FUEL CELLS – PHOSPHORIC ACID FUEL CELLS | Overview

FUEL CELLS – PHOSPHORIC ACID FUEL CELLS | Overview

FUEL CELLS – PHOSPHORIC ACID FUEL CELLS Contents Overview Anodes Cathodes Electrolytes Cells and Stacks Systems Performance and Operational Conditions...

641KB Sizes 9 Downloads 373 Views

FUEL CELLS – PHOSPHORIC ACID FUEL CELLS Contents Overview Anodes Cathodes Electrolytes Cells and Stacks Systems Performance and Operational Conditions Life-Limiting Considerations

Overview AJ Appleby, Texas A&M University, College Station, TX, USA & 2009 Elsevier B.V. All rights reserved.

Introduction As indicated in Fuel Cells – Overview: Introduction, the only electrolytes that may be satisfactorily used in fuel cells capable of high-current-density operation are those in which ions are produced at one electrode and consumed at the other are present as the major conductors. These ions must be responsible for conduction of the ionic current to or from the anode to the cathode. If this condition is not met, large concentration gradients will result, whose back electromotive force will nullify fuel cell output. For fuel cells using a hydrogen-containing feedstock and oxygen, as fuel at the anode and oxidant at the cathode, these ions are limited to the protons, Hþ ions, and the oxide ions, O2, or ionic compounds transporting these entities. Among aqueous electrolytes, these compounds are either strong acids, in which the ions are hydrated hydronium ions (H3Oþ), or strong alkalis, containing hydroxide ions (OH), i.e., the reaction product of O2 and water. In the late 1950s–60s , there were a number of attempts to develop an acid fuel cell operating on conventional hydrocarbon fuels, particularly at the General Electric Company Research Laboratories, Schenectady, NY, USA. However, to obtain measurable reaction rates on the most effective platinum-based high-surface-area electrocatalysts even at uneconomically high loadings (platinum electrocatalyst mass per unit area), high temperatures over 150 1C were required. The only common acid sufficiently stable and involatile in this temperature range was phosphoric acid, which was normally considered to be a weak and poorly conducting acid. In dilute aqueous solutions, orthophosphoric acid, H3PO4, does behave as a typical (although relatively weak) aqueous acid: its

dissociation constant K is defined for the process H3PO4 þ (n þ 1)H2O-H3Oþ  n(H2O) þ H2PO4 as K ¼

½H3 Oþ  nðH2 OÞ½H2 PO4   ½H3 PO4 

½1

in the concentration range 0.01–0.1 mol L1, in which the quantities in square brackets are molar concentrations (gram molecules per liter). The hydrated hydronium ion H3Oþ  n(H2O), in which n is 3 or more, (H9O4þ) is often written simply as Hþ. The value of this first dissociation constant is 7.5  103 at 298 K, meaning that it is dissociated to the extent of about 24%. In contrast, sulfuric acid, a strong acid, is almost completely dissociated under the same conditions, and even its second dissociation to H3Oþ and SO42 has a higher K value of 1.2  102 under the same conditions. However, phosphoric acid did have some apparently unique properties. Unlike common aqueous acids, typified by sulfuric acid, its proton conductivity in the presence of water vapor did not fall with rising temperature. This is because the internal concentration of liquid water required for proton conduction via hydrated hydronium ions falls with increasing temperature at constant water vapor pressure. In contrast, phosphoric acid showed a unique rise in proton conductivity with increasing temperature owing to a self-ionizing property requiring no free water via the reaction 2H3 PO4 þH4 PO4 þ þ H2 PO4 

½I

Anhydrous sulfuric acid shows a somewhat similar behavior, but it does not exist under fuel cell conditions in the presence of water vapor, when it forms hydrates. It is

533

534

Fuel Cells – Phosphoric Acid Fuel Cells | Overview

also unstable and too volatile for use as an electrolyte. Another property of phosphoric acid is the formation of condensed acids, typified by pyrophosphoric acid, which is produced by elimination of water vapor starting at temperatures around 200 1C: 2H3 PO4 -H4 P2 O7 þ H2 O

½II

This acid is considerably stronger than orthophosphoric acid (K ¼ 1.4  101 and 3.2  102 for the first and second ionizations in dilute aqueous solution at 25 1C: it reverts back to orthophosphoric acid in hot water). The apparently unique ability of phosphate groups to form linear chains, e.g., as in pyrophosphoric acid, (HO)2(P ¼ O)–O–(P ¼ O) (OH)2, has been exploited by nature in perhaps the most basic biochemical energy storage reaction, the formation of adenosine triphosphate from adenosine diphosphate and phosphate ion.

The Tafel Relationship The purpose of this section is to give a brief introduction to fairly complex electrochemical processes, which will aid in understanding how the dioxygen reduction reaction in low-temperature fuel cells functions. Electrochemical reactions may involve the transfer of several electrons (e.g., four for molecular oxygen reduction and two for hydrogen oxidation). Such reactions take place in separate steps, each at different rates. One step will inevitably be slower than the others, i.e., it is the rate-determining step (rds). Consider the schematic hypothetical reaction sequence, written in the cathodic direction: A þ ne -Bn ; Bn þ e -Cðnþ1Þ ðrdsÞ; ðnþn0 þ1Þ

Cðnþ1Þ þ n0 e -D

½III

These reactions are of first order for simplicity, but other orders may be included. Here, the A–B and C–D processes are sufficiently rapid to be considered to be reversible, i.e., in equilibrium, while the one-electron B–C process is rate-determining. Its rate (in terms of a current density j, kA m2) is j ¼ ½Bn ðn þ 1ÞFkc exp  þ lnðn þ 1Þ 

bFV RT

bFV ; or lnj ¼ ln½Bn  RT ½2

where [Bn] is the effective concentration or activity of the intermediate Bn, (n+1) represents the number of electrons transferred for each unit rds, kc is the cathodic rate constant, b is the symmetry factor for the electron transfer reaction, a number typically close to 0.5, R, T, and F are explained under reversible reaction or process,

and V is the electrode potential on an arbitrary scale. This is one form of the experimental exponential Tafel relationship between current density of an electrochemical reaction and electrode potential. The equilibrium [III] may be expressed as ½Bn  ¼ ½AKAB exp 

nFV RT

½3

where KAB is the equilibrium constant for the A–B process at V ¼ 0. Thus, eliminating the concentration of the intermediate j ¼ ½Aðn þ 1ÞKAB Fkc exp  ¼ X exp 

ðn þ bÞFV RT

ðn þ bÞFV RT

½4

The Tafel slope b ¼ dV/dlog10 j is therefore  2.302 6RT/ (n þ b)F. At 298 K, this is about  60/(n þ b)F in millivolts (mV) per decade of current density. Common values of the slope are  120  mV (n ¼ 0, b ¼ 0.5),  60 mV (n ¼ 1, b ¼ 0, i.e., a chemical, not electrochemical, rds), and  40 mV (n ¼ 1, b ¼ 0.5). These should be interpreted with care: When intermediates are adsorbed, their changing surface concentration with potential can yield other interpretations (see entry Chemistry, Electrochemistry, and Electrochemical Applications: Oxygen). When the forward and backward rds are in equilibrium, we can write bFV j ¼ ½Bn ðn þ 1ÞFkc exp  RT h i b0 FV ðnþ1Þ 0 ¼ C ðn þ 1ÞFka exp þ RT

½5

where ka is the rate constant in the anodic direction and b0 is the anodic symmetry factor. When rearranged, this equality must have the form [Bn–] ¼ [C(nþ1)–]KBC exp þ FV/RT. It follows that KBC ¼ [(n þ 1)kc/(n0 +1)ka] and b+b0 ¼ 1, i.e., b0 ¼ (1–b). 0 In terms of [Dn+n +1], the back reaction to (4) is h i FV 0 j ¼ Dnþn þ1 ðn0 þ 1ÞKDC Fka exp þ ðn0 þ 1  bÞ RT FV ¼ Y exp þ ðn0 þ 1  bÞ RT

½6

Equating (4) and (6) at equilibrium, the equilibrium potential E is found to be E¼

  RT ðln X  ln Y Þ NF

½7

where N=(n+n0 +1) is the number of electrons transferred in the overall process. Substituting E in either (4) or (6) gives the following expression for the logarithm of the

Fuel Cells – Phosphoric Acid Fuel Cells | Overview

exchange current density, jo, i.e., the current density when the anodic and cathodic reactions are equal: log10

 0    ðn þ 1  bÞ ðn þ bÞ jo ¼ log10 X þ log10 Y N N

½8

where the multipliers are sometimes called the electrochemical reaction orders. Another version of the Tafel equation may be written in terms of the exchange current for reversible reactions at given concentrations of reactants and products contained in X and Y: j ¼ jo exp  ðn þ bÞ

FZ ðcathodicÞ RT

j ¼ jo exp þ ðn0 þ 1  bÞ

FZ ðanodeÞ RT

½9

½10

where Z is the negative and positive overpotential for (9) and (10) respectively, that is, the potential measured from the reversible potential E. (n+b) and (n0 +1  b) are called the transfer coefficients. Their sum is N. For a small displacement Z from equilibrium, the exponentials become approximately 1–(n+b)FZ/RT and 1+(n0 +1–b)FZ/RT. The net reaction rate, i.e., the difference between the anodic and cathodic processes, becomes j ¼ joNFZ/RT, which provides a means to determine jo in such cases. The hydrogen electrode on effective electrocatalysts in both acid and alkaline solutions shows this reversible behavior, whereas oxygen reduction at temperatures o250 1C in acid media and o150 1C in alkaline media is completely irreversible even on the best platinum or platinum alloy electrocatalysts, which have jo values up to about 106 less than those for hydrogen oxidation. The anodic reaction for oxygen evolution may therefore be completely neglected at all attainable values of Z in these media. It is clear that as more electrons are transferred before the rds, the less steep is the Tafel slope. Thus, the trend will be that an rds occurring after such rapid electron transfers will become much more rapid with increasing Z, and will therefore cease to be the rds, which will be pushed to become earlier in the reaction sequence. In this way, the cathodic and anodic processes away from equilibrium (i.e., at significant 7Z values) may show different rate-determining steps.

Technology Overview Because it is a catalytic direct energy conversion device that transforms the free energy of a chemical reaction directly into direct current (DC) electricity, a fuel cell is often represented to students, journalists, and the general public as a specialized type of primary battery. This is in many respects true, but only up to a point. Like any other battery, the core element of a fuel cell has negatively

535

charged electrodes (anodes) and positively charged electrodes (cathodes), which are in contact with a suitable ionically conducting electrolyte and which are in separate contact with an oxidizable material (a ‘fuel’) and a reductive material (‘oxidant’), respectively. In a throwaway zinc manganese dioxide primary battery such as the familiar AA cell, the ‘fuel’ is powdered metallic zinc and the ‘oxidant’ is powdered manganese dioxide. These materials are stored within the current-collecting structure of the electrodes, which are separated by concentrated potassium hydroxide solution as electrolyte. In a typical fuel cell, the fuel is gaseous hydrogen and the oxidant is gaseous oxygen, rather than a reactive metal and a metal oxide as in the primary cell. Instead of being stored within the electrode structures as in a throwaway primary battery, the ‘fuel’ and ‘oxidant’ in the fuel cell are brought in from the outside as required from external reservoirs. The key to efficient fuel cell operation at temperatures between ambient and o200 1C is the use of hydrogen as the fuel, as common fuels such as hydrocarbons have negligible reactivity on common catalytic anode materials under these conditions. For convenience, the common oxidant is the 21 vol.% of oxygen in air, the remainder being inert gases (87% nitrogen), which act as a dilutant, making it less efficient as an oxidative reactant as the oxygen is consumed on passing through the cell reaction surfaces. Pure hydrogen–based fuel is only rarely available, and therefore hydrogen-rich fuel must generally be manufactured from an available carbonaceous fuel on-site for use in the fuel cell. This fuel is usually natural gas (NG). Because, like the oxygen from air at the cathode, the hydrogen-rich fuel becomes progressively diluted as it passes through the anode reaction areas, it cannot be completely used in the cell. In consequence, exiting anode exhaust gas is combusted to supply other needed heat inputs in the systems, particularly for the energy requirements for transformation of NG to hydrogen-rich gas. Because of the requirement to produce hydrogen-rich gas from a carbonaceous fuel on-site, what is commonly called the ‘phosphoric acid fuel cell’ (PAFC) in reality is mostly a chemical conversion plant whose volume may considerably exceed that of the electrochemical converter itself. This chemical plant must be as efficient as possible, which is difficult to achieve in a small unit. Because the reactors in the chemical plant are catalytic and are poisoned (i.e., rendered ineffective) by sulfur compounds, any such compounds present must first be removed from the fuel input. These include the odorants, which are added to NG before delivery to make leaks detectable, which is done after naturally occurring sulfur compounds are removed close to the wellhead. As explained in the section entitled ‘Fuel Processing’, initial fuel processing after desulfurization is performed in a heat-exchange steam-reforming reactor, which requires a

536

Fuel Cells – Phosphoric Acid Fuel Cells | Overview Exhaust

Air in

S

NG

H

R

RB

R

HS

LS

A

P

C

DC power

AC power I

AE WP W CE

Figure 1 Schematic arrangement of a generic atmospheric pressure phosphoric acid fuel cell (PAFC) system. Pumps, blowers, heat exchangers, and start-up systems with nitrogen and hydrogen storage are not shown. NG, natural gas entry; H, hydrodesulfurization system; R,R, heat-exchange steam reformer; RB, reformer burner; HS, high-temperature water–gas shift reactor; LS, low-temperature shift reactor; A, fuel cell stack assembly (CSA) anodes; P, CSA cooling plates; C, CSA cathodes; W, subsystem for water recovery from cathode exhaust gas; WP, water purification subsystem supplying pure liquid water to cooling plates, where it is converted into steam; S, steam supply to reformer; I, electronic inverter, converting direct (DC) to alternating current (AC) power. Cogeneration heat may be extracted from the reformer, shift converters, water condenser, and steam supply, at different temperatures. The system is shown with coflow anode–cathode systems. This is illustrative. In practice, systems may have counter- or crossflow configuration, the latter being most common.

considerable heat input after the desulfurization step. It then requires a chemical cleanup system using the water– gas shift reaction operating in two stages at lower temperatures to remove carbon monoxide, a catalytic poison in the electrochemical energy converter. Although this reaction does produce some heat, which may be collected, it does so at a temperature lower than the reformer, so that its energy can be recovered only for preheating. In addition, both the reforming and shift reactions require steam, which must be produced from liquid water and whose energy is not recoverable. This in turn requires an energy input, which in the case of the PAFC electrochemical energy converter operating at about 200 1C can be provided by its waste heat. The reformer requires a heat source to supply its chemical conversion reaction, which is provided by burning depleted anode exit gas containing some hydrogen mixed with carbon dioxide and a small amount of carbon monoxide. Finally (and in contrast to, e.g., a gas turbine turning an alternator), the DC electrical output of the electrochemical energy converter must be converted to alternating current (AC) power for normal domestic, commercial, or industrial use. This again involves an energy loss, further reducing the overall efficiency beyond that of the chemical energy plant and the electrochemical energy converter. Thus, the competitiveness of a PAFC unit as a power producer from NG compared with a competing technology (e.g., an advanced

gas turbine with an alternator) becomes a careful exercise in energy recovery and integration. The fact that its atmospheric emissions are very low compared, e.g., with a turbine will not by itself sell the technology. It must at the same time produce AC electrical power at a competitive price on-site for a given application. A PAFC is therefore a complex system, which adds to its overall cost. A simplified schematic of a PAFC system operating on NG is shown in Figure 1.

Phosphoric Acid Fuel Cell System Engineering General Electric’s (GE’s) phosphoric acid electrolyte cells were single laboratory units of not more than postage stamp size. Scale-up to practical systems would require a great deal of practical electrochemical, chemical, mechanical, and electrical engineering. Work on the PAFC had been taken up at the Pratt and Whitney Aircraft Division of United Aircraft Corporation (South Windsor, CT, USA; later United Technologies Corporation (UTC)) during the 1960s to examine it alongside the alkaline fuel cell used in the Apollo Moon program. This is discussed in Fuel Cells – Overview: Introduction. In 1966, the American Gas Association announced the TARGET (Team to Advance Research in Gas Energy

Fuel Cells – Phosphoric Acid Fuel Cells | Overview

Transformation) program to supply individual homes and businesses with utility-quality AC power from NG using fuel cells. Possible competitive technologies to small on-site fuel cell systems at the start of the TARGET program were seen to be internal combustion piston engines or small gas turbines. Even today, the latter are not commercially available products. The TARGET program was intended to exploit the large differential between delivered domestic electricity and NG costs. The aim was to use NG in small fuel cell systems, which could supply on-site power at higher efficiency than small internal combustion generators. Fuel cells would also have the advantage of relatively noiseless operation. Before the advent of the 1970 Clean Air Act, their potentially lower pollutant emissions were less of a consideration than they are today. The prime contractor was Pratt and Whitney Aircraft. A 3-year research program started in January 1967 to explore technologies, followed by a 3-year evaluation and finally a 3-year demonstration program before intended full deployment. The total cost was to be $27 million in then-year dollars, or about $136 million in third-quarter 2007 dollars. The PAFC program was undertaken with parallel studies on possible alternative fuel cell approaches, including alkaline cells of Bacon (or more advanced alkaline) type and high-temperature electrolyte cells. All systems considered would require fuel processing, i.e., conversion of NG to hydrogen-rich gas (a mixture of hydrogen, carbon monoxide, water vapor, with some residual carbon monoxide produced by catalytic steam-reforming followed by water–gas shifting). In spite of the advantages of Bacon-type alkaline cells (no noble metals, high performance under pressurized operating conditions), they were rapidly rejected because of carbonate formation if carbon dioxidecontaining fuel gases were used (as in the Jacques’ cell, see Fuel Cells – Overview: Introduction). The complex and expensive gas cleanup procedures they would require appeared impractical, at least in small units. As other hightemperature systems, notably the Westinghouse (Pittsburgh, PA, USA) solid oxide fuel cell (SOFC), seemed largely conceptual at the time, the only PAFC backup alternative identified was the molten carbonate fuel cell (MCFC), whose technology resulted from 1920s work in Europe, and had recently been made more feasible by the work of G. H. J. Broers and J. A. A. Ketelaar in the Netherlands since the 1950s. However, this technology existed only on a laboratory scale in 1967 at the Institute of Gas Technology (IGT, Chicago, IL, USA), a TARGET program subcontractor.

Fuel Processing As stated above, work on the direct oxidation of hydrocarbons at the PAFC anode had been conducted at the General Electric Company Research Laboratory

537

(Schenectady, NY, USA) in the late 1950s and early 1960s. However, reaction rates were impracticably low. In contrast, the oxidation of hydrogen was very rapid and efficient on these anode electrocatalysts. In consequence, the TARGET engineers examined the high-temperature (B800–1000 1C) catalytic reaction of desulfurized hydrocarbons with steam to give a mixture of hydrogen, carbon monoxide, carbon dioxide, and residual steam as a preliminary process. The carbon monoxide could then be catalytically converted using more steam as far as possible to carbon dioxide and more hydrogen at a lower temperature, where this process was favored by thermodynamics. Using the major constituent of NG (methane (CH4)) as an example, the successive reactions are CH4 þ H2 O-CO þ 3H2

½IV

CO þ H2 O-CO2 þ H2

½V

The first reaction is steam-reforming of hydrocarbons and the second is the water–gas shift reaction. Conversion of the chemical energy content of NG (or any other hydrocarbon) to hydrogen involves an upgrade in heating value per equivalent of fuel converted. The chemical or electrochemical term ‘equivalent’ as used here means an oxidative unit. An ‘equivalent’ is defined as one gram molecule of electrons, i.e., the molecular weight of a chemical material expressed in grams divided by the number of electrons involved in the transformation of each molecule in a given reaction. For one molecule of hydrogen, the number of electrons is two, or one per atom, whereas for one molecule of oxygen, it is four, or two per atom. Thus, to produce one molecule of water, two electrons must be given up by two hydrogen atoms and accepted by one atom of oxygen, to be shared between the individual atoms to give the stable water molecule. For methane, which requires two molecules of oxygen to complete the reaction CH4 þ 2O2-CO2 þ 2H2O, eight equivalents of electrons are required per molecule of methane. The number of atoms in a gram molecule is 6.022  1023, the same as the number of electrons involved in a gram equivalent. This, multiplied by the fundamental electronic charge, gives the total electrical charge in coulombs (C, which is amperes  seconds, A-s) per equivalent of electrons, which is the Faraday unit, F ¼ 96 485 C. The energy change per equivalent in heat (enthalpy) units, i.e., the value given as DH in joules per equivalent, is equal to the unit of electrical intensity, the volt, multiplied by the quantity of electricity in C per equivalent. This, divided by F, gives a V value equal to the quality of the energy per equivalent in units of electron-volts (eV). At 25 1C, for hydrogen-producing gaseous water product, the value is 1.254 eV. For methane, it is 1.040 eV. Thus, the conversion of one gram molecule or eight equivalents of methane to the corresponding eight equivalents of hydrogen via the

538

Fuel Cells – Phosphoric Acid Fuel Cells | Overview

steam-reforming reaction (i.e., combining reactions [IV] and [V]) is given by CH4 þ 2H2 O-CO2 þ 4H2

½VI

This requires an upgrade in energy of 0.214/1.040 per equivalent, or 22%. The situation is in practice even worse than this, because under the temperature conditions of the initial step in catalytic steam-reforming, roughly equal parts of carbon monoxide, carbon monoxide, and carbon dioxide are produced, i.e., the reaction is approximately CH4 þ 1:5H2 O-0:5CO2 þ 0:5CO þ 3:5H2

½VII

The quantity of energy per equivalent for carbon monoxide is higher than that for hydrogen (1.467 eV), so that the total energy on the right-hand side averages 1.281 eV. Thus, an upgrade of 0.241/1.040 per equivalent, or 25%, is needed. This must be supplied to the first reactor, the heat-exchange steam reformer in a PAFC system, which in practice is not 100% efficient. The carbon monoxide must be converted to as much hydrogen as possible in following two chemical reactors by the water–gas shift reaction [V]. These reactors are at two successively lower temperatures than that of the steam reformer, so the heat they produce when converting carbon monoxide to hydrogen can be used only for preheating. If the overall process goes to completion, the overall reaction will be reaction [IV]. In the catalytic steam-reforming reactor, [IV] and [V] proceed simultaneously in the usual operating temperature range of about 800–1000 1C, the result being an overall reaction absorbing heat (an endothermic reaction). Hence, heat must be provided in this reactor from another source to balance the difference in the combustion energy of reactants and products. The product gases are then converted via [V] in reactors operating at lower temperatures to thermodynamically favor the products as far as possible in a reaction producing heat (in an exothermic reaction). Finally, the steam required for the reformer must be produced from liquid water, which requires the addition of the latent heat of vaporization at the boiling point. This requires about 11% of the heating value of the methane fuel for the theoretical amount of two water molecules per molecule of methane: in practice, three or more water molecules are required to force the reaction to go as far as possible to completion. These heat balances pose a challenge to the system designer. The steam reformer requires injection of heat from the combustion of fuel, as does the water evaporator raising the steam. Heat is always lost to the environment; therefore, the designer’s job is to identify how to supply the heat and to minimize the losses. Similarly, one must recover as much heat as possible from the water–gas shift reactors, and use it in the most practical manner in the rest of the system.

As it is available at a temperature lower than that required for steam-reforming, it can be used only for preheating or steam-raising. For steam-reforming, a nickel-based catalyst, generally supported on alumina, is used. Because sulfur is a poison for this catalyst, any sulfur compounds in the incoming fuel must be reduced to very low levels. Natural gas at the wellhead usually contains significant amounts of sulfur, but this is removed before it is sent by pipeline to distribution centers. At this point, up to a few parts per million of malodorous organic sulfur compounds (thiols and/or thioethers) related to skunk odor agents are added as odorants for safety reasons. Thus, the first fuel treatment in the fuel cell system is the removal of the majority of these compounds. In PAFC systems, this is carried out by a catalytic hydrodesulfurization reaction with hydrogen to form hydrocarbons and hydrogen sulfide, and hydrogen sulfide is then absorbed in a zinc oxide bed to give zinc sulfide, which is recycled when spent. Heat slightly above the reformer temperature must be supplied by combustion of fuel with air through a stainless steel wall to supply the endothermic reaction in the reforming catalyst chamber. The latter must therefore be as thin as possible to give effective heat transfer to reaction sites. The heat is generally supplied by anode off-gas, containing hydrogen that cannot be profitably used in the stack . The off-gas is dilute in hydrogen and humid, so catalytic burners are required in the combustion chambers. The original reactors were tube and shell structures resembling water-tube steam boilers, but these have generally been replaced by welded flat-plate structures with sandwiches of combustion chambers and reforming chambers. For water–gas shift conversion via reaction [V], a copper–zinc oxide catalyst is generally used, and heat must be removed from the exothermic reaction. Like steam-reforming, this is conducted in heatexchange reactors with fuel, air, and steam in the cooling chambers, so that the waste heat from the exothermic reaction can be used for preheating. To take advantage of the thermodynamic equilibria as a function of temperature, and the fact that reaction rates are more rapid at higher temperatures, allowing smaller reactors to be used, a high-temperature shift reactor is used as a first stage. This is followed by a low-temperature shift to reduce carbon monoxide to levels at which it does not poison the fuel cell anode electrocatalyst. These levels vary with cell operating temperature. At today’s operating temperature of 200 1C, the PAFC anode can tolerate about 1.5% of carbon monoxide in the incoming anode fuel gas. Unlike a heat engine operating on NG or other hydrocarbon fuels, fuel cells of both low- and hightemperature types must use these conversion reactions to make hydrogen-rich gas from common fuels to generate

Fuel Cells – Phosphoric Acid Fuel Cells | Overview

DC electricity using direct electrochemical energy conversion. Therefore, the need to upgrade conventional fuels to produce hydrogen represents a first challenge in producing an efficient fuel cell system. There are two possible approaches to using hydrogenrich product gas from fuel processing. The first is to separate carbon dioxide, which will generally also result in separation of water to give a hydrogen stream of industrial purity. This has the advantage of allowing almost 100% of the hydrogen to be used in the cell, avoiding the problem associated with hydrogen-rich gas (reformate) discussed below. This would have permitted the use of alkaline electrolyte cells, but this approach was rejected by the TARGET engineers because of the complexity and energy losses associated with gas cleanup. In consequence, reformate would be used directly in the fuel cell.

Electrochemical Phosphoric Acid Fuel Cell Performance Issues The use of reformate results in a second type of loss because an efficient cell must consume as much fuel as possible, as only hydrogen consumed in the cell generates electrical power. Any unused hydrogen can, however, be burned to supply heat for the fuel-processing operations, i.e., endothermic steam-reforming. If the fuel gas is a mixture of hydrogen, carbon dioxide, and excess water vapor, all the hydrogen cannot be consumed in the cell as the mixture flows across the anode surface from entrance to exit. This is because the current per unit area (current density) falls as the concentration of hydrogen fuel falls, so the current will drop to zero at zero concentration, giving diminishing returns at high hydrogen utilizations. Because the electrocatalysts used for hydrogen oxidation in the PAFC have very high activity, the hydrogen oxidation reaction is thermodynamically reversible. High utilization has the immediate effect of making the negative value more positive because of the applicability of the Nernst equation, i.e., the thermodynamic expression " #   pH2;en RT  Eex ¼ Een  In  2F pH2;ex

½11

in which R is the universal gas constant, F is the Faraday constant (96 485 C per equivalent), T is the absolute temperature in kelvin (K), and the p values refer to the effective concentrations (expressed as fractional pressures or partial pressures) of the reactant at the electrode. The multiplier 2 in the denominator occurs because two electrons are transferred for each hydrogen molecule oxidized. In [VIII], the E terms are electrode potentials, and the subscripts en and ex refer to conditions at the anode entry and exit, respectively. Because it is a good

539

electronic conductor, the anode current collector is an equipotential surface, so the driving force (the potential difference) at the anode entry is greater than at the exit. This indicates how the current density falls from entry to exit as pH2 is progressively reduced. In an acid electrolyte fuel cell operating on hydrogen and humidified air, product water is formed at the cathode: H2 -2Hþ þ 2e ðanodeÞ  1 ðcathodeÞ 2O2 þ 2e -2H2 O

½VIII

Because the cathode reaction is irreversible, an equation similar to [11] does not apply. Suppose that hydrogen enters the cell at 1 atmosphere absolute (atma) pressure in a 4:1:1 H2/CO2/H2O mixture from a fuel-processing system at 1 atma total pressure, and 90% of the hydrogen is used in the cell in a single-pass operation. The hydrogen partial pressure is 0.67 atma at the anode entrance. At the exit, 90% of the hydrogen is consumed, so the ratios of hydrogen partial pressure to carbon dioxide are now 0.067:0.33 at a total pressure still at 1 atma, i.e., a hydrogen pressure of 0.17 (i.e., 0.067/0.397) atma. R/F is 8.6  105 V. At 100 1C (373 K), the logarithmic term for the hydrogen anode exit partial pressure compared to the standard 1.0 atma value is –0.028 V, which appears to be not a very serious loss. However, only 90% of the fuel is used in the fuel cell, which constitutes a further and more important loss. If this hypothetical fuel cell could operate at 1000 1C (1273 K), the corresponding loss would be much larger: about 0.098 V. This has important consequences for the SOFC, which are discussed in Fuel Cells – Overview: Introduction. In low-temperature fuel cells, a well-catalyzed hydrogen anode in the absence of catalytic poisons behaves more or less as indicated above, i.e., its potential generally follows the predictions of reversible thermodynamics unless the hydrogen partial pressure becomes very low. The reaction is then limited by concentration gradients and the effect of kinetics, which leads to further displacements from the local reversible potential values. Such displacements are called overpotentials. The cathodic oxygen reduction reaction in acid electrolytes is about 106 times less rapid than hydrogen oxidation, even on the best platinum-based electrocatalysts. For these reasons, it is said to be irreversible. Oxygen reactions, e.g., with organic compounds, are explosively rapid at high temperatures (as in simple combustion), but their rapid fall in rate with temperature has made possible the evolution of living molecules under ordinary conditions. Noble metal-based electrocatalysts are more stable than those that nature has developed, but perhaps they are not so effective. The net result is that typical oxygen reduction processes are so slow under equilibrium conditions (which would theoretically be at a cell voltage of 1.163 V at standard 1.0 atma pressure for hydrogen and

Fuel Cells – Phosphoric Acid Fuel Cells | Overview

oxygen reactants and product water vapor at 373 K) that they never reach their equilibrium potential even at zero current, i.e., at open circuit. Under these circumstances, oxygen reduction is opposed in an equal and opposite manner by parasitic oxidation reactions. The result is an open-circuit (zero-current) potential of close to 1.0 V versus hydrogen, about 0.2 V below the theoretical value. This is followed by a rapid fall in potential at small current densities, which is in turn followed by a roughly linear potential–current density range in the region of practical interest, between about 0.7 and 0.6 V versus hydrogen. This is explained by a variety of irreversible effects, the first (seen mostly above 0.8 V) being a kinetic effect due to the occurrence of a logarithmic relationship between current density and potential because of the effects of slow reaction kinetics, known as the Tafel relationship. The linear region occurs because other effects, mostly a combination of various ionic and electronic resistances and gaseous concentration gradients, overwhelm the Tafel relationship. This region appears to be an ohmic polarization in which cell potential falls almost linearly with current density, but in reality it is much more complex. This ohmic-like region generally extrapolates to about 0.75–0.8 V at zero current in the atmospheric pressure PAFC. Finally, at very high current densities, the oxygen supply cannot keep up with the local current density requirements because of the occurrence of concentration gradients, and the current density rapidly falls off to limiting values at low cell potentials. As a result, the real working potential difference at economical current densities is between 0.75 and 0.65 V. A representation of the single-cell performance (DC output) of an atmospheric pressure PAFC single cell is shown in Figure 2.

TARGET Program Economics The original objective of the TARGET program was for the American Gas Association (AGA) member gas companies to supply combined heat and power to individual new homes, providing competition to the delivery of electric power by centralized electric utilities, so TARGET homes were not originally intended to be gridconnected. In 1967, wellhead NG and residential electricity costs were about 20 MMBTU1 (higher heating value (HHV)) and 2.0 kWh1, or about $1.00 MMBTU1 and 10.0 b kWh1 in third-quarter 2007 dollars. Fuels containing hydrogen have two values for their heat of combustion or heating value. The HHV is for liquid water product, whereas the lower heating value (LHV) is for water vapor product. For NG (essentially methane), the HHV is about 10% higher than the LHV, the difference being the latent heat of evaporation of water. Natural gas is purchased based on HHV. However, the

1.2 1.0 Cell voltage (V)

540

0.8 0.6 0.4 0.2 0 0

0.1

0.2

0.3

0.4

Current density (A cm−2)

Figure 2 A typical performance curve for a 200 1C phosphoric acid fuel cell (PAFC) under standard hydrogen (in reformate) and oxygen (in air) utilization conditions. Note: To prevent graphite component corrosion, operation above 0.8 V is not permitted.

value used to determine efficiencies of engines is customarily the LHV. A new home at that time required about the same average electric power input as one today, i.e., about 1.2 kW, or about 10 000 kWh per year. Assume an HHV-AC fuel cell unit efficiency of 35%, with the remaining 42% of energy recovered as hot water, which would otherwise have been supplied by NG at a water heater efficiency of 85%. The cost of delivered residential NG is estimated at 25 MMBTU1. Thus, using a comparison in third-quarter 2007 dollars, the equivalent of $1000 of delivered grid electricity and 14 286 kWh (thermal, HHV) or 49 MMBTU of NG for water- and space-heating purposes worth $61 ($1061 total) could both be replaced by 97.5 MMBTU of delivered NG worth $122, which would be converted on-site to electricity and heat. The total savings would therefore be $939 per year. In spite of the relatively low mean power requirement for an average home, large power peaks occur when all lights and appliances are on, and when instantaneous loads, e.g., electric motor starting, are required. Thus, the nongrid-connected TARGET system for an individual home would be designed to provide 12.5 kW to cover peaking requirements. The unit cost target was $150(1967) kW1 in commercial production, i.e., $760 kW1 in third-quarter 2007 dollars. This was the only financial parameter given publicly at the beginning of the program, so the fiscal conclusions given in the following section have been recreated. This fiscal analysis is important, as it determines present and future economics of grid-connected and non-grid-connected fuel cell systems. Assuming $760 kW1 (2007) to be the TARGET program installed cost, each $9500 unit would have required an annual payback of 4.0% of initial cost per

Fuel Cells – Phosphoric Acid Fuel Cells | Overview

annum to cover energy savings. This percentage would have to include debt service, depreciation, operation and maintenance (O&M), and any other costs beyond those of fuel alone. How this could be achieved was not disclosed. However, it would have required a very long write-off of assets. In 1967–68, inflation (‘i ’) was about 4.5% p.a. (0.045), with an expected return ‘r ’ of 6.0% (0.060) for debt service or return on capital. The general expectation would also have been that inflation would increase delivered electrical energy costs on a year-to-year basis by 4.5%. In the most favorable case, the expected yearly savings in start-of-year-one constant dollars (‘c ’) can be related to the inflation-corrected mortgage formula: c¼

Pðr  iÞ    ð1 þ iÞ n ð1 þ iÞ 1  ð1 þ r Þ

½12

where ‘P’ is the principal ($9500 per 12.5 kW unit), ‘c ’ ¼ $369 per unit, ‘i ’ and ‘r ’ are expressed as fractions, i.e., 0.060 and 0.045, respectively, and ‘n ’ is the depreciation time, system lifetime, or write-off time in years. This expression may be adjusted as required for changes in ‘i ’ and ‘r ’ over time. It assumes no profits, possible taxes, tax rebates, or individual O&M costs. With the above fiscal assumptions, ‘n ’ for TARGET units would be 11.0 years. Any other assumptions will either require greater payback times or lead to overall losses over time. This 11.0 years assumed lifetime represented a challenge for an unproven technology, especially with the assumption of zero O&M costs. In any case, the replacement lifetime goal for the electrochemical converter (the PAFC stack itself, see below) was 40 000 hot hours. The stack was the day-to-day term for the ‘pile’ of individual square cells lying one on top of the next, in intimate electronic contact with each other so that the voltage of individual cells was additive. The developer called this the DC module (sometimes including the fuel processor) and later the cell stack assembly (CSA).

Phosphoric Acid Fuel Cell Engineering Challenges Stack Materials The PAFC was selected by the TARGET team with little experimental knowledge of the electrochemistry of the limiting oxygen reduction reaction in this medium. However, it was known that platinum electrocatalysts at 150 1C showed lower rates than in sulfuric acid at temperatures below 100 1C. However, under these conditions, sulfuric acid became unstable, being reduced at the fuel cell anode. The system engineering was better understood than the electrochemistry, although many uncertainties remained. The first uncertainty was

541

whether the fuel processor could be efficiently scaled down to that required for a 12.5 kW system. The principal problems of the PAFC cells were the cost of suitable electrocatalysts and cell construction materials in the highly corrosive acid environment. The system DC voltage required for electronic conversion to supply domestic AC loads would be achieved by assembling a series of flat single PAFC cells in electrical series, each in efficient electronic contact to form a stack. Between the cells was an electronically conducting bipolar plate with channels on each side to distribute reactants and remove reaction product water vapor. At 0.65 V electrical output per cell, the heat output of a stack producing water vapor product would be about 48% of the electrical output. The stack would be cooled by electronically conducting cooling plates fed by a cooling fluid (pure water) located at regular intervals (about 1 plate per 6–8 cells). The stack was completed by metal end-plates and/or power take-off terminals. The endplates would be connected by insulated bolts or other connectors to maintain electrical contact through the stack using pressure, or may themselves be insulated from the stack with separate electrical end-plates for power take-off. For the cooling system, it seemed clear that care should be taken concerning cooling water purity, so the water treatment subsystem used before its injection into the cooling plates was arranged to deionize the water and to remove dissolved oxygen to low levels. However, this alone turned out to be insufficient. Some early tests showed that great care had to be taken in the materials design of the intercell cooling system, in spite of the use of high-purity water to raise flash steam. The use of dissimilar metals in the system was shown to be impossible, as corrosion resulted. In consequence, an all stainless steel serpentine cooling plate circulation system was ultimately used. The PAFC had system engineering advantages over one operating on sulfuric acid at lower temperatures, which were seen as offsetting any catalytic advantages. Its operating temperature allowed it to be efficiently cooled by evaporation of pure water in the stack cooling plates. The resulting steam could be used in steam-forming NG, reducing the system heat requirements by using the waste heat. A further advantage became apparent as the program progressed. The higher the operating temperature of the fuel cell, the more resistant the anode became to the presence of carbon monoxide in the fuel gas. Poisoning of a hydrogen anode occurred at parts per million levels below 100 1C, whereas at 150 1C about 0.15% could be tolerated. Thus, a less complex water–gas shift gas cleanup system could be used in the PAFC. In the early research phase, the greatest requirement was for cost reduction, which meant cheaper materials in the PAFC stack and improved performance to better use

542

Fuel Cells – Phosphoric Acid Fuel Cells | Overview

them. In the first 3 years, it was discovered that carbon (especially if it was graphitized) was sufficiently stable in the PAFC electrolytes even at the cell potentials of the oxygen-reducing cathode to be used as a construction material. This was unexpected on the basis of conventional thermodynamic wisdom. However, it showed that materials that were apparently unstable in theory were not necessarily unstable in practice, owing to the effects of slow reaction kinetics, which made unstabilizing (e.g., corrosion) reactions so slow as to be negligible. This meant that carbons or graphites could be used for the bipolar plate instead of the boiler-plate gold-plated tantalum used in testing. Taking a cue from conventional gas-phase catalysis, where platinum electrocatalysts were stabilized in high-surface-area form on highly dispersed ceramic particles such as alumina, high-surface-area platinum was formed on a required high-area carbon support, conductive carbon black (later graphitized). The material of choice was Vulcan XC-72 R conducting furnace black (Cabot Corp., Boston, MA) with a surface area of about 250 m2 g1. The first patents for such preparations via sulfur chemistry (using bisulfites, not from Pratt & Whitney) date from 1973. These allowed platinum electrocatalyst surface areas of 100 m2 g1, five times that of the pure platinum metal black powders previously used. Oxygen electrodes (cathodes) could then be prepared with 5.0 g m2 of platinum, giving an equal performance to that of 200 g m2 of pure platinum black at 20 m2 g1 loading. Similar carbon-supported platinum electrodes at the much less demanding anode could be as low as 0.5 g m2, but typically were 2.0 g m2. By the early 1970s, somewhat more than 0.6 V could be achieved at 1.5 kA m2 under typical operating conditions. This corresponded to (then) a total of about $50 kW1 for platinum electrocatalyst (with platinum at $200 per Troy oz: in February 2008, it had reached $2300 per Troy oz after the great increase in commodity prices in 2007). As time progressed, operating temperatures were further increased, eventually to 200–205 1C, giving higher performance. This required all-graphite electrocatalyst supports and bipolar plates to resist corrosion. Initially, plates with solid machined ribs for reactant product flow were used, the anode and cathode ribs being at right angles in a crossflow arrangement. Attempts in the midto-late 1970s to use molded polymer (polyphenylene sulfide) composite plates were unsuccessful, and the next stage was the development of the ribbed substrate compound bipolar plate, consisting of a flat machined graphite separator plate prepared by molding 67 wt% graphite with 33% phenolic resin or pitch, heat-treating to 900 1C for carbonization, followed by high-temperature graphitization, with a final surface finishing step. On each side of this plate were porous ribbed substrates, with ribs facing the separator plate, carrying an electrocatalyst layer on the flat upper side. The ribbed porous substrates

were partially wet-proofed to allow gas transport, at the same time serving as reservoirs for phosphoric acid electrolyte, which slowly evaporates from the cells over time. The cell was completed by a polytetrafluoroethylene (PTFE)-bonded silicon carbide matrix of separator holding the electrolyte between the anode and cathode electrocatalyst layers. A schematic of the cell elements is shown in Figure 3. As in all UTC PAFC systems, the crossflow stack had four insulated external box-manifolds on the four sides to supply reactants and remove effluents. By the mid-1980s, three other developments occurred. The first was graphitized high-surface-area carbon-supported platinum base metal alloy cathode electrocatalysts (e.g., platinum–chromium–cobalt (Pt–Cr–Co)), which allowed an activity improvement corresponding to a 0.030 V increase in cathode potential at constant current. Normal thinking would lead to the belief that in the harsh corrosion environment of the PAFC cathode at 4200 1C, the base metal would rapidly dissolve. However, later work has shown that the alloy particles are in fact covered by a one-atom-thick protective layer of platinum atoms, and the performance improvement results from changes in electronic properties. The second Anode

S

Electrolyte

R

Cathode

S

B

C

B1

Figure 3 Schematic view of cell parts associated with cells in a generic planar bipolar phosphoric acid fuel cell (PAFC) stack. Electrolyte, anode, and cathode layers are shown. S,S are graphite anode and cathode gas ribbed substrates, which may be porous for electrolyte storage with polytetrafluoroethylene (PTFE) wet-proofing. R is a porous graphite electrolyte reservoir plate (containing some PTFE, shown here on the anode side only). B is a solid graphite bipolar separator plate. C (associated with its solid B1 separator plate) is a cell cooling plate (typically one each for every 6–8 active cells). The electrolyte is immobilized in a polyethersulfone-bonded silicon carbide matrix. For clarity, the gas channels are shown in parallel-flow configuration. In practice, they are usually in crossflow configuration.

Fuel Cells – Phosphoric Acid Fuel Cells | Overview

involved the development of variations on the ribbedsubstrate configuration, both at UTC and by other developers. The internal phosphoric acid reservoir ensured a 40 000 h stack lifetime, provided graphite corrosion was kept under control by never allowing the cell potential to exceed about 0.8 V under low-load or open-circuit conditions. Hot shutdown required a nitrogen supply for this, whereas a hydrogen supply was required for start-up. A final precaution to avoid damage was required on shutdown, when the stack will contain phosphoric acid of concentration corresponding to operating conditions (effluent water vapor pressure) at 205 1C. This concentration is greater than 100%, and pyrophosphoric acid (melting point (mp) 61 1C) has an effective concentration corresponding to 110% orthophosphoric acid (mp, 29.3 1C). If the electrolyte is allowed to solidify, damage to the electrodes may occur. Shutdown stacks must, therefore, be maintained at a temperature above the mean mp, generally at 38–41 1C. A third development was improvements to the matrix. The PTFE-bonded silicon carbide matrix had disadvantages, as it could use no more than 5–10 wt% PTFE, otherwise it was too hydrophobic and rejected the electrolyte. It therefore retained the electrolyte poorly and had a low bubble pressure (i.e., a tendency for gases to break through the porosity in the layer under a pressure differential). These problems were solved by the use of about 10–20 wt% of a new stable, hydrophilic polyethersulfone binder, in tests starting at the beginning of 1983, with a patent application in April 1985. Fuel Processing The fuel-processing system also turned out to be more challenging than expected. More high-cost heat exchangers (11–14) per plant were required than was originally thought. In addition, the trade-off between atmospheric pressure operation (with larger, higher cost piping diameters) and the use of higher pressures was not fully evaluated. Pressurization would give lower containment costs and higher stack performance, versus the negative aspect of parasitic energy requirements. Overall Efficiency Considerations A final major issue in fuel cell systems is the cascade of efficiency losses in each individual part of the system. Natural gas must be converted to hydrogen, which is then converted into DC electricity, which is in turn converted to AC. In spite of the fact that hydrogen can be converted in the fuel cell at rather high efficiency, all of the hydrogen in NG reformate cannot be converted, and must be profitably used elsewhere in the system. In a heat engine, all primary fuel can be directly converted into work, although at lower efficiency. This work then directly supplies AC power. Thus, fuel cell systems involve a cascade of losses, which

543

accumulate, and in some cases do not add much advantage to the efficiency of heat engines to give AC power. This represents an additional challenge to the fuel cell systems engineer. This is addressed in more detail in Fuel Cells – Overview: Introduction.

End of the TARGET Program, 1975 The result of the program was the testing of about 50 dispersed 12.5 kW units operating at 1.0 atma, designated as PowerCell 11 (PC11TM) during 1973–75. Their installation costs did not approach the program requirements, but they showed the possibilities of the technology. Their reliability did not approach the degree required, so it was considered that planned future units should have grid backup, as extra fuel cell backup would render them even more costly, and internal combustion engine or other conventional backup seemed to be pointless. However, 1973–75 included the Yom Yippur war, which was followed by a more than fivefold increase in oil prices, with a lesser escalation in the cost of US NG. This immediately changed the attitudes and consumers toward the possibilities of on-site generation. Before that war, US electricity use was increasing by 7.2% per year. Afterward, its growth was much less. The end of the 1975 TARGET testing and evaluation period showed that the technology should not be abandoned, as the concept of the cogeneration NG PAFC unit was a technical (if not an economic) success. However, certain changes of scale seemed appropriate to achieve stand-alone commercial use. Already, by 1972, Pratt and Whitney was proposing an immensely scaled-up 27 MW version of the system to utilities in modular packages. This would operate under pressurized conditions to increase current density (and operating cell potential) and thereby reduce cost and increase efficiency. For the PC11, it was also felt that for cost reduction, an increase in system size would be required. One aspect was the cost of the DC–AC converter. Another was the cost of cell repeat parts, as larger cell components were considered to be more cost effective than smaller ones. Experience had also shown that cell performance could be increased to 0.215 A cm2 at 0.65 V, with an increase in operating temperature to 190 1C, which resulted in an increase in carbon monoxide tolerance at the anode to 0.15%, further improving system economics. The above resulted in the abandonment of the 12.5 kW system for individual homes and its replacement by the atmospheric pressure 40 kW PC18TM system supported by the Gas Research Institute (Chicago, IL, USA) for application in individual commercial buildings. This included the latest catalytic and materials developments, together with larger area cells. Field tests of this system were conducted at 42 sites in the United States and Japan, starting in the late 1970s.

544

Fuel Cells – Phosphoric Acid Fuel Cells | Overview

Other Prototype Phosphoric Acid Fuel Cell Systems The 4.8 kW Demonstrators In parallel to the development of the PC18 system, which could in principle be grid-connected or not, a pressurized (3.4 atma, with 0.34 m2 cell area) higher efficiency system was offered in prototype form for electric utilities in centralized generation applications. Pressurization increases cell voltage at the same current density, thereby increasing electrical efficiency. In contrast, the pressurization of large amounts of nitrogen in the excess cathode air supply requires expenditure of considerable amounts of energy. Owing to compressor–expander inefficiencies, only part of this can be recovered on expansion back to 1 atma after use. In consequence, careful design with heat and steam recovery is required to give compression work, resulting in a net improvement in efficiency. Unlike the small atmospheric pressure prototypes, the 4.5 MW demonstrator could not be used for cogeneration applications, as all waste heat was required to supply energy for pressurization. A 1.0 MW prototype unit supplying grid power was successfully tested in 1976–77 at UTC. This used the unsuccessful polyphenylene sulfide-bonded graphite ribbed bipolar plates. This was followed by the construction of a 4.5 MW (AC, 4.8 MW DC) plant in lower Manhattan, which was intended to start in 1978, with testing to be complete by 1979. It was to be fueled by naphtha. Stack performance was expected to be 2.5 kA m2 and 0.65 V at 190 1C, which would have resulted in 36.7% overall naphtha-to-AC HHV efficiency (39.5% LHV). Its overall cost in third-quarter 2007 dollars was estimated at $47 000 kW1, of which the construction cost was $23 500 kW1 and the stack alone cost $6000 kW1. The US Department of Energy (DoE) provided 48% of the funding, the Electric Power Research Institute (EPRI, Palo Alto, CA, USA) 25%, UTC 20%, and the utility Consolidated Edison of New York 7%. The unit never generated electrical power, but its fuel processor and its corresponding emissions output were demonstrated after a prolonged licensing schedule terminated by heat exchanger failures after water pressure tests under freezing conditions at the end of 1983. By then, the stored PAFC stacks, which used the old solid ribbed bipolar plate configuration with no internal electrolyte reservoir, had reached end of life owing to loss of acid into the slightly porous plate, which had been corrected in the meantime by major redesign to a ribbedsubstrate configuration. The unit served to acquire a permit for fuel cells to operate under city conditions (including permitting the use of hydrogen storage for start-up). Permits for conventional combustion machinery would not be provided in future. This provided a model for the ‘blanket permitting’ without specific site permits for fuel cell units elsewhere in the United States, particularly in the South

Coast Air Quality Management District in the Los Angeles area and in the Bay Area in California. This demonstration was followed by that of a similar but less costly unit at the Tokyo Electric Power Company (TEPCO) site in Goi, Chiba Prefecture, Japan, up to 1985. The on-site engineering and construction was provided by Toshiba Corporation (Kawasaki, Japan). This contained stack improvements including the ribbedsubstrate configuration discussed earlier. Cell area remained at 0.34 m2. As already stated, with these pressurized units, all waste heat was used in the plant, with none available for cogeneration. Testing of this system was a success, which allowed UTC (since 1983 in a joint venture with Toshiba called International Fuel Cells (IFC)) to go forward to the next stage, which was a step in the direction of the proposed, but abandoned, modular 27 MW system of 1972. The PC23TM The initial code name for the proposed unit was Fuel Cell Generator No. 1 (FCG-1), which would operate at a mature value of 0.73 V at 2.5 kA m2 at 8.2 atma pressure using 0.93 m2 cells. The demonstration unit (and its expected successors) was called PC23TM. The first was planned to produce 11 MW at the TEPCO Goi site. This was achieved at an HHV efficiency of 41.8% (net AC) in 1991. However, a major problem occurred because of the use of reformer burner exit gas (depleted air with carbon dioxide and water vapor) rather than nitrogen (as was originally planned) as an inerting blanket in the stacks within their pressure vessels. The result was corrosion in six stacks, which had to be taken out of commission. This resulted in a life test at 67% of design capacity. The decay rates of the operating stacks were less than expected, indicating that expected stack life under optimum conditions might be as high as 100 000, rather than 40 000 operating hours. In March 1997, the plant was shutdown, without plans for a pressurized successor. Work at Westinghouse Electric Company and Successors In 1978–79, Westinghouse Electric Company (Large, PA, USA) had proposed a pressurized (3.4 atma, later 4.8 atma) air-cooled electric utility PAFC based on licensed DiGasTM cooling technology from Energy Research Corporation (Danbury, CT, USA). DiGas used a common air supply to both cathode and cooling manifolds. The more practical, modified system split the reactant and cooling air streams, using a separate cathode air supply to a bipolar plate with the so-called Z-configuration, in which anode and cathode gas entered the shorter side of one-half of a rectangular bipolar plate in parallel side-byside inlets and exited in opposite outlets. The cooling plates were supplied with a separate airflow along the long

Fuel Cells – Phosphoric Acid Fuel Cells | Overview

side for cooling air in an overall crossflow configuration to the combined reactant streams. The overall stack consisted of four 100 kW substacks, arranged so that the wider cooling air sides formed a central square manifold through which the hot exit cooling gas went downward, with the inlet coolant going upward on the outside of the stacks. By 1992, short stacks were under test, and a 375 kW module was tested at this time. A 16 MW unit (configurable from 3 to 50 MW) was proposed for use at sites where hydrogen was available. The first 400 kW test was to be at Norsk Hydro’s Rafnes (Norway) chlorine plant, with a second to be sent to Mitsubishi Heavy Industries for testing at the Kyushu Electric Power Company in Japan. However, this did not take place, because after $100 million had been spent by DoE and $50 million by Westinghouse (a total of about $220 million in thirdquarter 2007 dollars), DoE declared the product commercial and stopped funding. Westinghouse sold the technology in March 1993 to a new company, FuelCell Corporation of America. This in turn transferred it to another new company, HydroGen LLC (Jefferson Hills, PA, USA), in 2001 to aggressively market the technology. In 2005, this company merged for assets with Chiste Corporation (Vero Beach, FL, USA, ex-Dyna-Cam Engine Corporation, becoming HydroGen Corporation in 2005). It installed a production facility in Versailles, PA, USA, and will test a 400 kW, 1.0 m diameter stack assembly at the chlor-alkali plant of Ashta Chemical in Ashtabula, OH, USA, with a $1.25 million grant from the State of Ohio. Engelhard Industries Another US PAFC program at Engelhard Industries was intended to exploit its catalyst expertise. By 1985, it was oriented toward mobile use (e.g., forklift trucks, later experimental buses) and remote on-site applications to be operated on methanol. Its final stack design was generally similar to that of the PC18, but it used a unique bipolar plate design (the ABA configuration, consisting of two plates, each ribbed on one side, with the flat faces joined by a film of pyrolyzed resin). In addition, the stack was cooled by a dielectric fluid, rather than by water/ steam. The Engelhard effort was abandoned by the early 1990s. Work in Japan In the early 1980s, Japan was determined to aggressively develop the PAFC under its Moonlight energy independence program to reduce imported oil usage. However, the program would still (in this case) use imported fuel, namely LNG at a 1983 cost of about $11.00 (2006) MMBTU1. Its then 3-year-old national PAFC program reached a spending level of about $18 million (2007) in 1983. Its object was to develop Fuji Electric and Mitsubishi Electric (MELCO;

545

R&D Center and Works, Amagasaki, Hyogo, Japan) 1– 5 MW systems operating at 4 atma and 190 1C, and Hitachi (Hitachi, Japan) and Toshiba 10 þ MW units operating at 7–8 atma and 205 1C inspired by the UTC 4.5 MW demonstrator and their proposed FCG-1 design. These were not developed as anticipated. Meanwhile, Toshiba built a production line for 1.0 m2 cells of PC23 type for units that could be used in basements of downtown buildings. The PAFC Research Association (a consortium of gas and electric power companies) supported a 1.0 atma, 1.0 MW, 40% efficient Toshiba demonstrator with two 414-cell 1.0 m2 stacks operating at 0.25 A cm2 and 205 1C in Minatoku, Tokyo, and a 6 atma, 5.0 MW Fuji Electric plant with 0.8 m2 cells in six stacks at Amagasaki (Hyogo Province). Operation of both units started in March 1995. Neither had commercial results. Fuji Electric’s on-site programs are described in Applications – Stationary: Fuel Cells. During the 1980s, MELCO performed R&D with the object of producing a prototype 200 kW on-site unit. The 1989 version of the prototype at the Plaza hotel in Osaka had measured nitrogen dioxide (NO2) emissions of 4 ppmv at 7% oxygen in the reformer exhaust, with negligible sulfur dioxide (SO2) and particulates. This corresponds to 10.8 g NO2 per MWh. It operated at 0.65 V at 2.0 kA m2 at 80 and 50% fuel and air utilization at atmospheric pressure and 205 1C. Testing of the 25 metric ton unit started in March 1990, and 13 038 operating hours (1797 MWh) were acquired before Government and private program shut down on 31 October 1993. Testing included 60 start-ups and a longest continuous run of 2656 h. The unit’s electrical efficiency was 40% (LHV-AC) at full load and 35% at 25% load. The corresponding values for the total cogeneration efficiency were 89.5% and 76%. Performance was judged excellent: the trends showed performance reduction of 7% (0.65–0.6 V) at 40 000 h. Work in Italy In a parallel European development, after Italy had decided to curtail nuclear power after Chernobyl in 1986, an experimental plant was planned in 1988 that would use Fuji Electric Company (Yokosuka, Japan) stacks. It was constructed with two 8.2 atma, 670 kW PC23 stacks and Ansaldo srl (Genoa, Italy) site engineering and balance of plant in 1992. This 1.3 MW PRODE plant operated from 1995 to 1998, and was mothballed in 1999 after 6000 operating hours. The site also served to test a 0.5 MW Ansaldo MCFC using hydrogen-rich gas from the same reformer. Work Elsewhere Bharat Heavy Electrical Ltd (Ranipur, India) has developed and deployed two 25 kW PAFC modules, and Korea Gas is developing a 40 kW PAFC system.

546

Fuel Cells – Phosphoric Acid Fuel Cells | Overview

Summary of International Fuel Cell Phosphoric Acid Fuel Cell System Prototypes The apparent lesson of the pressurized PAFC systems for IFC was that the improved efficiency of pressurized operation was not worth the trouble. The next stage was the atmospheric pressure on-site PC25TM, a version of the 40 kW PC18 with larger area cells and other scaled-up components, with other improvements. Like the MELCO 200 kW unit, it was intended to service 200 kW locations, as grid- and non-grid-connected units providing cogeneration if clients desired. The 200 kW concept was first proposed to gas utilities in 1985–86 as a customer side of the gas-meter device. It was first intended to use 0.34 m2 cells (as in the 4.5 MW demonstrators), but with a new low-resistance substrate known as ‘Configuration B’ at 2.2 kA m1. This was later changed to 0.47 m2 and 2.5 kA m2. By improved cathode electrocatalyst development, this was later improved to 3.25 kA m1. It was expected that 0.65 V at atmospheric pressure would be much less aggressive from the corrosion viewpoint than operation at 0.73 V under pressurized conditions. However, units operating at open circuit were always protected by nitrogen-blanketing to eliminate the possibility of cathodes reaching potentials high enough to cause corrosion of cathode electrocatalyst supports. This lesson has been apparently lost on PEMFC developers. The first four prototype 200 kW units were designated PCX, and were successfully tested in Japan in various locations in 1988–89. They were followed in 1991–92 by the

ONSI PC25ATM, which had 55% of the volume of the PCX and 76% of the weight. They would have a total 90% LHV efficiency in cogeneration plants. For non-cogeneration units, a separate 3.2 metric ton, 12.5 m3 air-dump cooling unit was required. They were built in a plant with a capacity of 50 units per year, expandable to 200 units. It was hoped that 100 orders would be rapidly received, but this did not happen in a period of economic slowdown. In fact, only 56 orders were received worldwide, together with 18 of the similar PC25BTM used on military bases. This was in spite of the excellent performance of individual systems, including lower than predicted performance decay, individual continuous runs of more than 8000 h between minimal-time planned outages, and a low forced outage rate owing to mechanical and/or electrical problems translating to an availability often exceeding 90%. Its emissions (from the lean reformer burner) were about 2.4 g MWh1 of oxides of nitrogen and 3.2 g MWh1 of carbon monoxide, which are orders of magnitude below those of conventional equipment. Unlike conventional generating equipment, it could be operated within 7 days of being shipped to its concrete slab. The major problem was cost: the cost of the first 50 units was originally stated as $2500(1991) kW1 ($3560 kW1 in third-quarter 2007 dollars); however, this rose to about $4300(2007) kW1 in early units, then to $6250(2007) kW1 including installation, training, and O&M (for the Southern California Air Quality Management District in 1993). This occurred as the real assembly costs in limited production became known. A new model, the PC25CTM, was then announced, which became available in March 1995. This was intended

Figure 4 Photograph of a UTC Power 200 kW PC25 C PAFC system (renamed the 200 kW PureCell), decorated to illustrate its low environmental impact. Used with permission of UTC Power Corporation.

Fuel Cells – Phosphoric Acid Fuel Cells | Overview

to be a definitive precommercial device, incorporating many improvements compared with the PC25 A and B models, to give reduced production costs. It weighed 18 200 kg, compared with 27 300 kg for the PC25 A and 36 000 kg for the PCX prototypes. The PCX prototypes were 3.5 m high, 11.3 m long, and 3.5 m wide, whereas the PC25 A was 3.5  7.3  3.0 m and the PC25 C was 3.0  5.5  3.0 m. The changes involved a more compact reformer, more compact flat-sheet instead of tube-andshell heat exchangers, and an improved inverter. The optional air cooling unit used for non-cogeneration applications was 2.5  2.5  2 m and weighed 3200 kg. This was also reduced in size and weight. The 200 kW PC25 C was offered for sale by ONSI and its successor company UTC Power Corporation for almost 13 years, later under the designation 200 kW PureCellTM. At the end of 2008, it was to be replaced by a new 400 kW unit. A photograph of a 200 kW PC25 C decorated to illustrate its very low environmental impact is shown in Figure 4 (see Applications – Stationary: Fuel Cells).

Nomenclature Symbols and Units c E F

i K n P r R T DH

expected annual cost savings for total energy electrode potential Faraday constant, 96 450 C per mole of electrons or per molecular equivalent; 26.8 Ah per equivalent annual economic rate of inflation dissociation constant capital depreciation time (years) capital cost (principal) expected annual rate of return on capital the gas constant (8.314 J mol  1 K  1) absolute temperature (K) (= 273.16 + t 1C). change in enthalpy

Abbreviations and Acronyms AC AGA CSA DC DoE EPRI FCG-1 GE HHV

alternating current American Gas Association cell stack assembly direct current US Department of Energy Electric Power Research Institute Fuel Cell Generator No. 1 General Electric higher heating value

IFC IGT LHV MCFC mp NG O&M PAFC PC11TM PTFE SOFC TARGET

TEPCO UTC

547

International Fuel Cells institute of Gas Technology lower heating value molten carbonate fuel cell melting point natural gas operation and maintenance phosphoric acid fuel cell PowerCell 11 polytetrafluoroethylene solid oxide fuel cell Team to Advance Research in Gas Energy Transformation, a 1967 fuel cell research and development program Tokyo Electric Power Company United Technologies Corporation

See also: Applications – Stationary: Fuel Cells; Chemistry, Electrochemistry, and Electrochemical Applications: Oxygen; Fuel Cells – Direct Alcohol Fuel Cells: Direct Methanol: Overview; Fuel Cells – Exploratory Fuel Cells: Microbial Fuel Cells; Regenerative Fuel Cells; Fuel Cells – Molten Carbonate Fuel Cells: Overview; Fuel Cells – Overview: Introduction; Fuel Cells – Phosphoric Acid Fuel Cells: Anodes; Cathodes; Cells and Stacks; Electrolytes; LifeLimiting Considerations; Performance and Operational Conditions; Systems; Fuel Cells – Solid Oxide Fuel Cells: Overview; History: Fuel Cells.

Further Reading Appleby AJ (1986) Phosphoric acid fuel cells. Energy 11: 13--94. Appleby AJ (1996) Fuel cell technology: Status and future prospects. Energy 21: 521--653. Appleby AJ and Foulkes FR (1989) Fuel Cell Handbook. New York: Van Nostrand Reinhold. Blomen LJMJ and Mugerwa MN (eds.) (1993) Fuel Cell Systems. New York: Plenum Press. EG&G Technical Services, under contract to the US Department of Energy, Office of Fossil Energy, National Energy Technology Laboratory. Morgantown WV (2004) Fuel Cell Handbook, 7th edn. Springfield, MD: National Technical Information Service. Kordesch K and Simader G (1996) Fuel Cells and Their Applications. Weinheim: VCH Verlag. Larminie J and Dicks A (2003) Fuel Cell Systems Explained, 2nd edn. New York: Wiley. Liebhafsky HA and Cairns EJ (1963) Fuel Cells and Fuel Batteries, pp. 458–523. New York: Wiley. Penner SS, Appleby AJ, Baker BS, et al. (1995) Commercialization of fuel cells. Energy 20: 331--470. Vielstich W, Gasteiger HA, and Lamm A (eds.) (2003) Handbook of Fuel Cells: Fundamentals, Technology, and Applications. New York: Wiley.