Electrical Energy Storage for the Grid: A Battery of Choices

Bruce Dunn et al.

Science 334, 928 (2011);
DOI: 10.1126/science.1212741

This copy is for your personal, non-commercial use only.

If you wish to distribute this article to others, you can order high-quality copies for your
colleagues, clients, or customers by clicking here.

Downloaded from www.sciencemag.org on July 6, 2012

Permission to republish or repurpose articles or portions of articles can be obtained by
following the guidelines here.

The following resources related to this article are available online at

www.sciencemag.org (this information is current as of July 6, 2012 ):

Updated information and services, including high-resolution figures, can be found in the online
version of this article at:
http://www.sciencemag.org/content/334/6058/928.full.html
Supporting Online Material can be found at:
http://www.sciencemag.org/content/suppl/2011/11/16/334.6058.928.DC1.html
A list of selected additional articles on the Science Web sites related to this article can be
found at:
http://www.sciencemag.org/content/334/6058/928.full.html#related
This article cites 42 articles, 2 of which can be accessed free:
http://www.sciencemag.org/content/334/6058/928.full.html#ref-list-1
This article appears in the following subject collections:
Materials Science
http://www.sciencemag.org/cgi/collection/mat_sci

Science (print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by the
American Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005. Copyright
2011 by the American Association for the Advancement of Science; all rights reserved. The title Science is a
registered trademark of AAAS.
 to peak shaving and load shifting, both of which
REVIEW can lead to improvements in grid reliability, sta-
bility, and cost (4). The electric power profile

Electrical Energy Storage shown in fig. S1 indicates how storage can in-
tegrate renewable resources and be used to ac-
commodate peak loads. Load shifting represents
for the Grid: A Battery of Choices one of the more tantalizing opportunities for EES
because of the benefit in storing energy when
Bruce Dunn,1 Haresh Kamath,2 Jean-Marie Tarascon3,4 excess power is generated and releasing it at
times of greater demand. The technical require-
The increasing interest in energy storage for the grid can be attributed to multiple factors, ments, however, are quite rigorous.
including the capital costs of managing peak demands, the investments needed for grid reliability, As indicated in Fig. 1, there are several en-
and the integration of renewable energy sources. Although existing energy storage is dominated ergy storage technologies that are based on bat-
by pumped hydroelectric, there is the recognition that battery systems can offer a number of teries. In general, electrochemical energy storage
high-value opportunities, provided that lower costs can be obtained. The battery systems possesses a number of desirable features, includ-
reviewed here include sodium-sulfur batteries that are commercially available for grid applications, ing pollution-free operation, high round-trip effi-
redox-flow batteries that offer low cost, and lithium-ion batteries whose development for ciency, flexible power and energy characteristics

Downloaded from www.sciencemag.org on July 6, 2012

commercial electronics and electric vehicles is being applied to grid storage. to meet different grid functions, long cycle life,
and low maintenance. Batteries represent an ex-
he August 2003 blackout in the Northeast pumped hydroelectric storage. This is far below cellent energy storage technology for the integra-

T and the recent September 2011 power fail-

ure that extended from Southern Califor-
nia to Mexico and Arizona are two of the more
the energy storage levels in Europe (10%) and
Japan (15%), where more favorable economics
and policies are in place (2).
tion of renewable resources. Their compact size
makes them well suited for use at distributed
locations, and they can provide frequency control
widely publicized examples in which power Energy storage technologies available for to reduce variations in local solar output and to
outages affected many millions of consumers. large-scale applications can be divided into four mitigate output fluctuations at wind farms. Al-
From a broader perspective, such power out- types: mechanical, electrical, chemical, and elec- though high cost limits market penetration, the
age events underscore the complex set of issues trochemical (3). Pumped hydroelectric systems modularity and scalability of different battery
associated with the generation and use of elec- account for 99% of a worldwide storage capac- systems provide the promise of a drop in costs in
tricity: the reliability of the grid, the use of fossil ity of 127,000 MW of discharge power. Com- the coming years. Today, sodium/sulfur (Na/S)
fuels and related carbon emissions, the develop- pressed air storage is a distant second at 440 MW. battery technology is commercially available for
ment of electric vehicles to decrease dependence The characteristics for several of these EES sys- grid applications, with some 200 installations
on foreign oil, and the increased deployment of tems in terms of power rating, which identifies worldwide, accounting for 315 MW of discharge
renewable energy resources. Underlying these potential applications, and duration of discharge power capacity. Moreover, there are emerging
considerations is the recognition that inexpen- are illustrated in Fig. 1. Potential grid applications opportunities for other battery systems because of
sive and reliable energy is vital for economic range from frequency regulation and load fol- potential low cost (redox-flow) and enhanced
development. Moreover, most of these issues are lowing, for which short response times are needed, performance [lithium (Li)–ion]. In this Review,
international in scope, with the additional caveat
that worldwide demand for electricity is projected
to double by 2050. UPS T & D grid support Bulk power
Electrical energy storage (EES) cannot pos- Power quality Load shifting management
sibly address all of these matters. However, ener- Pumped
hydro
gy storage does offer a well-established approach
Hours

Compressed air
for improving grid reliability and utilization. Flow batteries: Zn-Cl, Zn-Br Energy storage
Discharge time at rated power

Whereas transmission and distribution systems Vanadium redox New chemistries

are responsible for moving electricity over dis- NaS battery
tances to end users, the EES systems involve a Advanced lead-acid battery
High-energy
time dimension, providing electricity when it supercapacitors NaNiCl2 battery
is needed. A recent study identified a number
Minutes

Li-ion battery
of high-value applications for energy storage,
Lead-acid battery
ranging from the integration of renewable energy
sources to power quality and reliability (1). De- NiCd
spite the anticipated benefits and needs, there NiMH
are relatively few storage installations in opera-
Seconds

tion in the United States. Only ~2.5% of the total High-power flywheels
electric power delivered in the United States
uses energy storage, most of which is limited to High-power supercapacitors

1
Department of Materials Science and Engineering and Cali- 1 kW 10 kW 100 kW 1 MW 10 MW 100 MW 1 GW
fornia NanoSystems Institute, University of California, Los System power ratings, module size
Angeles, Los Angeles, CA 90095, USA. 2Electric Power Re-
search Institute (EPRI), Palo Alto, CA 94304, USA. 3Université
de Picardie Jules Verne, Laboratoire de Réactivité Chimie des Fig. 1. Comparison of discharge time and power rating for various EES technologies. The comparisons
Solides, Amiens 80039, France. 4Collège de France, Paris are of a general nature because several of the technologies have broader power ratings and longer
75231, France. discharge times than illustrated (1). [Courtesy of EPRI]

928 18 NOVEMBER 2011 VOL 334 SCIENCE www.sciencemag.org

 SPECIALSECTION
we present some of the overarching issues facing peak and baseload generation (fig. S1), allowing energy (7). Batteries, regardless of their chemistry—
the integration of energy storage into the grid and loads at any time to be serviced by the lowest aqueous, nonaqueous, Li or Na-based—store en-
assess some of the key battery technologies for cost energy resources (6). ergy within the electrode structure through charge
energy storage, identify their challenges, and pro- Storage solutions based on the technologies transfer reactions. By comparison, fuel cells, which
vide perspectives on future directions. we have today are so expensive that historically are not rechargeable, store energy in the reactants
it has been far more cost-effective to expand that are externally fed to the cells. Both of these
Utility Perspective on Energy Storage generation as well as transmission and distribu- differ from redox-flow cells, which store energy
EES has often been described as the “Holy Grail” tion to serve the peak load and provide sufficient in the redox species that are continuously circu-
of the electric utility industry. This phrase evokes operating margin to meet consumer demands for lating through the cells. Supercapacitors offer
the eagerness of utilities and other stakeholders reliability. In those cases in which storage is yet a different energy storage mechanism, via a
to achieve cost-effective storage options, which used, pumped hydroelectric plants are general- capacitive process arising from an electrochem-
could potentially cure many of the ills faced by ly involved. These plants are composed of low- ical double layer at the electrode-electrolyte in-
the electric power enterprise. However, the phrase cost materials (dirt, concrete, and water) that terface. Each mechanism has different strategies
Holy Grail also suggests that the search for en- have a lifetime of over 40 years, minimal main- that can be used to improve the power and energy
ergy storage will be long, difficult, and perilous. tenance costs, and relatively high round-trip ef- densities of the EES approach.
We are unlikely to find, at least in
the near term, a single technology Table 1. Energy storage for utility transmission and distribution grid support. The megawatt- and kilowatt-scale energy

Downloaded from www.sciencemag.org on July 6, 2012

that can repeatedly and efficiently storage systems listed here have potential impact in several areas, including transmission and distribution substation
store large quantities of electric grid support, peak shaving, capital deferral, reliability, and frequency regulation (1). [Courtesy of EPRI]
energy at low cost. On the other
hand, a portfolio approach that is Technology option Capacity Power Duration % Efficiency Total cost Cost
Maturity
(MWh) (MW) (hours) (total cycles) ($/kW) ($/kWh)
based on using a combination of
technologies may be the most ef- CAES Demo 250 50 5 (>10,000) 1950–2150 390–430
fective means to introduce and in- (aboveground)
tegrate energy storage. Advanced Demo 3.2–48 1–12 3.2–4 75–90 2000–4600 625–1150
The usefulness of EES stems Pb-acid (4500)
from the operational character- Na/S Commercial 7.2 1 7.2 75 3200–4000 445–555
istics of the grid as a supply chain (4500)
of a commodity, electric power. Zn/Br flow Demo 5–50 1–10 5 60–65 1670–2015 340–1350
At present, the electric power in- (>10,000)
frastructure functions largely as V redox Demo 4–40 1–10 4 65–70 3000–3310 750–830
a just-in-time inventory system (>10,000)
in which a majority of energy is Fe/Cr flow R&D 4 1 4 75 1200–1600 300–400
generated and then transmitted (>10000)
to the user as it is consumed. Zn/air R&D 5.4 1 5.4 75 1750–1900 325–350
Without the ability to store energy, (4500)
there must be sufficient generation Li-ion Demo 4–24 1–10 2–4 90–94 1800–4100 900–1700
capacity on the grid to handle (4500)
peak demand requirements, de-
spite the likelihood that much of
that capacity sits idle daily as well as for large ficiency (between 65 and 75%). Although there Although not discussed here, capacitive ener-
portions of the year (fig. S2). Correspondingly, are obvious limitations because of geographical gy storage offers some promising opportunities for
the transmission and distribution system must considerations, pumped hydro will be the bench- grid-scale applications (Fig. 1). Supercapacitors
also be sized to handle peak power transfer re- mark for grid-scale storage for years to come. provide higher power and longer cycle life than
quirements, even if only a fraction of that power In the near term, utilities are aware of the that of batteries and are receiving renewed atten-
transfer capacity is used during most of the rising need for EES solutions but are skeptical of tion as researchers try to better understand funda-
year. Operationally, electrical power generation the technologies that have been proposed to date. mental interfacial processes and improve energy
must be continuously ramped up and down to Even in cases in which technology has substan- density (8). The technology is of interest for power
ensure that the delicate balance between supply tial merit, the absence of cost-effective products quality applications, such as alleviating short-term
and demand is maintained. The up and down with a track record of safe and reliable operation disruptions of a few minutes until a generator,
cycling reduces power plant efficiency, resulting has made the industry skittish about their use. fuel cell, or battery can be placed in service. Be-
in higher fuel consumption and higher emissions Table 1 lists some of the current maturity levels cause the lifetime costs for supercapacitors can
per kilowatt-hour (kWh) produced. This proce- for various energy storage technologies, their be attractive (6), there is the prospect that this
dure also causes more wear on the equipment operational characteristics, and cost estimates. If technology will be used in conjunction with bat-
and reduces the lifetime of power plants (5). successful, the outcomes from these projects may teries so as to provide future grid storage solutions.
By decoupling generation and load, grid en- alleviate industry concerns of matters such as per- A battery is composed of several electrochem-
ergy storage would simplify the balancing act formance, cycle life, economics, and risks. Another ical cells that are connected in series and/or in
between electricity supply and demand, and on promising development is that the industry has parallel in order to provide the required voltage
overall grid power flow. EES systems have po- begun working to establish standards and targets. and capacity, respectively. Each cell is composed
tential applications throughout the grid, from of a positive and a negative electrode, which are
bulk energy storage to distributed energy func- Electrochemical Energy Storage where the redox reactions take place. The elec-
tions (1). The availability of energy storage Electrochemical energy storage approaches can trodes are separated by an electrolyte, usually a
would help to eliminate the distinction between be distinguished by the mechanisms used to store solution containing dissociated salts so as to

www.sciencemag.org SCIENCE VOL 334 18 NOVEMBER 2011 929

 enable ion transfer between the two electrodes.
Once these electrodes are connected externally,
the chemical reactions proceed in tandem at both

300
electrodes, liberating electrons and providing the
current to be tapped by the user (9, 10). The en- Li ion Li ion
ergy storage properties for most of the common LiFePO4-C Li(TM)O2-C

Specific power (W kg-1)

rechargeable batteries are shown in Fig. 2, with
additional details provided in table S1.

200
Nickel-
Lithium Ion Batteries metal
Li metal-
Sodium- polymer
The Li-ion battery (LIB) technology commer- hydride
metal
cially introduced by Sony in the early 1990s is chloride
based on the use of Li-intercalation compounds. Lead

100
acid Nickel-
Li ions migrate across the electrolyte located cadmium
between the two host structures, which serve as
the positive and negative electrodes (Fig. 3). Li- Sodium-sulfur
ion batteries outperform, by at least a factor of V redox flow
0

Downloaded from www.sciencemag.org on July 6, 2012

2.5, competing technologies [nickel (Ni)–metal
0 50 100 150 200
hydride, Ni-cadmium (Cd), and lead (Pb)–acid)]
Specific energy (W h kg-1)
in terms of delivered energy while providing high
specific power (Fig. 2). The overwhelming ap-
peal of Li-electrochemistry lies in its low molec- Fig. 2. Gravimetric power and energy densities for different rechargeable batteries. Most of these
ular weight; small ionic radius, which is beneficial systems are currently being investigated for grid storage applications.
for diffusion; and low redox potential [E°(Li+/Li) =
−3.04 V vs standard hydrogen electrode (SHE)]
(11). The latter enables high-output voltages and
therefore high-energy densities. Such attractive
properties, coupled with its long cycle life and A
rate capability, have enabled Li-ion technology to
capture the portable electronics market and make
in-roads in the power tools equipment field. LIBs e e
Anode Cathode
are also regarded as the battery of choice for pow-
ering the next generation of hybrid electric vehi- - Electrolyte
+
cles (HEVs) as well as plug-in hybrids (PHEVs),
provided that improvements can be achieved in
terms of performance, cost, and safety (12). Be- e e
cause long-term stability, high-energy density,
safety, and low cost are common to developing
batteries for both automotive and grid applica- e e
tions, considerable synergy should exist between
the two areas, although there will be certain dif-
ferences. Figures of merit for electric vehicle ap- e
plications call for a reduction in the price per e
kilowatt-hour by a factor of 2 and a doubling of
the present energy density. The realization of e
such goals will be beneficial for grid storage e
systems, although with probably more emphasis
e
on cost and less on energy density. Other dif-
ferences between the two technologies include
safety, which is easier to achieve in stationary sit-
uations than in mobile ones, whereas long cycle
life is a key factor for grid applications. LIBs for Cu Al
current current
vehicles require versatility in their energy and
collector collector
power capabilities in order to meet the needs of
the various types of electric vehicles and the as- Graphene Li+ Solvent LiMO2 layer
sociated performance requirements, whereas LIBs structure molecule structure
for the grid are likely to be modular. Fig. 3. Schematic of a LIB. The negative electrode is a graphitic carbon that holds Li in its layers, whereas
A number of advances have been made in the positive electrode is a Li-intercalation compound—usually an oxide because of its higher potential—
the LIB field by controlling particle size in ad- that often is characterized by a layered structure. Both electrodes are able to reversibly insert and remove
dition to composition, structure, and morphology Li ions from their respective structures. On charging, Li ions are removed or deintercalated from the
in order to design better electrodes and electrolyte layered oxide compound and intercalated into the graphite layers. The process is reversed on discharge.
components (13). Decreasing electrochemically The electrodes are separated by a nonaqueous electrolyte that transports Li ions between the electrodes.
active materials to sub-micrometer and smaller [Derived from (4)]

930 18 NOVEMBER 2011 VOL 334 SCIENCE www.sciencemag.org

 SPECIALSECTION
sizes combined with carbon-coating approaches ly promising and suggests that the performance little doubt that rechargeable Li-air cells either
to achieve core-shell morphologies has led to of organic electrodes could become comparable for electric vehicles or grid storage applications
new directions in electrode materials (14). Reac- in gravimetric energy density, life cycle, and pow- still have a long research and development path.
tion mechanisms and materials systems that were er rate to today’s best inorganic electrodes, with The prospect of developing Li-ion technol-
previously discarded are being reconsidered for the distinct advantage of providing a botanic al- ogy for both transportation and stationary storage
the next generation of LIBs. Moving from bulk ternative to the mineral approach currently in raises the issue of whether the demand for lithium
materials to nanosize particles has enabled (i) practice. will affect the existing world reserves. Na is an
the ability to use new Li-reaction mechanisms, At the research level, there is interest in re- attractive alternative because its intercalation
in which conversion-reaction electrodes show chargeable LIB systems that have significantly chemistry is similar to that of Li, there are ample
enormous capacity gains (15); (ii) the use of neg- higher energy densities (22, 23). Although the reserves, and its cost is low. These advantages are
ative electrodes based on alloy reactions—Tin Li-O2 system has been available for many years partially offset by the gravimetric energy density
(Sn)–based LIB technologies have already reached as a primary battery, the prospect of developing it penalty for using Na, which is both heavier and
the marketplace (such as NEXELION), and Si- into a reversible (secondary) battery has become less electropositive than Li. The development of
based ones are emerging (16); (iii) the identifica- tantalizing because of a projected three- to four- room-temperature Na-ion cells that are cost-
tion of poorly conducting polyanionic compounds fold increase in gravimetric energy density as effective, sustainable, and environmentally benign
or fluorine-based compounds that exhibit excel- compared with the current Li-ion technology (24). will require a new generation of Na-intercalation
lent electrochemical performance (17); and (iv)

Downloaded from www.sciencemag.org on July 6, 2012

the transformation of the poorly conducting lith-
ium iron phosphate (LiFePO4) insertion electrode e- e-
into perhaps the most valued electrode material
for electric vehicle applications (18). LIBs based
on LiFePO4 are extremely attractive because of
- +
safety and cost. The former arises from the fact
that the operating voltage of the LiFePO4 system O2
is compatible with the thermodynamic stability
of the electrolyte, whereas the latter is based on
the use of abundant and low-cost constituents. Li+
In addition to being an attractive LIB for the elec-
tric vehicle market, LiFePO4-based batteries are
being evaluated in stationary energy storage dem-
onstration projects (1). Discharge
A substantial segment of the battery materials
O2
community is moving toward developing electrode
materials on the basis of abundance and availabil- +
Li
ity of the relevant chemicals. Materials centered
on sustainable 3d metal redox elements such
as manganese (Mn) [lithium-manganese oxide
(LiMn2O4)], Fe (LiFePO4, Li2FeSiO4) and ti-
tanium (Ti) (TiO2, Li4Ti5O12), and made via
eco-efficient processes, are receiving increased Lithium Electrolyte Composite
attention (19). In addition, there is resurging in- electrode
terest in low-temperature–solution chemistry routes
Li2O2 Catalyst Carbon
in which hydro(solvo)thermal, ionothermal, and
bio-mineralization processes are used to prepare
electrode materials at temperatures >500°C lower Fig. 4. The Li-air cell uses Li as the anode and a cathode consisting of a porous conductive composite,
than traditional powder synthesis (20). usually carbon and a catalyst, that is flooded with electrolyte. Oxygen from the atmosphere dissolves in the
Life cycle costs represent another important electrolyte and is reduced. On discharge, Li ions pass through the electrolyte and react with the reduced
consideration. A foreseeable strategy for battery oxygen. The process is reversed on charging. Either aqueous or nonaqueous electrolytes can be used. For
processing will involve the use of electro-active the former, a Li-ion–conducting solid electrolyte separates the metallic Li from the aqueous electrolyte.
organic electrode materials synthesized from
“green chemistry” concepts through low-cost pro- However, the volumetric energy density may not compounds (30). The knowledge gained from
cesses free of toxic solvents; this will also enlist be much greater than that of Li-ion batteries (25). developing Li-ion insertion electrodes should
the use of natural organic sources [carbon dioxide The renewed interest in this system can be be applicable here. Thus, the demonstration of
(CO2)–harvesting entities] as precursors, which traced to the rechargeable behavior demonstrated a viable Na-ion technology for stationary energy
will be biodegradable and easily destroyed by in a nonaqueous Li-O2 system (Fig. 4) (26). storage should come well before that of Li-air
combustion (providing CO2) so that the battery Although there has been considerable progress technology because of the accumulated experi-
assembly/recovery processes will have a mini- in the past 5 years in the area of electrode ma- ence with Li-ion technology and high-temperature
mal CO2 footprint. Proof of this concept was dem- terials and architectures (27, 28), a number of Na battery technologies.
onstrated with the development of renewable fundamental problems still need to be addressed,
organic electrodes belonging to the family of and it is difficult to anticipate which of the ad- Sodium-Sulfur and Sodium-Metal
oxocarbons (Li2C6O6) or carboxylates (Li2C8H4O4) vanced Li-O2 aqueous and/or Li-O2 nonaqueous Halide Batteries
and the assembly of the first eco-compatible LIB systems will be able to achieve capabilities be- High-temperature Na-based battery technologies
laboratory prototype (21). This work is extreme- yond today’s Li-ion batteries (29). Thus, there is can be traced back to the 1960s, when researchers

www.sciencemag.org SCIENCE VOL 334 18 NOVEMBER 2011 931

 at Ford discovered that a common ceramic re-
fractory, sodium b-alumina (NaAl11O17), ex-
hibited extremely high ionic conductivity for Na Beta Beta alumina
ions (31). At 300°C, the ionic conductivity for alumina
tube V
NaAl11O17 approaches that of the aqueous elec-
trolyte, H2SO4, suggesting the possibility of - +
Sulfur electrode
using NaAl11O17 as a solid electrolyte in a high-
temperature electrochemical cell. Although sol-
ids with high ionic conductivity had been known
previously, none had b-alumina’s combination Sulfur
container
of chemical and thermal stability and low elec-
tronic conductivity. The recognition that inor- Discharge
Molten
ganic materials with high vacancy concentrations Na
could exhibit “fast ion conduction”—many or-
+
Na
Na+ Charge Protection
ders of magnitude greater than traditional alkali layer
Discharge
halides—led to the development of the field
known as solid-state ionics.

Downloaded from www.sciencemag.org on July 6, 2012

The two high-temperature Na batteries, Na/S
and Na-metal chloride (Na/MeCl2), are based on
using b-alumina as a Na+-conducting membrane
between two liquid electrodes (32). The batteries
operate at temperatures of 270 to 350°C so as to
take advantage of the increased conductivity of
Fig. 5. Schematic of the Na/S battery. The central Na design has molten Na (negative electrode)
the b-alumina at elevated temperatures and en-
contained within a Na b″-alumina solid electrolyte tube with molten S (positive electrode) surrounding
sure that the active electrode materials are molten.
the tube. The S electrode includes carbon in order to provide sufficient electronic conduction to carry
During discharge in the Na/S battery, Na is ox- out the electrochemical reactions. The magnified cross section of the cell shows the direction of Na+
idized at the solid electrolyte interface, and the transport through the b″-alumina electrolyte. On discharge, Na combines with the S to form Na
resulting Na+ migrates through the electrolyte to polysulfides. These reactions are reversed during charge, and Na returns to the interior of the tube.
react with S that is reduced at the positive elec-
trode, forming Na2S5 (Fig. 5). Initially, a two-phase
liquid is formed because Na2S5 is immiscible in which “blocks” of closely packed Al-O are a vital concern because it leads to cell failure,
with S at these temperatures. Over half of the separated by “conduction planes” (35). The latter whereas poor control of the ceramic micro-
discharge occurs in the two-phase region, where are loosely packed layers that contain the mo- structure results in interfacial reactions with the
the open-circuit voltage is 2.08 V (33). During bile Na+ along with O2– ions that bridge adjacent reactants. Large-scale production of b″-alumina
charge, the Na polysulfides are oxidized, and blocks. Ion motion occurs in two-dimensional has been established, but production yields and
when the Na content falls below Na2S5, the two honeycomb-like pathways around the bridging costs are major concerns (38). Other critical bat-
phase-region of Na2S5 and S reappears. In this oxygen. The polycrystalline b″-alumina tubes tery components are seals, which must not only
case, the formation of S must be managed ap- used in the Na/S and Na/MeCl2 batteries do not be hermetic in the 300 to 350°C range but also
propriately, or else the S can deposit on or near exhibit the anisotropic transport properties of withstand the vapor and/or actual contact with
the electrolyte, increasing cell resistance and lim- single crystals because the fine-grained, ran- the highly reactive molten electrode materials.
iting the amount of charging. domly oriented microstructures effectively elim- Recent activities in this area have involved the
Early in its development in the 1980s, the inate the anisotropy. Nonetheless, there are grain development of glass-ceramic sealing materials
Na/MeCl2 battery was nicknamed the ZEBRA boundary and tortuosity effects so that the con- whose thermal expansion coefficient matches
battery partially because of its scientific origins in ductivity of single-crystal Na b″-alumina at 300°C, that of a- and b-alumina components (39). There
South Africa, although its acronym stands for ~1 S cm−1, is three to five times greater than the is also the issue of identifying a low-cost ma-
Zero-Emission Battery Research Activities. The corresponding polycrystalline material (32). A terial for containing the molten positive electrode.
positive electrode in this battery is a semisolid recent study suggests that tortuosity effects can The corrosion problem is particularly difficult
combination of an electrochemically active metal be diminished because Na b″-alumina electrolytes for Na/S batteries because both S and polysulfides
chloride such as NiCl2 and a molten secondary in a planar configuration exhibit higher ionic are highly corrosive. The deposition of corrosion-
electrolyte, NaAlCl4, which conducts Na+. Dur- conductivity than that of tubular materials (36). resistant coatings such as carbides onto inexpen-
ing discharge, metallic Na is oxidized at the solid From inception, development for both sys- sive substrates has proven successful (40).
electrolyte interface. Na+ ions are transported tems targeted stationary energy storage and Na/S battery technology has been commer-
through the b-alumina electrolyte to the cathode electric vehicles. As a result, the technologies cialized in Japan since 2002, where it is largely
via the molten NaAlCl4. The solid metal chloride share a number of common features (and chal- used in utility-based load-leveling and peak-
is converted into NaCl and the parent metal (Ni lenges), even though specific designs differ some- shaving applications. Among the advantages
in the case of NiCl2). The open-circuit voltage is what. In both cases, the b″-alumina ceramic identified for stationary storage are its relatively
2.58 V (34). On charge, the Ni is oxidized, and tubes are acknowledged to be the key element small footprint (a result of high energy density),
the charge capacity is determined by the amount for determining battery operation and cost. Con- high coulombic efficiency, cycling flexibility,
of NaCl available in the cathode. siderable development effort has gone into es- and low maintenance requirements (41). The
Both batteries are based on the ion trans- tablishing large-scale manufacturing processes production of megawatt-size energy storage bat-
port properties of the b-alumina family of ma- for automating the fabrication of high-quality teries has involved considerable effort on such
terials. The high ionic conductivity of these ceramics with appropriate mechanical and elec- interrelated issues as electrical networking, cell
materials is the result of an unusual structure trical properties (37). Fracture of the ceramic is reliability, thermal management, and safety (42).

932 18 NOVEMBER 2011 VOL 334 SCIENCE www.sciencemag.org

 SPECIALSECTION
To provide appropriate voltages, energy, and attractive feature is that power and energy are issues associated with the lack of appropriate mem-
power, cells are assembled in series-parallel con- uncoupled, a characteristic that many other elec- branes for controlling long-term ion cross-over ef-
figurations to form modules, and the modules trochemical energy storage approaches do not fects. Designing better membranes is necessary,
themselves are connected in series-parallel ar- have (48, 49). This gives considerable design but whether such membranes can be of low cost
rangements to form batteries. This networking flexibility for stationary energy storage applica- is far from certain. Another important issue with
approach is designed to minimize the effect of tions. The capacity can be increased by simply redox-flow systems is that the currently used redox
individual cell failures. Modules are thermally increasing either the size of the reservoirs hold- couples, even with enhanced solubility, are limited
insulated and equipped with auxiliary heaters in ing the reactants or increasing the concentration to concentrations of about 8 M. This feature is
order to maintain a minimum operating temper- of the electrolyte. In addition, the power of the largely responsible for the fact that redox-flow
ature. Thermal management is especially chal- system can be tuned by either (i) modifying the systems do not surpass 25 Wh kg−1 (Fig. 2). The
lenging. The internal temperature of a module numbers of cells in the stacks, (ii) using bipolar identification of lower-cost redox couples with
increases on discharge because of joule heating electrodes, or (iii) connecting stacks in either par- high solubility would seem to be an essential de-
and exothermic cell reactions, whereas during allel or series configurations. This provides mod- velopment in order for this technology to succeed.
charge, there is a gradual cooling largely be- ularity and flexible operation to the system. Researchers recognize that redox-flow ap-
cause of the cell endothermic reaction (41). Despite the apparent advantages for redox- proaches represent potentially new directions for
The Na/MeCl2 batteries were developed al- flow batteries, application of this technology to increasing energy density. The semisolid Li battery
most exclusively for electric vehicles. At the time

Downloaded from www.sciencemag.org on July 6, 2012

of their development, the technology seemed to
offer certain advantages over Na/S in terms of
tolerance to overcharge and overdischarge, the
Ion-
ability to assemble cells in the discharged state, selective
a safe low-resistance failure mode, and poten- Electrode membrane
tially easier solutions for corrosion and sealing
Electrolyte Electrolyte
(42). Only recently have these batteries been di- tank + tank
H
rected at potential utility applications (43). O2- V4+ H
+ H+

H+
+ 3+
H O 2 -H +
V
Redox-Flow Batteries +

+ 2- 4 H+
-
+

Redox-flow batteries also have their origins in the O V O 2-

H+ H
+ H
2-
O +
e
Catholyte V2 Anolyte
1960s, with the development of the zinc/chlorine 4+ 5+
H
+

V /V 2- H+ V2+/ V3+
(Zn/Cl) hydrate battery. As a general description, O + H +

V5
+
O2- H O 2-

a redox-flow cell uses two circulating soluble e +

H
+
V2
redox couples as electroactive species that are H
+

oxidized and reduced to store or deliver energy

(44). By comparison, batteries rely on internal
solid electrodes to store energy. e
The flow-cell assembly (Fig. 6) has an ion- e
selective membrane separating the positive and Pump Pump
negative redox species, which are contained in sep-
arate storage tanks. During operation, redox-active
ions undergo oxidation or reduction reactions when
they are in contact or close proximity to the cur-
Power
rent collector; the membrane allows the transport source-load
of non-reaction ions (such as H + and Na+ ) to
maintain electroneutrality and electrolyte balance.
Since the 1970s, numerous types of redox Fig. 6. Schematic of the various components for a redox-flow battery. The cell consists of two electrolyte flow
flow battery systems have been investigated (45). compartments separated by an ion-selective membrane. The electrolyte solutions, which are pumped con-
A partial list includes iron/chromium, vanadium/ tinuously from external tanks, contain soluble redox couples. The energy in redox-flow batteries is stored in the
bromine, bromine/polysulfide, zinc-cerium, zinc/ electrolyte, which is charged or discharged accordingly. In practice, individual cells are arranged in stacks by
bromine (Zn/Br), and all-vanadium. The all- using bipolar electrodes. The power of the system is determined by the number of cells in the stack, whereas the
vanadium (1.26 V) and Zn/Br (1.85 V) systems energy is determined by the concentration and volume of electrolyte. In the vanadium redox-flow battery
are the most advanced and have reached the shown here, the V(II)/V (III) redox couple circulates through the negative compartment (anolyte), whereas
demonstration stage for stationary energy stor- the V (IV)/V(V) redox couple circulates through the positive compartment (catholyte). [Derived from (38)]
age. Interest in the all-vanadium system is based
on having a single cationic element so that the stationary energy storage is still uncertain. One demonstrated by Massachusetts Institute of Tech-
cross-over of vanadium ions through the mem- principal reason is that redox-flow systems have nology researchers uses electrode materials identical
brane upon long-term cycling is less deleterious been limited to relatively few field trials. In con- to those found in the LIB, but now the electrode
than with other chemistries (46). trast, other battery technologies have benefited materials are conducting inks (for example, sus-
Redox-flow batteries possess a number of from extensive experience in the development of pensions of LiCoO2 and of Li4Ti5O12 powders
advantages (47). The simplicity of the electrode products for portable electronics and automotive in nonaqueous electrolyte solutions) rather than
reactions contrasts with those of many conven- applications. A related disadvantage of flow bat- solids (50). The inks circulate separately on either
tional batteries that involve, for example, phase teries is the system requirements of pumps, sen- side of a membrane that regulates the Li-ion trans-
transformations, electrolyte degradation, or elec- sors, reservoirs, and flow management (48, 49). port between positive and negative electrodes. Both
trode morphology changes. Perhaps their most From a technical standpoint, there are reliability half cells and full cells have been demonstrated.

www.sciencemag.org SCIENCE VOL 334 18 NOVEMBER 2011 933

 The novel feature here is the use of redox-active products through a relatively simple manufactur- Note added in proof: Na/S batteries were re-
materials in suspension so as to circumvent the ing process and installed with few special re- sponsible for a fire that occurred at a power plant
problem of the relatively low solubility of the quirements. Operations and maintenance costs in Joso City (Ibaraki Prefecture) on 21 September
metal ion redox couples in aqueous solution. are also important; these costs are often tied to the 2011 (www.ngk.co.jp/english/news/2011/1028_01.
The flowable inks will be in the 10 to 40 M range, durability and lifetime of the energy storage solu- html). Although the cause of the fire is still un-
which is at least 5 times higher than traditional re- tion, for which the lifetimes of most assets are der investigation, this event underscores the fact
dox flow systems. Combining the higher materials measured in decades. Last, a premium will be that safety issues for Na/S batteries have not been
concentration with the feasibility of achieving 4-V placed on energy-efficient systems that do not lose completely resolved.
working systems is likely to lead to considerable energy through self-discharge or parasitic losses.
improvement in energy density, perhaps without With so many potential financial considerations, References and Notes
substantially affecting power density. it is not surprising that cost is given as the reason 1. “Electrical energy storage technology options” (Report
1020676, Electric Power Research Institute, Palo Alto, CA,
Another Li-ion–based flow system was dem- that energy storage is not widely used on the grid. December 2010).
onstrated recently by Goodenough and colleagues. The battery systems reviewed here satisfy 2. EPRI-DOE Handbook of Energy Storage for Transmission and
In this design, an aqueous cathode operating in a several, but not all, of the energy storage criteria Distribution Applications (1001834, EPRI, Palo Alto, CA,
flow-through mode was separated from a me- mentioned above. Na/S is commercially viable, and the U.S. Department of Energy, Washington, DC, 2003).
3. G. L. Soloveichik, Annu. Rev. Chem. Biomol. Eng. 2,
tallic Li anode by a Li-ion–conducting solid elec- and if this emerging technology follows patterns 503 (2011).
trolyte and an organic liquid electrolyte (51). This similar to others, costs can be expected to de- 4. “Basic research needs for electrical energy storage”

Downloaded from www.sciencemag.org on July 6, 2012

redox-flow system used an aqueous cathode con- crease as more production and operational ex- (Office of Basic Energy Sciences, U.S. Department of
taining 0.1 M K3Fe(CN)6 and demonstrated high- perience is gained. The technology, which is more Energy, Washington, DC, July 2007).
5. “Power Generation from Coal: Measuring and Reporting
ly efficient energy storage at 3.4 V. The design than 30 years old, needs to integrate some of the Efficiency Performance and CO2 Emissions” (International
strategy presented here offers some noteworthy scientific advances that have taken place in the Energy Agency, October 2010).
advances: (i) Li+ ion transport in solution is en- design of materials, creating new electrode ar- 6. “Energy Storage for the Electricity Grid: Benefits and Market
hanced as compared with that in a solid insertion chitectures and identifying new chemistries to Potential Assessment” (Report SAND 2010-0815, Sandia
National Laboratories, Albuquerque, NM, February 2010).
cathode and (ii) the absence of structural changes provide safe operation. Lowering the Na/S oper- 7. M. Winter, R. J. Brodd, Chem. Rev. 104, 4245 (2004).
during charge/discharge is beneficial for long- ating temperature is one topic that will affect the 8. P. Simon, Y. Gogotsi, Nat. Mater. 7, 845 (2008).
term cycling. The first laboratory prototypes were technology. Moreover, these advances will ben- 9. J. B. Goodenough, Y. Kim, Chem. Mater. 22, 587 (2010).
10. J. M. Tarascon, M. Armand, Nature 414, 359 (2001).
limited by low solubility of the metal-ion redox efit Na-ion technology, which is of growing inter- 11. V. Etacheri, R. Marom, R. Elazari, G. Salitra, D. Aurbach,
couple in the aqueous solvent and the poor mo- est because of its promise as a low-cost approach Energy Environ. Sci. 4, 3243 (2011).
bility of Li+ in the solid electrolyte. It is expected for grid storage applications. Redox-flow batteries 12. M. Armand, J. M. Tarascon, Nature 451, 652 (2008).
that the performance of the rechargeable alkali- possess several promising attributes for energy 13. A. S. Aricò, P. Bruce, B. Scrosati, J. M. Tarascon,
W. van Schalkwijk, Nat. Mater. 4, 366 (2005).
ion cathode flow battery will improve substantially storage, with low cost being one of the important
14. H. Li, Z. X. Wang, L. Q. Chen, X. J. Huang, Adv. Mater.
through the use of a better solid electrolyte and the drivers for this technology. A number of demon- (Deerfield Beach Fla.) 21, 4593 (2009).
possibility of using cathode inks. But perhaps the stration projects, ranging in size from 5 to 50 MWh 15. P. Poizot, S. Laruelle, S. Grugeon, L. Dupont, J. M. Tarascon,
more important point illustrated in these studies is and using a variety of different chemistries, are Nature 407, 496 (2000).
that redox-flow concepts adapt to other chemistries under way (48). The outcomes from these pro- 16. M. N. Obrovac, L. Christensen, Electrochem. Solid-State
Lett. 7, A93 (2004).
and hold considerable promise for improving bat- jects over the next 2 to 4 years will have a sub- 17. P. Barpanda et al., Inorg. Chem. 49, 7401 (2010).
tery performance and especially energy density. stantial influence on the future of this technology. 18. A. Yamada, S. C. Chung, K. Hinokuma, J. Electrochem.
The recent developments involving Li-redox flow Soc. 148, A224 (2001).
Future Directions and alkali-redox flow batteries stand as great oppor- 19. J. M. Tarascon, Philos. Trans. R. Soc. A 368, 3227 (2010).
20. J. M. Tarascon et al., Chem. Mater. 22, 724 (2010).
There are two related questions that need to be tunities that leverage existing knowledge of Li-ion 21. H. Chen et al., J. Am. Chem. Soc. 131, 8984 (2009).
addressed: What are the expectations for EES batteries with the advantages of redox-flow systems. 22. G. Girishkumar, B. McCloskey, A. C. Luntz, S. Swanson,
in the future, and what role will batteries play in Energy storage systems based on Li-ion bat- W. Wilcke, J. Phys. Chem. Lett. 1, 2193 (2010).
this future? The first part is becoming clearer teries are expected to take a different route than 23. K. T. Lee et al., Adv. Energy Mater. 1, 34 (2011).
24. K. M. Abraham, Z. Jiang, J. Electrochem. Soc. 143, 1 (1996).
as the value of energy storage becomes increas- either Na/S or redox-flow batteries. The devel- 25. J. P. Zheng, R. Y. Liang, M. Hendrickson, E. J. Plichta,
ingly evident. A recent EPRI study identified a opment of Li-ion batteries for commercial elec- J. Electrochem. Soc. 155, A432 (2008).
number of high-value opportunities for energy tronics and automotive applications enabled this 26. T. Ogasawara, A. Débart, M. Holzapfel, P. Novák,
P. G. Bruce, J. Am. Chem. Soc. 128, 1390 (2006).
storage, including wholesale energy services, in- technology to address reliability, cycle life, safe- 27. A. Débart, A. J. Paterson, J. Bao, P. G. Bruce, Angew.
tegration of renewables, commercial and indus- ty, and other factors that are equally as important Chem. Int. Ed. 47, 4521 (2008).
trial power quality and reliability, transportable for stationary energy storage. The research envi- 28. Y. C. Lu et al., J. Am. Chem. Soc. 132, 12170 (2010).
systems for transmission and distribution grid ronment for developing new low-cost materials 29. G. Girishkumar, B. McCloskey, A. C. Luntz, S. Swanson,
W. Wilcke, J. Phys. Chem. Lett. 1, 2193 (2010).
support and energy management (1). Moreover, is well established, and recent efforts directed at 30. J. F. Whitacre, A. Tevar, S. Sharma, Electrochem. Commun.
some of these benefits are complementary, fur- low-temperature processing and renewable or- 12, 463 (2010).
ther improving the economics of energy storage. ganic electrodes provide the basis for future ad- 31. Y. F. Y. Yao, J. T. Kummer, J. Inorg. Nucl. Chem. 29,
The success of these applications of energy vances in the field. However, it is the volume 2453 (1967).
32. X. C. Lu, G. G. Xia, J. P. Lemmon, Z. G. Yang, J. Power
storage will depend on how well storage technol- production anticipated for the electric vehicle Sources 195, 2431 (2010).
ogies can meet key expectations. The most impor- market that can lead to improvements in manu- 33. J. L. Sudworth, A. R. Tilley, The Sodium Sulfur Battery
tant of these are low installed cost, high durability facturing process and provide an economy of (Chapman & Hall, London, 1985).
34. C. H. Dustmann, J. Power Sources 127, 85 (2004).
and reliability, long life, and high round-trip effi- scale that will bring about the lower costs required 35. G. C. Farrington, J. L. Briant, Science 204, 1371 (1979).
ciency. The installed cost comprises the materials to make this battery technology viable for EES. 36. D. La Rosa et al., ChemSusChem 3, 1390 (2010).
costs, production costs, and installation costs for Another interesting scenario is the prospect of 37. J. L. Sudworth et al., MRS Bull. 25, 22 (2000).
38. Z. Yang et al., Chem. Rev. 111, 3577 (2011).
the system. In the future, the preferred energy stor- recovering Li-ion batteries used in automotive 39. S. F. Song et al., J. Solid State Electrochem. 14, 1735 (2010).
age technologies will be composed of low-cost, industries and to give them a “second life” in 40. B. Dunn, M. W. Breiter, D. S. Park, J. Appl. Electrochem.
easily acquired materials that are developed into large-scale energy storage applications. 11, 103 (1981).

934 18 NOVEMBER 2011 VOL 334 SCIENCE www.sciencemag.org

 SPECIALSECTION
41. A. Bito, “Overview of the sodium-sulfur battery for the 47. M . Skyllas-Kazacos et al., J. Electrochem. Soc. 158, R55 and from the DOE Office of Electricity, Energy Storage
IEEE Stationary Battery Committee,” paper presented at (2011). Systems Program. The authors greatly appreciate the
the IEEE Power Engineering Society General Meeting, 48. D. H. Doughty, P. C. Butler, A. A. Akhil, N. H. Clark, insightful comments provided by G. Farrington and
San Francisco, CA, 12 to 16 June 2005. J. D. Boyes, Electrochem. Soc. Interface 19, 49 (2010). A. Shukla. We also thank E. Lan and L. Smith for their
42. J. W. Braithwaite, W. L. Auxer, in Handbook of Batteries, 49. T. Nguyen, R. F. Savinell, Electrochem. Soc. Interface 19, assistance with the manuscript.
D. Linden, T. B. Reddy, Eds. (McGraw-Hill, 2004), chap. 40. 54 (2010).
43. www.geenergystorage.com/ 50. M. Duduta et al., Adv. Energy Mater. 1, 511 (2011).
Supporting Online Material
44. C. P. de Leon et al., J. Power Sources 160, 716 51. Y. H. Lu, J. B. Goodenough, J. Mater. Chem. 21, 10113 (2011). www.sciencemag.org/cgi/content/full/334/6058/928/DC1
Figs. S1 to S3
(2006). Acknowledgments: Support (B.D.) is from the Center for
45. M. Bartolozzi, J. Power Sources 27, 219 (1989). Molecularly Engineered Energy Materials, an Energy Frontier Table S1
46. M. Skyllas-Kazacos, M. Rychcik, R. G. Robins, A. G. Fane, Research Center funded by the U.S. Department of Energy References (52–62)
M. A. Green, J. Electrochem. Soc. 133, 1057 (1986). (DOE) Office of Basic Energy Sciences (DE-SC001342) 10.1126/science.1212741

Why Reduce SOFC Operating Temperature?

REVIEW
The key technical issue that has limited the de-
velopment and deployment of this transformative
Lowering the Temperature of Solid technology is its high operating temperature, re-
sulting in higher systems costs and performance
Oxide Fuel Cells

Downloaded from www.sciencemag.org on July 6, 2012

degradation rates, as well as slow start-up and
shutdown cycles, the latter dramatically limiting
Eric D. Wachsman* and Kang Taek Lee applicability in portable power and transportation
markets. Over the past decade, considerable pro-
Fuel cells are uniquely capable of overcoming combustion efficiency limitations (e.g., the Carnot cycle). gress has been achieved in bringing the temper-
However, the linking of fuel cells (an energy conversion device) and hydrogen (an energy carrier) has ature down to an intermediate temperature (IT)
emphasized investment in proton-exchange membrane fuel cells as part of a larger hydrogen economy range of 650 to 800°C so that metallic intercon-
and thus relegated fuel cells to a future technology. In contrast, solid oxide fuel cells are capable of nects could be used to reduce cost.
operating on conventional fuels (as well as hydrogen) today. The main issue for solid oxide fuel cells is high Low-temperature (LT) SOFCs (≤650°C) can
operating temperature (about 800°C) and the resulting materials and cost limitations and operating further reduce system cost due to wider mate-
complexities (e.g., thermal cycling). Recent solid oxide fuel cells results have demonstrated extremely rial choices for interconnects and compressive
high power densities of about 2 watts per square centimeter at 650°C along with flexible fueling, thus enabling nonglass/ceramic seals, as well as reduced balance
higher efficiency within the current fuel infrastructure. Newly developed, high-conductivity electrolytes of plant (BOP) costs. Moreover, below 600°C,
and nanostructured electrode designs provide a path for further performance improvement at much lower both radiative heat transfer (Stefan-Boltzmann)
temperatures, down to ~350°C, thus providing opportunity to transform the way we convert and store energy. and sintering rates exponentially drop off, thus re-
ducing insulation costs and primary performance
uel cells are the most efficient means to transportation applications, to distributed gener- degradation mechanisms, respectively.

F directly convert stored chemical energy to

usable electrical energy (an electrochem-
ical reaction). Although the more common proton-
ation and large-scale power generation, in both
civilian and military sectors (Fig. 1B).
Among the technologies available to con-
At even lower temperatures (≤350°C), cheap
stamped stainless steel interconnects, elastomeric/
polymeric seals (e.g., Kapton), and off-the-shelf
exchange membrane fuel cells (PEMFCs) require vert hydrocarbon-based resources (which in- BOP are possible. In addition, rapid start-up and re-
hydrogen fueling, because they are based on pro- clude not only fossil fuels but also, potentially, peated thermal cycling, from ambient to operating
ton conducting electrolytes, solid oxide fuel cells biomass and municipal solid waste) to elec- temperature, becomes possible. These are critical
(SOFCs) can oxidize essentially any fuel, from tricity, SOFCs are unique in their potential ef- parameters for portable power and transportation
hydrogen to hydrocarbons to even carbon, because ficiency. For stand-alone applications, SOFC applications, and it was because of PEMFCs’ low-
the electrolyte transports an oxygen ion. chemical to electrical efficiency is 45 to 65%, er operating temperature (~100°C) that they were
An SOFC consists of three major compo- based on the lower heating value (LHV) of the chosen for these applications over SOFCs, even
nents: two porous electrodes (cathode and anode) fuel (1), which is twice that of an internal com- though PEMFCs require hydrogen fueling.
separated by a solid oxygen ion (O2–) conducting bustion (IC) engine’s ability to convert chemical Another reason to reduce operating temper-
electrolyte (Fig. 1A). At the cathode, O2 (from energy to mechanical work (2). In a combined ature is maximum theoretical efficiency. In con-
air) is reduced and the resulting O2– ions are cycle, there are numerous combined heat and trast to the Carnot cycle temperature dependence
transported through the electrolyte lattice to the power (CHP) applications using SOFC systems, of IC engines, theoretical fuel cell efficiency in-
anode where they react with gaseous fuel, yield- which have the potential to achieve efficiencies creases with decreasing temperature [fig. S1
ing heat, H2O, and (in the case of hydrocarbon of >85% LHV (3). and supporting online material text (SOM text)].
fuels) CO2, and releasing e– to the external circuit. Unfortunately, government policy, the popu- For example, the maximum theoretical efficiency
Multiple cells are combined in series via in- lar press, and many scientific publications have of an SOFC using CO as a fuel increases from
terconnects that provide both electrical contacts focused on fuel cells as part of a broader hydro- 63% at 900°C to 81% at 350°C.
and gas channels between individual cells. The gen economy, thereby relegating fuel cells to a At first glance, this would imply that PEMFCs
resulting “stacks” are then arranged in series and “future energy” solution due to the need for a are more efficient than SOFCs because of their
parallel configurations to provide desired volt- required overhaul of our current hydrocarbon- lower operating temperature. However, this ig-
age and power outputs from portable power and fueling infrastructure. Although this may be true nores two important contributors to overall sys-
for PEMFCs, SOFCs have the advantage of fuel tem efficiency. The first is that the vast majority
University of Maryland Energy Research Center, College Park, flexibility that allows them to be used on our ex- of all H2 produced today comes from hydro-
MD 20742, USA. isting hydrocarbon fuel infrastructure (4) while carbon resources (typically CH4), thus requiring
*To whom correspondence should be addressed. E-mail: simultaneously providing efficiency gains (and additional external processes [e.g., steam reform-
ewach@umd.edu corresponding CO2 emission reductions). ing or catalytic partial oxidation (CPOX), water

www.sciencemag.org SCIENCE VOL 334 18 NOVEMBER 2011 935