CHAPTER 4 FUEL CELL CHARGE TRANSPORT The previous chapter on reaction kinetics detailed one of the most pivotal steps in the electrochemical generation of electricity: the production and consumption of charge via electrochemical half reactions. In this chapter, we address an equally important step in the electrochemical generation of electricity: charge transport. Charge transport “completes the circuit” in an electrochemical system, moving charges from the electrode where they are produced to the electrode where they are consumed. There are two major types of charged species: electrons and ions. Since both electrons and ions are involved in electrochemical reactions, both types of charge must be transported. The transport of electrons versus ions is fundamentally different, primarily due to the large difference in mass between the two. In most fuel cells, ion charge transport is far more difficult than electron charge transport; therefore, we are mainly concerned with ionic conductivity. As you will discover, resistance to charge transport results in a voltage loss for fuel cells. Because this voltage loss obeys Ohm’s law, it is called an ohmic, or IR, loss. Ohmic fuel cell losses are minimized by making electrolytes as thin as possible and employing high-conductivity materials. The search for high-ionic-conductivity materials will lead to a discussion of the fundamental mechanisms of ionic charge transport and a review of the most important electrolyte material classes. 4.1 CHARGES MOVE IN RESPONSE TO FORCES The rate at which charges move through a material is quantified in terms of flux (denoted with the symbol J ). Flux measures how much of a given quantity flows through a material per unit area per unit time. Figure 4.1 illustrates the concept of flux: Imagine water flowing down this tube at a volumetric flow rate of 10 L/s. If we divide the flow rate by the Fuel Cell Fundamentals: Third Edition. Ryan O’Hayre, Suk-Won Cha, Whitney G. Colella and Fritz B. Prinz © 2016 by John Wiley & Sons, Inc. Published by John Wiley & Sons, Inc. 117 118 FUEL CELL CHARGE TRANSPORT A JA A Figure 4.1. Schematic of flux. Imagine water flowing down this tube at a volumetric flow rate of 10 L/s. Dividing this flow rate by the cross-sectional area of the tube (A) gives the flux JA of water moving down the tube. Generally, flux is measured in molar rather than volumetric quantities, so in this example the liters of water should be converted to moles. cross-sectional area of the tube (A), we get the volumetric flux JA of water moving down the tube. In other words, JA gives the per-unit-area flow rate of water through the tube. Be careful! Remember that flux and flow rate are not the same thing. By computing a flux, we are normalizing the flow rate by a cross-sectional area. The most common type of flux is a molar flux (typical units are mol/cm2 ⋅ s). Charge flux is a special type of flux that measures the amount of charge that flows through a material per unit area per unit time. Typical units for charge flux are C/cm2 ⋅ s = Aโcm2 . From these units, you may recognize that charge flux is the same thing as current density. To denote that charge flux represents a current density and carries different units than molar flux, we give it the symbol j. The quantity zi F is required to convert from molar flux J to charge flux j, where zi is the charge number for the charge-carrying species (e.g., zi is +1 for Na+ , –2 for O2– , etc.) and F is Faraday’s constant: j = zi FJ (4.1) ELIMINATE CONFUSION BETWEEN zi AND n As we move from the discussion of electrochemical kinetics (Chapter 3) to a discussion of charge transport (Chapter 4), it is important to recognize the difference between the quantities zi and n. The quantity n, which we have used throughout the book, refers to the number of electrons transferred during an electrochemical reaction. For example, in the electrochemical half reaction H2 → 2H+ + 2e− two electrons are transferred per mole of H2 gas reacted, and therefore n = 2. In contrast, the quantity zi , which we introduce here in Chapter 4, refers to the amount of charge carried by a charged species. For the charged species H+ , as an example, zi = +1, while for the charged species e– , zi = −1. CHARGES MOVE IN RESPONSE TO FORCES In all materials, a force must be acting on the charge carriers (i.e., the mobile electrons or ions in the material) for charge transport to occur. If there is no force acting on the charge carriers, there is no reason for them to move! The governing equation for transport can be generalized (in one dimension) as ∑ Ji = Mik Fk (4.2) k Where Ji represents a flux of species i, the Fk ’s represent the k different forces acting on i, and the Mik ’s are the coupling coefficients between force and flux. The coupling coefficients reflect the relative ability of a species to respond to a given force with movement as well as the effective strength of the driving force itself. The coupling coefficients are therefore a property both of the species that is moving and the material through which it is moving. This general equation is valid for any type of transport (charge, heat, mass, etc.). In fuel cells, there are three major driving forces that give rise to charge transport: electrical driving forces (as represented by an electrical potential gradient dVโdx), chemical driving forces (as represented by a chemical potential gradient d๐โdx), and mechanical driving forces (as represented by a pressure gradient dPโdx). As an example of how these forces give rise to charge transport in a fuel cell, consider our familiar hydrogen–oxygen PEMFC (see Figure 4.2). As hydrogen reacts in this fuel e– – + e– H+ –+ e– H+ –+ + H2 e– H+ Anode H – + O2 e– H+ –+ e– H+ –+ Electrolyte Cathode Figure 4.2. In a H2 –O2 fuel cell, accumulation of protons/electrons at the anode and depletion of protons/electrons at the cathode lead to voltage gradients which drive charge transport. The electrons move from the negatively charged anode electrode to the positively charged cathode electrode. The protons move from the (relatively) positively charged anode side of the electrolyte to the (relatively) negatively charged cathode side of the electrolyte. The relative charge in the electrolyte at the anode versus the cathode arises due to differences in the concentration of protons. This concentration difference can also contribute to proton transport between the anode and cathode. 119 120 FUEL CELL CHARGE TRANSPORT cell, protons and electrons accumulate at the anode, while protons and electrons are consumed at the cathode. The accumulation/depletion of electrons at the two electrodes creates a voltage gradient, which drives the transport of electrons from the anode to the cathode. In the electrolyte, accumulation/depletion of protons creates both a voltage gradient and a concentration gradient. These coupled gradients then drive the transport of protons from the anode to the cathode. In the metal electrodes, only a voltage gradient drives electron charge transport. However, in the electrolyte, both a concentration (chemical potential) gradient and a voltage (electrical potential) gradient drive ion transport. How do we know which of these two driving forces is more important? In almost all situations, the electrical driving force dominates fuel cell ion transport. In other words, the electrical effect of the accumulated/depleted protons is far more important for charge transport than the chemical concentration effect of the accumulated/depleted protons. The underlying reasons why electrical driving forces dominate fuel cell charge transport are explained for the interested reader in an optional section near the end of this chapter (see Section 4.7). For the case where charge transport is dominated by electrical driving forces, Equation 4.2 can be rewritten as dV (4.3) j=๐ dx where j represents the charge flux (not molar flux), dVโdx is the electric field providing the driving force for charge transport, and ๐ is the conductivity, which measures the propensity of a material to permit charge flow in response to an electric field. This important application of Equation 4.2 simplifies the terms of fuel cell charge transport. In certain rare situations, both the concentration effects and electric potential effects may become important; in these cases, the charge transport equations become considerably more difficult. Comparing Equation 4.3 to Equation 4.2, it is apparent that conductivity ๐ is nothing more than the name of the coupling coefficient that describes how flux and electrical driving forces are related. The relevant coupling coefficient that describes transport due to a chemical potential (concentration) gradient is called diffusivity. For transport due to a pressure gradient, the relevant coupling coefficient is called viscosity. These transport processes are summarized in Table 4.1 using molar flux quantities. TABLE 4.1. Summary of Transport Processes Relevant to Charge Transport Transport Process Driving Force Coupling Coefficient Equation Conduction Electrical potential gradient, dVโdx Conductivity ๐ J= Diffusion Concentration gradient, dcโdx Diffusivity D Convection Pressure gradient, dpโdx Viscosity ๐ dc dx Gc dp J= ๐ dx ๐ dV |zi |F dx J = −D Note: The transport equation for convection in this table is based on Poiseuille’s law, where G is a geometric constant and c is the concentration of the transported species. Convection flux is often calculated simply as J = ๐ฃci , where v is the transport velocity. CHARGE TRANSPORT RESULTS IN A VOLTAGE LOSS 4.2 CHARGE TRANSPORT RESULTS IN A VOLTAGE LOSS Unfortunately, charge transport is not a lossless process. It occurs at a cost. For fuel cells, the penalty for charge transport is a loss in cell voltage. Why does charge transport result in a voltage loss? The answer is because fuel cell conductors are not perfect—they have an intrinsic resistance to charge flow. Consider the uniform conductor pictured in Figure 4.3. This conductor has a constant cross-sectional area A and length L. Applying this example conductor geometry to our charge transport equation 4.3 produces V L (4.4) ( ) L ๐ (4.5) j=๐ Solving for V yields V=j You might recognize that this equation is similar to Ohm’s law: V = iR. In fact, since charge flux (current density) and current are related by i = jA, we can rewrite Equation 4.5 as ( ) L = iR (4.6) V=i A๐ where we identify the quantity LโA๐ as the resistance R of our conductor. The voltage V in this equation represents the voltage which must be applied in order to transport charge at a rate given by i. Thus, this voltage represents a loss: It is the voltage that is expended, or sacrificed, in order to accomplish charge transport. This voltage loss arises due to our conductor’s intrinsic resistance to charge transport, as embodied by 1/๐. Length = L Area = A j j R = L/Aσ V V 0 V = jL/σ = iR 0 x L Figure 4.3. Illustration of charge transport along a uniform conductor of cross-sectional area A, length L, and conductivity ๐. A voltage gradient dV/dx drives the transport of charge down the conductor. From the charge transport equation j = ๐(dVโdx) and the conductor geometry, we can derive Ohm’s law: V = iR. The resistance of the conductor is dependent on the conductor’s geometry and conductivity: R = Lโ๐A. 121 FUEL CELL CHARGE TRANSPORT Voltage (V) Because this voltage loss obey’s Ohm’s law, we call it an “ohmic” loss. Like the activation overvoltage loss (๐act ) introduced in the previous chapter, we give this voltage loss the symbol η. Specifically, we label it ๐ohmic to distinguish it from ๐act . Rewriting Equation 4.6 to reflect our nomenclature and explicitly including both the electronic (Relec ) and ionic Eo Anode Electrolyte Cathode Distance (x) Voltage (V) (a) η act,C η act,A Anode V Electrolyte Eo Cathode Distance (x) (b) Voltage (V) 122 η ohmic Anode Electrolyte o V E Cathode Distance (x) (c) Figure 4.4. (a) Hypothetical voltage profile of a fuel cell at thermodynamic equilibrium (recall Figure 3.7). The thermodynamic voltage of the fuel cell is given by E0 . (b) Effect of anode and cathode activation losses on the fuel cell voltage profile (recall Figure 3.9). (c) Effect of ohmic losses on fuel cell voltage profile. Although the overall fuel cell voltage increases from the anode to the cathode, the cell voltage must decrease between the anode side of the electrolyte and the cathode side of the electrolyte to provide a driving force for charge transport. CHARGE TRANSPORT RESULTS IN A VOLTAGE LOSS (Rionic ) contributions to fuel cell resistance gives ๐ohmic = iRohmic = i(Relec + Rionic ) (4.7) Because ionic charge transport tends to be more difficult than electronic charge transport, the ionic contribution to Rohmic tends to dominate. The direction of the voltage gradient in an operating fuel cell electrolyte can often seem nonintuitive. As Figure 4.4c illustrates, although overall fuel cell voltage increases from the anode to the cathode, the cell voltage must decrease between the anode side of the electrolyte and the cathode side of the electrolyte to provide a driving force for charge transport. Example 4.1 A 10-cm2 PEMFC employs an electrolyte membrane with a conductivity of 0.10 Ω−1 ⋅ cm−1 . For this fuel cell, Relec has been determined to be 0.005 Ω. Assuming the only other contribution to cell resistance comes from the electrolyte membrane, determine the ohmic voltage loss (๐ohmic ) for the fuel cell at a current density of 1 Aโcm2 in the following cases: (a) the electrolyte membrane is 100 ๐m thick; (b) the electrolyte membrane is 50 ๐m thick. Solution: We need to calculate Rionic based on the electrolyte dimensions and then use Equation 4.7 to calculate ๐ohmic . Since the fuel cell has an area of 10 cm2 , the current i of the fuel cell is 10 A: i = jA = 1 Aโcm2 × 10 cm2 = 10 A (4.8) From Equation 4.6 we can calculate Rionic for the two cases (a), (b) given in this problem: L 0.01 cm = = 0.01 Ω −1 ๐A (0.10 Ω ⋅ cm−1 )(10 cm2 ) 0.005 cm = = 0.005 Ω (0.10 Ω−1 ⋅ cm−1 )(10 cm2 ) Case (a): Rionic = Case (b): Rionic (4.9) Inserting these values into Equation 4.7 and using i = 10 A gives the following values for ๐ohmic : Case (a): ๐ohmic = i(Relec + Rionic ) = 10 A(0.005 Ω + 0.01 Ω) = 0.15 V Case (b): ๐ohmic = 10 A(0.005 Ω + 0.005 Ω) = 0.10 V (4.10) With everything else equal, making the membrane thinner reduces the ohmic loss! However, note that the payoff does not scale directly with membrane thickness. Although the membrane thickness was cut in half in this example, the ohmic loss was only reduced by one-third. This occurs because not all of the fuel cell’s resistance contributions come from the electrolyte. 123 FUEL CELL CHARGE TRANSPORT 4.3 CHARACTERISTICS OF FUEL CELL CHARGE TRANSPORT RESISTANCE As Equation 4.7 implies, charge transport linearly decreases fuel cell operating voltage as current increases. Figure 4.5 illustrates this effect. Obviously, if fuel cell resistance is decreased, fuel cell performance will improve. Fuel cell resistance exhibits several important properties. First, resistance is geometry dependent, as Equation 4.6 clearly implies. Fuel cell resistance scales with area: To normalize out this effect, area-specific resistances are used to compare fuel cells of different sizes. Fuel cell resistance also scales with thickness; for this reason, fuel cell electrolytes are generally made as thin as possible. Additionally, fuel cell resistances are additive; resistance losses occurring at different locations within a fuel cell can be summed together in series. An investigation of the various contributions to fuel cell resistance reveals that the ionic (electrolyte) component to fuel cell resistance usually dominates. Thus, performance improvements may be won by the development of better ion conductors. Each of these important points will now be addressed. 4.3.1 Resistance Scales with Area Since fuel cells are generally compared on a per-unit-area basis using current density instead of current, it is generally necessary to use area-normalized fuel cell resistances when discussing ohmic losses. Area-normalized resistance, also known as area-specific resistance (ASR), carries units of Ω ⋅ cm2 . By using ASR, ohmic losses can be calculated from current density via (4.11) ๐ohmic = j(ASRohmic ) 1.2 Cell voltage (V) 124 Theoretical EMF or ideal voltage Ohmic loss: ηohmic = iRohmic Rohmic = 0.50 โฆ Rohmic = 0.75 โฆ 0.5 Rohmic = 1.0 โฆ Current (A) 1.0 Figure 4.5. Effect of ohmic loss on fuel cell performance. Charge transport resistance contributes a linear decrease in fuel cell operating voltage as determined by Ohm’s law (Equation 4.7). The magnitude of this loss is determined by the size of Rohmic . (Curves calculated for Rohmic equal 0.50 Ω, 0.75 Ω, and 1.0 Ω, respectively.) CHARACTERISTICS OF FUEL CELL CHARGE TRANSPORT RESISTANCE where ASRohmic is the ASR of the fuel cell. Area-specific resistance accounts for the fact that fuel cell resistance scales with area, thus allowing fuel cells of different sizes to be compared. It is calculated by multiplying a fuel cell’s ohmic resistance Rohmic by its area: ASRohmic = Afuel cell Rohmic (4.12) Be careful, you must multiply resistance by area to get ASR, not divide! This calculation will probably seem unintuitive at first. Because a large fuel cell has so much more area to flow current through than a small fuel cell, its resistance is far lower. However, on a per-unit-area basis, their resistances should be about the same; therefore, the resistance of the large fuel cell must be multiplied by its area. This concept may be more understandable if you recall the original definition of resistance in Equation 4.6: R= L A๐ (4.13) Since resistance is inversely proportional to area, multiplication by area is necessary to get area-independent resistances. This point is reinforced by Example 4.2. Example 4.2 Consider the two fuel cells illustrated in Figure 4.6. At a current density of 1 Aโcm2 , calculate the ohmic voltage losses for both fuel cells. Which fuel cell incurs the larger ohmic voltage loss? Fuel cell 1 A 1 = 1 cm 2 R1 = 0.1 โฆ Fuel cell 2 A 2 = 10 cm2 R2 = 0.02 โฆ Fuel cell 1 ASR R1A1 = 0.1 โฆ . cm2 Fuel cell 2 ASR R2A2 = 0.2 โฆ . cm2 Figure 4.6. The importance of ASR is illustrated by these two fuel cells. Fuel cell 2 has lower total resistance than fuel cell 1 but yields a larger ohmic loss for a given current density. Fuel cell resistance is best compared using ASR rather than R. Solution: There are two ways to solve this problem. To calculate voltage loss based on current density, we can either convert the resistances of the fuel cells to ASRs and then use Equation 4.11 (solution 1) or convert the current densities into currents and use Equation 4.6 (solution 2). Solution 1: Calculating the ASRs for the two fuel cells gives ASR1 = R1 A1 = (0.1 Ω)(1 cm2 ) = 0.1 Ω ⋅ cm2 ASR2 = R2 A2 = (0.02 Ω)(10 cm2 ) = 0.2 Ω ⋅ cm2 (4.14) 125 126 FUEL CELL CHARGE TRANSPORT Then, the ohmic voltage losses for the two cells can be calculated using Equation 4.11: ๐1,ohmic = j(ASR1 ) = (1 Aโcm2 )(0.1 Ω ⋅ cm2 ) = 0.1 V ๐2,ohmic = j(ASR2 ) = (1 Aโcm2 )(0.2 Ω ⋅ cm2 ) = 0.2 V (4.15) Solution 2: Converting current densities for the two fuel cells into currents gives i1 = jA1 = (1 Aโcm2 )(1 cm2 ) = 1 A i2 = jA2 = (1 Aโcm2 )(10 cm2 ) = 10 A (4.16) Then, the ohmic voltage losses for the two cells can be calculated using Equation 4.6: ๐1,ohmic = i1 (R1 ) = (1 A)(0.1 Ω) = 0.1 V (4.17) ๐2,ohmic = i2 (R2 ) = (10 A)(0.02 Ω) = 0.2 V In both solutions, the same answer is obtained; cell 2 incurs a greater voltage loss. Although the total resistance of cell 2 is lower than cell 1 (0.02 Ω versus 0.1 Ω), the ASR of cell 2 is higher than that of cell 1. Thus, on an area-normalized basis, cell 2 is actually more “resistive” than cell 1 and leads to poorer fuel cell performance. 4.3.2 Resistance Scales with Thickness Referring again to Equation 4.6, it is apparent that resistance scales not only with the cross-sectional area of the conductor but also with the length (thickness) of the conductor. If we normalize resistance by using ASR, then ASR = L ๐ (4.18) The shorter the conductor length L, the lower the resistance. It is intuitive that a shorter path results in less resistance. Ionic conductivity is orders of magnitude lower than the electronic conductivity of metals, so minimizing the resistance of the fuel cell electrolyte is essential. Hence, we want the shortest path possible for ions between the anode and the cathode. Fuel cell electrolytes, therefore, are designed to be as thin as possible. Although reducing electrolyte thickness improves fuel cell performance, there are several practical issues that limit how thin the electrolyte can be made. The most important limitations are as follows: • Mechanical Integrity. For solid electrolytes, the membrane cannot be made so thin that it risks breaking or develops pinholes. Membrane failure can result in catastrophic mixing of the fuel and oxidant! CHARACTERISTICS OF FUEL CELL CHARGE TRANSPORT RESISTANCE • Nonuniformities. Even mechanically sound, pinhole-free electrolytes may fail if the thickness varies considerably across the fuel cell. Thin electrolyte areas may become “hot spots” that are subject to rapid deterioration and failure. • Shorting. Extremely thin electrolytes (solid or liquid) risk electrical shorting, especially when the electrolyte thickness is on the same order of magnitude as the electrode roughness. • Fuel Crossover. As the electrolyte thickness is reduced, the crossover of reactants may increase. This leads to an undesirable parasitic loss, which can eventually become so large that further thickness decreases are counterproductive. • Contact Resistance. Part of the electrolyte resistance is associated with the interface between the electrolyte and the electrode. This “contact” resistance is independent of electrolyte thickness. • Dielectric Breakdown. The ultimate physical limit to solid-electrolyte thickness is given by the electrolyte’s dielectric breakdown properties. This limit is reached when the electrolyte is made so thin that the electric field across the membrane exceeds the dielectric breakdown field for the material. For most solid-electrolyte materials, the ultimate limit on thickness, as predicted by the dielectric breakdown field, is on the order of several nanometers. However, the other practical limitations listed above currently limit achievable thickness to about 10–100 ๐m, depending on the electrolyte. 4.3.3 Fuel Cell Resistances Are Additive As Figure 4.7 illustrates, the total ohmic resistance presented by a fuel cell is actually a combination of resistances coming from different components of the device. Depending on how much precision is needed, it is possible to assign individual resistances to the electrical interconnections, anode electrode, cathode electrode, anode catalyst layer, cathode catalyst layer, electrolyte, and so on. It is also possible to ascribe contact resistances associated with the interfaces between the various layers in the fuel cell (e.g., a flow structure/electrode contact resistance). Because the current produced by the fuel cell must flow serially through all of these regions, the total fuel cell resistance is simply the sum of all the individual resistance contributions. Unfortunately, it is experimentally very difficult to distinguish between all the various sources of resistance loss. You might think that it should be a relatively easy experimental task to measure the resistance of each component in a fuel cell (e.g., the electrodes, the flow structures, the interconnections, the membrane) before assembling them together into a device. However, such measurements never completely reflect the true total resistance of a fuel cell device. Variations in contact resistances, assembly processes, and operating conditions make total fuel cell resistance difficult to predict. These factors make fuel cell characterization extremely challenging, as discussed in Chapter 7, and emphasize the necessity of in situ fuel cell characterization. Despite the experimental difficulties involved in pinpointing all the sources of fuel cell resistance loss, the electrolyte yields the biggest resistance loss for most fuel cell devices. 127 128 FUEL CELL CHARGE TRANSPORT Rinterconnect Anode Ranode Relectrolyte Rcathode Electrolyte Rinterconnect Cathode Figure 4.7. The total ohmic resistance presented by a fuel cell is actually a combination of resistances, each attributed to different components of the fuel cell. In this diagram, fuel cell resistance is divided into interconnect, anode, electrolyte, and cathode components. Since current flows serially through all components, total fuel cell resistance is given by the series sum of the individual resistance components. 4.3.4 lonic (Electrolyte) Resistance Usually Dominates The best electrolytes employed in fuel cells have ionic conductivities of around 0.10 Ω−1 ⋅ cm−1 . Even at a thickness of 50 ๐m (very thin), this produces an ASR of 0.05 Ω ⋅ cm2 . In contrast, a 50-๐m-thick porous carbon cloth electrode would have an ASR of less than 5 × 10−6 Ω ⋅ cm2 . This example illustrates how electrolyte resistance usually dominates fuel cells. Well-designed fuel cells have a total ASR in the range of 0.05–0.10 Ω ⋅ cm2 , and electrolyte resistance usually accounts for most of the total. If electrolyte thickness cannot be reduced, decreasing ohmic loss depends on finding high-๐ ionic conductors. Unfortunately, developing satisfactory ionic conductors is challenging. The three most widely used electrolyte classes, discussed in Sections 4.5.1– 4.5.3, are aqueous, polymer, and ceramic electrolytes. The conductivity mechanisms and materials properties of these three electrolyte classes are quite different. Before we get to that discussion, however, it is helpful to develop a clear physical picture of conductivity in general terms. 4.4 PHYSICAL MEANING OF CONDUCTIVITY Conductivity quantifies the ability of a material to permit the flow of charge when driven by an electric field. In other words, conductivity is a measure of how well a material accommodates charge transport. A material’s conductivity is influenced by two major factors: how many carriers are available to transport charge and the mobility of those carriers within the material. The following equation defines ๐ in those terms: ๐i = (|zi |F)ci ui (4.19) PHYSICAL MEANING OF CONDUCTIVITY where ci represents the molar concentration of charge carriers (how many moles of carrier are available per unit volume) and ui is the mobility of the charge carriers within the material. The quantity |zi |F is necessary to convert charge carrier concentration from units of moles to units of coulombs. Here, zi is the charge number for the carrier (e.g., zi = +2 for Cu2+ , zi = −1 for e– , etc.), the absolute-value function ensures that conductivity is always a positive number, and F is Faraday’s constant. A material’s conductivity is therefore determined by the product of carrier concentration ci and carrier mobility ui . These properties are, in turn, set by the structure and conduction mechanisms within the material. Up to this point, the charge transport equations we have learned apply equally well to both electronic and ionic conduction. Now, however, their paths will diverge. Because electronic and ionic conduction mechanisms are vastly different, electronic and ionic conductivities are also quite different. CONDUCTIVITY AND MOBILITY The difference between conductivity and mobility can be understood by an analogy. Pretend that we are studying the transport of people (in cars) down an interstate highway. Mobility describes how fast the cars are driving down the highway. Conductivity, however, would also include information about how many cars are on the highway and how many people each car can hold. This analogy is not perfect but may help keep the two terms straight. 4.4.1 Electronic versus Ionic Conductors Differences in the fundamental nature of electrons versus ions lead to differences in the mechanisms for electronic versus ionic conduction. Figure 4.8 schematically contrasts a typical electronic conductor (a metal) and a typical ionic conductor (a solid electrolyte). Figure 4.8a illustrates the free-electron model of a metallic electron conductor. In this model, the valence electrons associated with the atoms of the metal become detached from the atomic lattice and are free to move about the metal. Meanwhile, the metal ions remain intact and immobile. The free valence electrons constitute a “sea” of mobile charges, which are able to move in response to an applied field. By contrast, Figure 4.8b illustrates the hopping model of a solid-state ionic conductor. The crystalline lattice of this ion conductor consists of both positive and negative ions, all of which are fixed to specific crystallographic positions. Occasionally, defects such as missing atoms (“vacancies”) or extra atoms (“interstitials”) will occur in the material. Charge transport is accomplished by the site-to-site “hopping” of these defects through the material. The structural differences between the two kinds of conductors lead to dramatic differences in carrier concentrations. In a metal, free electrons are populous, while carriers in a crystalline solid electrolyte are rare. The differences in the charge transport mechanisms, as illustrated in Figure 4.8, also lead to dramatic differences in carrier mobility. Combined, the differences in carrier concentration and carrier mobility lead to a very different picture for electron conductivity in a metal versus ion conductivity in a solid electrolyte. Let us take a closer look. 129 130 FUEL CELL CHARGE TRANSPORT e– M+ e– e– M+ M+ e– M+ M+ M+ e– e– M+ e– M+ e– M+ e– e– M+ e– e– M+ M+ e– e– e– M+ e– M+ e– M+ M+ e– M+ e– M+ e– e– (a) A– C+ A– C+ A– C+ A– C+ A– C+ A– C+ C+ A– A– C+ A– C+ A– C+ A– C+ A– C+ A– C+ A– C+ A– A– C+ C+ A– C+ Vacancy A– C+ Interstitial (b) Figure 4.8. Illustration of charge transport mechanisms. (a) Electron transport in a free-electron metal. Valence electrons detach from immobile metal atom cores and move freely in response to an applied field. Their velocity is limited by scattering from the lattice. (b) Charge transport in this crystalline ionic conductor is accomplished by mobile anions, which “hop” from position to position within the lattice. The hopping process only occurs where lattice defects such as vacancies or interstitials are present. 4.4.2 Electron Conductivity in a Metal For a simple electron conductor, such as a metal, the Drude model predicts that the mobility of free electrons in the metal will be limited by scattering (from phonons, lattice imperfections, impurities, etc.): q๐ u= (4.20) m where ๐ gives the mean free time between scattering events, m is the mass of the electron (m = 9.11 × 10−31 kg), and q is the elementary electron charge in coulombs (q = 1.602 × 10−19 C). Inserting the results for electron mobility (Equation 4.20) into the expression for conductivity (Equation 4.19) gives |z F|c q๐ ๐= e e (4.21) m Carrier concentration in a metal may be calculated from the density of free electrons. In general, each metal atom will contribute approximately one free electron. Atomic packing PHYSICAL MEANING OF CONDUCTIVITY densities are generally on the order of 1028 atoms/m3 , which yields molar carrier concentrations on the order of 104 mol/m3 . Inserting typical numbers into Equation 4.21 allows us to calculate ballpark electronic conductivity values. The charge number on an electron is, of course, –1(|ze | = 1). Typical scattering times (in relatively pure metals) are 10−12 –10–14 s. Using ce ≈ 104 molโm3 yields typical electron conductivities for metals in the range of 106 –108 Ω–1 ⋅ cm–1 ). 4.4.3 Ion Conductivity in a Crystalline Solid Electrolyte The conduction hopping process illustrated in Figure 4.8b for a solid ion conductor leads to a very different expression for mobility than that used for a metallic electron conductor. Ion mobility for the material in Figure 4.8b is dependent on the rate at which ions can hop from position to position within the lattice. This hopping rate, like the reaction rates studied in the previous chapter, is exponentially activated. The effectiveness of the hopping process is characterized by the material’s diffusivity D: D = Do e−ΔGact โ(RT) (4.22) where Do is a constant reflecting the attempt frequency of the hopping process, ΔGact is the activation barrier for the hopping process, R is the gas constant, and T is the temperature (K). The overall mobility of ions in the solid electrolyte is then given by u= |zi |FD RT (4.23) Where |zi | is the charge number on the ion, F is Faraday’s constant, R is the gas constant, and T is the temperature (K). Inserting the expression for ion mobility (Equation 4.23) into the equation for conductivity (Equation 4.19) gives c(zi F)2 D (4.24) ๐= RT Carrier concentration in a crystalline electrolyte is controlled by the density of the mobile defect species. Most crystalline electrolytes conduct via a vacancy mechanism. These vacancies are intentionally introduced into the lattice by doping. Maximum effective vacancy doping levels are around 8–10%, leading to carrier concentrations of 102 –103 molโm3 . Typical ion diffusivities are on the order of 10–8 m2 โs for liquid and polymer electrolytes at room temperature, and on the order of 10–11 m2 โs for ceramic electrolytes at 700–1000โ C. Typical ion carrier concentrations are 103 –104 molโm3 for liquid electrolytes, 102 –103 molโm3 for polymer electrolytes, and 102 –103 molโm3 for ceramic electrolytes at 700–1000โ C. Inserting these values into Equation 4.24 yields ionic conductivity values of 10−4 –102 Ω–1 ⋅ m−1 (10−6 − 100 Ω–1 ⋅ cm−1 ). Note that solid-electrolyte ionic conductivity values are well below electronic conductivity values for metals. As has been previously stated, ionic charge transport tends to be far more difficult than electronic charge transport. Therefore, much of the focus in fuel cell research is placed on finding better electrolytes. 131 132 FUEL CELL CHARGE TRANSPORT 4.5 REVIEW OF FUEL CELL ELECTROLYTE CLASSES The search for better electrolytes has led to the development of three major candidate materials classes for fuel cells: aqueous, polymer, and ceramic electrolytes. Regardless of the class, however, any fuel cell electrolyte must meet the following requirements: • • • • • • High ionic conductivity Low electronic conductivity High stability (in both oxidizing and reducing environments) Low fuel crossover Reasonable mechanical strength (if solid) Ease of manufacturability Other than the high-conductivity requirement, the electrolyte stability requirement is often the hardest to fulfill. It is difficult to find an electrolyte that is stable in both the highly reducing environment of the anode and the highly oxidizing environment of the cathode. 4.5.1 Ionic Conduction in Aqueous Electrolytes/Ionic Liquids In this section, we discuss ionic conduction in aqueous electrolytes and ionic liquids. An aqueous electrolyte is a water-based solution containing dissolved ions that can transport charge. An ionic liquid is a material which is itself simultaneously liquid and ionic. Sodium chloride dissolved in water is an example of an aqueous electrolyte. Upon dissolution in water, the NaCl separates into mobile Na+ ions and mobile Cl– ions, which can transport charge by moving through the water solvent. Molten NaCl (when heated to high temperature) is an example of an ionic liquid. Pure H3 PO4 at 50โ C is another example of an ionic liquid. At room temperature, H3 PO4 is a somewhat waxy, white crystalline solid. However, when heated above 42โ C it becomes a viscous ionic liquid consisting of H+ ions, PO4 3– ions, and H3 PO4 molecules. Almost all aqueous/liquid electrolyte fuel cells use a matrix material to support or immobilize the electrolyte. The matrix generally accomplishes three tasks: 1. Provides mechanical strength to the electrolyte 2. Minimizes the distance between the electrodes while preventing shorts 3. Prevents crossover of reactant gases through the electrolyte Reactant crossover, the last task on this list, is a particular problem for aqueous/liquid electrolytes (much more so than for solid electrolytes). In an unsupported liquid electrolyte, reactant gas crossover can be severe; in these situations, unbalanced-pressure or high-pressure operation is impossible. The use of a matrix material provides mechanical integrity and reduces gas crossover problems, while still permitting thin (0.1–1.0-mm) electrolytes. Alkaline fuel cells use concentrated aqueous KOH electrolytes, while phosphoric acid fuel cells use either concentrated aqueous H3 PO4 electrolytes or pure H3 PO4 (an ionic liquid). Molten carbonate fuel cells use molten (K/Li)2 CO3 immobilized in a supporting REVIEW OF FUEL CELL ELECTROLYTE CLASSES matrix. The (K/Li)2 CO3 material melts at around 450โ C to become a liquid (“molten”) electrolyte. (MCFCs must therefore obviously be operated above 450โ C.) Ionic conductivity in aqueous/liquid environments can best be approached using a driving force/frictional force balance model. In liquids, an ion will accelerate under the force of an electric field until frictional drag exactly counteracts the electric field force. The balance between the electric field and frictional drag determines the terminal velocity of the ion. The electric field force, FE , is given by FE = zi q dV dx (4.25) where zi is the charge number of the ion and q is the fundamental electron charge (1.6 × 10–19 C). Although we do not show the derivation here, the frictional drag force FD may be approximated from Stokes’s law as FD = 6๐๐rv (4.26) where ๐ is the viscosity of the liquid, r is the radius of the ion, and v is the velocity of the ion. Equating the two forces allows us to determine the mobility, ui , which is defined as the ratio between the applied electric field and the resulting ion velocity (because mobility is defined as a positive quantity, inclusion of the absolute value is again required): | ๐ฃ | | = |zi |q ui = || | 6๐๐r dVโdx | | (4.27) Thus, mobility is determined by the ion size and the liquid viscosity. Intuitively, this expression makes sense: Bulky ions or highly viscous liquids should lead to lower mobilities, while nonviscous liquids and small ions should yield higher mobilities. The mobilities of a variety of ions in aqueous solution are given in Table 4.2. Note that in aqueous solutions the H+ ion tends to be hydrated by one or more water molecules. This ionic species is therefore better thought of as H3 O+ or H ⋅ (H2 O)x + , where x represents the number of water molecules “hydrating” the proton. Recall our expression for conductivity (Equation 4.19), which is repeated here for clarity: (4.28) ๐i = (|zi |F)ci ui If the values of ion mobilities in Table 4.2 are inserted into this expression, the ionic conductivity of various aqueous electrolytes may be calculated. Unfortunately, these TABLE 4.2. Selected Ionic Mobilities at Infinite Dilution in Aqueous Solutions at 25โ C Cation Mobility, u (cm2 /V ⋅ s) Anion Mobility, u (cm2 /V ⋅ s) H+ (H3 O+ ) 3.63 × 10−3 OH− 2.05 × 10−3 K+ 7.62 × 10−4 Br− 8.13 × 10−4 Ag+ 6.40 × 10−4 I− 7.96 × 10−4 Na+ 5.19 × 10−4 Cl− 7.91 × 10−4 Li+ 4.01 × 10−4 HCO3 − 4.61 × 10−4 Source: From Ref. [6a]. 133 134 FUEL CELL CHARGE TRANSPORT calculations are only accurate for dilute aqueous solutions when the ion concentration is low. At high ion concentration (or for ionic liquids) strong electrical interactions between the ions make conductivity far more difficult to calculate. In general, the conductivity of highly concentrated aqueous solutions or pure ionic liquids will be much lower than that predicted by Equation 4.28. For example, the conductivity of pure H3 PO4 is experimentally determined to be 0.1–1.0 Ω−1 ⋅ cm−1 (depending on the temperature), whereas Equation 4.28 predicts that the conductivity of pure H3 PO4 should be approximately 18 Ω−1 ⋅ cm−1 . Table 4.2 does offer some other useful insights. For example, it explains why KOH is the electrolyte of choice in alkaline fuel cells. Besides being extremely inexpensive, KOH exhibits the highest ionic conductivity of any of the hydroxide compounds. (Compare the u value for K+ to other candidate hydroxide cations such as Na+ or Li+ .) In alkaline fuel cells, fairly concentrated (30–65%) solutions of KOH are used, resulting in conductivities on the order of 0.1–0.5 Ω−1 ⋅ cm−1 . How much would the conductivity be reduced if a far more dilute electrolyte was used? To get an answer, refer to Example 4.3, where the approximate conductivity of a 0.1 M KOH electrolyte solution is calculated using Equation 4.28. Example 4.3 Calculate the approximate conductivity of a 0.1 M aqueous solution of KOH. Solution: We use Equation 4.28 as our guide. Assuming that 0.1 M KOH completely dissolves into K+ ions and OH– ions (it does), the concentration of K+ and OH– will also be 0.1 M. Converting these concentrations to units of moles per cubic centimeter gives cK+ = (0.1 molโL)(1 Lโ1000 cm3 ) = 1 × 10−4 molโcm3 (4.29) cOH− = (0.1 molโL)(1 Lโ1000 cm3 ) = 1 × 10−4 molโcm3 The mobilities of K+ and OH– are given in Table 4.2. Inserting these numbers into Equation 4.28 yields ๐K+ = (1)(96, 485)(1 × 10−4 molโcm3 )(7.62 × 10−4 cm2 โV ⋅ s) = 0.0073 Ω−1 ⋅ cm−1 ๐OH− = (1)(96, 485)(1 × 10−4 molโcm3 )(2.05 × 10−3 cm2 โV ⋅ s) = 0.0198 Ω−1 ⋅ cm−1 (4.30) The total ionic conductivity of the electrolyte is then given by the sum of the cation and anion conductivities: ๐total = ๐K+ + ๐OH− = 0.0073 + 0.0198 = 0.0271 Ω−1 ⋅ cm−1 (4.31) In reality, the conductivity of the 0.1 M KOH solution will likely be a little lower than this predicted value. Note that most of the conductivity is provided by the OH– ion, rather than the K+ ion. This is due to the higher mobility of the OH– ion. REVIEW OF FUEL CELL ELECTROLYTE CLASSES 4.5.2 Ionic Conduction in Polymer Electrolytes In general, ionic transport in polymer electrolytes follows the exponential relationship described by Equations 4.22 and 4.24. By combining these two equations, we can obtain (see problem 4.11) (4.32) ๐T = APEM e−Ea โkT where APEM is a preexponential factor and Ea represents the activation energy (eV/atom) (Ea = ΔGact โF, where F is Faraday’s constant). As this equation indicates, conductivity increases exponentially with increasing temperature. Most polymer and crystalline ion conductors obey this model quite well. For a polymer to be a good ion conductor, at a minimum it should possess the following structural properties: 1. The presence of fixed charge sites 2. The presence of free volume (“open space”) The fixed charge sites should be of opposite charge compared to the moving ions, ensuring that the net charge balance across the polymer is maintained. The fixed charge sites provide temporary centers where the moving ions can be accepted or released. In a polymer structure, maximizing the concentration of these charge sites is critical to ensuring high conductivity. However, excessive addition of ionically charged side chains will significantly degrade the mechanical stability of the polymer, making it unsuitable for fuel cell use. Free volume correlates with the spatial organization of the polymer. In general, a typical polymer structure is not fully dense. Small-pore structures (or free volumes) will almost always exist. Free volume improves the ability of ions to move across the polymer. Increasing the polymer free volume increases the range of small-scale structural vibrations and motions within the polymer. These motions can result in the physical transfer of ions from site to site across the polymer. (See Figure 4.9.) Because of these free-volume effects, polymer membranes exhibit relatively high ionic conductivities compared to other solid-state ion-conducting materials (such as ceramics). Polymer free volume also leads to another well-known transport mechanism, known as the vehicle mechanism. In the vehicle mechanism, ions are transported through free-volume – – + – – – Charged site – – – – – – –– – – + Ion + – – Polymer chain Figure 4.9. Schematic of ion transport between polymer chains. Polymer segments can move or vibrate in the free volume, thus inducing physical transfer of ions from one charged site to another. 135 136 FUEL CELL CHARGE TRANSPORT spaces by hitching a ride on certain free species (the “vehicles”) as these vehicles pass by. Water is a common vehicular species; as water molecules move through the free volumes in a polymer membrane, ions can go along for the ride. In this case, the conduction behavior of the ions in the polymer electrolyte is much like that in an aqueous electrolyte. Persulfonated polytetrafluoroethylene (PTFE)—more commonly known as Nafion—exhibits extremely high proton conductivity based on the vehicle mechanism. Since Nafion is the most popular and important electrolyte for PEMFC applications, we review its properties in the next section. Ionic Transport in Nafion. Nafion has a backbone structure similar to polytetrafluoroethylene (Teflon). However, unlike Teflon, Nafion includes sulfonic acid (SO3 – H+ ) functional groups. The Teflon backbone provides mechanical strength while the sulfonic acid (SO3 – H+ ) chains provide charge sites for proton transport. Figure 4.10 illustrates the structure of Nafion. It is believed that Nafion free volumes aggregate into interconnected nanometer-sized pores whose walls are lined by sulfonic acid (SO3 – H+ ) groups. In the presence of water, the protons (H+ ) in the pores form hydronium complexes (H3 O+ ) and detach from the sulfonic acid side chains. When sufficient water exists in the pores, the hydronium ions can transport in the aqueous phase. Under these circumstances, ionic conduction in Nafion is similar to conduction in liquid electrolytes (Section 4.5.1). As a bonus, the hydrophobic nature of the Teflon backbone further accelerates water transport through the membrane, since the hydrophobic pore surfaces tend to repel water. Because of these factors, Nafion exhibits proton conductivity comparable to that of a liquid electrolyte. To maintain this extraordinary conductivity, Nafion must be fully hydrated with liquid water. Usually, hydration is achieved by humidifying the fuel and oxidant gases provisioned to the fuel cell. In the following paragraphs, we review the key properties of Nafion in more detail.1 Nafion Absorbs Significant Amounts of Water. The pore structure in Nafion can hold significant amounts of water. In fact, Nafion can accommodate so much water that its volume will increase up to 22% when fully hydrated. (Strongly polar liquids, such as alcohols, can cause Nafion to swell up to 88%!) Since conductivity and water content are strongly related, determining water content is essential to determining the conductivity of the membrane. The water content λ in Nafion is defined as the ratio of the number of water molecules to the number of charged (SO3 – H+ ) sites. Experimental results suggest that λ can vary from almost 0 (for completely dehydrated Nafion) to 22 (for full saturation, under certain conditions). For fuel cells, experimental measurements have related the water content in Nafion to the humidity condition of the fuel cell, as shown in Figure 4.11. Thus, if the humidity condition of the fuel cell is known, the water content in the membrane can be estimated. Humidity in Figure 4.11 is quantified by water vapor activity a๐ค (essentially relative humidity): p a๐ค = ๐ค (4.33) pSAT 1 The Nafion model reviewed here was suggested by Springer et al. [8] REVIEW OF FUEL CELL ELECTROLYTE CLASSES Nafion Polytetraflouroethylene (PTFE) F F F F F F F F C C C C C C C C F F F F F O F F F C F F C CF3 n n O F C F C = O= S n m F F O– H+ O (a) H2O + H 3O – SO3 1nm (b) Figure 4.10. (a) Chemical structure of Nafion. Nafion has a PTFE backbone for mechanical stability with sulfonic groups to promote proton conduction. (b) Schematic microscopic view of proton conduction in Nafion. When hydrated, nanometer-sized pores swell and become largely interconnected. Protons bind with water molecules to form hydronium complexes. Sulfonic groups near the pore walls enable hydronium conduction. where p๐ค represents the actual partial pressure of water vapor in the system and pSAT represents the saturation water vapor pressure for the system at the temperature of operation. The data in Figure 4.11 can be represented mathematically as { λ = 0.043 + 17.18a๐ค − 39.85a2๐ค + 36.0a3๐ค ) ( 14 + 4 a๐ค − 1 for 0 < a๐ค ≤ 1 for 1 < a๐ค ≤ 3 (4.34) 137 FUEL CELL CHARGE TRANSPORT 14 12 10 − λ = H2O/SO3 138 8 6 4 2 0 0 0.2 0.4 0.6 0.8 1 Water vapor activity ( pw /pSAT) Figure 4.11. Water content versus water activity for Nafion 117 at 303 K (30โ C) according to Equation 4.34. Water vapor activity is defined as the ratio of the actual water vapor pressure (p๐ค ) for the system compared to the saturation water vapor pressure (pSAT ) for the system at the temperature of interest. Reprinted with permission from Ref. [8], Journal of the Electrochemical Society, 138: 2334, 1991. Copyright 1991 by the Electrochemical Society. Equation 4.34 does not consider the effects of temperature; however, it is reasonably accurate for PEMFCs operating near 80โ C. WATER VAPOR SATURATION PRESSURE When the partial pressure of water vapor (p๐ค ) within a gas stream reaches the water vapor saturation pressure pSAT for a given temperature, the water vapor will start to condense, generating water droplets. In other words, relative humidity is 100% when p๐ค = pSAT . Importantly, pSAT is a strong function of temperature: log10 pSAT = −2.1794 + 0.02953T − 9.1837 × 10−5 T 2 + 1.4454 × 10−7 T 3 (4.35) where pSAT is given in bars (1 bar = 100,000 Pa) and T is the temperature in degrees Celsius. For example, if fully humidified air at 80โ C and 3 atm is provided to a fuel cell, the water vapor pressure is [9] −5 ×802 +1.4454×10−7 ×803 pSAT = 10−2.1794+0.02953×80−9.1837×10 = 0.4669 bar (4.36) This gives the mole fraction of water in fully humidified air at 80โ C and 3 atm as 0.4669 bar/3 atm = 0.4669 bar/(3 × 1.0132501 bar) = 0.154 assuming an ideal gas. REVIEW OF FUEL CELL ELECTROLYTE CLASSES Under these same conditions, if the air is instead only partially humidified, such that the water mole fraction is 0.1, then the water vapor activity (or relative humidity) would be (again assuming an ideal gas) a๐ค = pH2 O๐ค pSAT = xH2 O × ptotal xH2 O,SAT × ptotal = 0.1 = 0.65 0.154 (4.37) Nafion Conductivity Is Highly Dependent on Water Content. As previously mentioned, conductivity and water content are strongly related in Nafion. Conductivity and temperature are also strongly related. In general, the proton conductivity of Nafion increases linearly with increasing water content and exponentially with increasing temperature, as shown by the experimental data in Figures 4.12 and 4.13. In equation form, these experimentally determined relationships may be summarized as )] [ ( 1 1 ๐(T, λ) = ๐303K (λ) exp 1268 − 303 T (4.38) ๐303K (λ) = 0.005193λ − 0.00326 (4.39) where where ๐ represents the conductivity (S/cm) of the membrane and T (K) is the temperature. Since the conductivity of Nafion can change locally depending on water content, the total area-specific resistance of a membrane is found by integrating the local resistivity over the 0.12 0.1 σ (S/cm) 0.08 0.06 0.04 0.02 0 0 5 10 15 20 25 λ = H O/SO 2 3 Figure 4.12. Ionic conductivity of Nafion versus water content λ according to Equations 4.38 and 4.39 at 303 K. 139 FUEL CELL CHARGE TRANSPORT 100หC 50หC 0หC –0.6 –0.7 log(σ) [log(S/cm)] 140 –0.8 –0.9 –1 –1.1 –1.2 –1.3 2.6 2.8 3 3.2 3.4 3.6 3.8 3 1/T (x10 K) Figure 4.13. Ionic conductivity of Nafion versus temperature according to Equation 4.38 when λ = 22. membrane thickness (tm ) as ASRm = ∫0 tm ๐(z)dz = ∫0 tm dz ๐[λ(z)] (4.40) Protons Drag Water with Them. Since conductivity in Nafion is dependent on water content, it is essential to know how water content varies across a Nafion membrane. During fuel cell operation, the water content across a Nafion membrane is generally not uniform. Water content varies across a Nafion membrane because of several factors. Perhaps most important is the fact that protons2 traveling through the pores of Nafion generally drag one or more water molecules along with them. This well-known phenomenon is called electro-osmotic drag. The degree to which proton movement causes water movement is quantified by the electro-osmotic drag coefficient ndrag , which is defined as the number of water molecules accompanying the movement of each proton (ndrag = nH2 O โH+ ). Obviously, how much water is dragged per proton depends on how much water exists in the Nafion membrane in the first place. It has been measured that ndrag = 2.5 ± 0.2 (between 30 and 50โ C) in fully hydrated Nafion (when λ = 22). When λ = 11, ndrag = ∼ 0.9. Commonly, it is assumed that ndrag changes linearly with λ as ndrag = nSAT drag λ 22 for 0 ≤ λ ≤ 22 (4.41) 2 Actually, protons travel in the form of hydronium complexes as explained in the text. For simplicity, however, we use the term “proton” in these discussions. Also, it is more straightforward to define the electro-osmotic drag coefficient in terms of the number of water molecules per proton (rather than per hydronium, which contains a water molecule already). REVIEW OF FUEL CELL ELECTROLYTE CLASSES where nSAT ≈ 2.5. Knowledge of the electro-osmotic drag coefficient allows us to estimate drag the water drag flux from anode to cathode when a net current j flows through the PEMFC: JH2 O,drag = 2ndrag j 2F (4.42) where J is the molar flux of water due to electro-osmotic drag (mol/cm2 ), j is the operating current density of the fuel cell (A/cm2 ), and the quantity 2F converts from current density to hydrogen flux. The factor of 2 in the front of the equation then converts from hydrogen flux to proton flux. As you will see in Chapter 6, the drag coefficient becomes very important in modeling the behavior of Nafion membranes in PEMFCs. Back Diffusion of Water. In a PEMFC, electro-osmotic water drag moves water from the anode to the cathode. As this water builds up at the cathode, however, back diffusion occurs, resulting in the transport of water from the cathode back to the anode. This back-diffusion phenomenon occurs because the concentration of water at the cathode is generally far higher than the concentration of water at the anode (exacerbated by the fact that water is produced at the cathode by the electrochemical reaction). Back diffusion counterbalances the effects of electro-osmotic drag. Driven by the anode/cathode water concentration gradient, the water back-diffusion flux can be determined by JH2 O,back diffusion =− ๐dry Mm Dλ dλ dz (4.43) where ๐dry is the dry density (kg/m3 ) of Nafion, Mm is the Nafion equivalent weight (kg/mol), and z is the direction through the membrane thickness. The key factor in this equation is the diffusivity of water in the Nafion membrane (Dλ ). Unfortunately, Dλ is not constant but is a function of water content λ. Since the total water flux in Nafion is simply the addition of electro-osmotic drag and back diffusion, we have JH2 O = 2nSAT drag ๐dry j λ dλ D (λ) − 2F 22 Mm λ dz (4.44) This combined expression makes it explicitly clear that the water flux in Nafion is a complex function of λ. [We state the water diffusivity as Dλ (λ) in this equation to emphasize its dependency on water content.] Summary. Based on the fuel cell operating conditions (humidity and current density), we can estimate the water content profile (λ(z)) in the membrane by using Equations 4.34 and 4.44. Once we have the water content profile, we can then calculate the ion conductivity of the membrane by using Equation 4.38. In this fashion, the ohmic losses in a PEMFC may be quantified. This procedure is demonstrated in Example 4.4. In Chapter 6 we will combine these equations with the other fuel cell loss terms to create a complete PEMFC model. 141 142 FUEL CELL CHARGE TRANSPORT Example 4.4 Consider a hydrogen PEMFC powering an external load at 0.7 A/cm2 . The activities of water vapor on the anode and cathode sides of the membrane are measured to be 0.8 and 1.0, respectively. The temperature of the fuel cell is 80โ C. If the Nafion membrane thickness is 0.125 mm, estimate the ohmic overvoltage loss across the membrane. Solution: We can convert the water activity on the Nafion surfaces to water contents using Equation 4.34: λA = 0.043 + 17.18 × 0.8 − 39.85 × 0.82 + 36.0 × 0.83 = 7.2 λC = 0.043 + 17.18 × 1.0 − 39.85 × 1.02 + 36.0 × 1.03 = 14.0 (4.45) With these values as boundary conditions, we then solve Equation 4.44. In this equation, we have two unknowns, JH2 O and λ. For convenience, we will set JH2 O = ๐ผNH2 = ๐ผ(jโ2F), where ๐ผ is an unknown that denotes the ratio of water flux to hydrogen flux. After rearrangement, Equation 4.44 becomes ( ) jM λ dλ m = 2nSAT − ๐ผ drag 22 dz 2F๐dry Dλ (4.46) EQUIVALENT WEIGHT The equivalent weight of a species is defined by its atomic weight or formula weight divided by its valence: Equivalent weight = atomic (formula) weight valence (4.47) Valence is defined by the number of electrons that the species can donate or accept. For example, hydrogen has a valence of 1 (H+ ). Oxygen has a valence of 2 (O2– ). Thus, hydrogen has an equivalent weight of 1.008 gโmolโ1 = 1.008 gโmol and oxygen has an equivalent weight of 15.9994 gโmolโ2 = 7.9997 gโmol. In the case of sulfate radicals (SO4 2– ), the formula weight is (1 × 32.06) + (4 × 15.9994) = 96.058 gโmol. Thus, the equivalent weight is (96.058 gโmol)โ2 = 48.029 gโmol. The sulfonic group (SO3 – H+ ) in Nafion has a valence of 1, since it can accept only one proton. Thus, the equivalent weight of Nafion is equal to the average weight of the polymer chain structure that can accept one proton. This number is very useful since it facilitates the calculation of sulfonic charge (SO3 – ) concentration in Nafion as CSO− (molโm3 ) = 3 ๐dry (kgโm3 ) Mm (kgโmol) (4.48) where ๐dry is the dry density of Nafion (kg/m3 ) and Mm is the Nafion equivalent weight (kg/mol). REVIEW OF FUEL CELL ELECTROLYTE CLASSES In a similar fashion, water content, λ (H2 OโSO3 – ), can be converted to water concentration in Nafion as CH2 O (molโm3 ) = λ ๐dry (kgโm3 ) (4.49) M m (kgโmol) Typically, Nafion has an equivalent weight of around ∼ 1–1.1 kgโmol and a dry density of ∼ 1970 kgโm3 . Thus, the estimated charge density for Nafion would be CSO− (molโm3 ) = 3 1970 kgโm3 = 1970 molโm3 1 kgโmol (4.50) WATER DIFFUSIVITY IN NAFION As emphasized above, water diffusivity in Nafion (Dλ ) is a function of water content λ. Experimentally (using magnetic resonance techniques), this dependence has been measured as )] [ ( 1 1 − Dλ = exp 2416 303 T × (2.563 − 0.33λ + 0.0264λ2 − 0.000671λ3 ) × 10−6 for λ > 4 (cm2 โs) (4.51) The exponential part describes the temperature dependence, while the polynomial portion describes the λ dependence at the reference temperature of 303 K. This equation is only valid for λ > 4. For λ < 4, values extrapolated from Figure 4.14 (dotted line) should be used instead. Water diffusivity, Dλ (cm2/s) 4 x 10 −6 3.5 3 2.5 2 1.5 1 0.5 0 0 5 10 λ (H2O/SO3-) 15 Figure 4.14. Water diffusivity Dλ in Nafion versus water content λ at 303 K. 143 144 FUEL CELL CHARGE TRANSPORT Even though this is an ordinary differential equation on λ, we may not solve it analytically since Dλ is a function of λ. However, if we assume λ in the membrane changes from 7.2 to 14.0 according to the boundary conditions, we can see from Figure 4.14 that the water diffusivity is fairly constant over this range. If we assume an average value of λ = 10, we can estimate Dλ from Equation 4.51 as )] [ ( 1 1 − Dλ = 10−6 exp 2416 303 353 × (2.563 − 0.33 × 10 + 0.0264 × 102 − 0.000671 × 103 ) = 3.81 × 10−6 cm2 โs (4.52) Now we can evaluate Equation 4.46, yielding the analytical solution [ ] jMm nSAT drag 11๐ผ 11๐ผ λ(z) SAT + C exp z = 22 F ๐dry Dλ 2.5 ndrag [ ] ( ) 0.7 Aโcm2 × (1.0 kgโmol) × 2.5 + C exp z (22 × 96, 485 Cโmol) × (0.00197 kgโcm3 ) × (3.81 cm2 โs) = 4.4๐ผ + C exp(109.8z) (4.53) where z is in centimeters and C is a constant to be determined from the boundary conditions. If we set the anode side as z = 0, we have λ(0) = 7.2 and λ(0.0125) = 14 from Equation 4.45. Accordingly, Equation 4.53 becomes λ(z) = 4.4๐ผ + 2.30 exp(109.8z) where ๐ผ = 1.12 (4.54) Now we know that about 1.12 water molecules are dragged per each hydrogen (or in other words, about 0.56 water molecules per proton). Figure 4.15a shows the result of how ๐ varies across the membrane in this example. At the start of the problem, we assumed a constant Dλ for λ in the range of 7.2–14. We can confirm that this assumption is reasonable from the results of Figure 4.15. From Equations 4.38 and 4.54, we can determine the conductivity profile of the membrane: ๐(z) = {0.005193[4.4๐ผ + 2.30 exp(109.8z)] − 0.00326} )] [ ( 1 1 − × exp 1268 303 353 = 0.0404 + 0.0216 exp(109.8z) (4.55) REVIEW OF FUEL CELL ELECTROLYTE CLASSES Figure 4.15b shows the result. Finally, we can determine the area-specific resistance of the membrane using Equation 4.40: ∫0 tm dz = ๐[λ(z)] ∫0 0.0125 dz = 0.15 Ω ⋅ cm2 0.0404 + 0.0216 exp(109.8z) (4.56) Thus, the ohmic overvoltage due to the membrane resistance in this PEMFC is approximately ASRm = Vohm = j × ASRm = (0.7 Aโcm2 ) × (0.15 Ω ⋅ cm2 ) = 0.105 V (4.57) This section has focused exclusively on the details of Nafion. However, the conduction properties and characteristics of other polymer electrolyte alternatives are discussed in Chapter 9 for the interested reader. 4.5.3 Ionic Conduction in Ceramic Electrolytes This section explains the underlying physics of ion transport in SOFC electrolytes. As their name implies, SOFC electrolytes are solid, crystalline oxide materials that can conduct ions. The most popular SOFC electrolyte material is yttria-stabilized zirconia (YSZ). A typical YSZ electrolyte contains 8% yttria mixed with zirconia. What is the meaning of zirconia and yttria? Zirconia is related to the metal zirconium, and yttria derives its name from another metal, yttrium. Zirconia has the chemical composition ZrO2 ; it is the oxide of zirconium. By analogy, yttria, or Y2 O3 , is the oxide of yttrium. A mixture of zirconia and yttria is called yttria-stabilized zirconia because the yttria stabilizes the zirconia crystal structure in the cubic phase (where it is most conductive). Even more importantly, however, the yttria introduces high concentrations of oxygen vacancies into the zirconia crystal structure. This high oxygen vacancy concentration allows YSZ to exhibit high ion conductivity. Adding yttria to zirconia introduces oxygen vacancies due to charge compensation effects. Pure ZrO2 forms an ionic lattice consisting of Zr4+ ions and O2– ions, as shown in Figure 4.16a. Addition of Y3+ ions to this lattice upsets the charge balance. As shown in Figure 4.16b, for every two Y3+ ions taking the place of Zr4+ ions, one oxygen vacancy is created to maintain overall charge neutrality. The addition of 8% (molar) yttria to zirconia causes about 4% of the oxygen sites to be vacant. At elevated temperatures, these oxygen vacancies facilitate the transport of oxygen ions in the lattice, as shown in Figure 4.8b. As discussed in Section 4.4, a material’s conductivity is determined by the combination of carrier concentration (c) and carrier mobility (u): ๐ = (|z|F)cu (4.58) In the case of YSZ, carrier concentration is determined by the strength of the yttria doping. Because a vacancy is required for ionic motion to occur within the YSZ lattice, the 145 FUEL CELL CHARGE TRANSPORT 15 14 −) Water content λ (HO/SO 2 3 13 12 11 10 9 8 7 0 0.002 Anode 0.004 0.006 0.008 Membrane thickness(cm) 0.01 0.012 Cathode (a) 0.13 0.12 Local conductivity (S/cm) 146 0.11 0.1 0.09 0.08 0.07 0.06 0 0.002 Anode 0.004 0.006 0.008 Membrane thickness(cm) 0.01 0.012 Cathode (b) Figure 4.15. Calculated properties of Nafion membrane for Example 4.4. (a) Water content profile across Nafion membrane. (b) Local conductivity profile across Nafion membrane. REVIEW OF FUEL CELL ELECTROLYTE CLASSES Vacancy Zr 4+ Zr 4+ Zr 4+ O 2– O2– O 2– Zr 4+ O2– O2– Zr 4+ O2– Zr 4+ Zr 4+ Zr 4+ O2– O2– O2– Zr 4+ Zr 4+ Zr 4+ Y 3+ O2– O2– O2– O2– O2– O2– ห Zr 4+ Zr 4+ O2– O2– O2– Zr 4+ Zr 4+ Zr 4+ Zr 4+ O 2– Y 3+ (a) O2– Zr 4+ O2– O2– Zr 4+ Zr 4+ Zr 4+ O2– Zr 4+ (b) Figure 4.16. View of the (110) plane in (a) pure ZrO2 and (b) YSZ. Charge compensation effects in YSZ lead to creation of oxygen vacancies. One oxygen vacancy is created for every two yttrium atoms doped into the lattice. oxygen vacancies can be considered to be the ionic charge “carriers.” Increasing the yttria content will result in increased oxygen vacancy concentration, improving the conductivity. Unfortunately, however, there is an upper limit to doping. Above a certain dopant or vacancy concentration, defects start to interact with each other, reducing their ability to move. Above this concentration, further doping is counterproductive and conductivity actually decreases. Plots of conductivity versus dopant concentration show a maximum at the point where defect interaction or “association” commences. For YSZ, this maximum occurs at about 8% molar yttria concentration. (See Figure 4.17.) log(σT ) (Ω–1 · cm–1 K) 2.4 2.3 2.2 2.1 2 1.9 1.8 1.7 1.6 6 7 8 9 10 11 12 13 14 15 %Y2O3 Figure 4.17. YSZ conductivity versus %Y2 O3 (molar basis) [10]; YSZ conductivity is displayed as σ(Ω–1 ⋅ cm–1 )times T (K). In the next section, Figure 4.18 will clarify why it is convenient to multiply ๐ with T. 147 148 FUEL CELL CHARGE TRANSPORT The complete expression for conductivity combines carrier concentration and carrier mobility, as described in Section 4.4.3: ๐= c(zF)2 D RT (4.59) where carrier mobility is described by D, the diffusivity of the carrier in the crystal lattice. Diffusivity describes the ability of a carrier to move, or diffuse, from site to site within a crystal lattice. High diffusivities translate into high conductivities because the carriers are able to move quickly through the crystal. The atomic origins and physical explanation behind diffusivity will be detailed in forthcoming sections. For now, however, it is sufficient to know that carrier diffusivity in SOFC electrolytes is exponentially temperature dependent: (4.60) D = D0 e−ΔGact โ(RT) where D0 is a constant (cm2 /s), ΔGact is the activation barrier for the diffusion process (J/mol), R is the gas constant, and T is the temperature (K). Combining Equations 4.59 and 4.61 provides a complete expression for conductivity in SOFC electrolytes: ๐= c(zF)2 D0 e−ΔGact โ(RT) RT (4.61) INTRINSIC CARRIERS VERSUS EXTRINSIC CARRIERS In YSZ and most other SOFC electrolytes, dopants are used to intentionally create high vacancy (or other charge carrier) concentrations. These carriers are known as extrinsic carriers because their presence is extrinsically created by intentional doping. However, any crystal, even an undoped one, will have at least some natural carrier population. These natural charge carriers are referred to as intrinsic carriers because they occur intrinsically due to the natural energetics of the crystal. Intrinsic carriers exist because no crystal is perfect (unless it is at absolute zero). All crystals will contain “mistakes” such as vacancies that can act as charge carriers for conduction. These mistakes are actually energetically favorable, because they increase the entropy of the crystal. (Recall Section 2.1.4.) For the case of vacancies, an energy balance may be developed that considers the enthalpy cost to create the vacancies versus the entropy benefit they deliver. Solving for this balance results in the following expression for intrinsic vacancy concentration as a function of temperature in an ionic crystal: xV ≈ e−Δh๐ฃ โ(2kT) (4.62) where xV represents the fractional vacancy concentration (expressed as the fraction of lattice sites of the species of interest that are vacant), Δh๐ฃ is the formation enthalpy for REVIEW OF FUEL CELL ELECTROLYTE CLASSES the vacancy in electron-volts (in other words, the enthalpy cost to “create” a vacancy), k is Boltzmann’s constant, and T is the temperature in Kelvin. This expression states that the intrinsic concentration of vacancies within a crystal increases exponentially with temperature. However, since Δh๐ฃ is typically on the order of 1 eV or larger, intrinsic vacancy concentrations are generally quite low, even at high temperatures. At 800โ C, the intrinsic vacancy concentration in pure ZrO2 is around 0.001, or about one vacancy per 1000 sites. Compare this to extrinsically doped crystal structures, which can attain vacancy concentrations as high as 0.1, or about one vacancy per 10 sites. This equation can be further refined depending on whether the charge carriers are extrinsic or intrinsic: • For extrinsic carriers, c is determined by the doping chemistry of the electrolyte. In this case, c is a constant and Equation 4.62 can be used as is. • For intrinsic carriers, c is exponentially dependent on temperature, and Equation 4.62 must be modified as follows: ๐= csites (zF)2 D0 e−Δh๐ฃ โ(2kT) e−ΔGact โ(RT) RT (4.63) where csites stands for the concentration of lattice sites for the species of interest in the material (moles of sites/cm3 ). Almost all useful fuel cell electrolyte materials are purposely doped to increase the number of charge carriers, and therefore the concentration of intrinsic carriers is usually insignificant compared to the concentration of extrinsic carriers (see text box on previous page). Thus, Equation 4.62 is far more important than Equation 4.63 for describing ionic conduction in practical electrolytes. Equation 4.62 is often simplified to a pseudo-empirical expression by lumping the various preexponential terms into a single factor, yielding ๐T = ASOFC e−ΔGact โRT (4.64) Similarly to Equation 4.32, the term ΔGact โRT can instead be written as Ea โkT, yielding ๐T = ASOFC e−Ea โkT (4.65) Experimental observations confirm the relationship described by Equation 4.64 (or 4.65). Figure 4.18 shows experimental plots of log(๐T) versus 1โT for both YSZ and gadolinia-doped ceria (GDC, another candidate SOFC electrolyte). The multiplication of ๐ with T ensures that the slopes in these plots are indicative of the activation energy for ion migration, ΔGact . The size of ΔGact is often critical for determining the conductivity 149 FUEL CELL CHARGE TRANSPORT 4 3 log(σT ) (Ω–1 · cm–1 K) 150 2 1 ΔGact=0.60eV 0 –1 –2 Gd-doped ceria Y-stabilized zirconia –3 –4 0.6 0.8 1.0 1.2 ΔGact=0.89eV 1.4 1000/T K 1.6 1.8 2.0 –1 Figure 4.18. Conductivity of YSZ and GDC electrolytes versus temperature. of SOFC electrolytes. Typically, its value ranges between about 50,000 and 120,000 J/mol (0.5–1.2 eV). Further details on specific fuel cell electrolyte materials properties, including a more in-depth discussion on YSZ and GDC, are provided in Chapter 9. CALCULATING EXTRINSIC DEFECT CONCENTRATIONS IN CRYSTALLINE CERAMIC MATERIALS As was pointed out earlier in this chapter, almost all useful ceramic fuel cell electrolyte materials are purposely doped to increase the number of charge carriers, and therefore extrinsically created carriers dominate the conduction process. In order to calculate the concentration of the extrinsically created charge carriers (c), which is needed in Equation 4.62, information about the material composition, the doping concentration, and the crystal structure or density is required. As an example, consider the classic case of 8YSZ, which is zirconia doped with 8 mol% yttria. As shown in Figure 4.16, for every 2 Y that are substituted into the ZrO2 lattice, one oxygen vacancy is created. These extrinsically created oxygen vacancies become the source of ionic conduction in this material. To create 8YSZ, 8 mol % Y2 O3 is combined with 92 mol % ZrO2 . The chemical formula of 8YSZ can therefore be represented as 0.92(ZrO2 ) + 0.08(Y2 O3 ) = Zr0.92 Y0.16 O2.08 . Because of the 2-to-1 relationship between Y dopants and the created oxygen vacancies, the number of oxygen vacancies can be explicitly shown by writing the formula as Zr0.92 Y0.16 O2.08 V0.08 . One REVIEW OF FUEL CELL ELECTROLYTE CLASSES mole of this material will therefore contain 0.08 mol of oxygen vacancies. The fraction of oxygen sites that are vacant, xv , is 0.08โ2.16 = 0.037. This vacancy fraction can be converted into a vacancy concentration (cv , units of vacancies/cm3 ) by applying knowledge about the molecular weight and density of the material or by applying knowledge about the molar volume of the material. If the density of the material is known, this information can be used to convert molar vacancy fraction to vacancy concentration as follows: no (4.66) V where co is the concentration of oxygen sites in the material (mol/cm3 ), no is the moles of oxygen atoms per mole of material, and V is the molar volume of the material (cm3 /mol). The molar volume can be calculated from the molecular weight (M, g/mol) and the density (๐, gโcm3 ) as M V= (4.67) ๐ c๐ฃ = x๐ฃ co = x๐ฃ For 8YSZ, ๐ = 6.15 gโcm3 and M = (91.22 gโmol × 0.92 + 88.9 gโmol × 0.16 + 16 gโmol × 2.08) = 131.4 gโmol. Thus V= 131.4 gโmol = 21.4 cm3 6.15 gโcm3 ( c๐ฃ = 0.037 vacanciesโO site 2.16 mol O sitesโmol YSZ ( ) 21.4 cm3 โmol YSZ (4.68) ) (4.69) 3 = 0.0037 mol vacanciesโcm If the lattice constant and crystal structure of the material are known, this information can be used to convert vacancy fraction to vacancy concentration in an analogous fashion. In this case, the molar volume can be calculated from the unit cell information. For example, 8YSZ has the cubic (fluorite-type) structure with a lattice constant a = 5.15 Å and a total of four ZrO2 formula units per unit cell (e.g., four cations and eight anions). Based on this information the molar volume can be estimated as V= (5.15A)3 × (6.022 × 1023 ) = 20.5 cm3 4 (4.70) which is reasonably close to the density-based value calculated from Equation 4.68. From this point, the vacancy concentration, cv , can be calculated as before using Equation 4.69. 151 152 FUEL CELL CHARGE TRANSPORT 4.5.4 Mixed Ionic–Electronic Conductors So far, this chapter has focused almost exclusively on pure ionic conductors. These are materials that conduct charged ionic species but do not conduct electrons. Beyond the traditional classes of pure ionic conductors and pure electronic conductors, however, there are also interesting classes of materials that can conduct both ions and electrons. These materials are known as “mixed ionic–electronic conductors” (MIECs) or, more simply, “mixed conductors.” Many doped metal oxide ceramic materials exhibit both electronic and ionic conductivity. This is because doping can introduce both ionic defects (like oxygen vacancies) and electronic defects (like free electrons or free holes). Both the ionic and electronic defects can then “wander” through the material, leading to simultaneous ionic and electronic conductivity. If an oxide material is a mixed conductor, it is unsuitable for use as a fuel cell electrolyte (since the electronic conductivity would essentially “short” the fuel cell). However, MIECs are extremely attractive for SOFC electrode structures, because they can dramatically increase electrochemical reactivity and thereby improve fuel cell performance. Why do MIECs increase electrochemical activity? As you may recall from Chapter 3 (Section 3.11), fuel cell reactions can only occur where the electrolyte, electrode, and gas phases are all in contact. This requirement is expressed by the concept of the “triple-phase zone,” which refers to regions or points where the gas pores, electrode, and electrolyte phases converge (see Figure 3.14). In order to maximize the number of these three-phase zones, most fuel cell electrode–electrolyte interfaces employ a highly nanostructured geometry with significant intermixing, or blending, of the electrode and electrolyte phases (along with gas porosity). However, another strategy to increase the number of reaction zones is to employ a mixed-conductor electrode. Because a MIEC conducts both ions and electrons, it can simultaneously provide both the ionic species and the electrons needed for an electrochemical reaction. In this case, only one additional phase (the gas phase) is needed for electrochemical reaction. Thus, fuel cell reactions can occur anywhere along the entire surface of the MIEC where it is in contact with the gas phase. Figure 4.19 schematically illustrates the difference between a standard fuel cell electrode (Figure 4.19a) and a MIEC electrode (Figure 4.19b). As you can imagine, MIECs are scientifically fascinating materials. Most MIECs are ceramic materials and are therefore employed in SOFC electrodes—particularly as cathode electrode materials. In contrast, there is very little research on MIECs for low-temperature PEMFCs, but perhaps this will be an interesting area for future work. The prototypical MIEC is (La,Sr)MnO3 (LSM). LSM is used as the cathode electrode in many SOFC designs. In LSM, Sr2+ is substituted for La3+ as a dopant in order to create oxygen vacancies and holes. Due to the charge difference between La3+ and Sr2+ , either oxygen vacancies or electron holes must be created to maintain charge neutrality, as illustrated by the following defect reactions: Oxygen vacancy formation: Electron hole formation: ′ 2Oxo → 2SrLa + Vo− ′ null → 2SrLa + 2h⋅ MORE ON DIFFUSIVITY AND CONDUCTIVITY (OPTIONAL) Standard Electrode: Only TPBs are active for reaction MIEC Electrode: Entire surface is active for reaction e– e– O2 O2 O2 O2 O2– O2– Electrolyte O2– Electrolyte (a) (b) Figure 4.19. A standard SOFC cathode electrode (a) versus a mixed ionic–electronic conducting (MIEC) SOFC cathode electrode (b). In the first reaction, one oxygen vacancy (Vo⋅⋅ ) is formed for every two Sr2+ dopant substitutions. This process is identical to the vacancy creation process in YSZ (see Section 4.5.3). In the second reaction, two holes (h⋅ ) are formed for every two Sr2+ dopant substitutions. Under typical SOFC conditions, hole conduction in LSM is dominant compared to oxygen vacancy conduction. Therefore, LSM is only a marginal MIEC (i.e., for all intents and purposes it is almost exclusively a p-type electronic conductor). Nevertheless, its remarkable stability and compatibility with other SOFC materials make it a popular choice in many SOFC designs. Significant recent research has been conducted to develop better MIEC materials, and there are several other La-based perovskites that show increased ionic conductivity, and therefore better mixed-conduction behavior, compared to LSM. These materials include (La,Sr)(Co,Mn)O3 , (La,Sr)FeO3 , and (La,Sr)CoO3 . These materials tend to provide much higher ionic conductivity compared to LSM and therefore function as true mixed ionic–electronic conductors. Unfortunately, these materials also tend to be less stable than LSM and have therefore proven difficult to deploy in functional SOFC designs. Nevertheless, the electrochemical benefits of MIEC electrodes are substantial, and therefore MIEC development remains an extremely intriguing area of research. Further details on these materials are provided for the interested reader in Chapter 9. 4.6 MORE ON DIFFUSIVITY AND CONDUCTIVITY (OPTIONAL) In this optional section, we develop an atomistic picture to explore conductivity and diffusivity in more detail. We find that for conductors where charge transport involves a “hopping”-type mechanism, conductivity and diffusivity are intimately related. Diffusivity measures the intrinsic rate of this hopping process. Conductivity incorporates how this 153 FUEL CELL CHARGE TRANSPORT hopping process is modified by the presence of an electric field driving force. Diffusivity is therefore actually the more fundamental parameter. Diffusivity is a more fundamental parameter of atomic motion because even in the absence of any driving force, hopping of ions from site to site within the lattice still occurs at a rate that is characterized by the diffusivity. Of course, without a driving force, the net movement of ions is zero, but they are still exchanging lattice sites with one another. This is another example of a dynamic equilibrium; compare it to the exchange current density phenomenon that we learned about in Chapter 3. 4.6.1 Atomistic Origins of Diffusivity Using the schematic in Figure 4.20b, we can derive an atomistic picture of diffusivity. The atoms in this figure are arranged in a series of parallel atomic planes. We would like to calculate the net flux (net movement) of gray atoms from left to right across the imaginary plane labeled A in Figure 4.20 (which lies between two real atomic planes in the material). Examining atomic plane 1 in the figure, we assume that the flux of gray atoms hopping in the forward direction (and therefore through plane A) is simply determined by the number Concentration of gray atoms 154 Jnet Distance (x) (a) ΔX A JA+ JA– (c1) (c2) A (b) Figure 4.20. (a) Macroscopic picture of diffusion. (b) Atomistic view of diffusion. The net flux of gray atoms across an imaginary plane A in this crystalline lattice is given by the flux of gray atoms hopping from plane 1 to plane 2 minus the flux of gray atoms hopping from plane 2 to plane 1. Since there are more gray atoms on plane 1 than plane 2, there is a net flux of gray atoms from plane 1 to plane 2. This net flux will be proportional to the concentration difference of gray atoms between the two planes. MORE ON DIFFUSIVITY AND CONDUCTIVITY (OPTIONAL) (concentration) of gray atoms available to hop times the hopping rate: JA+ = 21 ๐ฃc1 Δx (4.71) where JA+ is the forward flux through plane A (mol/cm2 ⋅ s), v is the hopping rate (s–1 ), c1 is the volume concentration (mol/cm3 ) of gray atoms in plane 1, Δx (cm) is the atomic spacing required to convert volume concentration to planar concentration (mol/cm2 ), and the 1/2 accounts for the fact that on average only half of the jumps will be “forward” jumps. (On average, half of the jumps will be to the left, half of the jumps will be to the right.) Similarly, the flux of gray atoms hopping from plane 2 backward through plane A will be given by JA− = 21 ๐ฃc2 Δx (4.72) where JA− is the backward flux through plane A and c2 is the volume concentration (mol/cm3 ) of gray atoms in plane 2. The net flux of gray atoms across plane A is therefore given by the difference between the forward and backward fluxes through plane A: Jnet = 21 ๐ฃΔx(c1 − c2 ) (4.73) We would like to make this expression look like our familiar equation for diffusion: J = −D(dcโdx) We can express Equation 4.73 in terms of a concentration gradient as (c1 − c2 ) Δx 1 2 Δc = − 2 ๐ฃ(Δx) Δx 1 2 dc = − 2 ๐ฃ(Δx) (for small x) dx Jnet = − 12 ๐ฃ(Δx)2 (4.74) Comparison with the traditional diffusion equation J = −D(dcโdx) allows us to identify what we call the diffusivity as (4.75) D = 12 ๐ฃ(Δx)2 We therefore recognize that the diffusivity embodies information about the intrinsic hopping rate for atoms in the material (v) and information about the atomic length scale (jump distance) associated with the material. As mentioned previously, the hopping rate embodied by v is exponentially activated. Consider Figure 4.21b, which shows the free-energy curve encountered by an atom as it hops from one lattice site to a neighboring lattice site. Because the two lattice sites are essentially equivalent, in the absence of a driving force a hopping atom will possess the same free energy in its initial and final positions. However, an activation barrier impedes the motion of the atom as it hops between positions. We might associate this energy barrier with the displacements that the atom causes as it squeezes through the crystal lattice between lattice sites. (See Figure 4.21a, which shows a physical picture of the hopping process.) 155 FUEL CELL CHARGE TRANSPORT (a) C+ C+ C+ C+ C+ – A + C (b) Free energy 156 โGact Distance Figure 4.21. Atomistic view of hopping process. (a) Physical picture of the hopping process. As the anion (A− ) hops from its original lattice site to an adjacent, vacant lattice site, it must squeeze through a tight spot in the crystal lattice. (b) Free-energy picture of the hopping process. The tight spot in the crystal lattice represents an energy barrier for the hopping process. In a treatment analogous to the reaction rate theory developed in the previous chapter, we can write the hopping rate as ๐ฃ = ๐ฃ0 e−ΔGact โ(RT) (4.76) where ΔGact is the activation barrier for the hopping process and v0 is the jump attempt frequency. Based on this activated model for diffusion, we can then write a complete expression for the diffusivity as (4.77) D = 12 (Δx)2 ๐ฃ0 e−ΔGact โ(RT) or, lumping all the preexponential constants into a D0 term. D = D0 e−ΔGact โ(RT) 4.6.2 (4.78) Relationship between Conductivity and Diffusivity (1) To understand how conductivity relates to diffusivity, we take a look at how an applied electric field will affect the hopping probabilities for diffusion. Consider Figure 4.22, which shows the effect of a linear voltage gradient on the activation barrier for the hopping process. From this picture, it is clear that the activation barrier for a “forward” hop is MORE ON DIFFUSIVITY AND CONDUCTIVITY (OPTIONAL) Free energy dV 1 zFโx dx 2 Voltage gradient dV zF dx โG’act zF โx 1 โx 2 dV dx โx Distance Figure 4.22. Effect of linear voltage gradient on activation barrier for hopping. The linear variation in voltage with distance causes a linear drop in free energy with distance. This reduces the forward activation barrier (ΔG′act < ΔGact ). Two adjacent lattice sites are separated by Δx; therefore, the total free-energy drop between them is given by zFΔx(dVโdx). If the activation barrier occurs halfway between the two lattice sites, ΔGact will be decreased by 21 zF Δx(dVโdx). [In other words, ΔG′act = ΔGact − 21 zF Δx(dVโdx).] reduced by 21 zF Δx(dVโdx) while the activation barrier for the “reverse” hop is increased by 12 zF Δx(dVโdx). (We are assuming that the activated state occurs exactly halfway between the two lattice positions, or in other words that ๐ผ = 12 .) The forward-(๐ฃ1 ) and reverse-(๐ฃ2 ) hopping-rate expressions are therefore ๐ฃ1 = ๐ฃ0 exp ] [ − ΔGact − 21 zF Δx (dVโdx) [ ๐ฃ2 = ๐ฃ0 exp − ΔGact + RT 1 zF Δx (dVโdx) 2 ] (4.79) RT This voltage gradient modification to the activation barrier turns out to be small. In fact, 1 2 zF dV Δx โช1 RT dx 157 158 FUEL CELL CHARGE TRANSPORT so we can use the approximation ex ≈ 1 + x for the second term in the exponentials. This allows us to rewrite the hopping rate expressions as ) ( zF dV Δx ๐ฃ1 ≈ ๐ฃ0 e−ΔGact โ(RT) 1 + 21 RT dx ) ( dV 1 zF −ΔGact โ(RT) ๐ฃ2 ≈ ๐ฃ0 e 1− 2 Δx RT dx (4.80) Proceeding as before, we can then write the net flux across an imaginary plane A in a material as (4.81) Jnet = JA+ − JA− = 21 Δx(c1 ๐ฃ1 − c2 ๐ฃ2 ) Since we are interested in conductivity this time, we would like to consider a flux that is driven purely by the potential gradient. In other words, we want to get rid of any effects of a concentration gradient by saying that c1 = c2 = c. Making this modification and inserting the formulas for v1 and v2 give ) czF dV Δx RT dx ) ( czF dV 1 = 2 (Δx)2 ๐ฃ0 e−ΔGact โ(RT) RT dx Jnet = 12 Δx ๐ฃ0 e−ΔGact โ(RT) ( (4.82) Recognizing the first group of terms as our diffusion coefficient D, we thus have Jnet = czFD dV RT dx (4.83) Comparing this to the conduction equation J= ๐ dV zF dx we see that ๐ and D are related by ๐= c(zF)2 D RT (4.84) For conductors that rely on a diffusive hopping-based charge transport mechanism, this important result relates the observed conductivity of the material to the atomistic diffusivity of the charge carriers. This equation is our key for understanding the atomistic underpinnings of ionic conductivity in crystalline materials. 4.6.3 Relationship between Diffusivity and Conductivity (2) Recall from Section 2.4.4 that the introduction of the electrochemical potential gave us an alternate way to understand the Nernst equation. In a similar fashion, looking at charge MORE ON DIFFUSIVITY AND CONDUCTIVITY (OPTIONAL) transport from the perspective of the electrochemical potential gives us an alternate way to understand the relationship between conductivity and diffusivity. Recall the definition of the electrochemical potential (Equation 2.99): ๐ฬi = ๐i0 + RT ln ai + zi F๐i If we assume that activity is purely related to concentration (ai = ci โc0 ), then the electrochemical potential can be written as ๐ฬi = ๐i0 + RT ln ci + zi F๐i c0 (4.85) The charge transport flux due to a gradient in the electrochemical potential will include both the flux contributions due to the concentration gradient and the flux contributions due to the potential gradient: ) ( )] [ ( d ln ci โc0 ๐ ๐ฬ dV = −Mi๐ RT + zi F (4.86) Ji = −Mi๐ ๐x dx dx The concentration term in the natural logarithm can be processed by remembering the chain rule of differentiation: d[ln(ci โc0 )] c0 d(ci โc0 ) 1 dci = = dx ci dx ci dx (4.87) Therefore, the total charge transport flux due to an electrochemical potential gradient is really made up of two fluxes, one driven by a concentration gradient and one driven by a voltage gradient: Mi๐ RT dci dV − Mi๐ zi F (4.88) Ji = − ci dx dx Comparing the concentration gradient term in this equation to our previous expression for diffusion allows us to identify Mi๐ in terms of diffusivity: Mi๐ RT ci =D Dci Mi๐ = RT (4.89) Comparing the voltage gradient term in this expression to our previous expression for conduction allows us to identify ๐ in terms of diffusivity: Mi๐ zF = c (zF)2 D ๐ , where ๐ = i |z|F RT (4.90) 159 160 FUEL CELL CHARGE TRANSPORT By using the electrochemical potential, we arrive at the same result as before. Interestingly, we did not have to make any assumptions about the mechanism of the transport process this time. Thus, we see that the relationship between diffusivity and conductivity is completely general. (In other words, it does not just apply to hopping mechanisms.) The conductivity and diffusivity of a material are related because the fundamental driving forces for diffusion and conduction are related via the electrochemical potential. 4.7 WHY ELECTRICAL DRIVING FORCES DOMINATE CHARGE TRANSPORT (OPTIONAL) Our relationship between conductivity and diffusivity allows us to explain why electrical driving forces dominate charge transport. In metallic electron conductors, the extremely high background concentration of free electrons means that electron concentration is basically invariant across the conductor. This means that there are no gradients in electron chemical potential across the conductor. Additionally, since metal conductors are solid materials, pressure gradients do not exist. Therefore, we find that electron conduction in metals is driven only by voltage gradients. What about for ion conductors? Like the metallic conductors, most fuel cell ion conductors are also solid state, therefore pressure gradients do not exist. (Even in fuel cells that employ liquid electrolytes, the electrolyte is usually so thin that convection does not contribute significantly). Similarly, the background concentration of ionic charge carriers is also usually large, so that significant concentration gradients do not arise. However, even if large concentration gradients were to arise, we find that the “effective strength” of a voltage gradient driving force is far greater than the effective strength of a concentration gradient driving force. To illustrate this point, let’s compare the charge flux generated by a concentration gradient to the charge flux generated by a voltage gradient. The charge flux generated by a concentration gradient (jc ) is given by jc = zFD dc dx (4.91) The charge flux generated by a voltage gradient (j๐ฃ ) is given by j๐ฃ = ๐ dV dx (4.92) Note that the quantity zF is required to convert moles in the diffusion equation into charge in coulombs. As we have learned, ๐ and D are related by ๐= c(zF)2 D RT (4.93) The maximum possible sustainable charge flux due to a concentration gradient across a material is c jc = zFD 0 (4.94) L QUANTUM MECHANICS–BASED SIMULATION OF ION CONDUCTION IN OXIDE ELECTROLYTES (OPTIONAL) where L is the thickness of the material and c0 is the bulk concentration of charge carriers. The voltage, V, that would be required to produce an equivalent charge flux can be calculated from j ๐ฃ = jc c0 (zF)2 D RT c V = zFD 0 L L Solving for V gives V= RT zF (4.95) (4.96) At room temperature, for z = 1, RTโzF = 0.0257 V. Therefore a voltage drop of 25.7 mV across the thickness of the material accomplishes the same thing as the maximum possible chemical driving force available from concentration effects. Effectively, the quantity RTโzF sets the strength of the electric driving force relative to the chemical (concentration) driving force. Because RTโzF is small (for the fuel cell temperature range of interest), fuel cell charge transport is dominated by electrical driving forces rather than chemical potential driving forces. 4.8 QUANTUM MECHANICS–BASED SIMULATION OF ION CONDUCTION IN OXIDE ELECTROLYTES (OPTIONAL) In the previous sections, we have discussed the atomistic mechanisms of conduction and diffusion. In particular, you have learned that diffusion (and hence conduction) in crystalline oxide electrolytes occurs by a hopping process and that the rate of this hopping process is determined by the size of the energy barrier for motion, ΔGact . In general, materials with a lower barrier height will yield higher ionic diffusivities and hence higher ionic conductivities. This is exemplified in Figure 4.18 where GDC displays higher ionic conductivity than YSZ (especially at lower temperatures) due to a smaller ΔGact . The quest for new solid-oxide electrolyte materials has therefore focused on creating materials with higher concentrations of mobile defects and lower activation barriers. New electrolyte development, like new catalyst development, is largely a trial-and-error process. Researchers first develop new candidate materials and then screen them for high ionic conductivity and stability. Recently, however, the same quantum mechanics techniques that have been developed to help identify new catalyst materials (recall Chapter 3.12) are also being applied to identify new oxide electrolyte materials. The basic idea is that quantum mechanics techniques can be used to directly calculate the size of activation barriers associated with atomic motion through a crystalline lattice. Based on these calculated barrier heights, the conductivity of potential new electrolyte materials can then be theoretically predicted. Consider a quantum simulation approach applied to YSZ. In YSZ the diffusing species are oxide ions, which must jump from an occupied site in the lattice to an adjacent (unoccupied) “vacancy.” The height of the barrier associated with this jump depends on the exact nature and symmetry of all the other atoms in the nearby vicinity. The exact neighborhood 161 FUEL CELL CHARGE TRANSPORT surrounding a single atom in the lattice can vary significantly—in fact, a detailed analysis reveals that there are 42 different atomic configurations that an oxide ion may encounter when jumping into a neighboring vacancy in YSZ [11]! (And this analysis considers only nearest neighbors and next-nearest neighbors.) The barrier heights for each of these 42 different atomic configurations will be different because the local environment associated with each of these configurations is different. These barrier heights can be calculated based on approximations to the Schrödinger equation (as discussed in Appendix D), which allows the determination of the energy “landscape” for a system of atoms at zero degrees Kelvin. The barrier height associated with moving an atom into a vacancy is calculated by determining the energy of the entire atomic configuration in a step-by-step fashion as the oxide ion moves into the vacancy. Figure 4.23 shows the concept of this barrier height calculation, performed step by step by considering atomic rearrangements, applied to one of the 42 possible configurations in YSZ. Once this process has been completed for the first configuration, it must then be repeated for the other 41 atomic configurations—a laborious and time-consuming process! After calculating each of the 42 possible barrier heights associated with moving an atom from its lattice position to an open vacancy, the next step is to employ the methods of statistical thermodynamics to calculate the overall macroscopic diffusivity. Statistical thermodynamics teaches us that barriers with lower height can be more easily overcome than those with a higher barrier height. Thus, the macroscopic diffusivity will largely be dominated by the atomic configurations that occur most frequently and that have the lowest barrier heights. Diffusion processes are typically simulated using kinetic Monte Carlo (KMC) techniques, which assume that all atoms move randomly, but that the probability of a successful move depends exponentially on the barrier height as we discussed in Section 4.5.3. In KMC methods, the rate of successful atomic jumps is proportional to a random number multiplied with an exponential Boltzmann factor that contains the barrier height for diffusion. By simulating hundreds of thousands (if not millions) of individual atomic jumps using this KMC technique, the averaged “macroscopic” diffusivity for a material can be estimated. This diffusivity information can then be used to predict the performance of new ion conductors or help in understanding the behavior of current ion conductors. Relative energy 162 ΔE m Migration path Figure 4.23. Illustration of the migration energy barrier. The middle point corresponds to the saddle where the oxygen ion and two cations such as zirconia align in the same plane before the oxide ion continues its path forward creating a vacancy in the location where it started. CHAPTER SUMMARY 2.4 –4.2 Experiment –4.3 log(σT ) (Ω –1 · cm–1 K) KMC 2.2 –4.4 2.1 –4.5 2.0 –4.6 1.9 –4.7 1.8 –4.8 1.7 1.6 log D/D0 2.3 –4.9 6 8 10 12 mole % Y2O3 14 16 Figure 4.24. Logarithmic plot of conductivity times T versus mol% Y2 O3 in YSZ comparing experiment (open squares) and calculation (closed circles). As an example of the power provided by this combined quantum–KMC technique, Figure 4.24 compares experimental measurements and theoretical predictions for the conductivity of YSZ as a function of yttria dopant concentration. As discussed in Section 4.5.3, adding excessive amounts of yttria to zirconia will actually decrease ionic conductivity because defects begin to interact with one another, reducing their ability to move. This subtle effect is captured beautifully by the combined quantum–KMC simulation approach. 4.9 CHAPTER SUMMARY • Charge transport in fuel cells is predominantly driven by a voltage gradient. This charge transport process is known as conduction. • The voltage that is expended to drive conductive charge transport represents a loss to fuel cell performance. Known as the ohmic overvoltage, this loss generally obeys Ohm’s law of conduction, V = iR, where R is the ohmic resistance of the fuel cell. • Fuel cell ohmic resistance includes the resistance from the electrodes, electrolyte, interconnects, and so on. However, it is usually dominated by the electrolyte resistance. • Resistance scales with conductor area A, thickness L, and conductivity σ: R = Lโ๐A. • Because resistance scales with area, area-specific fuel cell resistances (ASRs) are computed to make comparisons between different-size fuel cells possible (ASR = A × R). • Because resistance scales with thickness, fuel cell electrolytes are made as thin as possible. 163 164 FUEL CELL CHARGE TRANSPORT • Because resistance scales with conductivity, developing high-conductivity electrode and electrolyte materials is critical. • Conductivity is determined by carrier concentration and carrier mobility: ๐i = (|zi |F)ci ui . • Metals and ion conductors show vastly different structures and conduction mechanisms, leading to vastly different conductivities. • Ion conductivity even in good electrolytes is generally four to eight orders of magnitude lower than electron conductivity in metals. • In addition to having high ionic conductivity, electrolytes must be stable in both highly reducing and highly oxidizing environments. This can be a significant challenge. • The three major electrolyte classes employed in fuel cells are (1) liquid, (2) polymer, and (3) ceramic electrolytes. • Mobility (and hence conductivity) in aqueous electrolytes is determined by the balance between ion acceleration under an electric field and frictional drag due to fluid viscosity. In general, the smaller the ion and the greater its charge, the higher the mobility. • Conductivity in Nafion (a polymer electrolyte) is dominated by water content. High water content leads to high conductivity. Nafion conductivity may be determined by modeling the water content in the membrane. • Conductivity in ceramic electrolytes is controlled by defects (“mistakes”) in the crystal lattice. Natural (intrinsic) defect concentrations are generally low, so higher (extrinsic) defect concentrations are usually introduced into the lattice on purpose via doping. • Mixed ionic and electronic conductors (MIECs) conduct both electrons and ions. They are useful for SOFC electrodes, where simultaneous conduction of electrons and ions enables improved reactivity by extending three-phase boundaries into two-phase reaction zones. • (Optional section) At the atomistic level, we find that conductivity is determined by a more basic parameter known as diffusivity D. Diffusivity expresses the intrinsic rate of movement of atoms within a material. • (Optional section) By examining an atomistic picture of diffusion and conduction, we can explicitly relate diffusivity and conductivity: ๐ = c(zF)2 Dโ(RT). • (Optional section) Using the relationship between conductivity and diffusivity, we can understand why voltage driving forces (conduction) dominate charge transport. CHAPTER EXERCISES Review Questions 4.1 Why does charge transport result in a voltage loss in fuel cells? 4.2 If a fuel cell’s area is increased 10-fold and its resistance is decreased 9-fold, will the ohmic losses in the fuel cell increase or decrease (for a given current density, all else being equal)? CHAPTER EXERCISES 4.3 What are the two main factors that determine a material’s conductivity? 4.4 Why are the electron conductivities of metals so much larger than the ion conductivities of electrolytes? 4.5 List at least four important requirements for a candidate fuel cell electrolyte. Which requirement (other than high conductivity) is often the hardest to fulfill? Calculations 4.6 Redraw Figure 4.4c for a SOFC, where O2– is the mobile charge carrier in the electrolyte. Is there any change in the figure? 4.7 Draw a fuel cell voltage profile similar to those shown in Figure 4.4 that simultaneously shows the effects of both activation losses and ohmic losses. 4.8 Given that fuel cell voltages are typically around 1 V or less, what would be the absolute minimum possible functional electrolyte thickness for a SOFC if the dielectric breakdown strength of the electrolyte is 108 V/m? 4.9 In Section 4.3.2, we discussed how fuel cell electrolyte resistance scales with thickness (in general as Lโ๐). Several practical factors were listed that limit the useful range of electrolyte thickness. Fuel crossover was stated to cause an undesirable parasitic loss which can eventually become so large that further thickness decreases are counterproductive! In other words, at a given current density, an optimal electrolyte thickness may exist, and reducing the electrolyte thickness below this optimal value will actually increase the total fuel cell losses. We would like to model this phenomenon. Assume that the leak current jleak across an electrolyte gives rise to an additional fuel cell loss of the following form: ๐leak = A ln jleak . Furthermore, assume that jleak varies inversely with electrolyte thickness L as jleak = BโL. For a given current density j determine the optimal electrolyte thickness that minimizes ๐ohmic + ๐leak . 4.10 A 5-cm2 fuel cell has Relec = 0.01 Ω and ๐electrolyte = 0.10 Ω−1 ⋅ cm−1 . If the electrolyte is 100 ๐m thick, predict the ohmic voltage losses for this fuel cell at j = 50 mAโcm2 . 4.11 Derive Equation 4.32 using Equations 4.22 and 4.24. 4.12 Consider a PEMFC operating at 0.8 A/cm2 and 70โ C. Hydrogen gas at 90โ C and 80% relative humidity is provided to the fuel cell at the rate of 8 A. The fuel cell area is 8 cm2 and the drag ratio of water molecules to hydrogen, α, is 0.8. Find the water activity of the hydrogen exhaust. Assume that p = 1atm and that the hydrogen exhaust exits at the fuel cell temperature, 70โ C. 4.13 Consider two H2 –O2 PEMFCs powering an external load at 1 A/cm2 . The fuel cells are running with differently humidified gases: (a) aW,anode = 1.0, aW,cathode = 0.5; (b) aW,anode = 0.5, aW,cathode = 1.0. Estimate the ohmic overpotential for both fuel cells if they are both running at 80โ C. Assume that they both employ a 125-๐m-thick Nafion electrolyte. Based on your results, discuss the relative effects of humidity at the anode versus the cathode. 165 166 FUEL CELL CHARGE TRANSPORT 4.14 (a) Calculate the diffusion coefficient for oxygen ions in a pure ZrO2 electrolyte at T = 1000โ C given ΔGact = 100 kJโmol, ๐ฃ0 = 1013 Hz. ZrO2 has a cubic unit cell with a lattice constant a = 5 Å and contains four Zr atoms and eight O atoms. Assume that the oxygen–oxygen “jump”distance Δx = 12 a. (b) Calculate the intrinsic carrier concentration in the electrolyte given Δh๐ฃ = 1 eV. (Assume vacancies are the dominant carrier.) (c) From your answers in (a) and (b), calculate the intrinsic conductivity of this electrolyte at 1000โ C. 4.15 You have determined the resistance of a 100-๐m-thick, 1.0-cm2 -area YSZ electrolyte sample to be 47.7 Ω at T = 700 K and 0.680 Ω at T = 1000 K. Calculate D0 and ΔGact for this electrolyte material given that the material is doped with 8% molar Y2 O3 . Recall from problem 4.14 that pure ZrO2 has a cubic unit cell with a lattice constant of 5 Å and contains four Zr atoms and eight O atoms. Assume that the lattice constant does not change with doping. 4.16 Which of the following is a correct statement for the water behavior in a Nafion-based PEMFC operating on dry H2 /dry air at room temperature: (a) Both electro-osmotic drag and backdiffusion move water from the anode to the cathode. (b) Both electro-osmotic drag and backdiffusion move water from the cathode to the anode (c) Electro-osmotic drag moves water from the cathode to the anode while backdiffusion moves water from the anode to the cathode (d) Electro-osmotic drag moves water from the anode to the cathode while backdiffusion moves water from the cathode to the anode 4.17 A solid-oxide fuel cell electrolyte has ASR = 0.20 Ω ⋅ cm2 at T = 726.85โ C and ASR = 0.05 Ω ⋅ cm2 at T = 926.85โ C. What is the activation energy (ΔG ) for conduction in this electrolyte material? act