The following acronyms are in common use in the literature for oxide materials used in solid-oxide fuel cells (SOFCs) (xis the doping level).*
BSCF strontium-doped barium cobaltite ferrite (Ba1xSrxCoyFe1yO3)
CGO gadolinium-doped ceria (Ce1xGdxO2) CSO samarium-doped ceria (Ce1xSmxO2) LDC lanthanum-doped ceria (Ce1xLaxO2)
LSC strontium-doped lanthanum cobaltite (La1xSrxCoO3) LSCF strontium-doped lanthanum cobaltite ferrite
(LaxSr1xCoyFe1yO3)
LSGM strontium-doped lanthanum magnesite gallate (LaxSr1xMgyGa1yO3)
LSM strontium-doped lanthanum manganite (La1xSrxMnO3) SDZ scandium-doped zirconia (ZrO2)1x(Sc2O3)x
YSZ yttrium-doped (stabilized) zirconia (ZrO2)1x(Y2O3)x
Fuel Cells: Problems and Solutions,By Vladimir S. Bagotsky Copyrightr2009 John Wiley & Sons, Inc.
* These acronyms represent a compilation from recent literature. The nomenclature is neither systematic (it does not follow the compound names inChemical Abstracts) nor uniform (with different authors, the same letter may stand for different elements, or an element may appear in the ‘‘cationic’’ or ‘‘anionic’’ portion of a name).
135
Fuel cells of this class are built with solid electrolytes that have unipolar O2ion conduction. Best known among these electrolytes is yttria-stabilized zirconia (YSZ), that is, zirconium dioxide doped with the oxide of triva- lent yttrium: ZrO2+ 10% Y2O3or (ZrO2)0.92(Y2O3)0.08. This compound, the basis of Nernst’s (glower) lamp of 1897, became known as the Nernst mass and was regarded as a candidate electrolyte for fuel cells by Baur (1937) and Davtyan (1938) (see Section 2.2). It is commonly found in the oxygen sensors (lambda probes) used to optimize the operation of internal combustion engines.
The Y3+ ions introduced into the crystal lattice of ZrO2act as dopant ions by giving rise to the formation of oxygen vacancies in the lattice, and electrical conduction comes about by O2ions jumping from a current position into a neighboring vacancy, thus filling this vacancy while leaving a vacancy behind (one could also visualize this process as vacancy migration, which occurs in a direction opposite to that of the O2ion migration).
The conductivity of YSZ-type electrolytes becomes acceptable (with values of about 0.15 S/cm) only at temperatures above 9001C. For this reason the working temperature of fuel cells having such an electrolyte is between 900 and 10001C. Such fuel cells will be called conventional SOFCs in what follows.
8.1 SCHEMATIC DESIGN OF CONVENTIONAL SOFCs
Conventional SOFCs exist in several design variants. The basic variants are tubular and planar cells. Monolithic cells joined the first two variants around 1990. The specific design and operating features of these and other variants are described in subsequent sections of this chapter. Those factors that are common for all variants of conventional SOFCs are described in the present section.
The anodes of these cells consist of a cermet (ceramic–metal composite) of nickel and the zirconia electrolyte. This material is made from a mixture of nickel oxide (NiO) and the YSZ electrolyte. The nickel oxide is reduced in situ to metallic nickel, forming highly disperse particles that serve as the catalyst for anodic fuel gas oxidation reactions. These particles are distributed uniformly in the solid electrolyte, and are prevented from agglomerating during fuel cell operation by this electrolyte, thus retaining their catalytic activity. The YSZ material present in the anode also improves the contact between the nickel catalyst and the fuel cell’s electrolyte layer.
The cathodes consist of manganites or cobaltites of lanthanum doped with divalent metal ions [e.g., La1xSrxMnO3 (LSM) or La1xSrxCoO3 (LSC), where 0.15oxo0.25]. Apart from their O2ion conductivity, these cathode materials also have some electronic conductivity that secures a uniform current 136 SOLID-OXIDE FUEL CELLS
distribution over the entire electrode. LSC has a higher ionic conductivity than LSM but is more expensive and leads to problems, in particular because of a possible chemical interaction with the electrolyte.
These three components of the fuel cell: anode, cathode, and electro- lyte, form a membrane–electrolyte assembly (MEA), since by analogy with PEMFCs, one may regard the thin layer of solid electrolyte as a membrane.
Any one of the three MEA components can be selected as the entire fuel cell’s support, and is then made relatively thick (up to 2 mm) to provide mechanical stability. The other two components are then applied to this support in different ways as thin layers (tenths of a millimeter). Accordingly, one has anode-supported, electrolyte-supported, and cathode-supported cells. Some- times, though, an independent metal or ceramic substrate is used, to which the three functional layers are then applied (Figure 8.1).
An important component of SOFCs that governs their operating reliability to a large extent (and sometimes their manufacturing cost) is that of the interconnectors needed to combine the individual cells in a battery. These must be purely electronically conducting, chemically sufficiently stable toward the oxidizing and reducing atmospheres within the fuel cells, and free of chemical interactions with the active materials on the electrodes.
Hydrogen and carbon monoxide can be used as the reactive fuels in SOFCs.
The reactions occurring in such fuel cells and the thermodynamic parameters associated with the reactions (at 9001C) are
Anode: H2ỵO2!H2Oỵ2e E0ẳ0 V ð8:1ị
Cathode: 12O2ỵ2e!O2 E0ẳ0:89 V ð8:2ị
Overall: H2ỵ12O2!H2O E0ẳ0:89 V
DH0ẳ 248:8 kJ=mol ð8:3ị
Cathode Cathode
Electrolyte Electrolyte
Anode Anode
(a) (b) (c)
FIGURE 8.1 Different versions of SOFC: (a) anode-supported; (b) electrolye-sup- ported; (c) cathode-supported.
8.1 SCHEMATIC DESIGN OF CONVENTIONAL SOFCs 137
when hydrogen is the fuel, and
Anode: COỵO2!CO2ỵ2e E0ẳ0:2 V ð8:4ị Cathode: 12O2ỵ2e!O2 E0 ẳ0:89 V ð8:2ị
Overall: COỵ12O2!CO2 E0 ẳ0:87 V
DH0ẳ 283:0 kJ=mol ð8:5ị when carbon monoxide is the fuel.*
The open-cell voltage (OCV)U0in hydrogen–oxygen SOFCs at a temperature of 9001C has a value of about 0.9 V, the exact value depending on reactant composition. The relation between current density and voltage of an operating cell, Ui versus i, is practically linear; that is, the cell’s apparent internal resistanceRapphas a constant value that does not depend on the current density:
UiẳU0iRapp ð8:6ị
This does not constitute evidence for the voltage drop having merely ohmic origins (e.g., the electrolyte’s ohmic resistance). Rather, this function is due in part to special features of polarization of the electrodes. At a current density of 200 mA/cm2, the cell voltage typically has a value of about 0.7 V; that is, parameterRapphas a value on the order of 1Ocm2(its exact value will depend on the thickness of the electrolyte layer).
8.2 TUBULAR SOFCs
8.2.1 Tubular Cells of Siemens–Westinghouse
At the start of the new upswing in fuel cell development, Weissbart and Ruka (1962) of the Westinghouse Electric Corporation, Pittsburgh, Pennsylvania, conducted tests with a hydrogen–oxygen fuel cell built around a tube of ion- conducting electrolyte, (ZrO2)0.85(CaO)0.15(i.e., a material somewhat different from YSZ). Like a test tube, it was closed off on one end. In the bottom part, thin platinum electrodes (less than 25mm thick) were disposed on the inside and outside. Oxygen was fed into the tube; from the outside it was bathed in a hydrogen stream (Figure 8.2).
* The potentials are referred to the potential of a hydrogen electrode contacting the same electrolyte and kept at the same temperature, and to standard pressure (1 bar) for all components.
138 SOLID-OXIDE FUEL CELLS
At a temperature of 10101C and gas pressures of about 1 bar, the cell had an OCV of about 1.15 V; at a current density of 110 mA/cm2, the cell voltage was about 0.55 V. The current–voltage relation was strictly linear. The authors attributed the voltage drop seen with increasing current density to the ohmic resistance of the rather thick electrolyte layer between the electrodes.
Not long after building this first model of a tubular solid-oxide fuel cell of electrolyte-supported design that was associated with large ohmic losses, Westinghouse switched to a new cathode-supported design admitting a much thinner electrolyte layer and thus much lower ohmic losses. Also, the YSZ electrolyte was used for all subsequent work.
In 1998, Westinghouse joined forces with the German company Siemens, which up to then had worked on planar SOFCs, which gave rise to a new enterprise called Siemens–Westinghouse (S-W). This enterprise specialized in the further development and commercialization of tubular SOFCs and soon became the world leader in this field.
.002* Teflon Gasket Seal Cell Leads to Potentiometer and Milliammeter Water-Cooled Metal Flange Kovar to Glass to Ceramic Seal
Pt Wire Leads to Electrodes
Platinum Electrodes
Pt vs Pt-10%Rh Thermocouple O2 out O2 in
Fuel-Water Mixture in
Alumino Tube (ZrO2)0.85(CaO)0.15Tube
(ZrO2)0.85(CaO)0.15 Electrolyte Furnace
FIGURE 8.2 Schematic of a galvanic cell with a solid electrolyte. (From Weissbart and Ruka, 1962, with permission of The Electrochemical Society.)
8.2 TUBULAR SOFCs 139
In S-W cells, ceramic tubes produced by extrusion of the cathode material, lanthanum manganite (with some added alkaline-earth metal oxides), are used.
They have a porosity of 30%, length between 50 and 150 cm, and a diameter of 22 mm. From the outside, a thin layer of YSZ electrolyte (40mm) is applied by chemical vapor deposition (CVD) from mixed ZrCl4, YCl3, and water vapors and oxygen. From inside, a layer of lanthanum manganite with magnesium or strontium dopant is used as cathode material by plasma spraying. An anode material is deposited on top of the electrolyte from a slurry of Ni (or NiO) and YSZ material, and then sintered. A narrow band, 85mm thick, of lanthanum chromite doped with divalent cations (calcium, magnesium, or strontium), which is an electronically conducting semiconductor material, is plasma-sprayed along the tube’s outside to serve as a cell interconnector for series combination of the cells (Figure 8.3a). In a power plant built with such fuel cells, several tubes are combined into a bundle. Such a bundle is shown in Figure 8.3b; it consists of 24 cells (three rows in parallel, each consisting of eight cells in series).
A 100-kW power plant was built by S-W in Westervoort in the Netherlands from tubular cells (Figure 8.4). The fuel cell stacks used in this plant contained four bundles of this type combined in series to form a row, 12 rows then being placed in parallel. Between the rows, units for the conversion of natural gas
Interconnection
Fuel flow Porous support tube
Fuel electrode Electrolyte
Air flow Air electrode
(a)
(b)
FIGURE 8.3 (a) Single tubular SOFC. (From Yamamoto, 2000, with permission from Elsevier). (b) Bundle of 24 (38) tubular SOFCs. (Courtesy of Siemens AG, Energy Sector.)
140 SOLID-OXIDE FUEL CELLS
were installed. The plant also included units for desulfurization and pre- reforming of the natural gas.
During the years 1998 to 2000, the power plant operated for 16,000 hours, providing local grid power of 105 to 110 kW and an additional 65 kW equivalent as hot water to the district heating scheme. The system had an electrical efficiency of 46% and an overall efficiency of 75%. The operation of this plant gave proof of the high functional reliability of tubular SOFCs under real-world conditions of a large power plant (including several temperature cycles caused by temporary stoppages; the use of natural gas containing sulfur) (George, 2000). In March 2001 the system was moved from the Netherlands to a site in Essen, Germany, where it was operated by the German utility RWE for an additional 3700 hours, for a total of over 20,000 hours. Following the Essen experience, the system was brought to GTT-Turbo Care in Turin, Italy. So far the system has reached an operating time of about 37,000 hours with minimum degradation.
8.2.2 Other Versions with Tubular Electrodes
To achieve higher specific power, Sammes et al. (2005) and Suzuki et al. (2006) proposed using tubular cells of very small diameter (microtubular or submilli- meter tubular SOFCs). Tubes of smaller diameter have another important advantage. In fact, the mechanical stresses experienced by all ceramic parts under conditions of drastic temperature change (such as switching the fuel cell on or off) will lead to cracks when these changes occur repeatedly. The stresses FIGURE 8.4 100-kW SOFC CHP plant of Siemens–Westinghouse in The Netherlands.
(Courtesy of Siemens AG, Energy Sector.)
8.2 TUBULAR SOFCs 141
will, however, be less significant for the smaller of the linear dimensions (here, the diameter) of the ceramic part.
With the aim of drawing lower currents from the unit surface area of the electrodes, and bringing the cell design closer to the design with the flat electrodes and flat electrolyte generally adopted, Kim et al. (2003) suggested flat-tube SOFCs. A section of such cells is shown in Figure 8.5. Cells of this shape yield a higher specific power of the battery per unit weight and per unit volume.
Flat tubular electrodes are used even now by Siemens for building large SOFC-based power plants (see Chapter 15). There has been some gradual evolutionary change in the shape of these electrodes, as can be seen in Figure 8.6.
FIGURE 8.5 Anode-supported flat-tube SOFCs (From Kim et al., 2003, with permis- sion from Elsevier.)
FIGURE 8.6 Progress and sequence of advance in the design of Siemens fuel cell electrodes leading up to the 2007 Delta 8 reformation (displayed at the bottom of the stack). (Courtesy of Siemens AG, Energy Sector.)
142 SOLID-OXIDE FUEL CELLS
8.3 PLANAR SOFCs
Flat solid-oxide fuel cells are built analogously to other types of fuel cells, such as PEMFCs. Usually, one of the electrodes (the fuel anode or the oxygen cathode) serves as support for the membrane–electrode assembly (MEA). To this end it is relatively thick (up to 2 mm), and thin layers of the electrolyte and the second electrode are applied to it. In batteries of the filter-press design, the MEA alternates with bipolar plates having systems of channels through which reactant gases are supplied to the electrodes and reaction products are eliminated from them (Figure 8.7). In addition, the bipolar plates act as intercell connectors in the battery by passing the current from a given cell to its neighbor. Alternating with groups consisting of several cells each, special heat exchangers are installed for cooling the operating battery and heating an idle battery during startup.
The development of planar SOFCs started later than that of tubular SOFCs.
They have several advantages over the latter. In batteries of planar design, higher values of specific power per unit weight and volume can be realized than in batteries with tubular cells. This is due to the fact that in them, the path of the current from all surface segments of the electrodes to the current collector is shorter. The current path between the individual cells is also shorter. All this leads to an important decrease in the ohmic losses in the battery. Another advantage of the planar version is a much simpler, less expensive manufactur- ing technology. The technological processes used in making planar solid-oxide fuel cells are more flexible and allow different types of materials to be used for electrodes and electrolyte. For this reason, the basic work on entirely new SOFC versions, particularly those able to work at intermediate and low temperatures, has all been done with cells of the flat type.
Interconnect
Oxidant Cathode
Electrolyte Anode Fuel
FIGURE 8.7 Flat-plate design of SOFCs. (From Yamamoto, 2000, with permission from Elsevier.)
8.3 PLANAR SOFCs 143
Considerable difficulties turned up, however, in the development of flat SOFCs and have so far not been definitely resolved, thus preventing broad commercialization of such cells. The difficulties are related primarily to the fact that the selection of materials having sufficient chemical and mechanical strength for operation at temperatures of 900 to 10001C in the presence of oxygen and/or hydrogen is rather restricted. This holds true both for the materials of electrodes and electrolyte and for various structural materials.
The major problems arising in the development and manufacturing of planar-type SOFC are described below.
8.3.1 Sealing
A large problem in high-temperature fuel cells is careful sealing. The gas compartments of the fuel and oxygen (or air) electrodes should be protected against entry of the ‘‘wrong’’ gas. The joints and welds of all inner channels and gas manifolds should be free of gas leaks. For flat cells, this is a much more difficult problem than for tubular cells, since their entire perimeter must be sealed. Also, flat cells usually have a larger number of gas feeds.
Two types of sealing materials, rigid and compressive, are in use. The elastic materials require constant compression during fuel cell operation. Rigid materials, on the other hand, must meet certain requirements of adhesion (wetting) and of compatible thermal expansion coefficients. Glass or glass ceramics were used initially as the basic sealant. By changes in glass composition and in the conditions of crystallization of glass ceramics, it was possible to adapt the properties of these materials to the operating conditions prevalent in fuel cells, but they have the basic defect of brittleness. In recent years, therefore, development of rigid and elastic sealants on the basis of metals and ceramics was initiated. By using multiphase materials, one can adjust the elastic properties and wetting of surfaces in contact. More details concerning the various sealing materials under study today may be found in a review by Fergus (2005).
8.3.2 Bipolar Plates
The bipolar plates that function as intercell connectors are the most expensive and at the same time the most vulnerable component of planar SOFCs. The basic requirements that must be met by these plates are high chemical stability under fuel cell operating conditions, high electronic conductivity, and complete impermeability to gases. Usually, two types of material are used to make bipolar plates: ceramics and thermally stable high-alloy steels. The ceramics generally used for bipolar plates are oxides based on lanthanum chromite (LaCrO3) doped with MgO, CaO, or SrO. It is a defect of this material that at elevated temperatures, structural changes occur that cause internal stresses in the plates. Liu et al. (2006) suggest using ceramics based on praseodymium 144 SOLID-OXIDE FUEL CELLS
oxides (PrCrO3) of the perovskite type to overcome this defect. A common defect of ceramic materials is their brittleness and insufficient mechanical strength (high sensitivity) under mechanical and heat shocks.
Chemically and thermally resistant steels contain considerable quantities of chromium (more than 20%). On the side of the oxygen cathode, metallic chromium is oxidized to Cr2O3. Depending on the temperature and oxygen partial pressure, this oxide may oxidize further to volatile compounds CrO3and CrO2(OH)2, which could then settle at the cathode–electrolyte interface and hinder oxygen reduction. Such an ‘‘evaporation’’ of chromium is the basic difficulty in the use of metallic bipolar plates (Hilpert et al., 1996). It was found by Stanislowski et al. (2007) that the rate of this evaporation can be reduced substantially by covering the surface of the chrome steel with thin layers (about 10mm) of cobalt, nickel, or copper. These metal layers are converted completely to the corresponding oxides, which become firmly bound to the substrate and have good electronic conductivity.
8.3.3 Stresses in Planar SOFCs
Considerable internal stresses often develop when making and using ceramic parts. Strong temperature changes as well as temperature gradients within a given part are the most important reasons for the development of these stresses.
These factors develop in SOFCs operated at temperatures going up to 10001C.
When such fuel cells are started or stopped, the temperature may change at a rate of hundreds of degrees per minute. Local temperature gradients arise when colder reactant gases enter a hot fuel cell. Fischer et al. (2005) measured the values of these stresses quantitatively with x-ray powder diffraction. They found that the compressive stresses in anodically supported MEA would be as high as 500 MPa.
The risk that such stresses develop is particularly high in ceramic parts consisting of a number of layers of different materials having a mismatch of their coefficients of thermal expansion (CTEs). The membrane–electrode assemblies (MEAs) in SOFCs are exactly in this category. For this reason, all materials used to make high-temperature fuel cells should have identical or at least very close CTEs. For many of the materials used, this coefficient has values between 9 and 14 K1 (Apfel et al., 2006). A somewhat lower probability of internal stress development in ceramic parts can be achieved when making them from mixtures of coarse and fine powders of given materials.
All the problems and difficulties listed will arise precisely at the high working temperatures of conventional SOFCs. For this reason, many research groups working in different countries have tried over the last decade to develop new variants working at lower temperatures. This work is discussed in Sections 8.7 and 8.8.
8.3 PLANAR SOFCs 145