PROMISE AND PROBLEMS OF SOLID OXIDE FUEL CELLS FOR TRANSPORTATION*
Romesh Kumar, Shabbir Ahmed, Michael Krumpelt, and Xiaoping Wang
Electrochemical Technology Program, Chemical Technology Division, Argonne National Laboratory
Paper to be presented at The International Symposium on Fuel Cells for Vehicles The 41st Battery Symposium in Japan November 2022, 2000
ABSTRACT
Fuel cells are being developed to provide clean, efficient propulsion power in
transportation applications. This paper discusses the promise and problems of the solid oxide
fuel cell (SOFC) for the automotive application. The SOFC system requires a simple, compact
fuel processor, even for complex fuels such as gasoline and diesel fuel. Because of the similar
operating temperatures, the fuel processor and the fuel cell stack can be close-coupled,
offering a high degree of thermal integration for a high net power density for the system.
Although the startup of the SOFC system would not be new-instantaneous, it would not
necessarily require more time than for other types of fuel cell systems operating on
conventional or alternative transportation fuels. Simulations of system performance show that
high efficiencies (-40%) should be achievable. Developments in advanced interconnect and
other cell and stack materials are expected to greatly improve the thermal and mechanical
ruggedness of the fuel cell stack. Although the balance-of-plant is relatively simple, some
components still need development, such as an air preheater offering high heat transfer in a
small volume.
1. INTRODUCTION
Most of the world's major automobile companies are exploring andlor developing fuel
cells for providing clean, efficient propulsion power onboard light-duty and heavy-duty
vehicles. While the phosphoric acid and the alkaline fuel cells have been used in limited
demonstration vehicles, the majority of fuel cell cars and buses are powered by the polymer
electrolyte fuel cell (PEFC). The major driver for the selection of the PEFC for automotive
use is its relatively low operating temperature of 80"C. Even at room ternperatme, the PEFC
can deliver useful power, IF hydrogen is the on-board fuel. Thus, the PEFC has been perceived
as being capable of near-instant start, similar to today's internal combustion engine.
So far, the solid oxide fuel cell (SOFC), which operates at 700-1000°C, has been
dismissed as being unsuitable for automotive applications because it does not offer near-
instant start from room temperature. While this may be true, it is not necessarily true that a
PEFC system can begin to deliver useful power faster than an automotive SOFC system if the
on-board fuel is other than hydrogen. In particular, if the fuel is gasoline or diesel, the startup
time for the PEFC-based system may be controlled by the startup time for the fuel processor
rather than by that of the fuel cell stack. Furthermore, the fuel processor and the balance-of-
plant in an automotive SOFC system are likely to be much smaller and simpler than in an
equivalent PEFC system. This paper discusses these and other features that recommend
SOFCS for automotive propulsion power. Also covered are the issues that must still be
resolved before the solid oxide fuel cell can be successfully developed for this application.
2. PROMISE OF SOFC FOR TRANSPORTATION
The SOFC system requires a relatively simple fuel processor, even for complex fuels such
as gasoline and diesel fuel. The reforming reactor only needs to break down the larger
hydrocarbon molecules to hydrogen, carbon monoxide, and carbon dioxide, along with lighter
hydrocarbon species. Since the carbon monoxide is not a poison for the SOFC (indeed, CO
can be electrochemically oxidized in the SOFC and, therefore, can be used as a fuel by itself),
there is no need for one or more water-gas shift reactors or for final CO clean-up devices,
such as preferential oxidation or methanation reactors. In addition to being simpler than what
is needed for PEFC systems, this fuel processor is considerably smaller, as discussed below.
Furthermore, the reforming reactor in an SOFC system does not need to achieve complete
conversion of the fuel to H2 and CO to obtain high system efficiencies. Indeed, designing the
reformer for incomplete conversion of the fuel makes it possible to use the water and heat
produced by the electrochemical cell reaction to steam-reform (within the stack) the
hydrocarbons in the reformate from the fuel processor. Thereby, the cooling requirements in
the fuel cell stack cari be reduced. Operating in this mode would lead to lower cathode air
flow rates with a decreased air-blower power requirement, potentially resulting in increased
system operating efficiencies. Thus, the reforming reactor itself can be much smaller than
would be needed for complete conversion of the fuel.
The SOFC-based automotive propulsion system offers excellent potential for thermal
integration between an autothermal reformer (ATR) for gasoline or diesel and the fuel cell
stack. Recent developments in ATR technology for gasoline refomiing have driven the
reforming temperature to the range of 7001100°C [11. `rhe lower end of this temperature
range is preferred for enhanced PEFC system efficiencies [2]. This temperature range is
similar to the range of operating temperatures for the solid oxide fuel cells under
development. Since the reformer and the fuel cell stack are two of the major components in
the SOFC system, this match in operating temperatures permits close coupling of the two,
reduces thermodynamic irreversibilities, and eliminates the need for the temperature matching
that is required in PEFC systems.
Solid oxide "fuelcell systems are also attractive for automobile propulsion because of their
high component and system power densities. Our analyses have shown that cross-flow, planar
SOFC stacks can achieve power densities above 1 kW/L, particularly for designs based on
anode- or interconnect-supported cell and stack structures. Since the entire SOFC system
consists of only a few major components, system-level power densities of >250 W/L can be
achieved, with efficiencies approaching or exceeding 40% based on the lower heating value
of gasoline fuel [3].
Furthermore, the SOFC makes possible an energy conversion system that is essentially all
solid state [4]. Because of the elevated temperature of operation, there is no liquid water
within the fuel cell stack, and no need for water management for the purpose of achieving the
humidification levels needed to maintain electrolyte conductivity. The absence of liquid water
and the all-solid-state fuel cell stack lead to a high degree of flexibility in system
configuration and component placement and orientation.
3. ISSUES WITH SOFC FOR TRANSPORTATION
Several major issues must be resolved before SOFCS can be used effectively in
transportation applications. These include the system startup time and fuel energy
consumption, sizes of the fuel cells and stacks, sizes of the balance-of-plant components,
thermal and mechanical ruggedness of the ceramic components, and the materials and designs
of the thermal enclosure for the elevated temperature components of the SOFC system.
The major component that governs the startup time and fuel consumption for the SOFC
system is the fuel cell stack. Although a single SOFC can be heated up very quickly without
damage, the mechanical constraints imposed by the stack housing and manifolds could dictate
the maximum safe rate at which the fuel cell stack can be heated up. With the appropriate
stack designs and materials, however, these constraints can be greatly reduced. For example,
with a metallic interconnect that also provides the structural support for the fuel cell stack,
stack heat up rates of 2°C/s would yield a startup time of approximately 400s for a stack
operating at 800°C. Based on the electrochemically active fuel cell area, gravimetric stack
power densities of approximately 1.0 kW/kg may be achievable with advanced stack designs
and materials. Allowing for the manifolding and other stack hardware, the corresponding
stack-level power density would be 0.5 kW/kg. Assuming an average heat capacity of the
stack to be 0.54 J/g-K (similar to that of zirconia in the range 25900 C), and a net efficiency
of 36% at rated power, the average fuel feed rate during the startup time of 400s would be
-75% of the value at the rated power.
This represents a significant amount of fuel consumed during startup from room
temperature, and would have to be considered along with the fuel cell system's duty cycle to
assess the viability of the SOFC system for a particular application. For vehicles with a
minimal duty cycle, say, 1 h of operation with two start-stop cycles over 24 h (i.e., a very
light-duty vehicle used only for commuting to and from work), the startup fuel consumption
could be comparable to the amount of fuel used during vehicle operation. With longer duty
cycles, the startup fuel consumption would be a progressively smaller fraction of the total fuel
consumed. The startup fuel consumption can also be reduced by appropriate thermal
insulation to reduce heat loss from the system during shutdown, such that the startup is from
warm rather than cold conditions.
A second major issue is the size of the cells, stacks, and other components in the
automotive SOFC system. Much of the "developmental activity in various types of planar
SOFC cells and stacks is with -100-cm2 active area cells (e.g., 10x10-cm square, or 11-12 cm
diameter circular). With single-cell power densities of 0.4-0.5 W/cmz, this translates to 40-
50 W/cell. A l-kW stack would then require 20-25 such cells. In practice, stack-level power
densities are often lower, thus requiring a larger active area in the stack. For the light-duty
vehicle application, fuel cell systems of 30-50 kW or higher power rating may be needed for a
fuel cell/battery hybrid power plant. These power requirements would dictate the use of larger
active area cells. Planar single cells as large as 40x40 cm have been developed. Also,
"window-pane" cell designs have been developed that have multiple cells on one plate. These
larger sizes are comparable to the cell sizes developed for automotive PEFC stacks.
The solid oxide fuel cells are air cooled. Therefore, the larger cells require a significant
increase in the size of the air flow channels to limit the parasitic power needed to overcome
the air-side pressure drop to 10% (or less) of the gross power generated by the cell. Our
analyses have shown, as discussed below, that by increasing the size of a thin-electrolyte,
cross-flow planar cell from 6x6 cm to 15x 15 cm, the net power increases from 21 W/cell to
105 W/cell, but the net power density decreases from 2 kwm to 0.6 kW/L of the active
volume of the stack. Using a systems approach, however, and absorbing some of the stack
waste heat in the endothermic steam reforming of part of the fuel, the cooling air requirements
may be reduced substantially. This would decrease the required air channel height and
increase the effective net power density, greatly reducing the active stack volume. The
specific magnitude of this reduction would depend on the design of the close-coupled
autothermal reformer and the fuel cell stack, but combined ATRktack net power densities of
-1 kW/L should be achievable.
The ceramic structure of the SOFC raises issues of thermal and mechanical ruggedness of
the individual cells, stack, and the whole assembly. While these issues have not been fully
addressed, some of the SOFC developments underway are expected to alleviate the related
problems. For example, the use of ferritic steels or other metallic bipolar plates in the lower
temperature SOFCS (those operating at 700-800°C) would increase the mechanical strength
and vibration/sliock resistance of the stack. Additionally, the use of flexible or compressible
gasketing [5] would permit some differential movement of the stack components, thereby
increasing the stack's mechanical ruggedness. At the cell level, use of an electrolyte of
partially stabilized 3 mol% Yz03-ZrOZ rather than fully stabilized 8 mol% Y203-ZrOz offers
much greater mechanical strength at only a relatively small 10SSin cell performance.
Similarly, other approaches in cell and stack design, such as the "windowpane" approach of
making multiple cells on each cell plate, would help to reduce the incidence of failures from
. thermal and mechanical stresses.
4. FUEL PROCESSORS FOR GASOLINE/DIESEL
For a fuel cell system to operate on a conventional or alternative transportation fuel (i.e.,
gasoline, diesel, methanol, ethanol, natural gas, or propane), the on-board fuel must first be
converted to hydrogen in a fuel processor [6]. The first step in such a process is to break down
the carbonaceous fuel into a gaseous mixture containing Hz, CO, COZ,HzO, Nz, etc. This may
be achieved by steam reforming, e.g., by the unbalanced reactions:
or by a combination of partial oxidation and steam reforming (autothermal reforming),
There may also be other byproduct and intermediate species, as well as unconverted input
fuel. These reforming ;eactions are carried out at elevated temperatures, from a relatively low
temperature of -250°C (for the catalytic steam reforming of methanol) to a high temperature
of- 1300°C (for the non-catalytic partial oxidation reforming of the liquid hydrocarbon fuels).
Although a fuel cell can operate on the hydrogen in this gas mixture, the low-temperature
PEFC requires that this raw reformate be processed further to reduce the CO content to only a
few parts per million (ppm) to avoid poisoning of the anode catalyst. This is generally carried
out in multiple steps. The bulk of the CO is converted to C02 and additional H2 by reaction
with steam according to the water-gas shift reaction,
The water-gas shift reaction is generally carried out over suitable catalysts in two stages, with
cooling between the stages to remove the heat of reaction. The CO content in the reformate
after the shift reaction is -0.5% (by volume). The final reduction of CO to <50-100 ppm is
achieved by preferential oxidation (with the injection of a controlled small amount of air) or
preferential methanation in a catalytic process:
Because of the relative reaction kinetics and the heat exchange required, the major fraction of
the total weight and volume of the fuel processor (typically, >80% [7,8]) is due to these CO
reductionhemoval steps and the associated thermal management components.
For a solid oxide fuel cell, on the other hand, the fuel processor does not need any of the
CO removal steps. In addition, as discussed above, even the initial reforming reactions (l),
(2), or (3) do not need to be carried to complete conversion. Thus, the fuel processor for an
automotive SOFC system is expected to be one-fifth to one-tenth the size of a fuel processor
for a PEFC system having comparable fuel processing throughput.
Fuel processors for automotive PEFC systems operating on methanol and hydrocarbon
fuels are being developed by several organizations [1,6,7]. A much simplified version of these
can be readily adapted for use in SOFC systems, with considerable savings in weight, volume,
and cost. An added significant benefit of the SOFC fuel processor is the much reduced startup
time. As discussed in [7], the startup time for the initial reforming reactor is approximately
one-tenth of the startup time for the total PEFC fuel processor. Therefore, the startup time for
an SOFC fuel processor would be only one-tenth (or less) that of an equivalent" PEFC fuel
processor.
5. CELL AND STACK DESIGNS AND MATERIALS
Various cell and st;ck designs are being explored by the different SOFC developers. The
largest SOFC power generators to date, 25100 kW, have been built by Siemens-
Westinghouse. These generators are based on their seal-less tubular cell design and operate at
a nominal temperature of 1000°C. Smaller (10-25 kw) systems have been developed by
Ceramic Fuel Cells Limited using planar fuel cell and stack designs. Fuel cell stacks of l
10 kW have been built by Sulzer-Hexis, Honeywell, McDermott, Tokyo Gas, Global
Thermoelectric, and others, using square or circular planar cell geometries.
The - 1000°C operating temperature of conventional solid oxide fuel cells limits the
materials that can be used for the various cell components. Lowering the operating
temperature of the SOFC increases the flexibility of materials selection and reduces cost,
while maintaining fuel cell performance. Consequently, a considerable amount of the present
R&D activity is focused on developing SOFC materials for use at 800°C, which is low
enough that alternative materials become viable. Further reduction in operating temperature to
700°C or even 600"C would increase the materials selection flexibility while yielding shorter
startup times and added cost savings. In addition, the lower SOFC operating temperatures
result in higher Nernst and cell voltages, leading to increased energy conversion efficiencies.
The reduced operating temperature would also improve mechanical robustness and reduce
thermal stresses and chemical interactions among the various cell components.
Successfully lowering the SOFC operating temperature does, however, require h
development of new materials andlor designs. The conventional yttria-stabilized zirconia
electrolyte and the nickel-zirconia anode can be used at temperatures down to 650"C, but the
electrolyte layer thickness must be reduced to 510 pm. At this thickness, the electrolyte-can
no longer function as the `structural support for the cell, and electrode-supported or
interconnect-supported cell and stack structures must be used. The cathode performance
decreases at lower SOFC temperatures due to the lack of ionic conductivity in the manganite-
based perovskites now used. Alternative materials being investigated include mixed
conducting perovskites containing Co, Fe, or Ni on the B-site, and mixed conducting layered
structures such as the high-critical temperature superconductors. Additionally, the newer
cathode materials may be usable as the supporting structure in the SOFC; the added flexibility
would result in certain benefits during cell and stack fabrication.
A significant change that can be realized in SOFCS operating at 800"C or below is the
move from ceramic to metallic interconnects. In addition to lower costs, metallic
interconnects would improve the mechanical strength of the stack and allow better thermal
management during operation. Development of metallic interconnects is still at an early stage.
The most significant p~oblem that they face is oxidation of the metal or ~loy on the air side of
the bipolar plate. New alloys and/or protective coatings are being explored to overcome this
problem.
6. STACK AND SYSTEM PERFORMANCE
We have analyzed the performance of a cross-flow planar SOFC cell and stack using an
electrochemical performance model developed earlier [9]. The analysis examined the
influence of various design and operating parameters on fuel cell efficiency, net power, and
net power density. The parameters we examined incIude the operating cell voltage (0.6-
0.85 V), geometric parameters (e.g., electrolyte thickness of 25200 pm), materials properties
(e.g., cathode interracial resistances of 0.064-1.0 Q cmz), cell size (6x6 cm, 9x9 cm, and
15x 15 cm), and fuel utilization (from 20% to 80%). The cell operation was constrained by
using constant average and maximum cell temperatures of 800°C and 850°C, respectively, and
limiting the maximum temperature gradient in the plane of the cell to 100°C/cm. In addition,
the parasitic power consumption to overcome the pressure drop in the air flow channels was
maintained at 10% or less of the gross power generated by the cell.
The base case was a 9x9-cm cell supported by a 200-pm-thick electrolyte and operating on
Hz/air with 80% fuel utilization. The calculated performance is a maximum net power density
of 0.68 kW/L (of active stack volume) at a cell voltage of 0.65 V and an efllciency of
36% LHV. In contrast, for an anode-supported cell (9x9 cm) with a relatively thin (25 pm)
electrolyte, the maximum net power density increases to 1.4 kW/L at 0.77 V, yielding an
efficiency of 43%. Thus, the thin electrolyte design increases both the maximum net power
density and the stack efficiency at that power density. Reducing the electrolyte thickness
further, to 5 pm, increases the maximum net power density to -1.5 kW/L at 0.8 V, for an
efficiency of 45%. However, these maximum power densities and efficiencies are reduced
significantly (to 0.9 kW/L and 39%, respectively), if the cathode interracial resistance is
increased from 0.064 Q cm2 to 0.5 Q cmz. Increasing the cell size increases the power
generated per cell, but decreases the stack's maximum net power density. For example, by
increasing the size of the anode-supported cell from 6x6 cm to 15x 15 cm, the maximum net
power increases by a factor of 5 (from 21 W to 105 W per cell), but the net power density
decreases by a factor of 3 (from 2 kW/L to 0.6 kW/L).
The SOFC has a particular advantage for use in automotive applications where the power
demands vary widely and rapidly over time. By decreasing the fuel utilization (i.e., providing
excess fuel to the fuel cell), the cell's power output can be increased sharply, even when
operating at the same cell voltage. This is due, in part, to a significant increase in the Nernst
voltage for the exit fuel and air conditions. For example, for a cell operating at 800°C with
H2/air, the exit Nernst voltage with 20% fuel utilization is 998 mV versus 874 mv with 80%
fuel utilization. Thus, our analyses show that with a fixed cell and stack geometry, the power
output can be increased by 55% or more without decreasing the cell voltage or violating any
of the operational constraints discussed above. Of course, operating with low fuel utilization
would decrease the overall system efficiency and would not be recommended except for
meeting high power demands for short periods.
We have also analyzed the performance of a relatively simple SOFC system operating on
diesel fuel for an auxiliary power unit [3]. The analysis would apply as well to an automotive-
sized system operating on gasoline. In addition to the fuel cell stack, the major components in
this system are a small catalytic autothermal reformer, a combined spent gas burner-air
preheater, a fuel pump, and an air blower. Since the autothermal reformer requires some feed
water, the system also includes a small condenser to recover the water from the anode
exhaust. Calculations were made for a system having 80% fuel utilization and a relatively
high air flow rate to the cathode to provide all the stack cooling with the air temperature rise
limited to 50"C. In this case, the overall system efficiency was 31%. By increasing the fuel
utilization to 90% and reducing the air flow rate to one-third the original value, the calculated
system efficiency was increased to 38910.With increased thermal integration to use the fuel
cell exhaust to preheat the inlet air, system efficiencies of 4070 or higher were obtained. Thus,
these analyses show that the projected high efficiency of the automotive SOFC system
operating on hydrocarbon fuel would be attainable.
7. OTHER CONSIDERATIONS
Other considerations of SOFCS for transportation applications include the sizes of the
various components and the issues of water management, heat rejection, and sulfur in the fuel.
The fuel cell stack is likely to be the most expensive single item in the system. It would,
therefore, be designed for the highest practical power densities. Assuming 0.4 W/cm* of
active cell area, a 40-kW gross rated power (36 kW net) system would require an aggregate
active area of 10 mz. The largest component in the system would be the spent gas burner-air
preheater, which the above analyses showed could require an active heat transfer area of 30 to
60 m2 (i.e., three to six times the active area in the fuel cell stack), depending on the operating
temperature difference driving forces. If this heat exchanger is of a conventional design that
offers -0.1 m2 heat transfer area per liter of volume, a 300 to 600 L unit would be needed.
Novel designs for compact heat exchangers offering much higher active area per unit volume, such as some microchannel concepts, would be needed to reduce the size of this heat
exchanger. All of the other components are considerably smaller and would not add
significantly to the total volume of the SOFC system.
The catalytic autothermal reformer requires the use of water to avoid excessively high
temperatures that would otherwise be needed to prevent carbon formation. For use in
automotive systems, this amount of water must be recovered from within the system. Overall,
excess water results from the conversion of the gasoline or other fuel. Recovering the needed
water, however, involves the use of a suitably placed, air-cooled condenser. If reforming
processes can be developed that do not need water as a reactant, or the cell reaction water can
be recycled to the reformer without phase change, the resultant SOFC system would be
inherently simple and easily controllable.
On the other hand, rejection of the waste heat from the SOFC system is straightforward.
All of the waste heat is rejected either in the process exhaust stream or at the air-cooled
condenser. The latter operates with high temperature-difference driving forces and is quite
compact, approximately one-fiftieth the size of the air preheater. Thus, heat rejection from the
SOFC system can be easily handled, even at high ambient temperatures.
One other significant consideration for automotive SOFC systems is the potential sulfur
content of the fuel. While some of the alternative transportation fuels are essentially sulfur-
free (e.g., methanol, ethanol, and propane), the petroleum-derived gasoline and diesel have
sulfur contents ranging from 300 to 1000 ppm by weight. Even the reduced sulfur fuels of the
future are likely to contain 30 to 80 ppm of sulfur. The reformate derived from these fuels
would have a sulfur content of 3 to 8 ppm of H2S by volume. This concentration may still be
too high for SOFCS using conventional materials, and sulfur-tolerant fuel cell materials and
designs would be needed. Without those, the SOFC system would need to incorporate sulfur
removal, which would add a significant degree of complexity to the system.
8. CURRENT DEVELOPMENTS
Improved materials, processes, and components for use in low-temperature (700-800"C)
SOFCS are being developed at many universities, national laboratories, and commercial
organizations around the world. These include research groups at the universities of
Birmingham, California (Berkeley), Keele, Hannover, Tokyo, and Pennsylvania, as well as
Northwestern University, Imperial College, Forschungszentrum-JNich, ECN (Netherlands),
Argonne National Laboratory,. Honeywell, McDermott, Global Thermoelectric, Sulzer Hexis,
and Ceramic Fuel Cells. Many commercial organizations are also developing complete
prototype systems for a variety of applications, including distributed generation, portable and
transportation power, and auxiliary power units.
Recent advances include improved cathode and interconnect materials, enhanced single-
cell and stack performances, near-commercial manufacturing of cells and stacks of designs
tailored to customer specifications, and demonstrations of prototypic systems of power output
rangin tomer specifications, and demonstrations of prototypic systems of power output
ranging from <100 W to >100 kW. Specifically for transportation applications, Delphi
Automotive Systems and Global Thermoelectric have recently reported the successful testing
of an SOFC-based auxiliary power unit for automotive use.
osti.gov ... original report |
