and wireless MEMS. For such micro fuel cells, a MEMS fabrication
technology is essential source, for miniaturization and sophisticated fuel cell systems.
The summary of this study was discussed in chapter 1 to 5 is as follows:
Chapter 1 is general introduction of this study. Basic background of fuel cells, micro
fuel cell, and their characteristic are described. The key point of MEMS technology is
summarized.
In Chapter 2, the novel design of micro-DMFC based on MEMS fabrication
technology is proposed. The working principle which have proposed design are
described. A new concept micro-DMFC was fabricated by MEMS fabrication technique
including photolithography, etching, physical vapor deposition, and electro-plating.
In Chapter 3, The operation of the micro-DMFC with a new concept was
demonstrated. The cell performance of fabricated micro-DMFC was investigated and
the single electrode performance was evaluated using direct hydrogen electrode.
In Chapter 4, a Computational Fluid Dynamics (CFD) method has been adapted for
numerical calculation of the several parameters on the new concept micro-DMFC. The
characteristics were calculated.
In Chapter 5, the integration of micro-DMFC, so-called micro-DMFC array with 3
cells in series was designed and fabricated. The polymer based micro-DMFC was also
discussed about the fabrication process and the capability for practical application.
We discuss about the left problems and their solutions. Aspects of the design,
materials and fabrication of a micro-fabricated methanol fuel cell have been presented.
Our concept of a novel structure lies in that the anodic and cathodic micro-channels
arranged in a plane, which different from a conventional bipolar design. The first
objective of the experimental trials was simply to verify the feasibility of this novel
structure on basis of MEMS fabrication technology. Thus a micro-DMFC on a silicon
113

wafer has been successfully fabricated using photolithography, deep reactive ion
etching, alkali etching and electron beam deposition. However, the performance and
optimization of several parameters was not enough.
The problems in our proposed micro-DMFC are as follow.
1) Methanol crossover into the PEM
2) Low reliability towards flow sealing of surface on the substrate.
3) High ohmic resistance on the interconnection.
4) The optimization for several parameters of the channel and electrodes.
5) Poor catalytic activity.
In future prospects, in order to solve these problems, it is necessary to study them as
follows. First of all, in that case of a planar type, the development or an appropriate
choice of Polymer Electrolyte Membrane (PEM), which suitable for the characteristic,
should be need. Currently, micro-DMFCs are being studying about directly patterning
PEM method. The current approach to developing micro fuel cells is based on adoption
of micromachining technologies. The utilizing of silicon micromachining and related
thin film processes have been employed by several research groups as an attractive
route for fabrication [1-4]. The techniques enable simple stacking of fuel cell
components on a chip compact, light weight, scalable power generation. It has two
kinds of feature. First, it can be used as a portable power source as described above. As
second applications, it can be used as a simple and flexible development tool to explore
fuel cell materials and components. A number of technical issues such as cost,
performance, materials, etc, need to be resolved to commercialize fuel cells [5], hence,
extensive research efforts are necessary in order to develop novel, cheap, fuel cell
materials, to reduce electrocatalyst loadings, and to simplify manufacturing process.
This device consists of a base substrate with micro-channels to introduce and
distribute fuel, anode and cathode electrodes, PEM, and the external electric circuit. In
this case, silicon and poly-dimethyl siloxane (PDMS) as base substrate and their
associated fabrication techniques to create flow structures were used. A thin solid sheet
Nafion membrane was used as the PEM. Catalytic electrodes were selectively deposited
as anode and cathode on both sides of this membrane using vacuum sputtering. This
membrane was then bonded to the base substrate such that the anode side faced the flow
channels. The membrane was extended from the base on one edge so that it remained
free standing to provide electrical connections to both sides. The cathode side was kept
open to draw oxygen from the ambient air. Hydrogen enters through the inlet via of the
base substrate and distributes along the microchannels. The hydrogen then reacts in the
114

presence of the catalyst to form protons and electrons. Electrons conduct through the
external circuit producing electrical current and reach the cathode, while protons pass
through the Nafion membrane from anode to cathode, where they react with oxygen in
the presence to electrons to from water, which is allowed to vaporize into the ambient.
A novel membrane was proposed for fuel cell applications to overcome the problem
of fuel crossover and the costs. Pore-filling membrane [6] is membrane for liquid
separation, which is composed of two materials: a porous substrate and a filling
polymer that fills the pores of the substrate. The porous substrate is completely inert to
organic liquids, and the filling polymer exhibits solubility in one component of the feed.
The filling polymer exhibits perm-selectivity due to a solubility difference, and the
porous substrate matrix prevents the swelling of the filling polymer due to its
mechanical strength. These membranes show a high selectivity and permeability for
organic mixture separation of volatile organic compounds removal from water
controlling the membrane swelling [6, 7]. Recently, several research groups have
reported the use or pore-filling membranes for liquid separation, and these membranes
have also shown a high separation performance have also shown a high separation
performance and durability [8-11]. Pore-filling membranes with inorganic substrates
can offer thermal durability and an enhanced suppression of swelling [12, 13]. Moreover,
the new membrane concept also incorporates an electrolyte membrane-electrode
integrated system. To make a thin inorganic substrate, a ceramic sol-gel layer is formed
a porous electrode attached to a carbon backing. Another porous layered structure of
cathode, membrane substrate, anode, and backing is made initially. Then, an electrolyte
polymer is filled into the pores of the ceramic later, and pores to maintain gas diffusion
channels. Thus, the interfacial resistance for proton conduction between the membrane
and electrode is negligible, and the membrane area will not change with changes in
water uptake.
The results presented in this study indicated that new strategies for future micro
power sources were suggested. Although there are many points which are uncertain, that
is, some of these investigations are still needed to develop and optimize, the design
consideration and the fabrication process constructed in this study will contribute to the
realization of future technology such as new power sources for micro medical robot,
micro embedding device in human body. It is expected that these results might be
helpful for operating future micro electro-devices.
115

References
[1] C.Muller, M. Muller, and C. Hebling, “Microreactors for energy generation and
storage”, AIChE (2000)
[2] L. Mex, N. Phenath, and J. Muller, Fuel Cells Bull. 4(39), 9 (2001)
[3] J. P. Meyers, M. Helen, J. Power Sources, 109, 76 (2002)
[4] W. Y. Sim, G. Y. Kim, and S. S. Yang, Proceedings of the IEEE Micro Electro
mechanical Systems (MEMS), 341 (2001)
[5] H. Chang, P. Koschany,, C. Lim, J. New Mater. Electrochem. Soc., 3, 55 (2000)
[6] T. Yamaguchi, S. Nakao, and S.Kimura, Macromolecules, 24, 5522 (1991);
T.Yamaguchi, S. Nakao and S. Kimura, Ind. Eng. Chem. Res., 31, 1914 (1992); T.
Yamaguchi, S. Nakao and S. Kimura, Ind. Eng. Chem. Res., 32, 848 (1993)
[7] T. Yamaguchi, S. Yamahara, S. Yamahara, S. Nakao and S.Kimura, AIChE J., 42,
892 (1996)
[8] A. M. Mika, R. F. Childs, J. M. Dickson, B. E. McCarry, and D. R. Gagnon, J.
Membr. Sci., 135, 81 (1997)
[9] M. Ulbricht, H. H. Schwarz, J. Membr. Sci., 136, 25 (1997)
[10] H. Y. Wang, K. Tanaka, H. Kita and K. Okamoto, J. Membr. Sci., 154, 221 (1999)
[11] A. Wenzel, H. Yanagishita, D. Kitamoto, A. Endo, K. Haraya and T. Nakane, J.
Membr. Sci., 179, 69 (2000)
[12] J.D. Jou, W. Yoshida and Y. Cohen, J. Membr. Sci., 162, 269 (1999)
[13] T. Kai, T. Yamaguhchi, S. Nakao, Ind. Eng. Chem. Res., 39, 3284 (200)
116

List of Achievement
1. Original Articles
“MEMS-based design and fabrication of a new concept micro direct methanol fuel cell
(µ-DMFC)”, Electrochemistry Communications 6 (2004) 562-565.
Shinji Motokawa, Mohamed Mohamedi, Toshiyuki Momma, Shuichi Shoji, Tetsuya
Osaka
“A Micro Direct Methanol Fuel Cell Using Platinum and Platinum-Ruthenium
Electroplated Microchannel Electrodes”
Electrochemistry, in press
Shinji Motokawa, Mohamed Mohamedi, Toshiyuki Momma, Shuichi Shoji, and Tetsuya
Osaka
“Fabrication process of micro channel electrodes for µDMFC”
Electrochemistry, in press
Shinji Motokawa, Hiroyuki Obata, Mohamed Mohamedi, Toshiyuki Momma, Shuichi
Shoji, and Tetsuya Osaka
2. Oral Presentations (presented by S. Motokawa et al.)
“Fabrication of Small Direct Methanol Fuel Cell took advantage of MEMS technology”
The 81st Spring The Chemical Society of Japan
18 th Mar., 2002,Tokyo, Japan
Shinji Motokawa, Toshiyuki Momma, Eiichi Fujimoto, Masanori Ishizuka, Shuichi
Shoji, Tetsuya Osaka
“Fabrication of µDMFC using MEMS technology”
The Society of Chemical Engineers, Japan
14 th Jul., 2002, Gunma, Japan
Shjnji Motokawa, Toshiyuki Momma, Tetsuya Osaka
117

“MEMS-Based Design and Fabrication of a New Concept µ-DMFC”
The 5 th Korea-Japan Joint Seminar on Advanced Batteries
Sept. 25-27, 2003, Seoul, Korea
Shinji Motokawa, Hiroki Kitoh, Toshiyuki Momma, and Tetsuya Osaka
“Fabrication of a New Concept µDMFC using MEMS Technology”
The 44 th Battery Symposium in Japan
Nov. 4-6, 2003, Osaka, Japan
Shinji Motokawa, Hiroki Kitoh, Eiichi Fujimoto, Masanori Ishizuka, Mohamed
Mohamedi, Toshiyuki Momma, Shuichi Shoji, Tetsuya Osaka
“Properties of novel planar µDMFC fabricated by using MEMS technology”
The 71th The Electrochemical Society of Japan
Mar. 24-26, 2004, Kanagawa, Japan
Hiroki Kitoh, Shinji Motokawa, Masanori Ishizuka, Mohamed Mohamedi, Toshiyuki
Momma, Shuichi Shoji, Tetsuya Osaka
“MEMS-BASED COMPONENTS FOR MICRO DIRECT METHANOL FUEL CELL”
ELECTROCHEMICAL MICRO & NANO SYSTEM TECHNOLOGIES
Sept. 29, 2004, Tokyo, Japan
Shinji Motokawa, Hiroki Kitoh, Masanori Ishizuka,
Mohamed Mohamedi, Toshiyuki Momma, Shuichi Shoji, Tetsuya Osaka
118

Acknowledgements
I would like to express my sincere gratitude to Professor Dr. Tetsuya Osaka for his
supervision and continuous encouragement. I would extend my sincere appreciation to
Professor Dr. Kazuyuki Kuroda, Professor Dr. Yoshiyuki Sugawara, Professor Dr.
Shuichi Shoji and Associate Professor Dr. Takayuki Homma for their valuable
suggestions and encouragement. And I would like to thank Associate Professor Dr.
Toshiyuki Momma for his valuable suggestions and interesting discussion. It is a great
pleasure to express my sincere gratitude to Dr. Mohamed Mohamedi for valuable
suggestions in many aspects.
I wish to thank Professor Dr. Toru Asahi, Professor Dr. Soo-Gil Park, Chungbuk
National University, for their encouragement and valuable comments. I also wish to
thank Dr. Saito, Dr. K. Tsutsui, Dr. Mizuno, for their helpful advice, and supplying
materials in experimental.
Gratitude is offered to my senior and junior colleagues, Dr. J-E. Park, Dr. D. Niwa, J.
Kawaji, H. Sato, M. Ishizuka, T. Shimizu, N. Kubo, M. Hasegawa, H. Mukaibo, M.
Yoshino, T. Onishi, K. Omichi, Y. Ohinata, T. Sumi, N. Chihara, H. Nara, Y. Hondo, K.
Mori, H. Hojo, S. Tominaka, H. Kitoh, M. Naka, H. Fukunaga, M. Shoji, M. Iwasaki, M.
Ueda, H. Obata, K. Saigusa for their experimental assistance, discussions, and kind
friendship. And also, I would like to offer this thesis with all my thanks to all who
were/are on the Applied Physical Chemistry Laboratory, Waseda University.
Finally, I would like to send my many thanks to Hokushin Corporation for their
financial supports.
119
Shinji Motokawa