A Critical Review of LiAir Batteries

See
discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/267367426
A Critical Review of Li/Air Batteries
Article in Journal of The Electrochemical Society · January 2012

DOI: 10.1149/2.086202jes
CITATIONS READS
524 945
8 authors, including:
Jake Christensen Boris Kozinsky

Robert Bosch LLC Harvard University
36 PUBLICATIONS 2,219 CITATIONS 57 PUBLICATIONS 1,768 CITATIONS
SEE PROFILE SEE PROFILE
Jasim Ahmed
LG Chem
43 PUBLICATIONS 1,664 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
Battery heat generation View project
Electric Vehicle Batteries View project
All content following this page was uploaded by Jake Christensen on 20 January 2015.
The user has requested enhancement of the downloaded file.

Journal of The Electrochemical Society, 159 (2) R1-R30 (2012) R1
0013-4651/2012/159(2)/R1/30/$28.00 © The Electrochemical Society
A Critical Review of Li/Air Batteries

Jake Christensen,a,∗,z Paul Albertus,a,∗ Roel S. Sanchez-Carrera,b,∗ Timm Lohmann,a
Boris Kozinsky,b Ralf Liedtke,c Jasim Ahmed,a and Aleksandar Kojica
a Robert Bosch LLC, Research and Technology Center, Palo Alto, California 94304, USA
b Robert Bosch LLC, Research and Technology Center, Cambridge, Massachusetts 02142, USA
c Robert Bosch GmbH, Gerlingen-Schillerhöhe, Baden-Wuerttemberg 70839, Germany
Lithium/air batteries, based on their high theoretical specific energy, are an extremely attractive technology for electrical energy
storage that could make long-range electric vehicles widely affordable. However, the impact of this technology has so far fallen short
of its potential due to several daunting challenges. In nonaqueous Li/air cells, reversible chemistry with a high current efficiency
over several cycles has not yet been established, and the deposition of an electrically resistive discharge product appears to limit
the capacity. Aqueous cells require water-stable lithium-protection membranes that tend to be thick, heavy, and highly resistive.
Both types of cell suffer from poor oxygen redox kinetics at the positive electrode and deleterious volume and morphology changes
at the negative electrode. Closed Li/air systems that include oxygen storage are much larger and heavier than open systems, but
so far oxygen- and OH− -selective membranes are not effective in preventing contamination of cells. In this review we discuss the
most critical challenges to developing robust, high-energy Li/air batteries and suggest future research directions to understand and
overcome these challenges. We predict that Li/air batteries will primarily remain a research topic for the next several years. However,
if the fundamental challenges can be met, the Li/air battery has the potential to significantly surpass the energy storage capability of
today’s Li-ion batteries.
© 2011 The Electrochemical Society. [DOI: 10.1149/2.086202jes] All rights reserved.
Manuscript submitted April 15, 2011; revised manuscript received October 24, 2011. Published December 29, 2011. This article
was reviewed by Kuzhikalail Abraham ([email protected]) and Yang Shao-Horn ([email protected]).
The Li/air cell has received significant interest in the past several several fundamental challenges can be overcome (Figure 3 and the
years as researchers look at couples that may achieve a specific energy associated discussion address the issue of practical specific energy in
significantly higher than current lithium-ion cells with two intercala- detail). This specific energy could enable an electric driving range of
tion electrodes (e.g., C6 /LiMO2 , where “M” refers to a transition more than 380 miles on a single charge at the beginning of a battery’s
metal such as Ni, Mn, or Co). The main application driving interest life, a value approaching that of a gasoline-powered vehicle. Addi-
is transportation, where specific energy and energy densityd are most tionally, the cost of a system that achieves today’s driving range may
important, although applications in portable electronics and grid en- be reduced significantly with a much higher specific-energy battery.
ergy storage are also of interest. Of particular interest in the context of This could bring electric vehicles to the mass market, since consumer
transportation is the fact that, with the specific energy and energy den- demand is expected to be driven strongly by vehicle cost reduction,
sity of today’s automotive Li-ion cells, one’s driving range is limited provided adequate range is available.
to about 70 miles for a 200 kg pack, as shown in Figure 1. We assume
a specific energy of 150 Wh/kg at the cell level and 105 Wh/kg at the There are many possible reactions involving Li and air.— From
pack level (70% of a pack’s weight is the cells). This means that in or- the outset it is important to realize that several chemical products may
der to enable electric vehicles with a range similar to today’s vehicles result from the reaction of Li with O2 , depending on the chemical en-
powered by liquid fuels, a battery system with a specific energy and vironment and mode of operation. The main distinction is whether the
energy density much higher than today’s state-of-the-art is required. medium in which Li is combined with O2 is aqueous or nonaqueous,
Indeed, while some observers predict that Li-ion cells may eventually and this distinction will be noted throughout this review. Some re-
reach 400 Wh/kg through the use of high-capacity cathode materials searchers have also explored hybrid aqueous/nonaqueous Li/air cells,
(275 mAh/g) and alloy anode materials (2000 mAh/g), significantly a concept in which a Li-conducting ceramic is used to separate one
higher values can only be obtained with even higher capacity cathodes, compartment with a nonaqueous electrolyte containing Li metal and
Li metal, and improved packaging of active materials. A Li/air battery another compartment with an aqueous electrolyte.1
has the potential to truly surpass the battery technology used today, It is important to note that throughout this review we refer to
as well as that under development for deployment in the medium aprotic solvents as “nonaqueous,” which is not particularly precise but
term (i.e., that which may achieve 400 Wh/kg). A cell-level specific is consistent with much of the Li/air literature. We shall not explicitly
energy value for a Li/air cell remains uncertain, but our cell energy consider protic nonaqueous Li/air chemistry, as published work is
calculations show that 1000 Wh/kg or more should be attainable if lacking in this area, although it is expected that it shares many general
characteristics with aqueous Li/air chemistry. Likewise, mixtures of
water and other protic solvents are outside the scope of this review.
∗ Electrochemical Society Active Member.
z
We also note that while many researchers refer to Li/air batteries, in
E-mail: [email protected]
d By “specific energy” we mean energy per unit mass, and by “energy density” we fact most of the laboratory work has focused on Li/oxygen batteries,
mean energy per unit volume. Some authors instead use the terms “gravimetric energy as components in air such as H2 O and CO2 can interfere with the
density” and “volumetric energy density,” respectively. desired electrochemical behavior.
Downloaded 16 Jan 2012 to 209.221.240.193. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp
R2 Journal of The Electrochemical Society, 159 (2) R1-R30 (2012)
500 may form as a film on the surface of Li metal.8 An example of the

reaction of Li with O2 in a mildly acidic environment is the formation
1,000 Wh/kg, of LiCl:9
400 future Li/air 2 Li + 1/2 O2 + 2 NH4 Cl ↔ 2 LiCl + 2 NH3 + H2 O. [4]
Driving range (miles)
Another example of a reaction in an acidic solution is the formation

300 of Li2 SO4 from Li, O2 , and H2 SO4 .9 Although acidic solutions have
400 Wh/kg,
future Li-ion the advantage that carbonates are not formed as in basic solutions (e.g.,
the formation of K2 CO3 in KOH solutions), in the present review we
200 exclude from detailed consideration aqueous reactions involving Li
and O2 in acidic media because the examples known to the authors
have a lower specific energy than in basic media (e.g., reaction 4 has
100 200 Wh/kg, current a lower specific energy than reaction 3). Neutral solutions have also
state of the art Li-ion been discussed, and we exclude them from consideration here for
the same reason.10 There may also be other chemical considerations
0 for a given reaction, such as the fact that in reaction 4 NH3 has a
0 100 200 300 400 500 high vapor pressure, limiting the reversibility of an open cell due to
evaporation. We also note that if the only criterion for a “Li/air” cell is
Battery pack weight (kg) a reaction that includes Li and O2 , there are additional reactions that
fall into this broad classification. However, in this review we focus
Figure 1. Driving range and battery weight for different cell-level specific our attention on the reactions that have thus far received the most
energy values. It is assumed the battery cells weigh 70% of the battery pack, attention, 1–3. We also exclude from consideration the somewhat
the Li/air cell has an 83% energy efficiency, the Li-on cells have a 93% energy related reactions between Li and H2 O (e.g., the “seawater battery”
efficiency, and 300 Wh/mile are required from the battery. The range is given developed by PolyPlus Inc.), because O2 is not a reactant and H2 gas
at the beginning of a battery’s life and assumes 100% of the capacity can be
used; in practice not all the energy can be used, and the available energy falls
is evolved, significantly limiting the possibility of creating a secondary
with increasing battery age. The US Department of Energy has a goal for an system.
EV battery of 200 kg.176 What, then, are the specific energy and energy density values for
the Li/air cells included in our analysis? We approach this question
by first looking at the active materials alone and then including the
In nonaqueous Li/air batteries there are two principal electrode components of a typical cell sandwich.
reactions of interest:
2 Li + 1/2 O2 ↔ Li2 O, [1]
and Energy estimates for the active materials alone.— Calculations
of the specific energy and energy density based on the weight of
2 Li + O2 ↔ Li2 O2 . [2]
the active materials alone provide a benchmark for values that can
In the absence of practical considerations the full reduction of O2 be obtained by practical cells, although a practical cell should not
to Li2 O is desired because of its higher specific energy and energy be expected to achieve more than about half of the energy per mass
density, but it appears that Li2 O2 is a product that forms more readily or volume of the active materials alone. “Active materials” refers to
than Li2 O.2–4 In addition, when Li2 O2 is formed full cleavage of the Li, O2 , and H2 O in the charged state, and Li2 O, Li2 O2 , LiOH, and
O-O bond may not be necessary, which is important from a kinetic LiOH · H2 O in the discharged state. As mentioned above, to the best
point of view.5, 6 These reactions will be discussed in more detail in of our knowledge LiOH · H2 O and not LiOH will precipitate from an
later sections. aqueous solution, so its inclusion here is for the sake of comparison
Reactions involving Li and O2 in an aqueous medium depend on only. In Table I we summarize the physical properties of the Li/air
the pH. In a basic aqueous environment O2 reduction includes H2 O active materials in the discharged state we consider in this review, as
as a reactant and results in the formation of LiOH: well as a current and “next generation” Li-ion intercalation material
for comparison.11 Figure 2 shows specific energy and energy density
2 Li + 1/2 O2 + H2 O ↔ 2 LiOH. [3]
numbers based on active materials alone (i.e., excluding the mass or
The product of this reaction is aqueous LiOH, which has a solu- volume of all cell components besides the active materials defined
bility limit of about 5.25 M at standard temperature and pressure.7 If above); in the charged state the weight of O2 is excluded. For the
LiOH exceeds its solubility limit it will precipitate out of the solution LiMO2 material we assume a specific capacity of 275 mAh/g and
as a monohydrate, LiOH · H2 O, rather than LiOH.7 This is a critical a density of 4.25 g/cm3 , values appropriate for an advanced oxide
point for calculating the specific energy and energy density of aqueous material.12 Energy calculations for an open system like Li/air are
Li/air cells. We are presently unaware of any solvent system that leads different from those for other battery systems that are closed to the
to the precipitation of LiOH rather than LiOH · H2 O, although LiOH external environment because the mass of the battery increases during
Table I. Physical properties of select Li/air and Li-ion materials positive-electrode active materials in the discharged state, as well as Li metal.13
Specific Capacity Uθ vs. Li Theoretical specific Theoretical energy

Active capacity Density density metal energy (vs. Li metal) density (vs. Li metal)
material (mAh/g) (g/cm3 ) (mAh/cm3 ) (V) (kWh/kg) (kWh/L)
Li2 O 1794 2.01 3606 2.91 5.22 10.49

Li2 O2 1168 2.31 2698 2.96 3.46 7.99
LiOH · H2 O 639 1.51 965 3.45 2.20 3.33
LiOH 1119 1.46 1634 3.45 3.86 5.60
LiMO2 , M = Mn, Ni, Co 275 4.25 1169 3.75 1.03 4.36
LiFePO4 170 3.6 612 3.42 0.58 2.09
Li metal 3861 0.534 2062 0.0
14 14
Active-only specific energy (kWh/kg)
Active-only energy density (kWh/L)

(a) (b)
12 12
10 10
Charged
8 Charged w/ H2O weight 8 Figure 2. (a) specific energy and (b) energy
Discharged density values based on active materials alone
6 6 for selected Li/air active materials, and an in-
sertion reaction for comparison.
4 4
2 2
0 0
Li2O Li2O2 LiOH·H2O LiOH Li/LiMO2 Li2O Li2O2 LiOH·H2O LiOH Li/LiMO2
the discharge process (assuming oxygen is not carried on board). electrode volume with discharge product. While some authors have
For that reason, the specific energy and energy density numbers are suggested using a large excess of Li (100 to 300%), we consider
shown in Figure 2 for both the fully charged and the fully discharged this impractical, as it implies the tolerance of a significant degree of
states. As Figure 2a shows, a cell producing Li2 O or Li2 O2 has about parasitic reactions, likely involving electrolyte decomposition, over
the same specific energy in the charged state, but in the discharged the lifetime of the battery. The products generated by such significant
state the Li2 O2 cell has a lower energy per mass due to only one O2 parasitic reactions would likely impair cell performance well before
molecule being consumed per 2 Li atoms, rather than 1/2 O2 in the all of the excess Li was consumed. Assumptions about the source of
case of Li2 O. The specific energy calculations for the LiOH · H2 O and H2 O for the aqueous Li/air cells are very important. If humidity from
LiOH systems in the charged state depend on the assumption about
where the H2 O in the discharge product originates. In Figure 2a we
show one case in which the H2 O is provided by an external source so
that only the weight of Li metal is included in the charged state, and Table II. Cell and tank properties for practical cell energy
calculations. Thickness values are at full charge while volume
another case in which the H2 O in the product is stored in the charged fraction values are at the end of discharge. ε values indicate
cell along with the Li (e.g., in a water reservoir). The main assumption volume fractions, LPSL = Lithium protection separator layer.
for the LiOH and LiOH · H2 O values is that of a single equilibrium GDL = gas diffusion layer. CC = current collector. The practical
potential. In practice the cell potential will vary with the activity of cell energies shown here are nominal, that is they do not include
the reactants and products during the cycling process, but we use the practical energy efficiency.
only the standard cell potential here. Compared with a Li/LiMO2 cell,
all four Li/air discharge products have a significantly higher specific Property Value Units
energy.
While specific energy is important, energy density can be just as Lpositive 200 μm
important in automotive and other applications. Figure 2b shows the LLPSL 50 μm
energy density based on the active materials alone. The higher energy LGDL, positive 50 μm
density of the discharged cells than charged cells is partly a result of LCC, negative 5 μm
the low density of Li metal (0.534 g/cm3 ). The figure shows that the LCC, positive 7.5 μm
energy density of a discharged Li/LiMO2 cell is higher than that of an Amount of excess Li for all cells 20 %
aqueous Li/air cell, and within about a factor of two of a nonaqueous εactive material, pos at the end 0.65
of discharge
Li/air cell. Thus, the advantages of Li/air cells from a specific energy
εelectrolyte, pos, Li/air cells 0.20
point of view are more dramatic than from an energy density point of
εelectrolyte, pos, Li/LiMO2 cell 0.25
view because of the relatively low density of Li/air active materials εgas phase, pos at the end of discharge 0.05
compared to metal oxide intercalation materials. εinerts, pos (carbon) 0.10
εGDL, pos 70 %
Energy estimates for practical cells.— Energy estimates based Mass packing factor based 80 %
on the weight and volume of active materials alone should be fol- on charged cell w/o tank
lowed with energy estimates for practical cells. Such estimates for Volume packing factor based 70 %
Li/air cells are uncertain due to the absence of well developed cell on charged cell w/o tank
designs. However, in this section we make assumptions about pos- ρLiOH (aq) elyte. (saturated) 1.105 g/cm3
sible cell designs in order to arrive at initial estimates for practical ρLiPF6 in PC elyte. 1.2 g/cm3
cells that can later be refined. We focus on an optimistic practical cell ρinert (carbon) 2.2 g/cm3
design that will require additional materials development rather than ρLPSL 3.0 g/cm3
only looking at cell components that are available today. For example, ρGDL, positive 1.8 g/cm3
we assume that an ionically conductive lithium metal protection layer ρCC, negative 8.9 g/cm3
with a thickness of 50 μm and density of 3.0 g/cm3 will be developed ρCC, positive 2.7 g/cm3
Oxygen volume in tank 75 L
that can be manufactured and used in practical cells.
Battery pack energy from which 140 kWh
Our assumptions are summarized in Table II. We assume the Li/air oxygen pressure is calculated
charged cells have a significant volume fraction of gas in the positive Tensile strength of tank 460 MPa
electrode (70%) while in the discharged state a small amount of gas (Stainless steel)
remains (5%). A gas phase provides volume into which the solid Tank safety factor (additional wall 50 %
active materials can be deposited and provides good transport of O2 thickness beyond tensile strength)
into and out of the cell. We assume the cell contains 20% excess Additional mass and volume above 20 %
Li relative to the capacity obtained by filling 65% of the positive that of the tank for tank components
2.5 2.5
Practical specific energy (kWh/kg)
(a) (b)
Practical energy density (kWh/L)

Charged
2.0 Charged w/ tank 2.0
Discharged Figure 3. Practical (a) specific energy and
Discharged w/ tank (b) energy density values for selected Li/air
1.5 1.5 active materials, and an insertion reaction
for comparison. The energy of soluble LiOH
in the electrolyte phase is excluded for the
1.0 1.0 LiOH · H2 O and LiOH cases as it would lead
to an increase of less than 5% for this cell
design.
0.5 0.5
0.0 0.0
Li2O Li2O2 LiOH·H2O LiOH Li/LiMO2 Li2O Li2O2 LiOH·H2O LiOH Li/LiMO2
the external air can be used to supply some of the H2 O required by cell design is also about 1.0 kWh/kg. Note that such a thick electrode
reaction 3 and the LiOH · H2 O precipitate, that significantly reduces is more realistic for aqueous than nonaqueaous cells because of the
the weight and volume basis in the charged state. However, adding much higher conductivity of aqueous electrolyte solutions. Again, we
a system to capture water may add complexity and cost, and we stress that the “practical” cell energy numbers presented here will
therefore use a design in our calculations that would either involve certainly be revised as more detailed designs are developed, and are
filling a water reservoir (like filling current cars with fuel) or having a meant to represent optimistic estimates.
water-impermeable membrane that completely prevents water ingress Figure 3 also shows the results of calculations including an oxygen
and egress. Therefore, in our practical cell numbers for the charged tank. We include these numbers because it is important to see how
state we include the weight and volume of water present in LiOH Li/air cells compare with a Li/LiMO2 cell if the problems associated
and LiOH · H2 O. These assumptions can be revised as practical cell with making an open system cannot be solved. For these calculations
designs are more fully developed. the mass of the oxygen is included when the cell is charged, as it is
Energy results for “practical” cell designs are shown in stored in the tank. We assume the oxygen in the tank has a specified
Figure 3, with results for systems with and without an oxygen tank volume (75 L) and the tank is sized for a battery system that stores
shown. We exclude the energy content of soluble LiOH in the elec- 140 kWh. Additional specifications are given in Table II. Figure 3
trolyte for the LiOH and LiOH · H2 O energy calculations because, for shows that the use of an oxygen tank results in a significant reduc-
this cell design, including it will increase the energy values by less tion in the specific energy and energy density. In terms of specific
than 5%, and it is unclear whether it is better to cycle with a saturated energy, the Li2 O and LiOH cells with a tank still have a higher value
electrolyte solution or have a lower concentration in the fully charged than a Li/LiMO2 cell, but the Li2 O2 cell and LiOH · H2 O cells have
stage. The theoretical amount of energy stored when cycling between a slightly lower value than a Li/LiMO2 cell. In terms of energy den-
a 0 M and a saturated solution (5.25 M at 25◦ C) of aqueous LiOH is sity, if an oxygen tank is used the values for all the Li/air cells will
about 430 Wh/kg and 475 Wh/L. Although an oxygen tank is, strictly be lower than for a Li/LiMO2 cell. These calculations demonstrate
speaking, not part of a Li/air cell, including its mass and volume in the the importance of creating a Li/air battery system that is able to use
calculation underscores the potentially large disparity in the energy oxygen from the atmosphere rather than store it onboard, although the
density of closed vs. open systems. We assume the use of a stainless tank results may be improved if a lighter weight tank material (e.g.,
steel oxygen tank in the shape of a 1.25 m-long cylinder with two carbon fiber) or a higher pressure (and thus a smaller tank volume)
hemispherical ends. could be used. Table III shows the pressure of a fully charged oxygen
First considering systems without an oxygen tank, Figure 3a shows tank for each Li/air active material, as well as the isothermal energy
that a “practical” discharged Li2 O2 Li/air cell may achieve a specific of compression required to go from 1 bar to the final pressure. The
energy more than twice that of a Li/LiMO2 cell, while a LiOH · H2 O non-unity compressibility of oxygen was accounted for using the van
cell may have a “practical” specific energy only slightly higher than der Waals equation. Interestingly, the LiOH and LiOH · H2 O active
a Li/LiMO2 cell. In terms of energy density, Figure 3b shows that a materials require the smallest amount of oxygen because they react
Li/LiMO2 cell has a modestly lower energy density than a Li2 O or 4 electrons per mole of O2 (as does Li2 O) and have a higher equilib-
Li2 O2 cell, and a modestly higher value than a LiOH · H2 O cell. The rium potential than Li2 O. Li2 O2 requires significantly more oxygen,
major change that would allow the aqueous Li/air system to have a and therefore a higher pressure and heavier tank, than the other ac-
significantly higher specific energy and energy density would be the tive materials because only 2 electrons per mole of O2 react. The
formation of pure LiOH rather than LiOH · H2 O, which we also show isothermal work of compression is relatively small in each of these
for the sake of comparison in Figure 3. For comparison with our “prac- cases (<3% of the practical discharge energy) but isothermal compres-
tical” numbers here, PolyPlus, a company focused on the development sion is probably more practical if done electrochemically; however,
of protected lithium metal electrodes, has claimed a practical specific
energy of almost 1.0 kWh/kg for their basic-electrolyte aqueous Li/air
cells.14 The numbers given in Figure 3a (0.70 kWh/kg for charged, Table III. Details on the oxygen tank pressures and compression
0.66 kWh/kg for discharged), are about 30% below the number given energies that would enable a closed Li/oxygen battery system.
by PolyPlus. A number of factors may contribute to this difference,
including our use of a 80% packing weight factor (they may have a Isothermal compression
lower-weight packaging technique) and how much of the weight of Fully charged work (kWh/kWh practical
the water stored in the LiOH · H2 O is included in their weight basis. Active material O2 pressure (bar) discharge energy)
In particular, if they have a design that takes some water from the
Li2 O 134 0.0108
external environment, that could significantly lower their weight ba-
Li2 O2 275 0.0244
sis for the charged cell. Another factor is our use of a relatively thin
LiOH · H2 O 114 0.0088
200 μm positive electrode thickness; with a positive electrode 1 mm LiOH 114 0.0088
in thickness and all else the same, the specific energy for our practical
having 100 to 300 bar oxygen pressure in the stack presents chal- cell to the point that a Li/air cell is no longer compelling compared
lenges of its own, including safety. In addition, while electrochemical with a Li/LiMO2 cell, unless a low-weight and high-pressure tank
compression has been explored for hydrogen it is not well established is used. The magnitude of reductions in the practical numbers in
for oxygen.15 A major advantage of using electrochemical compres- Figure 3 when a tank is used will result from a constrained optimiza-
sion and having the oxygen electrode operate at the same pressure tion for the tank pressure that results in an acceptable tank volume
as the tank, besides avoiding the need for a mechanical compressor, and gives the lowest cost (taking into account compression losses).
is that the compression work can be returned on discharge, resulting
in a theoretical round-trip compression energy efficiency of 100%. If Overview of the critical challenges for the Li/air system.— Now
adiabatic mechanical compression is used (which is typical for me- that we have established the promising specific energy of both non-
chanical gas compression) the theoretical compression work is higher, aqueous and aqueous Li/air systems, we turn our attention to the
although again if the oxygen electrode is directly exposed to the oxy- significant challenges that stand between the Li/air concept and com-
gen tank pressure at least some of the compression energy can be mercialization. In the next section, we address the critical issues per-
returned on discharge. Mechanical compression seems more plausi- taining to nonaqueous Li/air cells, including those that are relevant to
ble if the compressor could be located and powered from outside the both nonaqueous and aqueous systems. In the subsequent section, we
vehicle. discuss issues that are relevant to aqueous systems only. Finally, we
The calculations presented in Figure 3 are for secondary cells. The summarize the main challenges and provide an outlook for the future
specific energy and energy density of primary cells may be higher, of this technology.
depending on the cell designs that an application allows. Because We consider issues that limit the practical achievable specific en-
many primary cells are intended for low-rate applications, a thicker ergy, specific power, and cycle life of Li/air systems, as well as the
piece of Li metal may be used, increasing the mass and volume of additional complexity involved in the open nature of the cell. Because
active materials compared to packaging and thereby increasing the we estimate that the commercialization of this technology will be pos-
specific energy and energy density. In addition, primary cells typically sible only with solution of several very difficult challenges we defer
do not require any (or as much) excess Li metal because a non-unity questions of cost until the design of a viable system becomes clear.
current efficiency does not affect later cycling. However, a significant increase in the specific energy will result in a
Along with our calculations of the cell energy, it is important to commensurate reduction in cost if the cost per mass remains constant;
present the capacity per area for the cells described above. In Figure 4 this relationship remains to be established. Similarly, we suspend dis-
we show the electrode capacity (in mAh/cm2 ) for each active material cussions of system safety. While nonaqueous systems, if achievable,
using the specifications given in Table II. The figure shows that each contain flammable liquid electrolyte and a Li anode with the potential
material has an electrode capacity above 10 mAh/cm2 , and in the case to generate dendritic shorts, there is evidence to suggest that basic-
of Li2 O it approaches 50 mAh/cm2 . Typical values for a Li-ion cell are electrolyte aqueous systems with protected Li metal have insufficient
in the range of 3–7 mAh/cm2 . Of course, a thicker electrode with the reactivity to create the conditions for explosion due to the formation
same active material and active material volume fraction has a higher of a film that limits the reaction rate.16, 17
area-specific capacity, but there is a limit to how thick an electrode Not every hurdle that lies in the path of commercialization can
can be made due to mechanical issues. In later sections we discuss the be discussed in this review, as it is impossible to predict all of the
importance of presenting not only the amount of capacity per, for ex- challenges considering that the ultimate design of a Li/air system may
ample, gram of carbon, but also the capacity stored per electrode area. be significantly different from present concepts. Here we shall focus
In conclusion, the specific energy of nonaqueous Li/air batteries on a set of significant challenges that have high priority within the next
that form Li2 O or Li2 O2 is extremely appealing, as in a “practical” five years. Depending upon the results of research being carried out
cell it may be about 2 to 4 times higher than a comparable cell with over that time period it may be that certain widely studied concepts are
a Li metal negative electrode and an advanced Li-ion intercalation shown to be impossible or unattractive, while other related systems
positive electrode. The specific energy of the aqueous Li/air cell is (e.g., aqueous Li cells18 ) showing more promise may be discovered.
somewhat higher than a comparable Li/LiMO2 cell, but its energy The issues we consider are listed here and correspond to the subsection
density is lower. The relatively low density of the Li/air active ma- headings.
terials means that their specific energy is more appealing than their
energy density when comparing against an advanced Li-ion interca- Issues that are broadly applicable to Li/air systems, or only
lation material. Storing oxygen onboard the vehicle in a tank will nonaqueous systems:
significantly reduce the specific energy and energy density of a Li/air r Establishing truly reversible electrochemical reactions.
r Obtaining high capacity in the positive electrode.
r Accommodating significant volume changes.
50 r
Electrode capacity (mAh/cm )
2
Stabilizing the Li metal negative electrode.

r Achieving adequate power capability and efficiency.
40 r Supplying contaminant-free O2 to the system.
30 Issues that apply only to aqueous systems:

r
Managing precipitation and dissolution of the discharge
20
product.
r Catalyzing discharge and charge when O-O bonds are broken.
10
Nonaqueous Li/Air Systems

0
Li2O Li2O2 LiOH·H2O LiOH Li/LiMO2 While the first paper on a Li/air cell came out in the 1970s,19
and the first paper on a nonaqueous Li/air cell in the late 1990s,2
Figure 4. Positive-electrode area-specific capacity for a thickness of 200 μm interest has grown significantly in the past ten years, with much of
and an active material volume fraction of 0.65. For comparison, most nonaque- the attention focused on nonaqueous cell designs.3–5, 20–26 Several re-
ous Li/air cells built thus far have an electrode capacity of less than 5 mAh/cm2 , view papers have already been published;3, 27–29 we avoid repeating
even though they may achieve a large amount of capacity per weight of carbon that task here and instead focus on the critical issues that must be
(≥1000 mAh/g carbon). understood and solved for the Li/air system to move toward commer-
Figure 5. Example of a Li/air gas diffusion cathode based on Ni-foam. The

coating consists of 50 wt% Super P carbon black, 40 wt% PVDF binder and
10 wt% MnO2 additive. (a) SEM-image of a coated Ni-foam. (b) Optical image
of uncoated Ni-foam. (c) Optical image of a coated Ni-foam-cathode.
Figure 6. Sample discharge and charge curve for a nonaqueous Li/air battery
with a hydrophobized gas diffusion cathode cycled at a current density of
cial viability. However, we first briefly discuss the components and 0.2 mA/cm2 . The electrolyte contains 1 M LiPF6 in TEGDME.). The cathode
performance of a nonaqueous Li/air cell. One advantage of nonaque- (50 wt% Super P carbon ± 40 wt% PVdF ± 10 wt% MnO2 ) loading in
ous Li/air cells over aqueous Li/air cells is that a solid-electrolyte each case is 2 mg/cm2 , and the electrode thickness is 50 μm. The capacity
interphase (SEI) forms on Li metal in many nonaqueous electrolytes, is normalized to the total mass of the cathode excluding the current collector.
allowing some experiments to be conducted without using a solid- For comparison the top x-axis shows the capacity normalized to the electrode
electrolyte separator. Thus, a typical experimental cell is composed surface.
of a piece of Li metal, a nonaqueous solvent, a separator (e.g., glass
fiber or a Celgard separator), and an air electrode. Air electrodes for idea being that the reaction happens on the carbon surface. We stress
Li/air batteries can be prepared following several techniques,30–33 and that it is also valuable to include the capacity in mAh/cm2 , as this
the most common used thus far involve carbon black (e.g., Super P, metric in combination with the electrode thickness can be compared
Ketjen Black, Active coal), a polymer binder (e.g. PVDF, PTFE, cellu- to Li-ion and other electrodes.
lose), and an organic solvent (e.g. NMP, Acetone) or water (in case of Figure 7 shows a typical result for the fall in capacity with cycling
PTFE-suspensions and cellulose binders) being homogenized to form for a nonaqueous Li/air cell containing a carbonate-based electrolyte.
a viscous slurry. The slurry is typically coated onto a metal grid, metal Some authors have shown more stable capacity, but no publication
foam or conducting fleece by a film-casting process30 or by ultrasonic has demonstrated 100 cycles with at least 90% of the initial capacity
treatment. The supporting and conducting grid or sheet influences retained. Different types of carbon material typically demonstrate
the positive electrode performance if a 3-phase boundary between different amounts of capacity and have different rates of capacity
oxygen, reaction product(s), and electrolyte is formed. Common pos- fade, as Figure 7 shows. For example, Super P carbon delivers the
itive electrode supports are Al grids,4, 34 Ni meshes35 or porous Ni best performance although its surface area is only 60 m2 /g. Active
foams.36 The resulting air cathode has to fulfill several requirements. coals like Supra 30 have more than 1700 m2 /g but deliver much lower
It should have a high surface area at reasonable pore volumes, good capacity. This demonstrates that a carbon with a higher surface area
electronic (>1 S/cm) and ionic (>10−2 S/cm) conductivities, and a
design that supports fast gas transport to the reaction centers during
discharge.
Figure 5 shows an SEM image of a coated Ni foam (a) together
with optical images of uncoated (b) and coated samples (c). The
coating consists of 50 wt% Super P carbon black, 40 wt% PVDF
binder and 10 wt% MnO2 particles. From the SEM image one can
visualize the functionality of the gas diffusion air electrode. The large
pores ensure gas transport to the coated branches (covered with active
slurry). The electrochemical reaction presumably takes place within
the mesopores of the coating. It is still an open question how the
reaction distributes itself over the pore volume and it likely depends
on the specific electrode formulation and operating conditions. In
order to provide good reactivity and efficiency, the wetting behavior,
oxygen solubility, ionic conductivity, and stability of the electrolyte
on the cathode surface are of great importance.
Figure 6 shows a sample discharge and charge cycle for a non-
aqueous Li/air cell. There is typically a significant offset between the
discharge and charge curves, often more than 1 V, indicating a high Figure 7. Variation of carbon materials for Ni-foam based cathodes in a
cell impedance and/or a different reaction mechanism on discharge propylene carbonate electrolyte with 1 M LiPF6 . Super P carbon black (60
and charge. As discussed in subsequent sections, several physical pro- m2 /g) shows the best performance in this comparison regarding maximum
cesses contribute to the high resistance. The abscissa label in Figure 6 capacity. The other materials are carbon black (Ensaco 350 G, 400 m2 /g),
shows that a common way to express the charge stored in a nonaqueous carbon fibers (CF), and two active coals, Supra 30 (1760 m2 /g) and PAK1000C
Li/air cell is in mAh/g of cathode material, including carbon, binder, (1000 m2 /g). The cathode (50 wt% carbon ± 40 wt% PVdF ± 10 wt% MnO2 )
and catalyst. Some authors define the capacity in mAh/g of carbon, the loading in each case is 2 mg/cm2 , and the electrode thickness is 50 μm.
Figure 8. Influence of three different cathode designs on the electrochemical

performance of a Li-O2 cell. The error bars indicate the capacity range observed
for many different samples of the respective type. The electrolyte is 1 M Figure 9. Potential vs. time of Li/air cells in three different nonaqueous elec-
LiPF6 in propylene carbonate. The Al-grid supports a solid coating and allows trolytes. All samples have been cycled at a current density of 0.2 mA/cm2 . The
gases to diffuse through the layer. The all-carbon-based electrode design is a air electrode consists of hydrophobized carbon, PVDF, and a MnO2 catalyst.
gas diffusion layer (GDL) consisting of a dense carbon fiber network with a It is important to mention that the plateau around 3.8 V for NMP is due to
hydrophobic coating (a wet chemical treatment using a sliane-based precursor) oxidative decomposition, and is not related to the desired charge reaction (i.e.
and a carbon black coating of about 50 μm. The cathode (50 wt% Super P reversible Li2 O2 decomposition). The same plateau is present during charging
carbon black ± 40 wt% PVdF ± 10 wt% MnO2 ) loading in each case is of NMP cells without a prior discharge. The cathode (50 wt% Super P carbon
2 mg/cm2 . The slurry was applied to the carbon black layer of the GDL using ± 40 wt% PVdF ± 10 wt% MnO2 ) loading in each case is 2 mg/cm2 and the
a coating knife. The Al-grid was coated by dipping it into the slurry similar to thickness is 50 μm.
Ni-foam. Note that the discharge capacity increases reproducibly between the
first and second cycle for two of the cathode compositions. This may be due
to some rearrangement or modification of the electrode surface. however, in the case of nonaqueous systems, it has not been shown
that they can be cycled with a current efficiency sufficiently high
does not necessarily have a higher capacity. Typical electrolytes most for a practical reversible cell (>99.95% is required to reach about
likely cannot penetrate the nanopores present in high-surface-area 450 cycles with 80% of the capacity remaining). The distinction be-
carbons; hence, reactive sites are restricted to the outer surface of tween cycling and cycling reversibly is crucial; a cycle is only re-
these carbons and the available surface area is reduced. In addition, versible if the chemical makeup of the system is the same at the start
the particular air electrode design (e.g., all carbon treated with a and end of the cycle. For the nonaqueous battery in particular, detailed
silane-based hydrophobic layer vs. carbon/Ni-foam vs. carbon/Al- investigation of the chemistry of discharge and charge products is a
grid (expanded metal) electrodes, see Figure 8) has a great influence critical issue that has thus far received insufficient attention. There is
on the capacity achieved. now conclusive evidence that the electrochemical reactions that oc-
The identity of the solvent also has a significant impact on the cur in nonaqueous cells containing carbonate solvents during charge
capacity obtained and the discharge and charge potentials. In our own are not, in part or whole, the reverse of those that occur during dis-
studies we have tested n-methyl pyrrolidone (NMP) and tetraethy- charge. Hence, the system can be cycled several times, but only at the
lene glycol dimethyl ether (TEGDME)e in comparison with propy- expense of some component, likely the electrolyte solvent, which is
lene carbonate (PC). Figure 9 shows data of the first cycle for PC, consumed irreversibly on each cycle. Without a continuous supply of
TEGDME, and NMP measured at a current density of 0.2 mA/cm2 . fresh reactant and a way to remove irreversibly generated discharge
The air electrode used in these experiments consists of silane-treated products, hundreds of cycles are unattainable. In addition, any num-
carbon black, PVDF binder and a MnO2 catalyst. The differences in ber of non-electrochemical reactions may also be occurring. In this
the capacity obtained and the potentials indicate possible differences section we provide a critical description of recent experimental and
in the reaction pathways, discharge product morphology and growth theoretical work that has examined the fundamental processes occur-
mechanism, electrode wetting, and oxygen solubility and transport. In ring in the nonaqueous Li/air electrochemical cell, as well as specify
this comparison NMP shows the highest discharge potential of about the types of studies that will be helpful in overcoming the critical
2.7 V and, at approximately 35 hours, the longest discharge time. issues.
In summary, while a basic Li/air cell can be assembled using Experimental evidence strongly supports irreversible reactions in
materials commonly available and often used in Li-ion cells, many carbonate-based solvents.—The first paper on a nonaqueous Li/air
experimental results and physical processes remain difficult to under- cell (published in 1996) described the use of an electrolyte with a
stand. In the coming sections we identify several of the challenges polymer solvent (polyacrylonitrile) as well as the plasticizers ethylene
that have been identified, and seek to delineate where clarity has been carbonate (EC) and propylene carbonate (PC) and a cobalt pthalocya-
established and where additional evidence is required before a clear nine catalyst.2 It identified Li2 O2 as the probable discharge product
explanation can be provided. based on qualitative analysis and Raman Spectroscopy. Although they
used some but not all of the same materials as Abraham and Jiang,
Truly reversible electrochemical reactions need to be
later authors appear to have taken the formation of Li2 O2 in carbon-
demonstrated.— Although efforts to commercialize primary Li/air
ate solvents as already established.2, 4, 20, 33, 34 However, several recent
batteries are already underway, the ultimate objective in automotive
papers have specifically focused on the reaction chemistry and estab-
electrification is to produce a high-specific-energy storage battery that
lished conclusively that in carbonate solvents (without the presence
can be cycled thousands of times. At this stage, limited cyclability of
of any polymers) Li2 O2 is not a principal discharge product, and on
both nonaqueous and aqueous Li/air cells has been demonstrated;
charge CO2 rather than O2 is evolved.38–42
In the most definitive article published so far, McCloskey et al.
e We note that recent interest in TEGDME follows early work on poly(ethylene glycol) carried out differential electrochemical mass spectroscopy (DEMS)
dimethyl ethers in Li/oxygen cells.37 experiments on Li/oxygen cells fed with isotopically labeled O2 as
eral other articles state that Li2 O2 forms in a Li/oxygen cell containing
a carbonate solvent (e.g., the original paper by Abraham and Jiang
and a paper by Thapa et al. who built carbon-free electrodes44 ), none
of them employed a set of analytical tools (including gas analysis and
spectroscopy) that could consistently account for the exclusive elec-
trochemical formation of Li2 O2 . We conclude from the articles that did
employ rigorous spectroelectrochemical techniques and demonstrated
the absence of significant Li2 O2 formation and reversible cycling that
carbonate solvents do not have sufficient stability for long-term Li/air
cell cycling.
While the spectroscopic techniques mentioned thus far have been
helpful for identifying the discharge and charge chemistry, Zhang and
co-workers demonstrated that the characteristic lithium NMR chem-
ical shift for Li2 O2 is difficult to distinguish from that of a reaction
by-product of their carbonate-electrolyte Li/air cells.43 Similar results
were also obtained in our recent theoretical work.45 Therefore, a direct
assignment for the chemical shift that indicates the presence of Li2 O2
could not be established by means of lithium NMR spectroscopy. The
sometimes conflicting results for the carbonate-based Li/air chemistry
underscore the importance of carefully selecting appropriate analyti-
Figure 10. Raman spectra of discharged carbon cathodes from a pure DME- cal tools to identify discharge and charge products.
based cell, a 1:1 (v:v) EC/DME-based cell, and a 1:2 (v:v) PC/DME-based
cell. Neat P50 carbon paper is included for comparison. Discharge condi- No solvent has yet demonstrated highly reversible cycling; further
tions: 0.09 mA/cm2 under O2 to 2 V. Caption and figure reproduced from exploration and quantification are necessary.—Some noncarbonate
reference 38. solvents have been explored for use in a Li/oxygen cell, including
ethers, ionic liquids, and acetonitrile. In this section we assess criti-
cally the results for each of these solvents. Considering, first, the class
well as spectroscopic analysis of discharged and charged electrodes.38 of ethers, McCloskey et al. also explored the use of dimethoxyethane
In a solvent composed of either 1:1 EC:DMC or 1:2 PC:DME, sig- (DME) in the same work that addressed carbonate solvents.38 Dur-
nificantly more than 2 e− /O2 molecule were consumed on discharge ing discharge in DME, 2.05 ± 0.05 e− /O2 were consumed and
(indicating reactions besides formation of Li2 O2 ), XRD of discharged the principal product identified by spectroscopy was Li2 O2 (see
air electrodes showed the absence of Li2 O2 , and Raman Spectroscopy Figure 10), while on charge the same isotopically labeled oxygen
showed both the absence of Li2 O2 and the presence of Li2 CO3 and consumed by the cell on discharge was released. However, during the
possibly Li alkyl carbonates (see Figure 10). Upon charging a cell charging process, in which the cell was charged by a capacity equiv-
that had undergone a first discharge, the principal gas evolved was alent to the discharge capacity, 3.2 e− /O2 released were consumed,
CO2 rather than isotopically labeled oxygen. The evidence in this indicating a current efficiency for O2 release of only 60%. Interest-
paper clearly establishes the absence of reversible cycling when, at ingly, even with such a low current efficiency, no Li2 O2 was found at
the very least, the specific carbonate solvents used by the authors the end of charge, which the authors suggest may be due to a chemical
are present. We consider the use of isotopically labeled oxygen in or electrochemical reaction of Li2 O2 during the charging process that
quantitative, on-line DEMS experiments to be a key method for estab- does not have O2 as a product. The authors conclude that while DME
lishing the reversibility of the Li/air system; in a reversible system the does provide for the formation of Li2 O2 on discharge, it is unstable
same isotopically labeled O2 consumed on discharge should emerge during the charging process and is therefore an unsuitable solvent for
on charge. If CO2 is evolved, isotopic labeling of carbon in the solvent reversible cycling. However, much of the charging process was car-
may help determine whether the solvent or, if carbon based, the elec- ried out at high potential (∼4.5 V vs. Li), at which DME is likely to
trode decomposes. We note that when evaluating new chemistries for decompose electrochemically,46, 47 and additional experiments (e.g.,
reversibility, it is important to quantify the number of electrons and charging at lower currents; using a cell design that ensures intimate
O2 molecules consumed on discharge and charge as well as the purity electrical contact between Li2 O2 and conductive carbon) should be
of the discharge product; reversible systems have charge/discharge performed before eliminating DME as a candidate solvent for non-
capacity ratios of very nearly 1 and e− /O2 ratios of very nearly 2 aqueous Li/oxygen cells. Peng et al.48 showed that electrodes packed
(assuming Li2 O2 is the discharge product). Other ratios and/or the with Li2 O2 could be charged in PC solvent, whose oxidation potential
presence of discharge products other than Li2 O2 indicate the presence is ∼5.1 to 6.0 V,4, 46, 47, 49, 50 without evidence of solvent decomposition.
of electrochemical or chemical side reactions. Quantifying electron Freunberger et al. also report on the chemistry of Li/oxygen cells
and O2 consumption and generation as functions of current density and that make use of ethereal solvents, and found that for linear ethers like
potential window can help identify the nature of these side reactions tetraglyme, as well as cyclic ethers like 1,3-dioxolane, some Li2 O2
if they exist. forms on discharge, although significant amounts of electrolyte de-
The conclusions of McCloskey et al. are supported by several composition products are also found.51 They did not report results
other publications. For example, Freunberger et al. also carried out for DME as a solvent, possibly explaining the difference between the
DEMS analysis on a carbonate-containing Li/air cell, as well as spec- fraction of Li2 O2 produced during the first discharge they found com-
troscopic analysis that included FTIR, NMR, and Surface-Enhanced pared with McCloskey et al. Like McCloskey et al.,38 they conclude
Raman Spectroscopy (SERS).39 They found that in an electrolyte of that ethers are unsuitable solvents for reversible Li/oxygen cells, al-
LiPF6 in PC on discharge the products include Li2 CO3 and several though they are more stable than carbonates as some Li2 O2 forms
alkyl carbonates, and on charge CO2 rather than O2 evolves. Their on the first discharge. Other authors have also discussed the fact that
spectroscopic and gas-analysis findings are consistent with those of Li2 O2 does form in ethereal solvents, and there is at least limited
McCloskey et al. and clearly point to irreversible reactions during the reversibility.26, 52 In our view, while additional work on ethereal sol-
cycling of a Li/oxygen cell with a PC solvent. In addition, similar re- vents is warranted, there are already indications that they too undergo
sults have been reported by Mizuno et al. who carried out FTIR and gas decomposition reactions.
analysis;40 Zhang et al. who carried out gas analysis during charging, A different class of solvents that has recently received attention is
FTIR, and XRD;41, 43 and Veith et al. who carried out XPS, FTIR, and ionic liquids. For example, Mizuno and Iba discuss the use of an ionic
Raman on Li/oxygen cells with an electrolyte composed of LiPF6 in liquid, N-methyl-N-propylpiperidinium bis-trifluoromethansulfonyl-
a mixture of ethylene carbonate and dimethyl carbonate.42 While sev- amide (PP13TFSA), as a solvent that demonstrates better reversibility
than PC, and results in O2 rather than CO2 evolution during charge.53 evidence for whether O2 − or LiO2 is more influential in unwanted
Other authors have also explored the use of ionic liquids for Li/air parasitic reactions is not yet definitive.
batteries, although to date there has been no paper that demonstrates Independent of whether the solvent decomposition reaction pro-
quantitatively a sufficiently high current efficiency for a practical ceeds via O− 2 , LiO2 , or another Li/air reactive species, it is now
cell.31, 54–56 clear that a solvent that is resistant to attack by reduced O2 species
Finally, several authors have explored the use of acetonitrile as a must be discovered in order to achieve a practical Li/air cell with
solvent for Li/air batteries.26, 48 While they have found that oxygen a long cycle life. Recent advances in surface-sensitive spectroscopy
reduction and evolution appear to be reversible, they have not ade- and microscopic analyzes38, 51 coupled with first-principles modeling
quately quantified the current efficiency of Li2 O2 formation, and point simulations58 could help unveil the fundamental elementary processes
out that the higher vapor pressure and inferior safety characteristics of the Li/air electrochemical cell and explain the reasons for the poor
of acetonitrile make it a poor choice for a practical Li/air cell. reversibility of the reactions in typical Li-ion battery solvents.
Laoire et al.26 have suggested that suitability of a solvent for a Li/air When screening electrolytes that do not form a stable SEI on Li metal
cell be evaluated in terms of Pearson’s “Hard Soft Acid Base” (HSAB) an appropriate counter electrode should be used.—The carbonate sol-
Theory.57 By comparing two cations (e.g., Bu4 N+ , a soft Lewis acid, vents of widespread use in Li-ion batteries can form solid-electrolyte
and Li+ , a hard Lewis acid), they showed that O2 − is more stable interface layers on Li metal electrodes that are sufficiently stable to
in a soft acid environment, but is reduced to O2 2− and ultimately allow tens or even hundreds of cycles. When noncarbonate solvents
O2− in a hard acid environment. Moreover, solvation decreases the are being screened for their stability toward positive-electrode reac-
hardness of the Li+ acidity in proportion to the donor number of tions, care must be taken that side reaction products generated at the
the solvent. This in turn leads to a longer lived superoxide and counter electrode do not diffuse across the cell and interfere. Ideally, a
reversible LiO2 formation. Such theories can help guide the solvent counter electrode and reference electrode that have a potential at which
exploration process. a solvent is stable should be used. For example, Li4 Ti5 O12 , with an
In conclusion, while several noncarbonate classes of solvents have equilibrium potential of 1.5 V vs. Li metal may be a good candidate,
been explored for use in Li/oxygen batteries, to date there has been no although it does not have cyclable Li as synthesized (active materials
report that clearly establishes highly reversible cycling, and finding are typically synthesized in the discharged state). To circumvent this
a salt and solvent combination that allow truly reversible cycling problem Li2 O or Li2 O2 could be packed into the positive electrode.
remains perhaps the first and most important challenge for Li/air Another possible counter and reference electrode is LiFePO4 . In any
batteries at this time. case, we stress the importance of a careful experimental design even
Hypothesized mechanistic routes for the formation of the desired for research cells in order to ensure the electrochemical stability of
reaction product include reactive intermediates.—There is now suffi- the reactions occurring at all electrodes and the avoidance of soluble
cient evidence that the solvent plays a critical role in determining the products that may obscure the interpretation of the electrochemical
nature of the Li/air reaction products and the cell’s electrochemical signals. A protected lithium electrode could likewise serve as a counter
cyclabilty. The search for a stable electrolyte could be greatly assisted electrode for screening nonaqueous solvents, and it may be necessary
by acquiring detailed knowledge of the prevalent reaction mechanisms for the function of certain solvents.
of the Li/air cell, including the identification of rate determining steps Outlook: Nonreactive nonaqueous solvents remain elusive and re-
and intermediate reactive species. For example, the experimental work quire a carefully designed screening procedure.—It is now clear that
of Abraham and co-workers25, 26 resulted in a proposed reaction mech- carbonate solvents are poor candidates for reversible Li/air systems
anism that includes the reduction of oxygen to lithium superoxide: because of their susceptibility to react with reaction intermediates.
O2 + Li+ + e− → LiO2 [5] The search for noncarbonate solvents should involve screening in
cells with appropriate negative electrodes (e.g., protected Li or high-
followed by potential intercalation electrodes). The amount and identity of O2
2LiO2 → Li2 O2 + O2 [6] consumed and generated during discharge and charge should be mea-
sured and compared to the discharge and charge capacity to quantify
or reversibility. The identity of the discharge product and its absence
LiO2 + Li+ + e− → Li2 O2 [7] after recharge should also be confirmed. Thus far linear and cyclic
ethers, acetonitrile, and some ionic liquids have been explored; while
and possibly Li2 O2 is formed on the first discharge in some cases, highly reversible
Li2 O2 + 2Li+ + 2e− → 2Li2 O [8] cycling has not been quantified and reported for any of these non-
carbonate solvents. In our view, finding a stable solvent that allows
truly reversible cycling remains perhaps the most important current
Equation 5 involves the irreversible reduction of O2 via a one- challenge for Li/air research. Using a variety of analytical techniques,
electron process to form LiO2 . This product disproportionates to Li2 O2 including gas analysis (preferably with isotopically labeled oxygen
and O2 (equation 6) or is further reduced to form Li2 O2 (equation 7). and possibly carbon), and spectroscopy such as Raman and XRD,
Li2 O is also probably formed as the ultimate reduction product of O2 , is required to ensure that cycling is truly reversible. Identifying and
as indicated in equation 8.25, 26 Recent in-situ spectroscopic data have understanding the reaction pathways that influence reversibility via a
provided direct experimental evidence of the formation of LiO2 as an combination of spectroscopy, microscopy, and first-principles simula-
intermediate species for the formation of Li2 O2 .51 LiO2 formation in- tions is a worthy effort that the scientific community should undertake.
volves reduction of O2 to superoxide, O− 2 , which has been implicated
both experimentally39 and theoretically58 as the species responsible for
the decomposition of the carbonate-based electrolytes. As discussed A high positive-electrode capacity needs to be achieved for a
extensively by Sawyer and Valetine, the superoxide ion is a powerful Li/air cell to achieve a high specific energy and energy density.—
reducing agent, which is thought to react very rapidly with a variety of Conventional positive-electrode materials like LiMn2 O4 , LiFePO4 ,61
organic substrates.59 Others have also ascribed significant reactivity to or LiNi1/3 Co1/3 Mn1/3 O2 have maximum capacities of 100 to
the electrochemically formed LiO2 species, which rapidly reduces the 200 mAh/g. A practical sulfur-electrode (elemental sulfur dispensed
carbonate solvent and forms solvent-decompositions products.38, 43 in a carbon or polymer matrix) reaches 400 to 900 mAh/g62, 63 referred
Although the rate of formation of LiO2 and its reactivity toward to the mass of carbon, binder and sulfur. The experimentally observed
carbonate solvents remain unmeasured, preliminary XRD results by capacity range for Li/air gas diffusion electrodes is about 600 to
Zhang and co-workers indicate that the attack of the carbonate-based 5000 mAh/g (the typical range of loadings is 3 to 6 mg/cm2 )
solvent molecule by the superoxide ion proceeds at a faster rate than depending on the electrolyte, carbon, binder, carrier material, and
the coordination of O− +
2 and Li to form LiO2 .
60
To summarize, the oxygen partial pressure.21, 22, 31 The capacity is referred to the total
mass of the carbon, binder, and additives in the cathode. This large is possible that its surface states and grain boundaries may be more
capacity range shows that many parameters may influence the specific conductive).
energy of a Li/air cell. Therefore the cathode has to be understood We stress that besides Li2 O2 , other reaction products such as Li2 O,
in terms of maximum capacity and relevant mechanisms that limit Li2 CO3 and Li-organic compounds may appear depending on the elec-
the capacity. Three major capacity-limiting issues, passivation, pore trolyte system. The mechanism of electrical passivation in noncarbon-
blockage, and O2 transport limitations, are discussed in the literature ate electrolyte systems is at an early stage of investigation.52
and are assessed critically in the following sections. The impedance rise associated with passivation leads to an in-
However, before moving to those capacity-limiting issues we stress creasing overpotential that may end the discharge before the available
that in order for a Li/air cell to achieve a high energy the cell, at the end pore volume is even moderately filled. It will therefore be necessary
of discharge, needs to have a high volume fraction of active material to understand the detailed growth mechanism of the desired reaction
(whether it be Li2 O, Li2 O2 , or LiOH · H2 O). Reporting the capacity in product, which is not yet well understood. Defects or grain boundaries
mAh/g carbon, mAh/g cathode, or even mAh/cm2 somewhat obscures in the Li2 O2 crystal could result in enhanced electronic conduction to
this extremely important point. Unlike standard Li-ion cells where the the Li2 O2 /electrolyte interface, or diffusion of Li and O through the
active material is built directly into the electrodes with a volume film could enable growth from the electrode/Li2 O2 interface. Theoreti-
fraction of around 50 to 75%, in a Li/air cell the active material cal computations of electronic conductivity through the film have been
is essentially being synthesized during the discharge process. While limited to the consideration of fully dense, monocrystalline films.23, 65
many authors report the capacity in mAh/g carbon, we emphasize that More detailed characterization of Li2 O2 films grown on porous elec-
carbon is not the true active material in the Li/air cell, as Table I shows. trodes bathed in noncarbonate electrolytes, using a combination of
To enable a reader to calculate the volume fraction of active material imaging (e.g., SEM, TEM, AFM) and electrochemical techniques
in an electrode or cell at the end of discharge it is important to include (e.g., use of ferrocene couples, GITT), is required to elucidate the
the information necessary to make that calculation. For example, if the Li2 O2 growth mechanism. Depending upon the results, models that
capacity is reported in mAh/g carbon, an author should also provide incorporate grain boundaries and/or defect chemistry could be devel-
the carbon loading (in mg-carbon/cm2 ) and the electrode thickness to oped to improve our understanding of Li2 O2 formation and growth.
allow the calculation of mAh/cm3 , and from that the volume fraction Pore blocking may also restrict the practical capacity.—Besides direct
of active material. passivation of the electrochemically active surface, blockage of micro-
Passivation by insulating discharge products appears to limit the pores and some mesopores by Li oxides or other products formed at
capacity.—Recent flat-electrode experiments and Li/air cell model- the beginning of discharge can limit the accessibility of some electrode
ing indicate that passivation of the electrode surface by electroni- surface for electrochemical reaction. Figure 12 shows schematically
cally insulating discharge products severely limits the capacity of how Li2 O2 growth on the surface of carbon in an electrode with re-
Li/air cells even at low rates of discharge, although the results may stricted pores could result in either passivation or pore blocking. Both
depend on the electrolyte used and the discharge rate.23 In partic- phenomena result in unused pore volume and hence limit the discharge
ular, Figure 11 shows the rapid passivation that occurs during dis- capacity.
charge on a flat glassy-carbon surface. The maximum thickness ob- In electrodes where passivation does not occur, pore blocking may
tained for the discharge product is less than 100 nm. The cell in be the mechanism that most severely limits capacity. To overcome
this case used a carbonate solvent; hence, the discharge reaction pore blocking the selection of a suitable carbon material with a suf-
picture is complicated by the fact that an array of discharge prod- ficiently large pore diameters to allow the entire electrochemically
ucts are generated. Researchers carrying out rotating-disk electrode active surface to react is important.
experiments have also commented on passivation.25 This limitation In order to generate innovative designs that prevent air electrode
may be general because the desired reaction product in the non- passivation and pore blocking one needs to understand the chem-
aqueous Li/air cell, Li2 O2 , has the electronic properties of an in- istry of the reaction process (discussed in Section 2.1) and how the
sulator (to be more specific, bulk Li2 O2 is insulating,5, 64 although it carbon/air electrode microstructure affects the discharge performance.
Hence, it is necessary to characterize the porous network of the real gas
diffusion electrode and relate it to its electrochemical performance. In
the literature N2 -adsorption measurements (BET66 ) are applied for the
selection of suitable carbon materials. Information about pore size dis-
Initial open-circuit potential
3.0 tribution, pore volume, and the available surface area can be related to
2 state-of-charge dependent impedance measurement data (EIS).67 This
0.75 µA/cm (~12.2 hr)
High 2
has been used to track the electrical passivation and relate the behavior
2.8 kinetic 1.50 uA/cm (~ 3.4 hr) to structural properties of the carbon material determined by BET. For
2
resistance 3.76 µA/cm (~1.0 hr) mesocellular carbon foam a capacity increase by 40% compared to
2
other carbon black materials has been achieved recently.68 The authors
Cell potential (V)
Fit to 0.75 µA/cm data

2.6 Exponential
resistance
rise
2.4
2.2
2.0
-3
0 2 4 6 8 10x10
2
Capacity (mAh/cm )
0 10 20 30 40 50 60 70
Thickness of discharge products (nm)
Figure 12. Schematic illustration of the pore filling during discharge. The
Figure 11. Demonstration of the passivation of a flat-electrode surface during growing Li2 O2 layer leads to cathode passivation by electrical isolation (top
a discharge on glassy carbon. Reproduced from reference 23. right) and pore blocking (bottom right).
attribute the enhanced performance to their large pore volumes and although the relationship is not monotonic. At low discharge rates (i.e.,
ultra-large mesoporous structures, which allow more lithium oxide the lines labeled f with a current density of 0.05 mA/cm2 ), the capacity
deposition during discharge. reaches its maximum at relatively low values of the inverse viscosity
It should be possible to explore in a quantitative way the degree 1/η and the Bunsen coefficient α, while at high rates (i.e., the lines
of pore blocking that develops. For example, an electrode may be labeled a with a current density of 0.5 mA/cm2 ) the capacity increases
harvested, its porosity checked, and a Gurley test conducted to deter- steadily over the measured range (Figure 13a and 13b). The viscosity
mine the resistance to flow through the porous structure for a defined and Bunsen coefficient influence oxygen transport in the electrolyte,
pressure drop. In addition, a careful design of experiments that sep- which becomes more critical at higher rates. Similarly, the capac-
arates the influence of total surface area, overall porosity, pore size, ity plateau shifts to higher oxygen partial pressures if the discharge
and pore size distribution would be helpful to truly quantify the de- rate increases (Figure 13c). The plateau’s position is related to the
gree of pore blocking that occurs. Such experiments should also take rate-dependent saturation concentration of oxygen in the respective
into account differences in electrode wetting among different pore electrolyte.
designs. Coupled with flat-electrode experiments that are useful for While the oxygen partial pressure, and the electrolyte’s oxygen
assessing passivation, such a study may help separate the effects of solubility and diffusion coefficient, have a strong influence on the
pore blocking and passivation, which may have a similar overall effect high-rate capacity, the fact that the low-rate capacity saturates at
on electrochemical performance. less than 2000 mAh/g carbon implies that O2 and Li+ transport in
Oxygen transport limitations may arise in flooded electrodes; gas the electrolyte do not ultimately limit the discharge capacity. How-
channels will improve oxygen transport; 3-phase percolation is ever, if passivation or other limitations can be avoided, selecting elec-
important.—Oxygen transport limitations are expected to become im- trolytes with high oxygen solubility, low viscosity, and high conduc-
portant even at low current densities (i.e., <1 mA/cm2 , although the tivity, and using a high oxygen partial pressure may afford higher
precise value depends on the electrode thickness and other parame- rate capability. The work of Read et al. suggests that ether-based
ters) in fully flooded Li/air cells because the solubility of oxygen in electrolytes and solvent blends made of propylene carbonate and
typical nonaqueous electrolytes is less than 5 mM, and thus at least tris(2,2,2-trifluoroethyl) phosphate possess favorable properties re-
a factor of 20 lower than the typical concentration of the salt in a lated to oxygen transport.70, 71
nonaqueous electrolyte. The diffusion coefficient of oxygen is prob- To further enhance oxygen transport, continuous gas-diffusion
ably somewhat higher than that of typical electrolyte salts, although paths could be created within the porous framework of the air
rigorous measurements are not available.21, 69 electrode.72, 73 Adjusting the O2 partial pressure slightly above am-
While optimizing the composition of the electrolyte involves a bient pressure may be sufficient to force the electrolyte into wetting
very large permutation of salts, solvents, and additives, the ability of pores and create a desirable 3-phase percolation throughout the air
an electrolyte to dissolve oxygen can be investigated systematically. electrode. This can also be achieved by employing suitable combi-
Read et al. measured the ionic conductivity σ, the viscosity η and nations of binders, carbons, and current collector grids, as well as
the Bunsen coefficient α of various electrolytes and compared these by fabricating the composite electrode with a particular pore-size
values to the capacities obtained from cycling experiments of lithium distribution.74 We performed measurements under systematic varia-
half-cells.20, 21 Furthermore they tested the influence of the oxygen tion of the amount of electrolyte in the cell, which suggest that the
partial pressure on the cell capacity. With respect to each parameter, cell performance goes through a maximum with the amount of elec-
the cell capacity increases and then saturates as shown in Figure 13, trolyte. While too much electrolyte impedes the supply of oxygen
Figure 13. Specific cell capacity (per g carbon) depending on four different electrolyte parameters at various discharge rates: (a) Electrolyte viscosity η; (b)
Bunsen coefficient α which represents the oxygen solubility; (c) oxygen partial pressure; (d) ionic conductivity σ. The letters a-f at the individual curves indicate
the discharge rate: a) 0.5, b) 0.4, c) 0.3, d) 0.2, e) 0.1, f) 0.05 mA/cm2 . Figures reproduced from reference 21.
is electrically insulating and remains at the reaction site that will be

Initial open-circuit potential the principal capacity-limiting mechanism.
3.0
With O2 transport limits, Enhancing Li2 O2 solubility could enable non-passivating Li2 O2
2.8 without passivation precipitation.—Passivation of the electrode surface occurs because
~20,500 mAh/g Super P carbon bulk Li2 O2 is electrically insulating and is insoluble in the elec-
Cell potential (V)
2.6 trolyte. Enhancing the solubility of Li2 O2 could potentially enable

nucleation and precipitation of solid Li2 O2 away from reactive sites
2.4 in the air electrode. PolyPlus has proposed the use of noncarbon-
With O2 transport limits, ate solvents with limited Li2 O2 solubility such as ethylene gly-
2.2 with passivation col and dimethylformamide, which could be enabled through the
use of a protected lithium anode.75 Researchers have also proposed
2.0 Without O2 transport limits, the use of boron-based anion receptor additives to significantly en-
with passivation hance the solubility of Li2 O2 ,76, 77 some of which may assist passi-
(a)
1.8 vation of the negative electrode at low potentials.77 However, these
studies have mainly been carried out in carbonate solvents, and one
Initial open-circuit potential must be careful to distinguish between Li2 O2 solubility in and reac-
3.0
tivity with the solvent.
2.8 Novel designs may address passivation and improve capacity.—To
With O2 transport limits, overcome capacity limitations related to passivation, it may be neces-
Cell potential (V)
without passivation sary to find radically new solutions that address the specific properties
2.6 ~7,350 mAh/g Super P carbon
of the Li/air system. Very recent work indicates that tailored electrode
2.4 structures,78 surface treatments,35, 79 introduction of defects into the
discharge product,5 and elevating the operating temperature may lead
2.2 to higher discharge capacities without passivation.
Mitchell et al. recently described an electrode consisting of a hol-
2.0 low carbon-nanofiber “carpet” grown via chemical vapor deposition
With O2 transport limits, Without O2 transport limits,
on porous anodized alumina coated with thin Ta and Fe layers.78
with passivation with passivation (b)
1.8 High discharge capacities (7200 mAh/g carbon at 63 mA/g carbon)
0 200 400 600 800 were obtained, and the Li2 O2 discharge product, identified by XRD,
grew as nodules on the fibers and developed into toroids (up to
Specific capacity (mAh/g Super P carbon) 1 μm) over the course of discharge before eventually forming a
monolithic mass (see Figure 15). It should be noted, however, that
Figure 14. Simulation-only results demonstrating the relative impacts of elim- the carbon loading was rather low (∼0.1 mg/cm2 ), and that the ef-
inating oxygen transport limitations and eliminating the electronic resistance of fective low-rate capacity (0.7 mAh/cm2 ) is roughly a factor of 2–10
the discharge products. (a) shows results at a current density of 0.08 mA/cm2 ; lower than in Li-ion cells. Additional work with higher carbon load-
the two simulations with passivation superpose. (b) shows results at a current ings would be very illuminating. Similar air electrode designs might
density of 0.47 mA/cm2 . Caption and figure reproduced from reference 23.
be realized using carbon nanotubes grown on catalyst particles (e.g.,
nickel) which are placed on the electrode surface. Alternatively one
could grow metal nanowires by chemical vapor deposition80 or elec-
gas, too little electrolyte results in poor effective ionic conductivity trochemically. We also note that a systematic study of the morphology
and diffusivity. If, as described in the introduction, a Li/air system of Li2 O2 that forms during discharge, including as a function of sub-
is designed with a pressurized oxygen tank and the stack is exposed strate type, current density, and electrolyte, is a high-impact area for
directly to those high pressures, the quantity of oxygen in the elec- research.
trolyte of a flooded cell would increase significantly, which would Surface treatment (e.g., hydrophobization35 ) of the electrode sur-
have the effect of minimizing oxygen transport limitations. In other face may also be an appropriate method for mitigating passiva-
words, running the stack at a high oxygen pressure could have the tion. For example, C. Tran et al. recently demonstrated that sur-
practical effect of increasing the capacity at a given current density, face treatment of the carbon material leads to an enhanced capac-
or moving to the right in Figure 13c. ity and a discharge curve that implies diminished passivation (see
Albertus et al. carried out modeling work to assess the principal Figure 16).79 In our own experiments on electrodes with a hydropho-
limits on the capacity of a Li/air cell. While the experimental system bized carbon surface, we found that the oxygen reduction products
on which their model was based contains carbonates, they neverthe- form small crystallites on the hydrophobized surface instead of a
less concluded that electrical passivation by the discharge product is dense film. Open pores between the crystallites may ensure charge
a key limitation in nonaqueous systems, as both Li2 O and Li2 O2 are and mass transfer to the reaction site. Nevertheless, conclusive ev-
electrically resistive and have a minimal solubility in typical nonaque- idence of this coating-passivation correlation is still lacking in the
ous electrolytes. Figure 14 shows the effect of electrical passivation literature.
by the discharge product and oxygen transport limitations for a non- Besides sophisticated air electrode geometries it may be possible
aqueous Li/air cell. The figure shows that at low current densities to increase the conductivity of the reaction product by doping. Hum-
(<0.5 mA/cm2 ) removing the passivation limitation leads to a more melshøj et al. predict by DFT calculation of the Li2 O2 density of states
significant increase in capacity than removing oxygen transport limi- that the introduction of Li-defects (vacancies) increases the density
tations. At current densities at and above about 0.5 mA/cm2 oxygen of states around the Fermi energy.5 The calculated density of states
transport limitations are important for the particular cell design and for pure Li2 O2 and Li2 O2 with Li vacancies is shown in Figure 17.
assumptions used by the authors, but again the removal of passiva- If one can find a way to create Li-vacancies in-situ during discharge
tion results in a more significant increase in capacity (i.e., removing the maximum capacity could be increased and the subsequent charge
oxygen transport limitations while keeping passivation increases the is believed to be facilitated. Similarly, introduction of dopants dur-
capacity by less than a factor of two; removing passivation while ing discharge could facilitate continuous Li2 O2 growth. A practical
keeping oxygen transport limitations increases the capacity by about solution for achieving such defects is unavailable.
a factor of 30). This result demonstrates that while efforts to improve Another possibility to make the discharge product more conduc-
gas flow channels can make notable improvements to the capacity that tive is to operate the Li/air cell at elevated temperatures (e.g., 60◦ C)
can be achieved by nonaqueous Li/air cells, if the discharge product because the conductivity of the insulating or semiconducting dis-
Figure 15. Evolution of Li2 Ox discharge product morphology on carbon nanofibers (electrolyte = 0.1 M LiClO4 in DME). Insets show the corresponding
discharge voltage profile. (a–b) Galvanostatic discharge to a capacity of 350 mAh/g carbon at 68 mA/g carbon. Li2 O2 particles appear to first form on the carbon
nanofiber sidewalls as small spheres with ≤100 nm diameters. (c–d) Intermediate galvanostatic discharge to 1880 mAh/g carbon at 64 mA/g carbon. Particles
appear to develop a toroidal shape as the average particle size increases to 400 nm. (e–f) Full discharge to 7200 mAh/g carbon at 63 mA/g carbon. The discrete
particles merge to form a monolithic Li2 O2 mass with low porosity. Caption and figure reproduced from reference 78.
charge product increases with temperature. First measurements in our continuum scale modeling of film conductance and growth are new
laboratories indicated an increase of the discharge capacity by 50% and open areas of research that could help improve this understand-
compared to room temperature. For certain electrolytes a significantly ing. In addition, there is currently no systematic understanding of the
(∼500 mV) lower charge potential was also observed. While O2 sol- growth process of Li2 O2 and the morphology of particles. This is a
ubility and transport of both O2 and Li+ would also be improved, key area for exploration.
operating Li/air cells at elevated temperature could make the search The use of solubilising electrolyte solvents or additives may reduce
for a stable electrolyte even more challenging. or eliminate the product resistivity problem, while the development of
Outlook: Overcoming product resistivity and transport limitations is novel electrode architectures could enable high pore volume utiliza-
the key to achieving high-energy nonaqueous Li/air cells.—Under- tion by avoiding or in spite of passivation. Appropriate pore structures
standing the morphology and conductivity of the discharge product are required to avoid pore blockage, and a combination of wetting
in Li/air cells will help pave the way to high-capacity nonaqueous
Li/air batteries. Experimental measurements of film conductance, us-
ing both flat and roughened substrates, as well as first-principles and
Figure 17. Calculated density of states for (a) pure Li2 O2 and (b) Li2 O2 with
Figure 16. Discharge curve for uncoated and coated carbon in a gas diffusion a concentration of 1/16 Li vacancies. The black curve shows the DFT single-
electrode. The different slopes, indicated by the black lines, suggest that the particle spectrum and the red curve shows the GW quasiparticle spectrum. In
coating mitigates electrical passivation, leading to a larger discharge capacity. (a), the top of the valence bands have been aligned and in (b) the Fermi levels
Reproduced from reference 79. have been aligned. Reproduced from reference 5.
Li2O 2 cell before dis charge Li/LiMO 2 cell before dis charge
CC CC CC CC
Li metal LP S L P os itive electrode G DL Li metal LP S L P os itive electrode
(w/ 20% (70 vol. % gas ) (w/ 20% (65 vol % MO 2)
exces s ) exces s )
Li2O 2 cell after dis charge Li /LiMO 2 cell after dis charge
CC CC CC CC
20% LP S L P os itive electrode G DL 20% LP S L P os itive electrode
exces s (65 vol. % Li2O 2) exces s (65 vol. % LiMO 2)
Li metal Li metal
Figure 18. Schematic showing the significant amount of volume change that occurs when Li metal cells are discharged. The cell on the left has a high-capacity
Li/air active material (Li2 O2 ) while the cell on the right has a lower-capacity intercalation active material (LiMO2 ). LPSL = Lithium Protection Separator Layer.
CC = current collector. GDL = gas diffusion layer. The layer thicknesses are drawn in proprtion.
and nonwetting pores may enable high transport rates for both lithium region of the cell undergoing significant volume changes. However, in
ions and oxygen to the reaction sites during discharge. We reiterate a Li/air cell the discharge process involves the creation of a new phase
that it is important to report directly the volume fraction of product rather than the incorporation of Li atoms into a pre-existing phase.
in the discharged electrode (or cell), or provide sufficient information The creation of a new phase in Li/air cells implies that electrolyte
for the reader to do so. Reporting only the capacity in mAh/g carbon displacement will occur unless the new phase displaces a gas phase.
is not enough to determine whether an experimental cell can actually Figure 18 shows a diagram of the volume changes in a Li2 O2 and
achieve a high capacity and energy. However, reporting the mAh/cm2 a Li/LiMO2 cell for the cell specifications given in Table II; for the
and the electrode thickness is sufficient. Li2 O2 cell it is assumed that the positive electrode begins with a
volume fraction of gas of 0.70 that is pushed out of the cell sandwich
during the discharge process by the Li2 O2 that forms.
Significant volume changes in Li/air cells need to be A more quantitative depiction of volume change is given in
accommodated.— In a Li/air cell both the Li metal electrode and Figure 19. The results in this figure depend strongly on assump-
the positive electrode have materials that undergo significant volume tions about the initial components in the positive electrode, partic-
changes. In earlier sections we discussed the formation of new phases ularly on the quantity of electrolyte vs. gas phase in the charged
in the positive electrode, and in the section on aqueous Li/air cells we cell. As discussed above, for our “practical” nonaqueous Li/air cells
will discuss the formation of a solid LiOH · H2 O phase. However, in we assume the active materials that are produced displace a gas
this section we focus on quantifying the magnitude of volume changes phase rather than electrolyte, while in the aqueous cases we in-
for a cell with the parameters given in Table II, and then critically re- clude an initial reservoir of H2 O. The results in Figure 19 show
viewing solutions that help manage those significant volume changes. that the volume change is more significant for Li2 O than Li2 O2
Volume changes in Li/air cells are particularly challenging because
the cell sandwich accumulates mass (oxygen) during discharge and
releases mass during charge. Note that it is important to distinguish 1.0
between electrodes that undergo significant volume change and a cell
that undergoes significant volume change. Balanced volume changes
at each electrode are possible if the density of the products and reac- 0.8
Final / initial volume
tants are matched, but this is not true for the active materials alone
in the case of Li/air cells. We consider the general topic of reversibly
accommodating major volume changes in solid systems to be a high- 0.6
impact area for research.
The high capacity of the cathode materials and the formation of new
phases make volume changes significant.—A cell with a Li metal 0.4
negative electrode undergoes significant changes in the thickness of Cell sandwich alone
the cell sandwich during a cycle, provided a significant amount of Including H2O reservoir
the Li metal is actually cycled. Volume changes are a feature of any 0.2
metal electrode in which cycling involves plating/striping, but are
particularly significant in the case of Li metal because the density of
Li metal is so low (0.534 g/cm3 at 25◦ C). Remarkably, the density of 0.0
the Li (in mol Li/cm3 , which is directly proportional to the capacity Li2O Li2O2 LiOH·H2O LiOH Li/LiMO2
density in mAh/cm3 ) stored in Li2 O and Li2 O2 is significantly higher
than in pure Li metal, as shown in Table I. Indeed, the Li/air cell is Figure 19. Plot showing the volume change of four Li/air cells and a Li/LiMO2
an interesting case of a cell that increases in mass and decreases in cell. LiOH has not been observed to precipitate and is included only for compar-
volume during discharge. In a Li metal cell with a positive electrode ison. Volume change is calculated as discharged volume / charged volume. The
that intercalates Li and has minor volume changes (<15% in many LiOH · H2 O case shows the volume change when an H2 O reservoir is excluded
positive-electrode intercalation materials), the Li metal is the only and included (it is assumed the H2 O is stored outside the cell sandwich).
because of the higher capacity density of Li2 O. The volume changes or SEI) must form at the surface in order to stabilize the interface. The
for the LiOH · H2 O, LiOH,and Li/LiMO2 cases are similar. The vol- SEI imparts additional ohmic and mass-transfer resistance to dissolu-
ume change for LiOH · H2 O and LiOH are smaller than the Li2 O and tion and deposition of the metal. Second, the plating and stripping of
Li2 O2 cases because their capacity density is lower than that of Li2 O or Li inherently involves change in volume of the active material. This
Li2 O2 (see Table I) and therefore less Li is required. In the LiOH · H2 O volume change often involves roughening of the Li surface, evolu-
and LiOH cases, when taking into account the volume of water in a tion of the metal morphology (i.e., formation of grain boundaries),
reservoir in a charged cell the overall volume change of the cell is and macroscopic shape change. In addition, the contact resistance be-
larger. Note that we exclude the additional volume from packaging in tween the Li and the electrolyte or separator may change depending
Figures 18 and 19. on the direction of current, electrode state of charge, age of the cell,
Pressure should be applied to cells changing volume to maintain good and pressure applied to the cell sandwich. Although SEI formation
contact.—Given the significant volume changes that occur in Li-metal allows for research on nonaqueous Li/air cells without ex-situ appli-
cells it is important to maintain good contact between the layers of the cation of protection layers, SEIs are not robust against volume and
cell sandwich in order to prevent significant contact resistances from surface changes of the Li anode.85 Hence, formation of an SEI should
arising. In particular, maintaining a good solid-solid contact between be inhibited by using a solid electrolyte separator that is stable against
Li metal and the lithium protection separator layer is important, as Li. This solid electrolyte should also provide mechanical resistance to
there will be a great deal of lithium transport across that interface. Li morphology development and roughening.
Application of external pressure on the cell, depending on its design, We note that despite decades of interest and development, thus far
may alleviate this variability.62 Another way of dealing with this issue there has been no successful and widespread introduction of secondary
is to use a block copolymer that maintains good adhesion through the Li-metal cells into the market. Research on methods to enable the use
ability of the conducting block to flow.81 of Li metal in secondary cells remains a high priority.
Minimizing changes in contact resistance may prove more diffi- It is challenging to form stable SEIs on Li metal.—Li is an alkaline
cult for wound cells. Establishing relationships between mechanical metal and hence quite electropositive, chemically reactive, and rather
design and cell performance and durability will become an important susceptible to oxidation. This property, which makes it very attractive
area of research after fundamental chemical challenges are overcome. for use as a negative electrode, also makes it prone to react with other
It is important to quantify the evolution of contact resistances with components in the cell. Although passivation layers on Li were pre-
cycling and the influence of pressure on this evolution. viously examined in aqueous16, 17, 86, 87 and nonaqueous88 electrolytes,
Volume change can be accommodated through the use of a flexi- Peled coined the now popular term “Solid Electrolyte Interphase”
ble seal.—How can cell packaging respond to significant volume (SEI) for the layer that forms on the surface of Li metal via decom-
changes? One promising idea for dealing with volume change is to position of the electrolyte,89 which is unstable at the Li potential.
use a flexible seal that allows the cell sandwich to move without pro- Comprehensive reviews of the SEI, which forms on lithium, graphite,
viding significant stress on the cell packaging. Such a sealing method and other Li-insertion anodes, include those of Aurbach,90 Balbuena
has been outlined by the PolyPlus battery company in a patent.82 The and Wang,91 Winter et al.,85 and Ogumi and Inaba.92
patent outlines a method to maintain good ionic, mechanical, and The SEI on Li differs significantly from that on graphite in that it
electrical connectivity between a Li metal electrode and both the Li forms immediately upon contact between the electrode and electrolyte.
metal backplane and a Li protection layer. However, this approach has Hence, it is much more challenging to control the chemistry and
so far been applied only to planar double-sandwich cells. Wound and morphology of the SEI during formation on Li. Rather, the electrolyte
stacked cells may require more intricate packaging. salt, solvents, and additives must be selected carefully to achieve a
Volume change can be accommodated through the use of an electrolyte highly stable SEI.
reservoir.—If a Li/air cell is discharged to the point that the active Evolution of the SEI on Li is also very different from that on
material being formed (e.g., Li2 O2 ) displaces all of the gas volume, it graphite due to the dramatic volume change of the electrode. De-
will begin to move electrolyte within or out of the cell sandwich. If pending on the chemistry, temperature, mechanical design of the cell,
the electrolyte remains within the cell sandwich and is pushed toward and cycling conditions, the Li-metal SEI may evolve more or less
the Li metal, it may lead to poor contact between the Li metal and dramatically along with morphology changes in the underlying metal
lithium protection separator layer or between the lithium protection (see Figure 20). If films are not elastic enough to accommodate the
separator layer and the positive electrode. Alternatively, it may be volume change of the active material, fracture and reformation or
drawn off in a direction perpendicular to the direction of current flow. thickening of the film may occur as the cell is cycled.93 Spotnitz mod-
Pollard and Newman treated an electrolyte reservoir in the context of eled SEI growth on graphite using an empirical relationship that takes
the Lithium-Aluminum, Iron Sulfide cell.83 into account the rate-dependence of film fracture.94 However, detailed
modeling of the SEI during evolution of the Li metal morphology is a
Outlook: The operating principle and durability of cells that repeat- challenging endeavor that has apparently not yet been undertaken.
edly undergo significant volume changes have to be evaluated.—Ac-
commodating the significant volume changes in both the Li metal Dendrites and morphology changes limit cyclability.—The most com-
electrode and within the positive electrode in a highly reversible man- mon failure mode for cells with Li metal anodes is that of dendrite
ner is one of the principal challenges for the successful commercial- growth and increase in electrode surface area. Needle-like dendrites
ization of Li/air cells. Most Li/air cells demonstrated thus far have can grow through the separator during charging of the cell, resulting
cycled only a small amount of Li, although primary cells have dis- in an internal short. “Soft shorts” that burn out rapidly result in a tem-
charged significant thicknesses.9, 14, 84 While there are promising ideas porary self discharge of the cell, while “strong shorts” consisting of a
for maintaining good seals and contact between cell layers, this is an higher, more stable contact area can lead to complete discharge of the
area in which major progress will need to be made. cell, cell failure, and possibly thermal runaway. While dendrites grow
through the separator during charge, shorts can sometimes develop
Li metal electrodes are chemically and morphologically un- during discharge depending on the external pressure placed on the
stable in most electrolytes and must be protected.— Reversible charge cell and/or internal volume changes that occur in both the negative
storage in the Li metal electrode is accomplished by plating (deposit- and positive electrodes.
ing) and stripping (dissolving) Li on/from the surface of the electrode. Because Li metal is highly electronically conductive, surfaces tend
This deposition-dissolution reaction can be simply modeled by an to roughen as the metal is plated and stripped. Peaks in the surface
appropriate kinetic expression, such as the Butler-Volmer equation. grow as dendrites during charge; while the surface is smoothed during
However, two factors complicate the process. First, the nonaqueous discharge, some roughness typically remains at the end of discharge,
liquid electrolyte commonly used in Li metal cells is unstable at the and, depending on the depth of discharge, the overall roughness can be
potential of Li, and a passivating film (the solid electrolyte interphase, amplified from one cycle to the next. Because the metal is essentially
Figure 20. Illustration of differences in SEI formation and evolution on the surfaces of (a) graphite and (b) metal (Li or Li-alloys). Reproduced from reference 85.
at the same electrochemical potential throughout, potential and, to a chanical models that account for gradual swelling of the anode with
lesser extent, concentration gradients in the electrolyte phase drive the cycling.
change in morphology. Previous Li dendrite growth modeling work Alloying Li with other elements such as Mg may help elevate
has shown that the moving front of a dendrite tends to accelerate during the surface energy and reduce morphology changes,99, 100 and more
cell charge due to the higher current density localized at the dendrite studies are needed in this direction.
tip relative to its base.95 Application of thermodynamic models has Shape change can occur due to nonuniform current densities.—Pro-
shown that dendrite initiation (i.e., initial roughening of an almost vided that Li dendrites and other forms of microscopic morphology
perfectly smooth surface) can be suppressed by applying mechanical development can be suppressed, macroscopic changes in the shape of
stress and selecting solid electrolytes with shear moduli on the order the anode may occur due to nonuniform current density distribution
of 10 GPa at room temperature.96, 97 The same models indicate that throughout the cell. When there is high resistance to electronic current
surface tension at metal-fluid interfaces is insufficient to suppress flow from the tab to the outermost edges of the current collector (e.g.,
dendrite initiation.96, 97 if the current collector thickness is too small relative to its area), the
Related to dendrite initiation and growth is development of the current density may be very nonuniform, with a higher current den-
Li morphology, which tends to increase the electrode surface area sity near the tabs. Hence, Li will deposit and dissolve preferentially
with cycling and consumes solvent to generate fresh passivation near the tabs, as depicted conceptually in Figure 22. Application of
layers. Formation of high-surface-area mossy Li tends to occur dur- pressure may alleviate this problem.
ing low-rate deposition from a liquid electrolyte, especially if the salt Li-conducting solid electrolytes can provide chemical and mechanical
concentration is high.98 The high surface area combined with high protection of Li metal.—Because of the enormous challenge involved
reactivity of Li and flammability of the organic solvent makes for a in stabilizing the Li surface chemically and mechanically through
very reactive and dangerous cell. the use of electrolyte additives, such that passivation remains in ef-
Sion Power has reported a diminishing grain size in its Li/S cells fect over hundreds to thousands of cycles, the preferred treatment for
without application of external pressure to the cell. This results in rechargeable Li-based cells is the use of a solid-electrolyte membrane
a gradual increase in the volume of the anode as the cell is cycled. that is mechanically robust and chemically stable against both elec-
Moderate pressures (∼10 bar) tend to mitigate this type of morphology trodes. Such a barrier removes several simultaneous constraints that
development (see Figure 21). Sion Power also proposed a strategy to the liquid electrolyte otherwise must satisfy, but the requirements for
minimize the surface morphology changes in the Li anode, which its properties are nonetheless multifaceted and challenging to obtain in
consists of ensuring complete stripping and re-plating of all lithium a single material. The barrier must be chemically stable with respect to
on the rigid current collector in an anode-capacity-limited cell.62 some or all of the following: the liquid electrolyte in the positive elec-
Modeling needs in the area of morphology change include models trode, electronic conductors and catalysts in the positive electrode,
of grain nucleation and growth during Li deposition, which include the metallic Li negative electrode, reactive species such as oxygen
the influence of pressure; cell-level aging models that account for molecules and reaction intermediates, and (in aqueous cells) water.
solvent consumption at the anode; and cell-level electrochemome- Solid electrolytes must also have sufficient Li+ conductivity over the
operating temperature range of the cell, negligible electronic con-
ductivity, and high elastic modulus to prevent Li dendrite initiation.
In order to provide cheap, robust, lightweight protection, a method
must be developed to produce relatively thin (< 50 μm), pinhole-free
solid-electrolyte layers at a reasonable cost.
Al tab Al tab
Cathode Cathode
1000 cycles
Separator Separator
Li metal
Li metal
Cu tab Cu tab
Figure 21. SEM images of Li foil anodes, cycled 50 times, (a) with and
(b) without application of 10 kg/m2 nonisotropic pressure. Reproduced from Figure 22. Depiction of macroscopic Li redistribution in a cycled Li/air cell
reference 62. at the end of charge.
Figure 23. Ionic conductivities of several classes

of solid electrolytes (LLTO = (La,Li)TiO3 ; LAGP
= Li1+x Alx Ge2−x (PO4 )3 ; PVdF-HFP = poly
(vinylidene fluoride)-hexafluoropropylene; PEO
= polyethylene oxide). Reproduced from reference
103. Figure numbers refer to those in the original
paper.
Mechanical stability is another important aspect of Li-metal elec- dio frequency magnetron sputtering of lithium silicates, phosphates,
trode protection that influences the age and safety of the cell. The or phosphosilicates resulted in N2 incorporation to form LiPON, an
two primary classes of Li solid electrolytes, inorganic ceramics and amorphous analog to LiSICON.110 Thin-film batteries with Li anodes
solid organic polymers, have different mechanical properties. Ceram- and Lithium Phosphate OxyNitride (LiPON) separators have demon-
ics have high elastic moduli and are thus more suitable for rigid flat strated thousands of cycles.111, 112 However, mechanical stability suf-
cell designs. Polymers are softer and more flexible, but at the same ficient for long cycle life has not been established in a thick-electrode
time less chemically and thermally stable, and also susceptible to cell design, and the low conductivity of LiPON (2×10−6 S/cm at
absorption and transport of liquids. Flexibility is an advantage due 25◦ C110 ) precludes the development of cells with thick LiPON mem-
to the high anode volume change, and in wound designs in which branes.
changes in the separator curvature could cause fracturing of ceramic Thio-LiSICON, the S analog to LiSICON (e.g., Li4 GeS4 113 ), can
membranes. At the same time, the lower shear moduli of polymers achieve high room-temperature conductivity and low activation en-
are likely insufficient to prevent Li metal dendrite initiation,97 leading ergy (e.g., 2.2×10−3 S/cm and 20 kJ/mol for Li3.25 Ge0.25 P0.75 S4 102 ).
to safety concerns due to possible electrical shorts. Elasticity may be Glass ceramics with structures related to thio-LiSICON exhibit even
important in order to reduce fracture and tearing after multiple cycles better performance (3.2×10−3 S/cm, Ea = 12 kJ/mol for 70Li2 S-
and to achieve wound cell designs. 30P2 S5 102, 114 ). For these systems glass ceramics have roughly an or-
A number of candidates that satisfy some of the requisite properties der of magnitude higher conductivity at room temperature than their
for Li protection have been proposed.10, 82, 101–128 For extensive reviews amorphous counterparts (glasses), although there are several excep-
see references101–103 and references therein. tions to this trend.103 Despite high conductivities (10−3 S/cm at room
A variety of inorganic compounds (sulfides, oxides, phosphates) temperature), ceramic films are not easy to fabricate and often have
in crystalline, polycrystalline or amorphous morphologies, as well poor chemical durability.115
as solid dense polymer-based materials, have been investigated with Early work on the Li analog to NaSICON, LiM2 (PO4)3 , led to
conductivities at room temperature ranging from 10−8 to 10−2 S/cm discoveries of some materials with high conductivity, but poor chem-
(see Figure 23).10, 82, 101–128 Most inorganic crystalline and glass mate- ical stability against Li (M = Ti116, 117 ), and others with good sta-
rials have lower conductivities than liquid electrolytes, by at least 1-2 bility, but poor conductivity (M = Ge118, 119 ). It was found that Al
orders of magnitude. It is worth mentioning that because Li motion substitution via solid state reactions resulted in NaSICON structures
in solid state systems is a thermally activated Arrhenius-type process, (Li1+x Alx Ge2−x (PO4 )3 , or LAGP) with 4 orders of magnitude higher
conductivity increases with temperature, sometimes by two orders of conductivity.120
magnitude or more over the range 0 to 200◦ C. While operating a Li/air Workers at Ohara found that by heat treating glasses they
cell at elevated temperature (>80◦ C) may increase the rate capability formed glass ceramics including a NaSICON crystalline phase
and capacity, this presents numerous engineering challenges at the (Li1+x Alx Ti2−x (PO4 )3 , or LTAP), with a high conductivity up to
system level. 1.3×10−3 S/cm, depending on heat-treatment temperature.115, 121 Not
Li3 N has high conductivity (∼10−3 S/cm at room temperature), only is the material stable against Li metal, it is also stable in aque-
but is unstable at high potentials (>0.445 V vs. Li).101, 104 Li3 P ous solutions, making it a suitable candidate for aqueous Li/air cells,
has an order of magnitude lower conductivity, but is stable up to and was recently shown to be stable in acidic solutions.122 However,
2.2 V.101, 105 The Li analog to sodium β-alumina, Li2 O·11Al2 O3 , has a this material tends to be difficult to manufacture with large area and
high room-temperature single-crystal conductivity of 3×10−3 S/cm, low thickness. The general composition of water-stable glass ceram-
but is extremely hygroscopic and challenging to prepare dry.101, 106 ics produced by Ohara and used by PolyPlus in its aqueous Li/air
The so-called Li Super-Ionic CONductors (LiSICONs) are γII - and Li/water cells is Li1+x (M,Al,Ga)x (Ge1-y Tiy )2−x (PO4 )3 , where x<
Li3 PO4 type oxysalts101, 102 (e.g., Li14 Zn(GeO4 )4 107, 108 ) that contain = 0.8, 0< = y< = 1, and M is a member of the Lanthanide series;
interstitial Li ions, but show a drop in conductivity over time at low and/or Li1+x+y Qx Ti2−x Siy P3−y O12 , where 0 < x < = 0.4 and 0 < y <
temperature because of Li trapping by the immobile sublattice via de- = 0.6, Q = Al or Ga.10, 82, 121, 123
fect complex formation.101, 109 Room temperature conductivities are The garnet family of ceramics have lower conductivity than
generally less than 10−4 S/cm.101 Bates and coworkers found that ra- NaSICON-type phosphates and perovskite-type oxides (8×10−4 S/cm
at 300 K124 ), but they have a lower grain boundary resistance and are Cell impedance must be significantly reduced to achieve adequate
stable against Li metal.102 The garnet chemical space is still being power density and efficiency.— In addition to providing high specific
widely explored. energy and cyclability, the specific power (or power density) and
Grain boundaries often limit the conductivity of solid electrolytes; energy efficiency of the Li/air cell must be acceptable for the intended
for instance, perovskites (ABO3 ) such as Li0.5−3x La0.5+x TiO3 125–128 application. We shall assume that this potentially high-energy system
have bulk conductivities as high as 1×10−3 S/cm, but with grain is mainly used in high-energy applications, such as electric vehicles,
boundaries the composite conductivity is only 2 × 10−5 S/cm.125, 126 portable electronics, stationary storage, and potentially hybridized
The grain boundary contribution to total resistance is commonly sen- applications, in which the Li/air battery provides the base load while
sitive to the ceramic composition and depends on foreign phases that a second battery or a capacitor provides load leveling.
precipitate at the boundaries. Thus making progress in ionic conduc- A long-range EV with advanced high-energy batteries can be ex-
tivity of a ceramic polycrystalline material requires understanding pected to travel roughly 387 miles on a single charge. Assuming
optimization of each contribution separately. To get a clear picture on 300 Wh of energy are required to propel the vehicle one mile, and
the limitations of bulk transport, one needs to synthesize and perform that only 83% of the battery’s energy is utilized, this range can be
transport measurements on single crystal samples, which is often very satisfied by a 140-kWh battery. If the maximum pulse discharge (i.e.,
challenging. First-principles computations of hopping barriers and acceleration) and charge (i.e., regeneration) power is 140 kW, then
molecular dynamics offer an alternative route which is proving useful the maximum pulsing rate of the battery should be 1C, while the av-
to understand atomic-level mechanisms. erage discharge rate, assuming a maximum sustained freeway speed
Once the bottlenecks in a given material are identified, it is critical of 77 mph, could be as high as C/5. Average charging rates could
to engineer the composition and fine-tune the synthesis process to be even higher, but we assume that overnight or partial charging can
optimize the microstructure and density. Even more difficult is to be tolerated for drivers demanding long range. A hybrid storage sys-
achieve these optima for a film that is only 50 μm thick. Thin-film tem, consisting of a high-energy, low-rate battery and a low-energy,
manufacturing processes, which vary from sputtering to chemical high-rate battery or capacitor could limit the high-energy battery rate
vapor deposition, tend to be expensive or result in inhomogeneous requirement to the average discharge C rate. It is clear that the required
coatings. power capability for a long-range battery strongly depends upon the
Multi-layer encapsulation of Li is needed to meet stability and vehicle range, charging strategy, and level of hybridization. However,
impedance requirements.—It is unlikely that a single material will be we can conclude that EV storage systems incapable of achieving ∼C/5
discovered that satisfies all the requirements for protecting Li metal continuous discharge are uninteresting, and that systems that achieve
in Li/air cells. Most Li/air cell designs to date involve the use of inter- > 2C pulsing are not required. Hence, the target battery-level specific
layers between Li and the solid-electrolyte membrane that are stable power is 140 to 1400 W/kg for a 700-Wh/kg battery, depending on the
at the Li potential and reduce considerably the contact resistance. degree of hybridization. Our cell energy calculations indicate that, to
PolyPlus has patented a method for producing a thin layer (∼0.2 to 1 achieve this high specific energy, an area-specific capacity of approx-
μm) of Li3 N between Li and the water stable NaSICON-type separa- imately 40 mAh/cm2 (similar to the capacities for Li2 O and Li2 O2
tor (LTAP) supplied by Ohara.123 This interlayer reduces the contact cells showin in Figure 4, and corresponding to 194 μm of cycled Li)
resistance that results from decomposition of the LTAP surface ex- is required. Hence, the target current density for Li/air cells should be
posed to Li. Electricite de France (EDF) has deposited a thin layer in the range 8 to 80 mA/cm2 .
(0.5 to 2 μm) of LiPON onto LTAPf to prevent degradation of the sep- The power density of a cell can be evaluated at the cell-sandwich
arator; however, imperfections in the coating allowed reaction of Li level as the product of the voltage (the open-circuit potential, U, minus
with the LTAP during cycling, ultimately resulting in the formation of the cell overpotential, ηcell ) and the current density, i (in A/m2 ), divided
cracks in the separator, and Li morphology development and detach- by the cell-sandwich thickness, Lcs :
ment from the separator were observed.129 Nonaqueous electrolytes i (U − ηcell )
have also been used to eliminate contact issues between Li and the P̃ =
separator.1, 27, 130 L cs
In its Li/S cells, Sion Power uses multi-layer membrane assemblies The specific power is obtained from the power density and the
including ceramic and polymer components to prevent reaction of the average material density, ρ:
electrolyte with Li,62 which may contribute to morphology develop-
ment and capacity fade. Polymer components may provide sufficient P̃
P̂ =
flexibility to achieve wound cell designs, while ceramic components ρ
may provide improved stiffness to prevent dendrite initiation.
For a typical Li/air cell, the average material density is
Outlook: A low-resistance multi-layer solid-electrolyte encapsulation approximately 1 g/mL (0.53 g/mL for Li, ∼3 g/mL for a
of Li metal is required for stable operation and good performance.— solid-electrolyte separator,120 8.9 g/mL for the Cu current col-
We identify three critical issues in Li metal protection: good ionic lector, 2.7 g/mL for the Al current collector, ∼1 g/mL for
conductance, chemical stability, and mechanical stability. There is polymers, water, and nonaqueous electrolytes); by comparison,
often a tradeoff between the three properties, depending on the cell Li-ion cells (without Li metal) have an average density of
design and positive-electrode electrolyte chemistry. In-situ protection ∼2 g/mL.
via an SEI is unlikely to prevent morphology development during The cell overpotential is a function of the current density, cell
cycling. Solid-electrolyte membranes that are flexible, strong, thin, design, state of charge, and materials used. As a function of cur-
pinhole free, easy to manufacture, and have high conductivities are still rent density, it is approximately linear at low current density and at
missing. More experimental and computational work to understand high current density increases exponentially due to increasing mass-
and design such materials is needed in the direction of mechanical transfer limitations. Hence, the optimum power is a balance between
strength and ionic conductivity. Interlayers between Li metal and overpotential and current density. Requirements of acceptable energy
the solid-electrolyte membrane are required to improve the stability efficiency make this a constrained optimization (see Figure 24). Even
and decrease the contact resistance of the interface. Application of if the cost of energy is not of concern, a lower system efficiency im-
pressure may be required to prevent Li metal morphology changes plies a higher rate of heat generation during charge and discharge, and
during cycling. therefore a higher cost, weight, and volume of the thermal manage-
f While the authors of reference 129 refer to the material as “Lisicon,” their reference
ment system.
to an article from PolyPlus and the fact that no other Li-protection material besides Several phenomena limit the rate capability of Li/air cells.—While
Ohara’s LTAP is known to be stable in water suggest that this is a misnomer. We shall the major rate-limiting processes include poor oxygen reduction and
assume that LTAP was intended. evolution reaction (ORR and OER) kinetics and large ohmic drops
changes that occur during charge and discharge, some contact re-
1.0 sistances, especially at the Li/separator interface, may evolve during
charge/discharge and as a function of cycle number. Likewise, gradual
physical separation of the discharge product from the carbon matrix
0.8 may result in significant changes in contact resistance as the cell is
cycled.
efficiency
Mass transfer limitations are those that involve concentration gra-

0.6 dients resulting from finite diffusivities in multicomponent solutions.
~ For instance, when oxygen is consumed at positive electrode reaction
P sites during discharge, it must be replenished by oxygen diffusing
through the liquid electrolyte. Because the solubility of oxygen in
0.4 nonaqueous electrolytes is limited,20, 21 even the relatively high oxy-
gen diffusivity is insufficient to maintain the reaction site oxygen
0.2 η cell concentration near its nominal bulk value in flooded cells discharged
at high rates. Hence, the ORR kinetics may suffer from a diminished
local O2 concentration. Limitations on the gas-phase oxygen transport
in diffusion media and Li-ion transport in the electrolyte can also im-
0.0 pact the cell impedance, but these contributions are typically smaller
than that of oxygen transport limitations in the electrolyte in flooded
current density cells.
Passivation of the positive electrode by an electronically insulating
Figure 24. Representative qualitative dependence of overpotential (ηcell ), discharge product is a complex phenomenon that may involve kinetic,
power density ( P̃), and discharge voltage efficiency upon current density. The ohmic, and mass-transfer processes. At this stage, the phenomenon is
vertical dashed line indicates the current density below which the efficiency is not well understood, although it has been identified as a main con-
at least 80%. tributor to the limited capacity and power capability of nonaqueous
Li/air cells. Empirically, the impedance of the discharge product ap-
pears to increase exponentially with film thickness, at least in some
through protection layers on top of Li metal, there are a number of solvent-salt combinations.23
processes that contribute to the poor rate capability and efficiency of In nonaqueous Li/air cells, current densities as high as 1 mA/cm2
Li/air cells. Impedance contributions in a battery can be classified as have been demonstrated.135 However, the capacity begins to fall off
kinetic, ohmic, or mass-transport limitations. The kinetic impedance at current densities above about 0.1 mA/cm2 , depending on the cell
for elementary electron-transfer reactions tends to obey the Butler- design and other parameters.23, 135 Moreover, most measurements in
Volmer expression, with approximately linear behavior at low surface the literature have been carried out in carbonate-based solvent sys-
overpotentials and logarithmic (or Tafel) behavior at high overpoten- tems, which show no evidence of reversible chemistry. There are
tials. However, ORR and OER in the Li/air system consist of multi-step limited data available for noncarbonate systems. Passivation of the
mechanisms that exhibit more complex behavior. electrode surface has the strongest influence on rate capability in
At the negative electrode, the kinetics of Li deposition and dis- carbonate systems.23 There is some evidence that it also plays a
solution are typically described by a Butler-Volmer expression, with role in noncarbonate systems.136 If passivation can be overcome
an exchange current density that can depend on the electrolyte. De- then the introduction of oxygen gas channels should allow suffi-
pending on the electrolyte, electrode preparation, and measurement cient quantities of oxygen into the cell to support any reasonable
technique, reported values of the Li exchange current density are gen- current density for a Li/air cell, as hydrogen-oxygen fuel cells can
erally on the order of 1 to 10 mA/cm2 .131–133 Assuming charge transfer operate at over 1 A/cm2 while much lower current densities are ex-
is described by Butler-Volmer kinetics with a symmetry factor of 0.5, pected for Li/air cells given their significantly higher area-specific
these current exchange densities imply that 100 to 250 mV of over- impedance.
potential will drive a current density of 100 mA/cm2 , which is much Aqueous systems, without O2 transport limitations or passivation,
higher than typical Li/air current densities. have achieved current densities of 15 mA/cm2 in primary applica-
In aqueous Li/air cells, the kinetics of LiOH · H2 O precipitation and tions with protected lithium electrodes137 and 2 mA/cm2 (at 30◦ C)
dissolution could limit the achievable specific power, as is discussed in secondary applications.129 However, these high current densities
in a later section. have not been demonstrated over large capacity windows. In pri-
Ohmic limitations include the high-frequency impedance of the mary applications, where cells are designed to discharge over weeks
electrolyte, the ionic resistance of separators or membranes that are or months, typical average discharge currents and specific power
single-ion conductors, and contact resistances. Nonaqueous elec- are 0.5 mA/cm2 and 2 W/kg, respectively.14, 84 By way of compar-
trolytes exhibit a conductivity of about 0.01 S/cm (1M LiPF6 in ison, more than 100 mA/cm2 has been demonstrated for rechargeable
PC/EC/DMC) at 25◦ C,134 while aqueous systems exhibit a conductiv- Zn/air systems using a bifunctional air electrode.138 Likewise, cur-
ity of 0.35 S/cm (saturated LiOH in H2 O) at 25◦ C.7 Li-conducting rent densities well over 100 mA/cm2 have been achieved in primary
ceramic separators have conductivities of 10−6 to 10−2 S/cm at Li/air cells without protection layers.87 This underscores the lim-
25◦ C.110, 115, 120, 121 Some Li-conducting layers can be made thin itation associated with the water-stable Li protection layer, which
(< 10 um)110, 111 so as to minimize the overpotential associated with has a Li conductivity of up to 3.5 × 10−4 S/cm at 25◦ C, but
the separator; however, such thin separators have not been successfully due to its fragility is difficult to manufacture at thicknesses below
employed in Li/air cells. Conduction of electrons through the carbon ∼150 μm.139 At 50 mA/cm2 , this represents a voltage drop of 2.15 V
matrix in the positive electrode, Li metal, current collectors, and tabs through the separator.
can also contribute to the cell impedance, but these contributions are Oxygen reduction and evolution are kinetically slow; the nature of
minor for a well-designed cell. electrocatalysis is unclear given that a solid product is generated.—
Contact resistances also tend to be ohmic in nature (i.e., with In order to improve air electrode performance and selectively as-
overpotential proportional to current density). These resistances can sist the formation of the desired reaction products, recent research
appear at a number of different interfaces within the cell, including has concentrated on the use of various metal and metal oxide par-
weld points, the Li/separator interface, interparticle contacts in the ticles in the carbon matrix.22, 24, 33, 34, 44, 141 Despite these efforts, the
carbon matrix of the positive electrode, and between the insulating molecular-level mechanistic processes that could explain the experi-
discharge product and carbon matrix. Because of the large volume mental observations (e.g., reduction of overpotentials for the charge
Mass breakdown: Total = 1.5963 kg/m2

Separator too Separator too Binder, etc.
-2 heavy heavy
10 Electrolyte
Conductivity (S/cm)
150 m Separator
100 m
-3
10 50 m
Anode cc
10-4 Cathode cc
Anode
-5
10
0 0.2 0.4 0.6 0.8 1 Cathode
C rate
Figure 25. (Left) The ionic conductivity of the separator required to attain a cell-sandwich specific energy of 1250 Wh/kg at the given C rate and separator
thickness. The conductivity of Ohara’s LTAP is indicated as a dashed line at 3.5e-4 S/cm. For a 150-μm separator, there is a C rate above which no cell design,
even if the separator had infinite conductivity, will achieve 1250 Wh/kg because the separator is too heavy. (Right) The mass breakdown by component in the
discharged state is shown for this limit. Assumptions are described in the text. cc = Current Collector.
and discharge reactions and/or extended capacity upon discharge of fied values of C rate and separator thickness.g The positive electrode
the Li/air cells) are not yet clearly understood. In particular, if a thickness and carbon loading were free to change in order to obtain
solid, insoluble product forms and remains at the reaction site the this optimum. Losses in the cell, which increased with C rate, were
catalyst may be quickly covered, such that the catalytic activity is re- included in the calculation of delivered specific energy (i.e., the dis-
duced. Nevertheless, several reports on catalysis have been published. charge energy was 1250 Wh/kg, while the nominal specific energy
Débart et al.34 have shown that a cell containing α-MnO2 as the catalyst was higher). To underscore the critical influence of the separator, only
significantly increases the discharge capacity of the Li/air cell. The au- ohmic losses were considered in the optimization. In a real cell, ki-
thors suggested that the improvement in the capacity of the cell is due netic losses, and potentially mass-transfer limitations, would further
to the presence of a tunneling structure that has the ability to accom- lower the delivered specific energy of the cell. The resulting discharge
modate in close proximity the Li+ and O2− ions, which in turn leads to efficiency for all optimized cell-sandwich designs was between 75
a subsequent incorporation of Li+ and O2− 2 into a compact film. Such and 85%.
proximity and incorporation is not possible in other manganese ox- Figure 25 shows the conductivity of the separator required to
ide materials. More recently, Lu et al.33, 142 incorporated bifunctional attain a cell-sandwich specific energy of 1250 Wh/kg at various C
catalysts into Li/air cells. By implementing a nano-structured PtAu/C rates in a Li/air cell with a solid-state separator. We assume that the
bifunctional catalyst into their cells, the authors were able to lower the density and conductivity of the electrolyte in the positive electrode
charge voltage and raise the discharge voltage, in order to obtain one are 1 g/mL and 0.01 S/cm, respectively, that the density of the
of the highest round-trip efficiencies of rechargeable Li/air batteries separator is 3 g/mL, and that the current collectors are Cu and Ni. An
reported to date. While there have been reports on catalysis, a recent anode/cathode capacity ratio of 1.2 was fixed for the optimization.
paper by McCloskey et al gives compelling evidence that catalysis The results imply that decreasing the thickness of a low-
observed in carbonate systems only aids electrolyte decomposition, conductivity separator may be a more effective route to attaining
while in DME no catalysis is observed for Au, Pt, or MnO2 .177 high practical specific energy at moderate C rates than increasing the
Thus far, the ORR and OER kinetics have not been well char- conductivity of a thick separator. For a 150-μm (100-μm) separator,
acterized for this system, although Xu and Shelton143 have recently it is impossible to achieve the target practical specific energy at a
applied DFT methods to investigate the Li-based oxygen reduction discharge rate above 0.4 C (0.7 C) with any cell design (i.e., the max-
reaction (Li-ORR) on two different metallic surfaces, Au (111) and Pt imum falls below 1250 Wh/kg). Put simply, a thick separator, even
(111), and found that Au(111) is the most active surface for Li-ORR. if highly conductive, requires correspondingly thick electrodes in or-
Their results also indicated that on both metallic surfaces, lithiation der to achieve high energy density, but thick electrodes imply a poor
significantly weakens the O–O bond and most likely will lead to the power density. On the other hand cells with thin separators and thin
formation of monoxides (Lix O), which will tend to aggregate to form electrodes could achieve both high energy density and power density.
cluster-like oxides. To the best of our knowledge, this is so far the only Outlook: Power density improvements require eliminating passiva-
study that has used theoretical approaches to understand the mech- tion in nonaqueous systems and reducing the thickness of lithium-
anistic details of the electrochemical reactivity of the Li/air cell on protection layers.—In nonaqueous systems, passivation phenomena
metallic surfaces, suggesting the need of further reports on this topic. appear to be the most restrictive with regard to achieving high cur-
Reducing the thickness of lithium-protection layers can significantly rent density. If passivation can be eliminated, poor solubility of O2 in
improve rate capability.—Solid Li-ion conductors generally have Li- nonaqueous solvents may require new cell designs with continuous
ion conductivities at least an order of magnitude lower than liquid Li gas, electrolyte, and electronically-conductive solid networks in order
electrolytes. One of the most attractive Li-protection layers, due to to achieve even higher current density. Catalysts that promote facile
its relatively high conductivity (>10−4 S/cm) and stability against a ORR and OER in nonaqueous systems may be required to improve
variety of solvents, including water, is the class of glass ceramics from the voltage efficiency of nonaqueous Li/air cells.
Ohara.115, 121, 139, 140 Unfortunately, this material is brittle and difficult In aqueous systems, the high thickness and low room-temperature
to manufacture at thicknesses below 150 μm. ionic conductivity of available water-stable Li-protection layers are
We used a 0D Matlab model to provide a rough estimate of the what limit the current density far below that of other metal-air sys-
separator materials design required to attain cell-level performance
targets. An optimization routine in Matlab (fmincon) was used to g For a cell-sandwich specific energy of 1250 Wh/kg, an 80% cell packaging factor
compute the minimum ionic conductivity in the separator required implies a cell level specifc energy of 1000 Wh/kg. A further 43% increase in mass at
to obtain a cell-sandwich specific energy of 1250 Wh/kg at speci- the pack level implies a system energy of 700 Wh/kg.
tems. Preliminary calculations indicate that reducing the protection Table IV. Gibbs free energies of formation for important lithium
layer thickness would have the greatest impact with regard to the compounds, and the reactants from which they are formed that
practically achievable specific energy. However, the mechanical was used to calculate the Gibbs formation energy. Data taken
stability of thinner layers must be adequate for the hundreds to from.11
thousands of large-volume-change cycles expected for automotive
applications. Similar requirements may also apply to solid separators Compound and reactants Gr (kJ/mol)
for nonaqueous systems.
Li2 CO3 (Li, CO2 , O2 ) −809
Supplying oxygen to the cells requires new membranes or on- Li2 O2 (Li, O2 ) −571
board storage.— As a required reactant for discharge, oxygen must Li2 O (Li, O2 ) −562
be drawn in from the environment or stored on board the vehicle. The LiOH (Li, O2 , H2 O) −320
most attractive scenario is an open system in which the cell “breathes” LiOH·H2 O (Li, O2 , H2 O) −329
oxygen from the ambient air. In this case, specific energies higher than Li3 N (Li, N2 ) −129
1000 Wh/kg could be achieved for a Li/air cell, as shown in Figure 3.
Conversely, storing oxygen inside a closed system (directly in the
cells or in a tank) leads to practical specific energies between 500 and NH3 ) may lead to a hazardous overpressure and flammable gases may
1000 Wh/kg. While a truly “air-based” system would afford higher react with oxygen in a chain reaction.
specific energy, the system most widely investigated in the literature so From a comparison of the Gibbs free energies of formation of the
far is the lithium-oxygen cell. In this section we discuss the challenges relevant compounds (see Table IV) one can see that lithium carbonate
related to ambient air operation of Li/air batteries and present some is more stable than the lithium oxides while lithium nitride has a
concepts from recent literature, including membrane solutions as well relatively small formation enthalpy. This implies that nitrogen entering
as closed systems using pure-oxygen tanks. Note that for an aqueous an open Li/air cell may be less critical than CO2 , although the reaction
Li/air cell it may be desirable for H2 O to enter the cell from ambient kinetics must also be considered.
air. Ideally the cell would include a gas selective membrane that allows
Open systems are susceptible to contaminants and electrolyte evapo- only oxygen to enter and exit the positive electrode. Optimization of
ration, which result in severe performance degradation.—The design the system implies a balance between effective separation of enter-
of an open lithium/air system presents many challenges associated ing gases and low weight and cost achieved by minimizing system
with preventing contaminants from entering the cell and keeping the complexity.
electrolyte in the cell. Any molecules that can conceivably be found A second critical issue in open Li/air systems is electrolyte evap-
in the atmosphere may enter the system and affect the cell chemistry, oration. Most recent studies of Li/air batteries use propylene car-
through catalyst poisoning, corrosion, or side reactions. Several un- bonate as the electrolyte solvent because of its high boiling point
desired side reactions may occur if carbon dioxide, nitrogen, or water (BP = 240◦ C) and good polarity, although it is chemically unstable,
enter the cell. The most important side reactions include: while some apparently more stable electrolytes26, 48 have lower boiling
points (e.g., acetonitrile, BP = 82◦ C). In order to use the electrolyte
CO2 : with the best properties it is preferable to avoid the evaporation issue
4Li + O2 + 2CO2 → 2Li2 CO3 [9] entirely. A gas selective membrane may serve this function as well.
Besides the fundamental aspect of gas separation and elec-
trolyte enclosure a realistic solution must be able to handle the
Li2 O + CO2 → Li2 CO3 [10] expected gas flows at realistic current densities. A 90-kWh elec-
tric vehicle Li/air system discharged at 1 C has an oxygen con-
sumption of 3.85 L/s at 25◦ C and 1 atm. Such practical constraints
Li2 O2 + CO2 → 1/2 O2 + Li2 CO3 [11] should be considered together with the gas selectivity of potential
membranes.
Membranes may be an effective means of keeping contaminants out
LiOH + CO2 → Li2 CO3 + H2 O [12] of the cell and keeping the electrolyte in the cell.—In an open system,
contaminants must be separated from the air that enters the positive
H2 O: electrode. We preclude membraneless gas separation processes such
as adsorption, absorption, and distillation, the components of which
2Li + 2H2 O → 2LiOH + H2 [13] would likely be too complex, costly, large, and inefficient for automo-
tive applications. Most researchers are instead exploring membranes
implemented at the cell level that prevent ingress of contaminants
LiOH + H2 O → LiOH • H2 O [14] present in the air. A further advantage of membranes over external
N2 : air filters is that they often prevent evaporation of electrolyte from
the cell. In primary aqueous Li/air cells, desiccants inside the cell
6Li + N2 → 2Li3 N [15] have been used to absorb water vapor from the atmosphere and retain
water,9 but this concept does not appear suitable for rechargeable sys-
Li3 N + 3H2 O → 3 LiOH + NH3 [16] tems. If membranes are used to purify the air their weight and volume
will need to be included in energy calculations.
If the Li metal anode is protected from the electrolyte in the pos- Membranes with O2 -selective silicone oils.—Zhang et al. demon-
itive electrode (e.g., with a solid-electrolyte separator), some of the strated an O2 -selective membrane for operation in ambient air of
reactions involving contaminants (9, 13, 15, and 16) can be prevented. 20–30% relative humidity, as shown in Figure 26.147 The cell de-
In this case, introduction of CO2 can still result in the formation of sign contained a nonaqueous electrolyte and the membrane allowed
electronically insulating bulk lithium carbonate and, in nonaqueous permeation of O2 while blocking moisture. Such membranes can be
electrolytes, H2 O may result in solvent degradation or the formation of prepared by loading O2 -selective silicone oils into porous supports
corrosive acids.144–146 Li2 CO3 formation may be partially irreversible, such as porous metal sheets and Teflon (PTFE) films. The O2 gas was
resulting in capacity fade. continuously supplied through a membrane barrier layer at the inter-
While the reactions with CO2 and N2 only degrade the cell perfor- face of the cathode and ambient air. The authors reported that silicone
mance, introduction of water is more critical if the negative electrode oil of high viscosity showed reasonable performance. The immobi-
is not protected. In this case, additional gas evolution (e.g., H2 and lized silicone oil membrane (see Figure 26) in the porous PTFE film
thickness of the polymer membrane was 25 μm, representing ∼1%

of the cell volume (see Figure 27). Current densities between 0.05
and 0.1 mA/cm2 were demonstrated, but at higher current densities
oxygen permeation through the membrane was insufficient.
A promising step toward more suitable membranes has been very
recently reported by MaxPower Inc.150 The authors describe the de-
velopment of oxygen-selective membranes based on polysiloxane and
methacrylate–polysiloxane copolymers which allow current densi-
ties of up to 2 mA/cm2 and protect the cell against humidity and
moisture. Thinner membranes with high O2 selectivity must be de-
veloped in order to meet the high power requirements of practical
applications.
Anion exchange membranes for aqueous systems.—For aqueous Li/air
cells, EDF has developed an anion exchange membrane (AEM) be-
tween the ORR electrode and discharge product reservoir that allows
transport of OH− , but mitigates influx of O2 or contaminants such as
CO2 .129 Their AEM is composed of interpenetrating polymer networks
(IPN): hydroxyl-conducting polycationic crosslinked polyepichlorhy-
drine (PECH) provides ionic conduction and poly(hydroxyl ethyl
Figure 26. Schematic of two kinds of O2 membrane structures. (a) Liquid methacrylate) provides mechanical strength and reduces swelling.
immobilized in homogeneous porous substrate, and (b) liquid immobilized in Current densities up to 6 mA/cm2 (∼C/12) were obtained at 60◦ C.
silicalite membrane coated porous metal substrate (asymmetrical). Reproduced However, some air can permeate the polymer, resulting in slow
from reference 147. degradation. Improvement of the membrane’s ability to screen CO2
could enable the use of more alkaline electrolytes, which are par-
ticularly susceptible to Li2 CO3 formation but have a lower OER
enabled Li/air batteries with carbon black air electrodes to operate overpotential.29
in ambient air (at 20% RH) for 16.3 days with a specific capacity Use of tanks and compressors results in a closed system, protected
of 789 mAh/g carbon and a specific energy of 2182 Wh/kg carbon. from contaminants.—As an alternative to membrane-enabled open
Its performance was much better than a reference battery assembled systems, Li/air batteries can be designed as fully closed or partially
with a commercial, porous PTFE diffusion membrane as the mois- closed systems that use oxygen from tanks rather than from the at-
ture barrier layer on the cathode, which only had a discharge time of mosphere. This idea was already discussed in the introduction in the
5.5 days, corresponding to a specific capacity of 267 mAh/g carbon context of energy calculations, and it was concluded that the added
and a specific energy of 704 Wh/kg carbon. mass and volume of a tank significantly dampen the appeal of a Li/air
The silicone based membrane prepared by Zhang et al. obtained system. The major advantage of using an oxygen tank is that all of the
an oxygen permeance of 1.6 · 10−6 mol m−2 s−1 Pa−1 at an O2 /H2 O complexities associated with air purification can be avoided. While
selectivity between 1.5 and 3.6. In a different work Reynolds et al. the ultimate commercial viability of a Li/oxygen, rather than a Li/air,
used a polyperfluorocarbon liquid in a porous Celgard 2500 polymer system is perhaps limited, the use of oxygen tanks will be critical dur-
substrate but they obtained a four orders of magnitude lower oxygen ing research and development phases to ensure well-defined chemistry
permeance of 1.7 · 10−10 mol m−2 s−1 Pa−1 at an O2 /H2 O selectivity free of possible contaminants from air.
of 3.9.148
Outlook: Until effective membranes are developed, a bulky closed sys-
Polymer protection membranes.—In another study, Zhang et al. used a tem may be required.—Thus far no membrane has been completely
heat-sealable polymer membrane (Melinex 301H) as both an oxygen- effective in preventing the introduction of contaminants into the cell
diffusion membrane and a moisture barrier in the operation of non- when ambient air is used as an oxygen source. O2 -selective mem-
aqueous Li/air cells under ambient conditions with an oxygen partial branes are preferred for nonaqueous cells, whereas anion-exchange
pressure of 0.21 atm and relative humidity of ∼20%.149 A schematic membranes with good OH− selectivity are an attractive option for
is shown in Figure 27. The membrane also minimized the evaporation aqueous cells. In addition to achieving good selectivity and prevent-
of the electrolyte solvent. These batteries operated in ambient condi- ing solvent evaporation, candidate membranes must enable high rates
tions for more than one month with a specific energy of 362 Wh/kg, of O2 or OH− transport.
based on the total weight of the battery including packaging. The
Figure 27. Schematic of a sealed test cell used by Zhang et al. for ambient operation of Li/air. Reproduced from reference 149.
Onboard storage of O2 provides a more reliable but bulkier solution The reaction mechanisms for aqueous Li/air batteries are different
to the contaminant problem. In this case, the O2 tank and compressors from those of the nonaqueous Li/air batteries and can be summarized
add weight and complexity to the system, and the power required to as follows:
run the compressors impacts the overall system efficiency. However,
Positive electrode : O2 + 2H2 O + 4e− ↔ 4OH− [17]
at least theoretically the efficiency loss may be small, as shown in
Table III, and in open systems efficiency can also be diminished by
resistive separation membranes. The added weight and volume are
Negative electrode : Li ↔ Li+ + e− [18]
a significant barrier to the development of closed systems, as the
achievable specific energy falls close to the threshold that can be The overall cell reaction is shown in equation 3.
obtained by a Li/LiMO2 cell (∼500 Wh/kg), while the energy density During the discharge process, O2 is electrocatalytically reduced to
could be much worse (see Figure 3). produce hydroxyl ions (OH− ) at the positive electrode (Equation 17),
while Li+ ions are generated (Equation 18) at the negative electrode.
LiOH is soluble in water, but precipitates to form LiOH·H2 O above
Aqueous Li/Air Systems the solubility limit (5.25 M at 25◦ C).7 During the charge process, O2
One of the major advantages of the aqueous Li/air system is that the gas is generated at the positive electrode and Li is plated at the negative
discharge reaction product is soluble in water, eliminating an appar- electrode. Catalytic materials are needed to accelerate the kinetics of
ent shortcoming of the nonaqueous Li/air batteries, namely formation both ORR and OER. The electrocatalytic reduction of O2 in aqueous
of electrically resistive products that may passivate the air electrode. solution often requires the use of expensive catalysts such as Pt,152, 153
Moreover, selection of an electrolyte system that achieves reversible particularly in acidic electrolytes.29
chemistry does not appear to be a bottleneck for the development of Like the nonaqueous Li/air cell, the rechargeable aqueous system is
aqueous Li/air cells. However, effective storage of the discharge prod- also at a development stage, and new concepts continue to be explored.
uct, stability of the separator against both the Li anode and the aqueous Here we discuss two challenges specific to the aqueous Li/air system:
air electrode, and catalysis of both the ORR and OER reactions remain controlling the location and morphology of the precipitated discharge
considerable challenges for rechargeable aqueous Li/air cells. product, and durably catalyzing both the ORR and OER.
The operating principle of the aqueous Li/air cell is shown
schematically in Figure 28. Under the proposed scheme, the metallic Precipitation and dissolution of the discharge product must be
lithium anode is protected by a Li-ion conducting ceramic film, which managed to enable high energy and acceptable power.— The dis-
prevents the vigorous reaction of metallic lithium with water. Another charge process in the aqueous Li/air cell differs from that of the non-
important characteristic of these cells is the need for catalysts in the aqueous cell in that the discharge product, LiOH, has a relatively high
positive electrode that reduce the activation barriers for both ORR solubility. Rather than forming a film directly on the surface of the
and OER. positive electrode, the dissolved product can reach a concentration of
The idea of using a protected lithium electrode (PLE) was intro- approximately 5.25 M (at room temperature) before it precipitates as
duced in 2004 by the PolyPlus Battery Company,10, 82, 151 who demon- LiOH · H2 O.154 A phase diagram is given in Figure 29. When cycled
strated long-term stability, high-discharge capacity, and some cyclabil-
ity of lithium metal in aqueous solutions when the anode was protected
by a water-stable glass-ceramic electrolyte (LTAP) from Ohara.115, 121
Imanishi et al. have also looked at the concept of using protected
Li anodes to fabricate rechargeable Li/air batteries using an aque-
ous electrolyte.152, 153 More recently, Zhou and co-workers introduced
the concept of a Li/air system including an organic/aqueous hybrid
electrolyte. In this case, the catalytic reduction of O2 occurs in an
aqueous electrolyte, while the metallic lithium remains in a nonaque-
ous (organic) electrolyte; the two electrolyte systems are separated
by a LiSICON separator.1, 27 Several other groups have adopted ap-
proaches similar to PolyPlus’ PLE, many of which rely on NASICON-
type lithium conductors.129, 130, 140
Figure 29. Phase diagram for the LiOH ± H2 O system, reproduced from
reference 154. The solid lines are fits to the data, which comes from various
sources given in reference 154. The eutectic is at a temperature of 255.15 K
and a composition of 5.27 mol/kg, while the peritectic is at a temperature of
382.05 K and a composition of 7.70 mol/kg. The phase diagram shows that at
temperatures below the peritectic the monohydrate LiOH · H2 O is the favored
Figure 28. Schematic of an aqueous Li/air cell with electrically isolated cath- form of LiOH, while above that temperature anhydrous LiOH is favored. The
ode and anode in the positive electrode. Reproduced from reference 27. use of supporting salts (such as LiCl or NaOH) could change the phase diagram.
to the solubility limit, the theoretical specific energy of the system is While this decreases the mass of the cell (in the charged state) and
430 Wh/kg; hence significant formation of LiOH · H2 O is key to at- prevents evaporation of catholyte, the use of such salts may not be
taining an attractive specific energy. Because precipitation and dis- appropriate for rechargeable cells.
solution of LiOH · H2 O does not involve electron transfer, the solid Toyota has demonstrated reversibility of aqueous Li/air system in-
product need not be in contact with the positive electrode. This poses a cluding precipitation/dissolution of LiOH.130 A symmetric aqueous
challenge, as impedance rise and restriction of accessible porosity can cell with Pt electrodes and Ohara separator was used to show the re-
result from an uneven product distribution, but also provides many versible exchange of 200 mAh capacity in 5M saturated solution. The
opportunities for novel cell and system designs as well as battery 200 mAh of charge passed corresponded to precipitation/dissolution
management system algorithms155 that control the discharge process. of solid LiOH in/from one of the two electrodes. Raman spectra of the
In order to achieve high cell specific energy, it is desirable to fill solids confirmed LiOH as the product, and formation of O2 bubbles
a significant fraction of the available reservoir or positive-electrode was observed during dissolution of the LiOH. In a second experiment,
pore volume with the solid discharge product. Figure 3 shows that two nonsaturated solutions were used, and the electrodes were cycled
when 65% of the volume of a 200 μm positive electrode is filled between 4 and 5 M without precipitation. In a second cell type, the
with LiOH · H2 O a specific energy of about 700 Wh/kg results at the air electrode consisted of LaSrCoO3 , carbon, and PTFE on carbon
cell level for a practical design. If only smaller volume fractions are paper, and the anode consisted of Li metal in a LiTFSA/PC elec-
achieved the cell-level specific energy will be even lower. The ob- trolyte, with an Ohara separator between the electrodes. In this case,
jective of maximizing the discharge capacity of the cell by going to 34 mAh were passed, corresponding to 30 mg of LiOH (composition
a high volume fraction of LiOH · H2 O must be balanced against the confirmed by XRD). During charge (from a second cell constructed in
undesired impedance rise resulting from restriction of the pores. Ide- the discharged state, with solid LiOH in the cathode), the solid LiOH
ally, the pores should fill uniformly with discharge product in order to was consumed.
maintain a homogeneous current distribution through the cell separa- EDF has proposed solutions to several critical LiOH storage issues
tor. Preferential precipitation of the product in certain locations due that enable up to 100 reversible charge/discharge cycles (100 cycles
to gravitational fields, concentration and thermal gradients, current at 0.1 mAh/cm2 , 40 cycles at 2 mAh/cm2 ).129 Improvements include
heterogeneity, or electrolyte flow would tend to amplify nonunifor- application of an anionic polymer layer on the air electrode side of their
mity of the current distribution and result in more rapid impedance LTAP separator, which prevents nucleation and preferential deposition
rise, and at worst could reduce the accessibility of large regions of of a resistive LiOH layer at the LTAP/reservoir interface. Without this
the reservoir or electrode. Current nonuniformity would also result layer an exponential rise in overpotential with time results. Humidified
in uneven Li plating/stripping during charge/discharge. Coverage of air is supplied to the air electrode, where oxygen is reduced and
the anode protection layer or positive electrode with a dense layer of reacted with water to form hydroxyl ions (Equation 17) which in turn
monohydrate can likewise result in impedance rise and a premature are conducted through the IPN to the reservoir. The IPN mitigates
end of discharge. influx of CO2 , which reacts with LiOH to form Li2 CO3 , resulting in
As with nonaqueous systems, aqueous Li/air cells involve sub- cell degradation. An oxygen evolution electrode is embedded in the
stantial volume change, and the precipitation of large amounts of reservoir to enable recharge of the cell (see Figure 31).
solid LiOH · H2 O within the pores of a reservoir or electrode could By using a transparent cell housing EDF observed the formation of
impart deleterious mechanical stress on the system. So far high cell a solid discharge product at the bottom of the reservoir (see Figure 31).
reversibility involving large volumes of LiOH · H2 O (more than a few Thus far, only a small amount of solid LiOH · H2 O, relative to the
percent of the porosity) has not been demonstrated. total pore volume of the reservoir, has been cycled. In cells that were
Some existing prototype cells demonstrate high discharge capacity initially charged to 73 mAh/cm2 (for an available cell specific energy
but have limited reversible capacity.—While much of the recent Li/air of 500 Wh/kg), 40 cycles were achieved at a depth of discharge (DOD)
literature has focused on nonaqueous systems, several companies have
published the results of their aqueous Li/air research and development.
Although each company describes a unique cell design, a common
feature is the concept of a discharge product reservoir that is distinct
from the positive electrode. PolyPlus has demonstrated the discharge
of primary aqueous Li/air cells in which discharge product is stored
in a porous reservoir of the cathode (see Figure 30).84 One of their
patents describes a porous zirconia felt reservoir that expands with the
discharge product as it precipitates and is filled with a catholyte that
contains hygroscopic salts used to draw in water from the atmosphere.9
Figure 30. Cross section of PolyPlus’ aqueous Li/air cell. Reproduced from Figure 31. Schematic of EDF’s rechargeable aqueous Li/air cell. Reproduced
reference 84. from reference 129.
of 2 mAh/cm2 (13 Wh/kg) and 100 cycles were achieved at a DOD of product reservoir) and durability (polymer protection of the reservoir
0.1 mAh/cm2 .129 to separate it from the anode protection layer and the air electrode) of
While the current EDF cell has a vertical orientation, such that aqueous Li/air cells.
LiOH · H2 O settles to the bottom of the cell and evolved gases exit the Success of companies like Sion and PolyPlus in protecting Li metal
top of the cell, this has the disadvantage that a non-uniform current could be leveraged to increase cyclability of the negative electrode,
distribution may result, with more current flowing in the top region which appears to limit cell life. Other cycle-life limitations involving
of the cell that does not contain LiOH · H2 O. This would lead to a storage of LiOH · H2 O may become apparent as the anode reversibility
non-uniform thickness of Li metal as a function of height in the cell. is improved.
An alternative orientation would be to place the cell in a horizontal The solutions proposed here should be attempted in rechargeable
configuration such that the LiOH · H2 O precipitate would settle either Li/air cells in which a much higher fraction of the theoretical capacity
on the AEM or the LTAP. This should result in a more uniform current is cycled. Volume changes associated with high-capacity charge and
distribution but may make it more difficult for oxygen produced at the discharge may require significantly different cell designs from those
OER electrode to exit the cell. The incorporation of the OER electrode proposed today.
in the cell sandwich may also be more difficult.
High reversible capacities must be demonstrated.—While Toyota and Charge and discharge need to be catalyzed to reduce kinetic
EDF have shown good reversibility of the Li/air system involving a losses.— In the aqueous Li/air system, where OH− is the cathodic
small amount of solid discharge product, PolyPlus has demonstrated reaction product, the O-O bond must be broken during oxygen reduc-
large discharge capacities in a primary cell. Further research on the tion and reformed during oxygen evolution. These bond-breaking and
reversibility of LiOH · H2 O reservoirs may pave the way to a cy- bond-forming reactions may result in a considerable overpotential in-
clable high-energy Li/air cell. There is as yet limited understanding crease relative to the ORR and OER in nonaqueous systems, in which
of where the solid product precipitates and how this affects current Li2 O2 may be formed and one of the O-O bonds possibly remains
distribution in the cell. The mechanical stability of candidate porous intact.
reservoirs, such as expandable zirconia felt, over many cycles should In order to lower the overpotential of the ORR, there are few al-
be investigated. The influence of current density, reservoir geometry ternatives to Pt with comparable performance, particularly in acidic
(porosity and thickness), and cell orientation with respect to gravita- environments.29 In alkaline electrolytes, most materials are unstable
tional field on the product distribution requires a more comprehensive at the anodic potential required for the OER, and even Pt dissolves.157
analysis. Hence, to avoid expensive catalysts, it may be necessary to use an
Discharge product morphology impacts recharge rate capability.— alkaline electrolyte and electrically isolate the ORR and OER elec-
During charge the solid particles of LiOH · H2 O must dissolve and the trodes.
LiOH must diffuse to the OER electrode. The rate of particle disso- Alternatively, a bifunctional electrode consisting of two different
lution will depend on their size and shape, as well as the porosity of catalysts for ORR and OER may be feasible, provided the active
the precipitate layer that forms. Providing forced convection in some components for the reduction and evolution of the O2 molecule are
way could significantly improve the dissolution rate. One important stable over a wide range of potentials, from 0.6–0.7 V (RHE) dur-
question is whether, if the LiOH · H2 O particles that form are very ing discharge (ORR), to over 2.1 V (RHE) during charge (OER).158
small, they will undergo Ostwald ripening during a rest step. If this Such bifunctional electrodes are already employed in metal-air
were to occur the larger particles would take longer to dissolve. There systems.138, 159, 160
could also be changes to the porosity of the layer that would impact Work on PEM and other fuel cells has significantly reduced the Pt
the dissolution rate. We consider the issue of the physical form of loading required for ORR.—Investigation of ORR mechanisms in fuel
LiOH · H2 O deposits to be a critical area for research for aqueous cell cathodes is much more advanced than what has been recently pro-
Li/air cells. posed for metal-air batteries. ORR has been extensively studied in the
Novel system concepts may involve external LiOH · H2 O reservoirs.— field of fuel cells because of its fundamental complexity, sensitivity
While reservoirs internal to the cell sandwich have been the focus to the electrode surface properties, and slow reaction kinetics. It has
of recent aqueous Li/air cell developments, earlier work on metal-air also been demonstrated that the ORR is a significant component of the
systems included electrolyte circulation,29, 87 which may allow for ex- cell overpotential and limits the efficiency of electrochemical energy
ternal reservoirs. Storing the product outside the path of ionic current devices that use air as the oxidant.161 Despite the complexity of this re-
flow would ensure that the pores of the separator and positive elec- action, two mechanisms have been proposed for oxygen reduction: the
trode remain open for ionic transport. Sufficiently high flow rates, so-called direct 4-electron pathway, in which peroxide is not formed,
relative to the applied current density, should minimize LiOH concen- except as a possible adsorbate; and the peroxide pathway, involving
tration gradients orthogonal to the applied electric field and maintain a 2-electron reduction of O2 to form peroxide, which subsequently
a uniform current density. reacts electrochemically to form H2 O or chemically to form H2 O
External storage of the product could be enabled by a flow-through and O2 .162 The 2-electron pathway is more common in alkaline solu-
cell connected to a storage reservoir, in which the aqueous LiOH tions and on mercury, gold, carbon, and most transition metal oxides,
solution is continually circulated between the two. Several general whereas the 4-electron pathway proceeds via dissociative adsorption
concepts could be explored for separating the monohydrate product on noble metallic catalysts and some transition metal catalysts and
from the circulating stream, including gravitational (e.g., the use of a macrocycles.162, 163 To obtain the maximum redox efficiency and to
settling tank as proposed for Al/air cells156 ), mechanical, evaporative, avoid the corrosion of the electrode material by the peroxide interme-
and thermal separation, as well as filtration. The energy requirements diate of the 2-electron reaction mechanism, it is preferable to achieve
for the separation process and for maintaining flow should be included a 4-electron ORR. Therefore, finding suitable inexpensive catalysts
in overall energy efficiency calculations. that promote direct 4-electron reduction of the O2 molecule is still an
Controls strategies involving thermal management and current pro- active area of research.
files could be explored as a means to improve the uniformity of product Bifunctional electrodes are used in Zn/air systems.—Metal-air sys-
distribution in the reservoir. tems involve a significant leap in complexity compared to fuel-cell
Outlook: LiOH · H2 O formation appears to be reversible; high- systems because OER must be catalyzed in addition to ORR. Since
capacity and long-term cycling need to be demonstrated.—There has the ORR reaction mechanisms have been discussed in the previous
recently been rapid progress in demonstrating reversibility and iden- section, here we describe the reaction pathways involved in oxygen
tifying challenges to long-term cyclability of aqueous Li/air systems. evolution. According to Jörissen,162 the OER mechanisms are rather
Several companies have proposed promising engineering approaches complex and susceptible to change depending on the electrode poten-
to improve the energy density (adding desiccants to the discharge tial. On metallic surfaces, the rate-determining step for the OER is the
electron transfer from H2 O in acidic media or OH− in basic media

to form adsorbed OH radicals at the electrode surface. OER has also
been investigated on metallic-oxide surfaces;164 in these systems the
intermediate formation of H2 O2 from metal-desorbed OH species has
been suggested to be the rate-determining step. Upon formation of the
physisorbed H2 O2 species, O2 evolves in the final step of the proposed
OER mechanism.164
Without a third electrode in the cell, the reduction and evolution
of oxygen in rechargeable (secondary) metal-air batteries requires
bifunctional air electrodes, which contain separate ORR and OER
catalysts connected electronically. Dating back several decades,159, 160
bifunctional electrodes have emerged as a practical solution to the
lack of highly active and stable bifunctional catalysts for both ORR
and OER. For example, it is well-known that the most effective ORR
catalysts are those based on platinum (Pt),165 but Pt has only moderate
activity for the OER. On the other hand, ruthenium (Ru) and iridium
(Ir) oxides are among the best OER catalysts,166 but they are not
as active as Pt for ORR. Although, alloys of these compounds have Figure 32. Activity versus the experimentally measured d-band center relative
shown a better bifunctional-catalytic performance,167 the development to platinum. The theoretical results are shown in black while the experimental
of bifunctional catalysts for practical applications still represents a results are shown in red. Reproduced from reference 173.
challenge as the best catalytic materials still consist of precious metals,
which are in turn scarce and expensive.
Since the composition and design principles used in bifunctional However, the use of two oxygen electrodes adds mass to the cell
electrocatalysts for metal-air batteries have been recently reviewed by and may significantly lower the system specific energy. It also requires
Neburchilov et al. and Jörissen et al.162, 168, 169 here we make only a increased complexity and cost of the electronics used to control the
few remarks relevant to the Zn/air cell. system.
While several metal-air chemistries have been proposed, Zn/air Computational screening could enable the discovery of suitable ORR,
batteries are the most mature technology and have contributed sig- OER, and bifunctional catalysts.—First-principles modeling of elec-
nificantly toward the development of metal-air batteries.158, 169 The trocatalytic reactions is fast becoming an indispensable tool for achiev-
overall discharge reaction can be summarized as follows:170 ing insight into the reaction mechanisms of complex electrochemical
environments.171, 172 Significant efforts in this area have led to de-
Positive electrode : O2 + 2H2 O + 4e− → 4OH− tailed interpretations of the experimental data and to the computa-
tional design of new catalytically active materials for the ORR in
Negative electrode : Zn → Zn2+ + 2e− [19] fuel cells.173, 174 In this work, many authors have extensively used
the concept of “volcano” plots to predict the activity and selectivity
Zn2+ + 4OH− → Zn(OH)2− [20] of catalysts for ORR. Volcano plots (shown in Figure 32) illustrate
4
very well Sabatier’s principle for catalytic materials, which states that
−
an effective catalytic surface should reach a compromise between
4 → ZnO + H2 O + 2OH
Zn(OH)2− [21] having enough strength to break the bonds of the molecular adsor-
bates and yet generate weakly-bonded intermediate species that are
Net reaction : 2Zn + O2 ↔ 2ZnO [22] not overly stabilized by the surface.175 As an example, Figure 32
(reproduced from 173 and 174) presents the volcano plot for the elec-
In contrast to the Li/air cell, hydroxyl ions generated at the air trocatalytic reduction of O2 on various Pt-complexes. The peaks in
electrode migrate to the Zn anode compartment to complete the cell Figure 32 represent those catalytic surfaces with the largest electro-
reaction, with a cell equilibrium potential of 1.65 V.170 chemical activity for ORR in fuel cells.
In spite of the challenges involved in catalysis for secondary Zn/air The values used to construct such plots are now readily attainable
batteries, some success (150 cycles) with cells that contain bifunc- from first-principles modeling techniques;174 it is therefore conceiv-
tional air electrodes has been reported by the ReVolt Company.138 In able that such computational-based approaches could be used to ex-
their work, a bifunctional electrode is constructed with a catalyst that plore the catalytic activity of various metallic surfaces for both the
promotes ORR (MnSO4 , Fe, Pt, and Pd among other materials) and ORR and OER. In turn, this possibility provides some hope for a
a bifunctional catalyst that shows high reaction rates for both ORR more versatile design strategy for bifunctional electrodes or bifunc-
and OER (for example, La2 O3 , Ag2 O, Ag, spinels and perovskites). tional catalysts for the aqueous Li/air system.
According to Neburchilov et al., although several companies have Outlook: Bifunctional electrodes or two-positive-electrode architec-
patented bifunctional electrodes (ReVolt among others), all of them tures are required.—Unless bifunctional catalysts are discovered
have insufficient ORR catalytic lifetimes at the high OER potential that promote facile kinetics for both the ORR and OER in aque-
(∼2.1 eV).169 It is therefore reasonable to expect some delay in the ous air electrodes and are stable over the potential window 0.6 to
commercialization of the Zn/air batteries while inexpensive, abundant, 2.1 V vs. RHE, the only viable approaches to making aqueous Li/air
active, and stable combinations of catalytic materials are sought. systems rechargeable are bifunctional-electrode and two-positive-
3-electrode cells remove restrictions on ORR and OER catalysts, but electrode architectures. The former design has already been used with
complicate the system design.—Perhaps the most promising solution some success in rechargeable Zn/air cells, although it remains chal-
to the catalysis of ORR and OER in aqueous Li/air cells is the use lenging to find a combination of catalysts with fast ORR kinetics, sta-
of two separate and electrically isolated electrodes for charge and bility at the OER potential, and low cost. Two-electrode designs, which
discharge. This sidesteps the catalyst stability issue, as the catalysts avoid the kinetics-stability compromise, have been implemented in
appropriate for ORR are never polarized to the high anodic potentials Li/air cells at the expense of higher complexity and weight.
required for OER. Several groups have designed aqueous Li/air cells
with this configuration.27, 129 An additional advantage of this configu-
Summary and Outlook
ration in the EDF cell is that the OER electrode can be located between
the IPN and product reservoir, thereby avoiding the overpotential as- Lithium/air batteries are an attractive technology, particularly for
sociated with OH− transport through the IPN during charge.129 long-range electric vehicles, because of their high theoretical specific
Table V. Requirements for durable, high-energy automotive Li/air batteries (N = nonaqueous only, A = aqueous only).
Requirements for durable, high-energy automotive Li/air batteries
Li anode
Robust and flexible containment, including flexible seals
Li protection layer(s)
transport properties
Sufficient conductivity over T range of interest
Negligible electronic conductivity
manufacturability
Sufficiently thin (<50 μm preferable)
Pinhole free
Low cost
mechanical properties
High elastic modulus (to prevent dendrite initiation)
Flexibility required for wound concepts
stability
Stable against O2 , contaminants (CO2 , N2 , trace H2 O)
Stable against Li
Water stable (A)
Air electrode
Continuous gas and electrolyte networks (N)
High surface area (N)
High pore volume (N)
Bifunctional or 2-electrode design
Membrane
Highly selective for OH- (A)
High OH- transport rates (A)
Good CO2 screening
Good H2 O screening (N)
Product reservoir
Promotes uniform LiOH·H2 O precipitation (A)
Flexible (A)
High pore volume (A)
Maintains electrolyte transport pathways (A)
Catalysts
High activity/mass ratio
High activity/cost ratio
Abundant materials
ORR catalyst
Breaks O-O bond (A, and for Li2 O)
Robust to poisoning (A)
OER catalyst
Promotes O-O bond formation (A, and for Li2 O)
Electrolyte
Adequate Li+ conductivity at all temperatures
Good stability at high temperatures
Low viscosity
Chemically inert in presence of O2 over operating voltage and temperature (N)
System
Closed system with tanks and compressors or open system with highly selective gas or anion-exchange membranes
In flow configuration, highly efficient separation of LiOH · H2 O from electrolyte (A)
energy. While researchers have been aware of and worked sparingly propriate noncarbonate solvents is underway, but so far adequate re-
on Li/air batteries for decades, no one has yet demonstrated a Li/air versibility remains elusive.
cell that is reversible and can be cycled over a significant fraction Once a reversible chemistry is established, the next major chal-
of its theoretical capacity. The only short-term commercially viable lenge that should be addressed for nonaqueous systems is the de-
products are primary Li/air cells that are designed for high specific position of electrically resistive products in the air electrode dur-
energy but not rechargeability. ing discharge. Bulk Li2 O2 is electronically insulating and difficult
Recent investigations of both aqueous and nonaqueous Li/air sys- to solvate; hence it may form an electrically resistive film on the
tems have resulted in a significantly improved understanding of the electrode surface, limiting the achievable capacity far below what
main challenges for this technology. Despite several research groups could be obtained by completely filling the pore volume. Appro-
demonstrating limited cyclability of nonaqueous cells using carbonate priate nanostructuring of the air electrode may enable higher ca-
solvents, it is now clear that these solvents participate in the reactions pacities in spite of resistive products, while the use of new sol-
that consume lithium and oxygen during discharge and are not re- vents or additives may enable Li2 O2 precipitation away from active
versibly generated during charge. Hence, the cycle life of carbonate- sites.
based Li/air cells is limited by the quantity of solvent available for Aqueous Li/air chemistry appears to be much more reversible than
reaction and the buildup of side reaction products. A search for ap- nonaqueous Li/air chemistry. LiOH is very soluble (up to 5.25 M at
25◦ C) and precipitates as a monohydrate above its solubility limit mercialization of Li/air EVs. However, as a long-term solution to
when using pure H2 O as a solvent. The monohydrate can be dis- the daunting challenge of low-cost, high-range electromobility, Li/air
solved reversibly. However, the molar mass of the discharge product, batteries remain one of the few and most promising, and funding of
LiOH · H2 O, is much higher than the discharge product in nonaque- research to accelerate their introduction into the marketplace should
ous systems, Li2 O2 , resulting in a comparatively lower theoretical be a top priority for government, academic, and industrial research
specific energy. In fact, practical aqueous Li/air cell designs may not institutions.
achieve higher energy density than Li/metal-oxide systems. More-
over, breaking and forming of the O-O bond on discharge and charge, References
respectively, suggests the need for rate-enhancing ORR and OER cat-
alysts, the stability of which may preclude the use of a bifunctional 1. Y. Wang and H. Zhou, Journal of Power Sources, 195, 358 (2010).
2. K. M. Abraham and Z. Jiang, Journal of the Electrochemical Society, 143, 1 (1996).
electrode. 3. G. Girishkumar, B. McCloskey, A. Luntz, S. Swanson, and W. Wilcke, The Journal
Controlling the distribution of reaction product is critical for of Physical Chemistry Letters, 1, 2193 (2010).
achieving high capacity and rate capability in nonaqueous and aque- 4. T. Ogasawara, A. Debart, M. Holzapfel, P. Novak, and P. Bruce, Journal of the
ous systems. Provided passivation does not restrict accumulation of American Chemical Society, 128, 1390 (2006).
5. J. S. Hummelshøj, J. Blomqvist, S. Datta, T. Vegge, J. Rossmeisl, K. S. Thygesen,
Li2 O2 to the electrode surface in nonaqueous cells, pore blocking A. Luntz, K. W. Jacobsen, and J. K. Nørskov, The Journal of Chemical Physics,
may restrict the accessibility of smaller pores. Poor oxygen transport 132, 071101 (2010).
could result in higher current density near the gas diffusion layer and 6. J. Hassoun, F. Croce, M. Armand, and B. Scrosati, Angewandte Chemie International
subsequent accumulation of Li2 O2 to block the ingress of oxygen. Edition, 50, 2999 (2011).
7. P. Gierszewski, P. Finn, and D. Kirk, Fusion Engineering and Design, 13, 59 (1990).
Appropriate pore structure and use of both wetting and nonwetting 8. M. Urquidi-Macdonald, D. D. Macdonald, O. Pensado, and J. R. Flores,
pores may be necessary to circumvent these issues. LiOH · H2 O tends Electrochimica Acta, 47, 833 (2001).
to precipitate on the surface of the LTAP separator preferred for aque- 9. S. Visco, L. De Jonghe, Y. Nimon, A. Petrov, and K. Pridatko, Aqueous Lithium/Air
ous Li/air cells. Polymer coatings on the water side of the separator Battery Cells, WO Patent Application WO/2010/005,686 (2010).
10. S. J. Visco and Y. S. Nimon, Active Metal/Aqueous Electrochemical Cells and
can prevent this from occurring. Gravitational effects (i.e., sedimen- Systems, U.S. Patent 7,666,233 (2010).
tation) may require special orientation of nonaqueous cells to control 11. D. R. Lide, CRC Handbook of Chemistry and Physics (1996).
the current density distribution and minimize ohmic and transport 12. J. S. Kim, C. S. Johnson, J. T. Vaughey, M. M. Thackeray, S. A. Hackney, W. Yoon,
losses. and C. P. Grey, Chemistry of Materials, 16, 1996 (2004).
13. W. M. Haynes, ed., CRC Handbook of Chemistry and Physics, 91st edition (2010).
Massive volume changes at the electrode and cell-sandwich level 14. S. J. Visco, E. Nimon, and L. De Jonghe, Lithium-Air, Lithium-Water and Lithium-
may require special cell-design features, including electrolyte or sol- Sulfur Batteries Based on Protected Lithium Electrodes (PLEs), in 15th International
vent reservoirs, flexible seals, application of pressure, and the use Meeting on Lithium Batteries, Montreal (2010).
of components with particular mechanical properties. Recirculating 15. R. Strobel, M. Oszcipok, M. Fasil, B. Rohland, L. Jorissen, and J. Garche, Journal
of Power Sources, 105, 208 (2002).
flow of the catholyte in aqueous cells may alleviate volume change 16. E. Littauer and K. Tsai, Journal of The Electrochemical Society, 124, 850 (1977).
in the positive electrode. Some components, such as flexible seals for 17. E. L. Littauer and K. C. Tsai, Journal of The Electrochemical Society, 123, 964
cells with protected Li metal, have already been developed for pri- (1976).
mary cells, but reversible cycling must also be demonstrated. Wound 18. Y. Lu, J. B. Goodenough, and Y. Kim, Journal of the American Chemical Society,
133, 5756 (2011).
jellyroll designs may be precluded by high volume changes in Li/air 19. A. Galbraith, The Lithium-Water-Air Battery for Automotive Propulsion, in 4th
cells. International Electric Vehicle Symposium, Duesseldorf, West Germany (1976).
Another critical system-level issue involves the open nature of 20. J. Read, Journal of the Electrochemical Society, 149, A1190 (2002).
tankless metal-air cells. Air contains contaminants, particularly H2 O 21. J. Read, K. Mutolo, M. Ervin, W. Behl, J. Wolfenstine, A. Driedger, and D. Foster,
Journal of the Electrochemical Society, 150, A1351 (2003).
and CO2 , that are very reactive against Li and Li2 O2 . CO2 also re- 22. A. Débart, J. Bao, G. Armstrong, and P. Bruce, Journal of Power Sources, 174, 1177
acts with LiOH to from Li2 CO3 . Moreover, evaporation of the solvent (2007).
from the positive electrode compartment can occur in an open sys- 23. P. Albertus, G. Girishkumar, B. McCloskey, R. S. Sanchez-Carrera, B. Kozinsky,
tem. Several membranes have been proposed to avoid contamination J. Christensen, and A. C. Luntz, Journal of The Electrochemical Society, 158, A343
(2011).
of the cell and evaporation, but there are no reports yet on total ef- 24. Y. Lu, Z. Xu, H. Gasteiger, S. Chen, K. Hamad-Schifferli, and Y. Shao-Horn, Journal
fectiveness or the long-term stability of these membranes. In case a of the American Chemical Society, 132, 12170 (2010).
membrane solution is not adequate, a tank and compressor solution 25. C. Laoire, S. Mukerjee, K. Abraham, E. Plichta, and M. Hendrickson, The Journal
seems feasible, at the expense of some specific energy and energy of Physical Chemistry C, 113, 20127 (2009).
26. C. Laoire, S. Mukerjee, K. Abraham, E. Plichta, and M. Hendrickson, The Journal
density. of Physical Chemistry C, 114, 9178 (2010).
Li metal is itself one of the most challenging components of the 27. H. Zhou, Y. Wang, H. Li, and P. He, ChemSusChem, 3, 1009 (2010).
Li/air cell, as it tends to roughen and develop dendrites with cycling. 28. J. S. Lee, S. Tai Kim, R. Cao, N. S. Choi, M. Liu, K. T. Lee, and J. Cho, Advanced
Application of pressure and the use of stiff solid-electrolyte separa- Energy Materials, 1, 34 (2011).
29. K. F. Blurton and A. F. Sammells, Journal of Power Sources, 4, 263 (1979).
tors can diminish this morphology development, but this results in 30. J. M. Tarascon, A. S. Gozdz, C. Schmutz, F. Shokoohi, and P. C. Warren, Solid State
an additional challenge for the Li/air system. The separator must be Ionics, 86, 49 (1996).
chemically stable against both electrodes and provide sufficient ionic 31. T. Kuboki, T. Okuyama, T. Ohsaki, and N. Takami, Journal of Power Sources, 146,
conductance over all operating conditions. It should not significantly 766 (2005).
32. D. Wang, J. Xiao, W. Xu, and J. G. Zhang, Journal of The Electrochemical Society,
reduce the specific energy and energy density of the system (i.e., it 157, A760 (2010).
should preferably be thin), but it should also be mechanically ro- 33. Y. Lu, H. Gasteiger, M. Parent, V. Chiloyan, and Y. Shao-Horn, Electrochemical
bust to thousands of cycles with significant electrode volume change. and Solid-State Letters, 13, A69 (2010).
It seems lilkely that a multi-layer composite will be the ultimate 34. A. Débart, A. J. Paterson, J. Bao, and P. G. Bruce, Angewandte Chemie, 120, 4597
(2008).
solution for Li/air cells, particularly aqueous cells. Several compos- 35. C. Tran, X. Q. Yang, and D. Qu, Journal of Power Sources, 195, 2057 (2010).
ites meet some of these requirements, but manufacturability remains 36. S. Beattie, D. Manolescu, and S. Blair, Journal of the Electrochemical Society, 156,
uncertain. A44 (2009).
Table V summarizes the requirements for the various components 37. K. M. Abraham, Z. Jiang, and B. Carroll, Chemistry of Materials, 9, 1978 (1997).
38. B. McCloskey, D. Bethune, R. Shelby, G. Girishkumar, and A. Luntz, The Journal
in Li/air cells. Those requirements that apply only to nonaqueous or of Physical Chemistry Letters, 2, 1161 (2011).
aqueous systems are labeled N and A, respectively. 39. S. A. Freunberger, Y. Chen, Z. Peng, J. M. Griffin, L. J. Hardwick, F. Barde,
Our prognosis for the rechargeable Li/air system is that it will P. Novak, and P. G. Bruce, Journal of the American Chemical Society, 133, 8040
primarily remain a research topic for at least the next five years. (2011).
40. F. Mizuno, S. Nakanishi, Y. Kotani, S. Yokoishi, and H. Iba, Electrochemistry, 78,
Even if the challenges discussed in this review are successfully ad- 403 (2010).
dressed in that time period, the stringent requirements of durable and 41. W. Xu, V. V. Viswanathan, D. Wang, S. A. Towne, J. Xiao, Z. Nie, D. Hu, and J. G.
safe operation in an automotive environment will further delay com- Zhang, Journal of Power Sources, 196, 3894 (2011).
42. G. M. Veith, J. Nanda, J. Y. Howe, and N. J. Dudney, Journal of Physical Chemistry 86. D. N. Bennion and E. L. Littauer, Journal of the Electrochemical Society, 123, 1462
C, 115, 14325 (2011). (1976).
43. J. Xiao, J. Hu, D. Wang, D. Hu, W. Xu, G. L. Graff, Z. Nie, J. Liu, and J.-G. Zhang, 87. E. L. Littauer and K. C. Tsai, Journal of the Electrochemical Society, 123, 771
Journal of Power Sources, 196, 5674 (2011). (1976).
44. A. K. Thapa, K. Saimen, and T. Ishihara, Electrochemical and Solid-State Letters, 88. A. N. Dey, Thin Solid Films, 43, 131 (1977).
13, A165 (2010). 89. E. Peled, Journal of the Electrochemical Society, 126, 2047 (1979).
45. R. S. Sanchez-Carrera and B. Kozinsky, First-Principles Theoretical Approach for 90. D. Aurbach, in Advances in Lithium-Ion Batteries, W. A. van Schalkwijk and B.
the Design and Characterization of Rechargeable Li-Air Batteries, in Materials Scrosati, editors, p. 7, Kluwer Academic/Plenem Publishers, New York, (2002).
Research Society, Boston (2010). 91. P. B. Balbuena and Y. Wang, Lithium-ion Batteries: Solid-Electrolyte Interphase,
46. X. Zhang, J. K. Pugh, and P. N. Ross, Journal of The Electrochemical Society, 148, Imperial College Press, London (2004).
E183 (2001). 92. Z. Ogumi and M. Inaba, in Advances in Lithium-Ion Batteries, W. A. van Schalkwijk
47. F. Ossola, G. Pistoia, R. Seeber, and P. Ugo, Electrochimica Acta, 33, 47 (1988). and B. Scrosati, editors, p. 79, Kluwer Academic/Plenum Publishers, New York,
48. Z. Peng, S. A. Freunberger, L. J. Hardwick, Y. Chen, V. Giordani, F. Bardé, P. (2002).
Novák, D. Graham, J. M. Tarascon, and P. G. Bruce, Angewandte Chemie, 123, 93. D. Aurbach, B. Markovsky, A. Rodkin, M. Cojocaru, E. Levi, and H. J. Kim,
6475 (2011). Electrochimica Acta, 47, 1899 (2002).
49. K. Xu, S. P. Ding, and T. R. Jow, Journal of The Electrochemical Society, 146, 4172 94. R. Spotnitz, Journal of Power Sources, 113, 72 (2003).
(1999). 95. C. Monroe and J. Newman, Journal of The Electrochemical Society, 150, A1377
50. K. Kanamura, T. Okagawa, and Z. Takehara, Journal of Power Sources, 57, 119 (2003).
(1995). 96. C. Monroe and J. Newman, Journal of The Electrochemical Society, 151, A880
51. S. A. Freunberger, Y. Chen, N. E. Drewett, L. J. Hardwick, F. Bardé, and P. G. (2004).
Bruce, Angewandte Chemie International Edition, 50, 1 (2011). 97. C. Monroe and J. Newman, Journal of The Electrochemical Society, 152, A396
52. C. O. Laoire, S. Mukerjee, E. J. Plichta, M. A. Hendrickson, and K. M. Abraham, (2005).
Journal of The Electrochemical Society, 158, A302 (2011). 98. J.-i. Yamaki, S.-i. Tobishima, K. Hayashi, S. Keiichi, Y. Nemoto, and M. Arakawa,
53. F. Mizuno and H. Iba, Control of Cathode Reaction on Lithium-Oxygen Recharge- Journal of Power Sources, 74, 219 (1998).
able Battery, in 219th Meeting of the Electrochemical Society, Montreal (2011). 99. T. J. Richardson and G. Chen, Journal of Power Sources, 174, 810 (2007).
54. K. Takechi, E. Sudo, T. Inaba, F. Mizuno, H. Nishikoori, and T. Shiga, Solvent 100. T. J. Richardson, Solid Solution Lithium Alloy Cermet Anodes, U.S. Patent Appli-
screening of the electrolyte for non-aqueous Li-air batteries, in 218th Meeting of cation 2010/0181528 (2010).
the Electrochemical Society, Las Vegas, NV (2010). 101. A. D. Robertson, A. R. West, and A. G. Ritchie, Solid State Ionics, 104, 1 (1997).
55. F. De Giorgio, F. Soavi, and M. Mastragostino, Electrochemistry Communications, 102. K. Takada, Encyclopedia of Electrochemical Power Sources (2009).
13, 1090 (2011). 103. J. W. Fergus, Journal of Power Sources, 195, 4554 (2010).
56. D. Zhang, R. Li, T. Huang, and A. Yu, Journal of Power Sources, 195, 1202 (2010). 104. U. Alpen, A. Rabenau, and G. H. Talat, Applied Physics Letters, 30, 621 (1977).
57. R. G. Pearson, Journal of the American Chemical Society, 85, 3533 (1963). 105. G. Nazri, Solid State Ionics, 34, 97 (1989).
58. V. S. Bryantsev and M. Blanco, The Journal of Physical Chemistry Letters, 2, 379 106. G. C. Farrington, B. S. Dunn, and J. L. Briant, Solid State Ionics, 3, 405 (1981).
(2011). 107. H. Y. P. Hong, Materials Research Bulletin, 13, 117 (1978).
59. D. T. Sawyer and J. S. Valentine, Accounts of Chemical Research, 14, 393 (1981). 108. P. G. Bruce and A. R. West, Materials Research Bulletin, 15, 379 (1980).
60. W. Xu, K. Xu, V. V. Viswanathan, S. A. Towne, J. S. Hardy, J. Xiao, Z. Nie, D. Hu, 109. P. G. Bruce and A. R. West, Journal of Solid State Chemistry, 53, 430 (1984).
D. Wang, and J.-G. Zhang, Journal of Power Sources, 196, 9631 (2011). 110. J. B. Bates, N. J. Dudney, G. R. Gruzalski, R. A. Zuhr, A. Choudhury, C. F. Luck,
61. A. K. Padhi, K. S. Nanjundaswamy, and J. B. Goodenough, Journal of the Electro- and J. D. Robertson, Journal of Power Sources, 43, 103 (1993).
chemical Society, 144, 1188 (1997). 111. J. B. Bates, N. J. Dudney, D. C. Lubben, G. R. Gruzalski, B. S. Kwak, X. Yu, and
62. Y. V. Mikhaylik, I. Kovalev, R. Schock, K. Kumaresan, J. Xu, and J. Affinito, ECS R. A. Zuhr, Journal of Power Sources, 54, 58 (1995).
Transactions, 25, 23 (2010). 112. J. B. Bates, N. J. Dudney, B. Neudecker, A. Ueda, and C. D. Evans, Solid State
63. V. S. Kolosnitsyn and E. V. Karaseva, Russian Journal of Electrochemistry, 44, 506 Ionics, 135, 33 (2000).
(2008). 113. R. Kanno, T. Hata, Y. Kawamoto, and M. Irie, Solid State Ionics, 130, 97 (2000).
64. J. M. Garcia-Lastra, Journal of Chemical Physics, 135, 121101 (2011). 114. F. Mizuno, A. Hayashi, K. Tadanaga, and M. Tatsumisago, Advanced Materials, 17,
65. J. Chen, J. S. Hummelshoj, K. S. Thygesen, J. S. G. Myrdal, J. K. Norskov, and 918 (2005).
T. Vegge, Catalysis Today, 165, 2 (2011). 115. J. Fu, Solid State Ionics, 96, 195 (1997).
66. S. Brunauer, P. H. Emmett, and E. Teller, Journal of the American Chemical Society, 116. M. A. Subramanian, R. Subramanian, and A. Clearfield, Solid State Ionics, 18-19,
60, 309 (1938). 562 (1986).
67. M. Mirzaeian and P. J. Hall, Journal of Power Sources, 195, 6817 (2010). 117. H. Aono, E. Sugimoto, Y. Sadaoka, N. Imanaka, and G.-Y. Adachi, Journal of The
68. X. Yang, P. He, and Y. Xia, Electrochemistry Communications, 11, 1127 (2009). Electrochemical Society, 136, 590 (1989).
69. S. Stewart, Determination of Transport Properties and Optimization of Lithium-ion 118. H. Aono, E. Sugimoto, Y. Sadaoka, N. Imanaka, and G.-Y. Adachi, Bulletin of the
Batteries, University of California, Berkeley, PhD Thesis, 2008. Chemical Society of Japan, 65, 2200 (1992).
70. J. Read, Journal of the Electrochemical Society, 153, A96 (2006). 119. H. Aono, E. Sugimoto, Y. Sadaoka, N. Imanaka, and G.-Y. Adachi, Journal of The
71. S. S. Zhang, K. Xu, and J. Read, Journal of Power Sources, 196, 3906 (2011). Electrochemical Society, 140, 1827 (1993).
72. S. J. Visco and Y. S. Nimon, Li/Air Non-Aqueous Batteries, U.S. Patent Application 120. H. Yamamoto, M. Tabuchi, T. Takeuchi, H. Kageyama, and O. Nakamura, Journal
2007/0117007 (2007). of Power Sources, 68, 397 (1997).
73. J. Xiao, W. Xu, D. Wang, and J. Zhang, Journal of the Electrochemical Society, 157, 121. J. Fu, Lithium Ion Conductive Glass-Ceramics, U.S. Patent 5,702,995 (1997).
A294 (2010). 122. Y. Shimonishi, T. Zhang, P. Johnson, N. Imanishi, A. Hirano, Y. Takeda,
74. A. Z. Weber, R. M. Darling, and J. Newman, Journal of The Electrochemical Society, O. Yamamoto, and N. Sammes, Journal of Power Sources, 195, 6187 (2010).
151, 1715 (2004). 123. S. J. Visco and Y. S. Nimon, Active Metal/Aqueous Electrochemical Cells and
75. S. J. Visco, E. Nimon, and L. C. De Jonghe, in Encyclopedia of Electrochemical Systems, U.S. Patent 7,645,543 (2010).
Power Sources, p. 376, Elsevier, Amsterdam (2009). 124. R. Murugan, V. Thangadurai, and W. Weppner, Ionics, 13, 195 (2007).
76. B. Xie, H. S. Lee, H. Li, X. Q. Yang, J. McBreen, and L. Q. Chen, Electrochemistry 125. Y. Inaguma, C. Liquan, M. Itoh, T. Nakamura, T. Uchida, H. Ikuta, and M. Wakihara,
Communications, 10, 1195 (2008). Solid State Communications, 86, 689 (1993).
77. L. F. Li, H. S. Lee, H. Li, X. Q. Yang, and X. J. Huang, Electrochemistry Commu- 126. H. Kawai and J. Kuwano, Journal of The Electrochemical Society, 141, L78
nications, 11, 2296 (2009). (1994).
78. R. R. Mitchell, B. M. Gallant, C. V. Thompson, and Y. Shao-Horn, Energy & 127. A. D. Robertson, S. G. Martin, A. Coats, and A. R. West, Journal of Materials
Environmental Science, 4, 2952 (2011). Chemistry, 5, 1405 (1995).
79. C. Tran, J. Kafle, X.-Q. Yang, and D. Qu, Carbon, 49, 1266 (2011). 128. O. Bohnke, C. Bohnke, and J. L. Fourquet, Solid State Ionics, 91, 21 (1996).
80. C. K. Chan, H. Peng, G. Liu, K. McIlwrath, X. F. Zhang, R. A. Huggins, and Y. Cui, 129. P. Stevens, G. Toussaint, G. Caillon, P. Viaud, P. Vinatier, C. Cantau, O. Fichet,
Nature Nanotechnology, 3, 31 (2007). C. Sarrazin, and M. Mallouki, ECS Transactions, 28, 1 (2010).
81. N. Balsara, Solid-state Batteries with Lithium Metal Electrodes, in Symposium on 130. K. Suto, S. Nakanishi, H. Iba, and K. Nishio, An Aqueous Li-Air Battery Based
Energy Storage Beyond Lithium Ion: Materials Perspective, Oak Ridge National on a Novel Reservoir Concept, in 15th International Meeting on Lithium Batteries,
Laboratory (2010). Montreal (2010).
82. S. J. Visco, Y. S. Nimon, L. De Jonghe, B. D. Katz, and A. Petrov, Compli- 131. R. V. Moshtev, Journal of Power Sources, 11, 93 (1984).
ant Seal Structures for Protected Active Metal Anodes, U.S. Patent Application 132. H. Gan and E. S. Takeuchi, Journal of Power Sources, 62, 45 (1996).
2007/0037058 (2007). 133. J. Xu and G. C. Farrington, Solid State Ionics, 74, 125 (1994).
83. R. Pollard and J. Newman, Journal of the Electrochemical Society, 128, 491 134. L. O. Valøen, and J. N. Reimers, Journal of The Electrochemical Society, 152, A882
(1981). (2005).
84. S. J. Visco, E. Nimon, B. Katz, L. C. D. Jonghe, and M. Y. Chu, The Devel- 135. X. Yang and Y. Xia, Journal of Solid State Electrochemistry, 14, 109 (2010).
opment of High Energy Density Lithium/Air and Lithium/Water Batteries with 136. B. McCloskey and G. Gopalakrishnan, Investigating the Electrochemistry of Li-O2
No Self-Discharge, in 210th Meeting of the Electrochemical Society, Cancun Battery Using DEMS and Surface Characterization Techniques, in Symposium on
(2006). Energy Storage Beyond Lithium Ion: Materials Perspectives, Oak Ridge National
85. M. Winter, W. K. Appel, B. Evers, T. Hodal, K. C. Möller, I. Schneider, M. Wachtler, Laboratory (2010).
M. R. Wagner, G. H. Wrodnigg, and J. O. Besenhard, Monatshefte für Chemie, 132, 137. S. J. Visco Darpa Palm Power: Advanced Lithium Anodes for Li/Air and Li/Water
473 (2001). Batteries, Polyplus Battery Company Inc, 2005.
138. T. Burchardt, Bifunctional Air Electrode, U.S. Patent Application 2007/0166602 158. J. J. Martin, V. Neburchilov, H. Wang, and W. Qu, Air cathodes for metal-air
(2007). batteries and fuel cells, in Electrical Power & Energy Conference, Montreal (2009).
139. K. Nakajima, T. Katoh, Y. Inda, and B. Hoffman, Lithium Ion Conductive Glass 159. H. Cnobloch, G. Siemsen, and F. V. Sturm, in Power sources 4: Research and
Ceramics: Properties and Application in Lithium Metal Batteries, in Symposium development in non-mechanical electrical power sources, D. H. Collins, editor,
on Energy Storage Beyond Lithium Ion: Materials Perspective, Oak Ridge National p. 311 (1973).
Laboratory (2010). 160. D. Groppel and G. Siemsen, Hybrid Electrode for Metal/Air Cells, U.S. Patent
140. T. Katoh, Y. Inda, K. Nakajima, M. Baba, and R. Ye, Lithium/water Battery With 3,944,434 (1976).
Lithium-ion Conducting Glass Ceramics, in 15th International Meeting on Lithium 161. W. Vielstich, A. Lamm, and H. A. Gasteiger, editors, Handbook of Fuel Cells:
Batteries, Montreal (2010). Fundamentals, Technology, and Applications, Wiley, Chichester, England, New
141. A. K. Thapa and T. Ishihara, Journal of Power Sources, 196, 7016 York (2003).
(2011). 162. L. Jörissen, Journal of Power Sources, 155, 23 (2006).
142. Y.-C. Lu, Z. Xu, H. A. Gasteiger, S. Chen, K. Hamad-Schifferli, and Y. Shao-Horn, 163. B. Viswanathan, C. V. Rao, and U. V. Varadaraju, in Photo/Electrochemistry & Pho-
Journal of the American Chemical Society, 132, 12170 (2010). tobiology in the Environment, Energy and Fuel, S. Kaneco, editor, p. 43, Research
143. Y. Xu and W. A. Shelton, The Journal of Chemical Physics, 133, 024703 Signpost, Kerala (2006).
(2010). 164. J. O. M. Bockris and T. Otagawa, Journal of The Electrochemical Society, 131, 290
144. E. Wang, D. Ofer, W. Bowden, N. Iltchev, R. Moses, and K. Brandt, Journal of the (1984).
Electrochemical Society, 147, 4023 (2000). 165. H. A. Gasteiger, S. S. Kocha, B. Sompalli, and F. T. Wagner, Applied Catalysis B:
145. S. E. Sloop, J. B. Kerr, and K. Kinoshita, Journal of Power Sources, 119, 330 Environmental, 56, 9 (2005).
(2003). 166. S. Trasatti, Electrochimica Acta, 29, 1503 (1984).
146. G. G. Amatucci, N. Pereira, T. Zheng, and J. M. Tarascon, Journal of the Electro- 167. G. Chen, S. R. Bare, and T. E. Mallouk, Journal of The Electrochemical Society,
chemical Society, 148, A171 (2001). 149, A1092 (2002).
147. J. Zhang, W. Xu, and W. Liu, Journal of Power Sources, 195, 7438 (2010). 168. L. Jörissen and G. Jürgen, in Encyclopedia of Electrochemical Power Sources, p.
148. T. A. Reynolds, D. J. Brose, and M. N. Golovin, Membrane for Selective Transport 356, Elsevier, Amsterdam, (2009).
of Oxygen Over Water Vapor and Metal-Air Electrochemical Cell Including Said 169. V. Neburchilov, H. Wang, J. J. Martin, and W. Qu, Journal of Power Sources, 195,
Membrane, U.S. Patent 5,985,475 (1999). 1271 (2010).
149. J. Zhang, D. Wang, W. Xu, J. Xiao, and R. Williford, Journal of Power Sources, 170. T. R. Crompton, Battery Reference Book, 3rd edition, Newnes, Oxford, England,
195, 4332 (2010). Boston (2000).
150. O. Crowther, B. Meyer, M. Morgan, and M. Salomon, Journal of Power Sources, 171. I. Dabo, Y. Li, N. Bonnet, and N. Marzari, Ab Initio Electrochemical Properties of
196, 1498 (2011). Electrode Surfaces, in Fuel Cell Science: Theory, Fundamentals, and Biocatalysis,
151. S. J. Visco, Protected Li anodes for use in Conventional & Aggressive Electrolytes, A. Wieckowski and J. K. Nørskov, editors, John Wiley & Sons, Inc., Hoboken, NJ
in Advanced Anode Workshop, Lawrence Berkeley National Laboratory, Berkeley, (2010).
CA (2004). 172. S. A. Wasileski, C. D. Taylor, and M. Neurock, in Device and Materials Modeling
152. S. Hasegawa, N. Imanishi, T. Zhang, J. Xie, A. Hirano, Y. Takeda, and O. Yamamoto, in PEM Fuel Cells, p. 551, Springer-Verlag, Berlin (2009).
Journal of Power Sources, 189, 371 (2009). 173. V. Stamenkovic, B. S. Mun, K. J. J. Mayrhofer, P. N. Ross, N. M. Markovic,
153. T. Zhang, N. Imanishi, S. Hasegawa, A. Hirano, J. Xie, Y. Takeda, O. Ya- J. Rossmeisl, J. Greeley, and J. K. Norskov, Angewandte Chemie-International
mamoto, and N. Sammes, Journal of The Electrochemical Society, 155, A965 Edition 45, 2897 (2006).
(2008). 174. J. K. Norskov, T. Bligaard, J. Rossmeisl, and C. H. Christensen, Nature Chemistry,
154. C. Monnin and M. Dubois, Journal of Chemical & Engineering Data, 50, 1109 1, 37 (2009).
(2005). 175. P. Sabatier, Ber. Deutschen Chemischen Gesellschaft, 44, 1984 (1911).
155. N. Chaturvedi, R. Klein, J. Christensen, J. Ahmed, and A. Kojic, IEEE Control 176. U. S. Advanced Battery Consortium http://www.uscar.org/commands/files_
Systems Magazine, 30, 49 (2010). download.php?files_id=27, http://www.uscar.org/commands/files_download.php?
156. J. F. Cooper and E. L. Littauer, Mechanically Rechargeable, Metal-Air Batteries files_id=83, http://www.uscar.org/commands/files_download.php?files_id=25
for Automotive Propulsion, in 13th Intersociety Energy Conversion Engineering (2002).
Conference, San Diego (1978). 177. B. D. McCloskey, R. Scheffler, A. Speidel, D. S. Bethune, R. M. Shelby, and A.
157. R. Thacker, Electrochimica Acta, 14, 433 (1969). C. Luntz, J. Am. Chem., 133(45), 18038 (2011).
View publication stats Downloaded 16 Jan 2012 to 209.221.240.193. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp

A Critical Review of LiAir Batteries

Uploaded by

Copyright:

Available Formats

A Critical Review of LiAir Batteries

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

A Critical Review of LiAir Batteries

Uploaded by

Copyright:

Available Formats

See

A Critical Review of Li/Air Batteries

Article in Journal of The Electrochemical Society · January 2012

Jake Christensen Boris Kozinsky

SEE PROFILE SEE PROFILE

Battery heat generation View project

Electric Vehicle Batteries View project

The user has requested enhancement of the downloaded file.

A Critical Review of Li/Air Batteries

500 may form as a film on the surface of Li metal.8 An example of the

Another example of a reaction in an acidic solution is the formation

Specific Capacity Uθ vs. Li Theoretical specific Theoretical energy

Li2 O 1794 2.01 3606 2.91 5.22 10.49

Active-only energy density (kWh/L)

Practical energy density (kWh/L)

Stabilizing the Li metal negative electrode.

30 Issues that apply only to aqueous systems:

Nonaqueous Li/Air Systems

Figure 5. Example of a Li/air gas diffusion cathode based on Ni-foam. The

Figure 8. Influence of three different cathode designs on the electrochemical

Fit to 0.75 µA/cm data

is electrically insulating and remains at the reaction site that will be

2.6 trolyte. Enhancing the solubility of Li2 O2 could potentially enable

Figure 23. Ionic conductivities of several classes

Mass transfer limitations are those that involve concentration gra-

Mass breakdown: Total = 1.5963 kg/m2

thickness of the polymer membrane was 25 μm, representing ∼1%

electron transfer from H2 O in acidic media or OH− in basic media

Requirements for durable, high-energy automotive Li/air batteries

You might also like