1. INTRODUCTION
1.3 Motivation
Silicon is an ideal material for constructing future micro fuel cell, because the silicon based micro fuel cells can be mass produced at a low cost via existing MEMS technology.
Hence, several investigations have attempted to fabricate the silicon based micro fuel cells [13]-[19]. In its early development, the anode and cathode structures (ex. current collectors, flow fields, fuel reservoir, etc.) were micro-fabricated using silicon wafer followed by stacking them with Nafion® membrane. However, Nafion® is polymer and, unlike silicon, is not compatible with conventional MEMS technology. To mitigate this problem, there have been attempts to replace this polymer membrane with acid functionalized inorganic membrane. For example, the use of porous silicon filled with Nafion® as a proton exchange membrane had been demonstrated [20].
In general, the porous silicon membrane is fabricated by anodizing a silicon wafer in a hydrofluoric solution. This fabrication process produces long narrow channels with a typical diameter of 30 nm. These channels need to be functionalized with acid groups in order to conduct protons. However, due to the long channels with very small diameters of porous membrane, it is almost impossible to wet the entire channel surfaces with these acid-functional-group solutions. For example, if the silicon porous membrane is functionalized with Nafion® by filling the channels with Nafion® solutions, only the top and bottom portions of each channel would be filled with Nafion® due to high surface tension. In addition, once this Nafion® filled porous silicon membrane is hydrated, swelled Nafion® can easily crack the channels.
Therefore, we propose a different way of making porous silicon membrane using a KOH anisotropic etching process. This new process provides a porous membrane with the V-shaped (or inverse pyramid shaped) channels, which is advantageous for avoiding aforementioned surface tension and swelling problems. Based on this new approach, we expect to achieve a
higher performance of future micro fuel cells.
1.4 Polymer Electrolyte Membrane Fuel Cells
In Polymer Electrolyte Membranefuel cells, a thin ion-conducting polymer membrane is utilized as the electrolyte. Benefits of solid electrolyte include high power density, reduced corrosion and fewer electrolyte management problems compared to liquid electrolytes. PEM fuel cells operate in temperatures where water is in liquid form. Low operating temperature guarantees quick startup from ambient conditions and extremely low nitrogen oxide emission, but then again requires the use of expensive platinum metal catalysts.
Polymer electrolyte membrane fuel cells use hydrogen or hydrogen-rich gas as the fuel and oxygen as the oxidant. Oxygen may be supplied pure or as air, depending on the application. Hydrogen molecules are split on the anode into protons and electrons. The protons travel through the electrolyte membrane to the cathode while the electrons are conducted to the cathode through the external circuit and the load. On the cathode oxygen, protons and electrons combine to form water.
1.4.1 PEM fuel cell structure and reactions
1.4.1.1 PEM fuel cell reactions and stack components
A PEM fuel cell consists of a negatively charged electrode (cathode), a positively charged electrode (anode), and an electrolyte membrane. Hydrogen is oxidized on the anode and oxygen is reduced on the cathode. Protons are transported from the anode to the cathode through the electrolyte membrane and electrons are carried to the cathode over an external circuit. On the cathode, oxygen reacts with protons and electrons forming water and producing heat. Both the anode and the cathode contain a catalyst to speed up the
electrochemical processes. Basic reactions of a PEM fuel cell are depicted in Figure 1-2. Half cell reactions and the total reaction of the PEM fuel cell are presented below.
The electrical and heat energy are produced by the cathode reaction. Theoretically, the Gibbs energy of the reaction is available as electrical energy and the rest of the reaction enthalpy is released as heat. In practice, a part of the Gibbs energy is also converted into heat via the loss mechanisms.
Single fuel cell produces a limited voltage, usually less than one volt. In order to produce a useful voltage for practical applications, several unit cells are connected in series to form a
Fig.1-2. Basic description of a PEM fuel cell operation.
anode reaction cathode reaction total reaction
H2
½O2 + 2e- + 2H+ H2 + ½O2
2H+ + 2e- H2O (l) H2O (l)
(1) (2) (3)
fuel cell stack. The output voltage depends on the number of unit cells in the stack. An exploded view of a PEM fuel cell and a PEMFC stack are presented in Figure 1-3.
1.4.1.2 Electrolyte membrane
The electrolyte membrane allows protons to pass through to the cathode side, but separates hydrogen and oxygen molecules and therefore prevents direct combustion. The membrane also acts as an electronic insulator between the bipolar plates.
The proton conducting membrane usually consists of a PTFE-based polymer backbone, to which sulfonic acid groups are attached. The acid molecules are fixed to the polymer and cannot leak out, but the protons on these acid groups are free to migrate through the membrane. The most common membrane material in PEMFC prototypes is Nafion®.
Fig.1-3. An exploded view of a PEM unit cell structure and a PEMFC stack.
MEA is an abbreviation for Membrane Electrode Assembly, the structure, which consists of an ionomer membrane and two electrodes.
The membrane must remain hydrated in order to be proton conductive. This limits the operating temperature of PEM fuel cells to under the boiling point of water and makes water management a key issue in PEM fuel cell development. However, materials, which are proton conductive in temperatures over 100.°C are being developed. A summary of proton conducting materials for both low and medium temperatures and the present state of art has been presented by Alberti and Casciola [21]. The conductivity of the membrane is sensitive to contamination. For example, if the membrane is exposed to metallic impurities, metal ions diffuse into the membrane and displace protons as charge carriers, which lower the membrane conductivity.
1.4.1.3 The electrodes
All electrochemical reactions take place on the electrode surfaces. To speed up cell reactions, electrodes contain catalysts particles, virtually always platinum or an alloy of platinum and other noble metals. Low operating temperature and the low pH makes the use of catalysts necessary [22], especially oxygen reduction reaction (ORR) on the cathode is very slow if catalyst is not present.
The electrodes are usually made of a porous mixture of carbon supported platinum and ionomer. In order to be able to catalyze reactions, catalyst particles must have contact to both protonic and electronic conductors. Furthermore, there must be passages for reactants to reach the catalyst sites and for reaction products to exit. The contacting point of the reactants, catalyst and electrolyte is conventionally referred to as the three-phase interface.
In order to achieve acceptable reaction rates, the effective area of active catalyst sites must be several times higher than the geometrical area of the electrode. Therefore, the electrodes are made porous to form a three-dimensional network, in which the threephase interfaces are located.
Nowadays, most PEMFC developers have chosen the thin-film approach, in which the electrodes are manufactured directly on the membrane surface, for their product prototypes.
The benefits of thin-film electrodes include lower price, better utilization of catalyst and improved mass transport [23]. The thickness of a thin-film electrode is typically 5-15 µm and the catalyst loading is between 0.1 to 0.3 mgcm-2. The other option is to manufacture the electrode on the surface of the porous gas diffusion layer by impregnating the layer with a mixture of carbon supported catalyst and ionomer.
1.4.1.4 Gas diffusion layers
In a PEM fuel cell, the MEA is sandwiched between the flow field plates. On each side of the MEA, between the electrode and the flow field plate, there are gas diffusion layers. They provide electrical contact between the electrodes and the bipolar plates, and distribute reactants to the electrodes. They also allow reaction product water to exit the electrode surface and permit passage of water between the electrodes and the flow channels.
Gas diffusion layers are made of a porous, electrically conductive material, usually carbon cloth or carbon paper. The substrate is usually treated with a fluoropolymer and carbon black to improve water management and electrical properties.
1.4.1.5 Bipolar plates
In a fuel cell stack, bipolar plates separate the reactant gases of adjacent cells, connect the cells electrically, and act as a support structure. Furthermore, bipolar plates have reactant flow channels on both sides, forming the anode and the cathode compartments of the unit cells on the opposing sides of the bipolar plate. In a unit cell, separator plates have flow channels only on one side and are sometimes called monopolar plates.
Flow channel geometry has an effect on reactant flow velocities and mass transfer and thus on fuel cell performance. Basic channel geometries and performance test results have been presented for example by Mennola [24].
Bipolar plate materials must have high conductivity and be impermeable to gases. Due to the presence of reactant gases and catalyst, the material should be corrosion resistant and chemically inert. For most of the applications, also low weight and high strength are important. For commercial applicability, the material should be cheap and suitable for high-volume manufacturing methods.
Most PEMFC bipolar plates are made of resin-impregnated graphite, but also stainless steel has been used [25], [26]. Solid graphite is highly conductive, chemically inert and resistant to corrosion, but expensive and costly to manufacture. Stainless steel is very affordable, but expensive to machine. In addition, stainless steel must often be coated to prevent corrosion and reduce contact resistance.
Flow channels are machined or electrochemically etched to the graphite or stainless steel bipolar plate surfaces. These methods are not suitable for mass production and therefore new bipolar materials are being researched. Best results have been achieved with carbon-polymer composites, which can be molded [27], [28].
1.4.2 Theory of operation
1.4.2.1 Open circuit voltage
The reversible standard potential E0 of an electrochemical reaction is defined as
zF E G
0
0 ∆
−
= (4)
where ∆G0is the Gibbs energy change for the reaction under NTP condition, z is the reaction
where ∆H is the reaction enthalpy change, T is the temperature, and ∆S is the reaction entropy change. Both reaction enthalpy and entropy are also dependent on the temperature.
Substituting the standard conditions values for hydrogen and oxygen [22] into Equations (4) and (5), gives
Outside standard conditions, the theoretical potential Et for an electrochemical reaction is expressed by the Nernst equation [29]
given by the Equation (4). By assuming that the gases are ideal, i.e. the activities of the gases are equal to their partial pressures, and that the activity of the water phase is equal to unity7, Equation (6) gives gases, partial pressure of species i, pi can be expressed as a product of total pressure p andp x
p
i = i (8)For a hydrogen-air fuel cell operated on dry gases, Equations (7) and (8) give
( )
At standard temperature and pressure, the theoretical potential of hydrogen-air fuel cell is
( 0 . 21 ) 1 . 219 V
same time, cell voltage drops due to various irreversible loss mechanisms. The loss, which is often called overpotential η, is defined as the deviation of the cell potential E from the theoretical potential Et,Et
E−
η = (10) In a fuel cell, overpotentials originate from three sources:
– activation overpotential arises from the kinetics of charge transfer reactions, – ohmic losses arise from component resistances,
– diffusion overpotential arises from the limited rate of mass transfer.
Activation overpotential arises from the kinetics of charge transfer reaction across the electrode-electrolyte interface. In other words, a portion of the electrode potential is lost in driving the electron transfer reaction. Activation overpotential is directly related to the nature of the electrochemical reactions and represents the magnitude of activation energy, when the reaction propagates at the rate demanded by the current.
In a hydrogen-oxygen fuel cell, the contribution of anode activation overpotential is negligible, since the oxygen reduction on the cathode is significantly slower than the hydrogen oxidation on the anode [22]. The oxygen reduction reaction (ORR) activation overpotential decreases as the potential shifts to the negative direction, making the activation overpotential significant only at low current densities.
Expressions for the activation overpotentials can be derived from the Butler-Volmer equation. The anode activation overpotential increases with current density and can be expressed as
Exchange current density represents the reaction rate. The equation for cathode activation overpotential is density. The right hand side of the Equation (12) can be expressed as
i
cathode exchange current is about 10-5 times smaller [22].Ohmic overpotential, also known as IR-losses, result from electrical resistance losses in the cell. These resistances can be found in practically all fuel cell components: ionic
resistance in the gas diffusion layers, bipolar plates and terminal connections. The dominant contribution to the IR-losses arises from the electrolyte membrane [31]. The IR-loss ηIR can be expressed by the equation reactants near the electrodes. The electrode reactions require a constant supply of reactants in order to sustain the current flow. When the diffusion limitations reduce the availability of a reactant, part of the available reaction energy is used to drive the mass transfer, thus creating a corresponding loss in output voltage. Similar problems can develop if a reaction product accumulates near the electrode surface and obstructs the diffusion paths or dilutes the reactants.
The diffusion overpotential can be expressed as
C
sO2 are the hydrogen and oxygen electrode surface concentrations, respectively [31].If the concentration profile between the flow channel and the electrode surface is assumed linear (Nernst diffusion layer model), the relation between the current density and the concentration can be solved using Fick’s I law and Faraday’s law. As a result, the diffusion overpotential can be expressed as
where il,a and il,c are the anode and cathode limiting currents, respectively. Limiting currents, which can be readily measured, occur when the reactant concentration on the electrode surface is zero.
The main source for diffusion overpotential is considered to be the diffusion of oxygen through the partially flooded gas diffusion layer on the cathode side [32]. Using pure oxygen removes the diffusion problems related to reactant availability, but mass transfer can still be affected by obstruction of diffusion paths and reactant dilution. Diffusion overpotential is much smaller on the anode than on the cathode, since the diffusion of hydrogen is much faster than that of oxygen. If pure hydrogen is used, the only diluting component is the water transported to the anode by back diffusion. Reactant availability problems can still occur, if the flow stoichiometry is very low.
Expressions for anode and cathode potentials can be derived from the Equation (10) and overpotential Equations (12) to (15), giving
C
potential of an electrochemical cell is defined as the potential difference between the cathode and the anode. Using Equations (17) and (18), and taking IR-loss, Equation (14), into account, the fuel cell voltage can be expressed as a function of current density:zF ir
Concentrations in Equation (19) can be expressed as functions of current density using the Nernst diffusion layer model and Equation (16).
i ir
Characteristics of a typical PEM fuel cell current-voltage, or polarization curve are presented in Figure1-4. The curve can be divided into three regions, which are governed by a different overpotential. Activation overpotential dominates at low current densities in region I.
Region II is governed by the IR-losses and the bending down of the polarization curve in region III is due to the diffusion overpotential. Relative contributions of different overpotentials are illustrated in Figure 1-5.
Fig.1-4. Characteristics of a typical PEMFC polarization curve.
The effect of activation overpotential is seen in the Figure 1-5 as a rapid drop of the voltage at low current densities. The middle region in Figure 1-4 is nearly linear and is governed by IR-losses. The ionomer conductivity depends on the humidification level, and thus drying out of the MEA increases total cell resistance, making the voltage drop more rapid.
Diffusion overpotentials are the dominating loss mechanism at higher current densities, where the reaction rate is mass transfer limited. Water management is of key importance in controlling the diffusion overpotentials. Excess water in the gas diffusion layer or on the electrode obstructs the reactant diffusion to the active area. Diffusion overpotentials bend polarization curves down at high current densities.
On many occasions, it is convenient to study polarization curves that are free from the IR-loss, i.e. IR-compensated. For IR-compensation, the polarization voltage and resistance are
Fig.1-5. Relative contributions of different voltage loss mechanisms.
effect of IR-loss is removed by compensating the polarization voltage at each point by the size of the IR-loss. The advantage of IR-compensation is that the polarization curves show only the contributions of activation and diffusion overpotentials and the resistance can be studied separately by plotting the current-resistance curves.
1.4.2.3 Identifying overpotential
Measuring polarization curves is a well known electrochemical characterization method and it is widely used in fuel cell performance testing. Together with resistance measurements, polarization curves provide information on the overpotentials. Polarization curves are measured by generating a current or voltage sweep with a load unit and recording the cell voltage as a function of current density. To achieve steady state operation at each current density, the sweep is discrete, i.e. the voltage is measured for a given time after each step.
A method for acquiring accurate data on the contributions of all the individual overpotentials simultaneously has not been presented, but overpotentials can be sorted out by comparing the polarization and resistance plots measured using air and pure oxygen. The downside of this approach is that the different overpotentials are determined at different points of time.
The contribution of IR-loss is the easiest one to measure. Many commercial fuel cell test stations feature a built-in resistance measurement function, and instructions for building resistance measurement apparatuses has been published for example by Wruck et al. [33] and Büchi et al. [34]. The relative contribution of IR-loss can be seen by comparing uncompensated and IR-compensated polarization curves, and the absolute cell resistance values can be plotted as a function of current density.
The effect of activation overpotential is best visible in IR-compensated polarization curves measured with pure oxygen. The absolute magnitude of activation overpotential cannot
be obtained, since the open circuit voltage is ca. 0.2 V lower than the theoretical cell potential [3]. According to Bockris and Srinivasan, this discrepancy is due to presence of impurities which undergo anodic oxidation causing a mixed potential [2]. Comparing IR-compensated polarization curves measured with pure oxygen provides information on the changes in the activation overpotential, i.e. condition of the MEA. Increase in activation overpotential indicates a reduction in active catalyst area, implying MEA drying out or catalyst contamination. Drying out of the electrodes decreases the active area as the protonic resistance of the ionomer part of the electrode rises [35].
The effect of diffusion overpotential can be seen by comparing the IR-compensated polarization curves measured with air and pure oxygen. Using pure oxygen removes the mass transfer limitations related to dilution of oxygen by nitrogen, but obstruction of diffusion paths or dilution by water vapor is still present to some extent. Furthermore, using pure
The effect of diffusion overpotential can be seen by comparing the IR-compensated polarization curves measured with air and pure oxygen. Using pure oxygen removes the mass transfer limitations related to dilution of oxygen by nitrogen, but obstruction of diffusion paths or dilution by water vapor is still present to some extent. Furthermore, using pure