You are on page 1of 6

Perspective

https://doi.org/10.1038/s41560-017-0047-2

Status and challenges in enabling the lithium


metal electrode for high-energy and low-cost
rechargeable batteries
Paul Albertus   1*, Susan Babinec1, Scott Litzelman2 and Aron Newman2

Enabling the reversible lithium metal electrode is essential for surpassing the energy content of today’s lithium-ion cells.
Although lithium metal cells for niche applications have been developed already, efforts are underway to create recharge-
able lithium metal batteries that can significantly advance vehicle electrification and grid energy storage. In this Perspective,
we focus on three tasks to guide and further advance the reversible lithium metal electrode. First, we summarize the state of
research and commercial efforts in terms of four key performance parameters, and identify additional performance parameters
of interest. We then advocate for the use of limited lithium (≤​30 μ​m) to ensure early identification of technical challenges asso-
ciated with stable and dendrite-free cycling and a more rapid transition to commercially relevant designs. Finally, we provide a
cost target and outline material costs and manufacturing methods that could allow lithium metal cells to reach 100 US$ kWh–1.

L
i-ion batteries are transforming the transportation and grid cathodes operating at <​4 V versus Li (refs  22,23). Because neither of
sectors. Their scale up is truly historic: Li-ion is now the these material platforms currently surpass Li-ion for vehicle or grid
only rechargeable battery other than lead acid produced at applications, development efforts continue. Unfortunately, such
>​5  GWh  y–1, with a worldwide manufacturing expansion reach- efforts are hampered by the lack of a systematic understanding —
ing hundreds of GWh  yr–1 over the next five years1,2. Li-ion bat- incorporating both theory and experimental validation — of the
tery packs achieve long cycle life (in the thousands), high charge/ conditions under which lithium dendrites do and do not form18,24–29.
discharge rates (>​1  C), high energy content (specific energy of As just one example, the physical form of lithium dendrites (or
~150 Wh kg–1 and energy density of 250 Wh L–1), and low capital more generically, lithium penetrations) differs significantly based
costs (<​300  US$  kWh–1)3,4. However, the present Li-ion material on the medium through which they are growing; lithium penetra-
platform (a graphite negative electrode coupled with a metal oxide tion through a polycrystalline inorganic solid electrolyte has a much
positive electrode) is not expected to reach the US Department different morphology than growth through a liquid30,31. Regardless
of Energy’s (DOE) electric vehicle pack goals of 235  Wh  kg–1, of the approach, challenges for high-rate, high-capacity cycling
500 Wh L–1 and 125 US$ kWh–1 (ref.  5). The intercalation mecha- of a metal electrode are profound. Metal dendrites are known
nism that fundamentally enables the excellent cycling of Li-ion also to grow in numerous systems, and observations of Ag dendrite
places an upper limit on energy content because of the weight and growth through a solid inorganic ion conductors appear to go back
volume of the hosts into which Li+ intercalates. Thus, there remains nearly 100 years32,33. In the case of lithium, dendrites are known to
an acute need for higher-energy alternatives to the Li-ion material grow through many classes of matter, including liquids, polymers,
platform6–8. Replacing the graphite electrode with lithium metal glasses, and polycrystalline inorganic solid electrolytes19,28,30,34.
(Fig.  1), which results in a ~35% increase in specific energy and In this Perspective, we provide a new figure that includes four
~50% increase in energy density at the cell level, provides a path to parameters of critical importance for the practical development
reach those goals, especially if the introduction of lithium metal is of a lithium metal electrode, and use this figure to summarize
combined with reduction of the liquid electrolytes, which impose the status of both academic and commercial efforts. We find that
both safety and thermal management mass and volume require- although impressive results have been achieved for subsets of the
ments at the pack level5,9–12. four critical parameters, none have met targets for all four, and we
Stable cycling of lithium metal requires high Coulombic effi- suggest researchers focus their attention on the complete set of criti-
ciency, a low and stable resistance, and the avoidance of lithium cal parameters when cycling cells. Perhaps the biggest gap in the
dendrites. To meet these requirements a myriad of approaches have parameters used for lithium metal cycling today is a large excess of
been pursued (starting in the 1960s), most of them focused on the lithium — far more than could be included in a high-energy cell —
use of solid, liquid, composite and other electrolytes13–19. Vapour- and an amount that makes it difficult to perform the careful experi-
deposited lithium phosphorus oxynitride (LiPON) solid separa- ments needed to ensure stable, dendrite-free cycling. In particular,
tors, and thin-film cells based thereon, were a major advance in we believe that numerous reports of stable lithium metal cycling
the late 1990s and remain the definitive example of stable cycling of in symmetric Li/Li cells are in fact the result of ‘soft shorts’ (stable
lithium metal at room temperature20,21. Another key advance came electronic connections between the electrodes). These carry elec-
with the development of polyethylene oxide polymer electrolytes, tronic current; fortunately, the use of a limited amount of lithium
which unfortunately require elevated temperature (~80  °C) and (≤​30 μ​m, now available commercially) can help identify soft shorts

1
Advanced Research Projects Agency – Energy, US Department of Energy, Washington, DC, USA. 2Booz Allen Hamilton, Washington, DC, USA.
*e-mail: paul.albertus@hq.doe.gov

16 Nature Energy | VOL 3 | JANUARY 2018 | 16–21 | www.nature.com/natureenergy

© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
Nature Energy Perspective
Cu current Porous polymer Al current
collector separator collector
Graphite porous Metal oxide
electrode porous electrode

Typical
lithium-ion cell
at the pack level:
~150 Wh kg–1
~250 Wh L–1

~20 μm ~85 μm

Prospective
lithium-metal cell
at the pack level:
~250 Wh kg–1
~750 Wh L–1

Lithium Metal oxide


metal porous electrode

Cu current Solid Al current


collector separator collector

Fig. 1 | Repeat layers of Li-ion and lithium metal cells. A typical Li-ion cell (top) and one conception of a lithium metal cell (bottom), containing a solid
separator and a dense layer of metallic lithium. The lithium metal cell is in the charged state, with 20% excess lithium. Layer thicknesses are shown to
scale. The reduction in volume and mass associated with replacing the graphite electrode with lithium metal is evident. Alternative conceptions of a
lithium metal electrode include the use of a host structure that limits electrode volume change as the lithium is passed to and from the positive electrode.
Assumptions for the lithium metal cell: cells have 80% of the specific energy and energy density of the repeat layers, and packs achieve 70% of the
specific energy and energy density of the cells.

in symmetric cell tests. Finally, we present a cost target for a lithium with a static host structure in which the lithium is deposited and
metal electrode and separator, and examine a wide range of com- removed35,36), the change in electrode thickness during a full plate
mon sheet materials and industrial processes to assess the poten- and/or strip.
tial of reaching that cost target. Plastic and metal sheets easily meet Several of the points in Fig. 2 merit special discussion. First, Fig. 2
the target, while several inorganic sheets are also at or close to the shows two goals for cycling lithium metal. Point  1, the ‘ARPA-E
target. It is clear that exceptional attention to material choice and IONICS goal’, comes from a recent US Department of Energy
process design will be needed in the early stages of development to programme that targets at least 80% of the lithium present passed
ensure that, if the reversible lithium metal electrode is enabled in a per cycle is targeted — a value that is rarely explored at present37.
high-energy cell format, it will also offer a cost that is transforma- Point 2 represents a goal for a lithium metal cell with a loading of
tional for automotive and grid storage applications. 5  mAh  cm–2 that is capable of fast charging at 2  C. Second, Fig. 2
includes data points from both commercial and research efforts.
The state of reversible lithium metal development Point 3 shows the state of the art for a LiPON thin-film cell; the areal
Figure 2 presents the four key parameters we believe are most rel- capacity is low, but thousands of cycles are achieved with about one
evant to cycling in high-energy cells and enables researchers to third of the lithium present passed per cycle and a plating current
readily compare their efforts to the published state of commercial density of ~1 mA cm–2. Points 7 and 8 are PEO-based cells, whereas
and research efforts. The first parameter is the cumulative capacity point 10 shows results from a cell with an inorganic solid separator
plated prior to cell shorting or excessive impedance rise; the and a liquid interlayer between the inorganic solid separator and the
cumulative capacity is the summation of the capacity plated on lithium metal. The LiPON and PEO cells achieve stable lithium plat-
each individual cycle, and can be obtained by multiplying the ing for well over 1,000 hours. Approaches using liquid electrolytes
capacity plated per cycle and the number of cycles. The second (points 15–25) have achieved impressive results in terms of current
parameter is the plating current density, and the third is the density and per-cycle areal capacity, but typically with <​1,000 hours
capacity of lithium metal plated on each cycle. The fourth parame- of plating and a large excess of lithium. Points 13, 18 and 26 approxi-
ter is the fraction of lithium metal initially in the cell that is plated mately meet the ARPA-E IONICS goal in terms of plating current
on each cycle; this is a measure of the lithium present that is actu- density, areal capacity and cumulative plating capacity, though each
ally utilized during cycling. Further information on each of the has a large excess of lithium. We believe that the impressive cycling
data points is given in Supplementary Table 1. Additional impor- in these reports may reflect some soft shorting, a possibility that is
tant parameters not shown in Fig. 2 include the separator thick- generally not evaluated by the current lithium metal research com-
ness, the type of porous separator and the volume of electrolyte munity. The following section advocates the use of a limited amount
present (for liquid electrolytes) and, for approaches that seek to of lithium in test cells, which could help identify electronic conduc-
limit volume change in the electrode during cycling (for example, tion through soft shorts.

Nature Energy | VOL 3 | JANUARY 2018 | 16–21 | www.nature.com/natureenergy 17


© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
Perspective Nature Energy

2 (Fast-charge goal) Per-cycle areal


capacity (mAh cm–2)
8 10
6 18 1
4 13
Cumulative capacity plated (Ah cm–2) 23
7 3
2 1 (ARPA-E
10,000 h 5
4 22 IONICS goal)
8
8 26
6 6
17
4 1.0
9 21 15

Fraction of Li passed per cycle


20 14 0.8
2 11
12
19
1,000 h 0.6
25
8 16
6 0.4
5
4 0.2
100 h 24
2 0.0
89 2 3 4 5 6 7 89 2 3 4 5 6 7 89
0.1 1 10
Plating current density (mA cm–2)

Fig. 2 | Status of published efforts on the cycling of lithium metal. The data is analysed in terms of four parameters: cumulative areal capacity plated, per-
cycle areal capacity, plating current density, and the fraction of the initial lithium metal present that is plated per cycle. The size of each point indicates the
per-cycle areal capacity (see the key in the upper-right corner), and the colour indicates the fraction of Li metal initially in the cell that is plated on each
cycle (see the key in the lower-right corner). Points 1 and 2 are goals, 3–6 are for LiPON thin-film cells, 7–9 are PEO-based solid polymer electrolytes, 10–12
are solid inorganic separators, 13 and 14 are custom nanostructures, and 15–26 are liquid electrolytes. See Supplementary Table 1 for detailed references
for each point. Some of the results shown in the figure likely include electronic soft shorts, for which the reported amount of lithium plating would be
incorrect, as discussed in the text.

Necessity of using limited lithium in test cells density in a matter of hours, and a full strip can be done peri-
The use of a much higher loading of lithium metal in a research odically during a long-term cycling test. In addition to the use of
cell than would be used in a commercial cell has a practical origin: limited lithium and a full strip, three additional experiments are
the ready availability and low cost of lithium foils that are hundreds helpful to check for the presence of electronic shorts: (1) measure
of micrometres thick at laboratory supply companies. However, we the activation energy of the conductivity (low or negative values
believe the lithium metal research community should shift to the indicate electronic conductivity); (2) check for inductive loops,
use of limited lithium (defined here as ≤​30 μ​m, or ≤​ ~6 mAh cm–2) or the absence of an imaginary loop, with impedance spectros-
even in experimental test cells, for reasons described below. This copy30,38,39; and (3) monitor the open-circuit voltage of a full cell to
shift can be made easily, as lithium foils with a thickness of 20 μ​m check for electronic shorting.
are now available commercially, although surface passivation may Second, the use of limited lithium in research cells will more
need to be handled carefully. closely reflect viable commercial designs with high energy density,
First, we believe that some of the results from Li/Li symmetric and ensure that technical challenges for those designs are identified
cells in Fig.  2 (for example, points  13, 18 and 26) reflect mixed at an early point. Several groups are now reporting current densities,
ionic/electronic conduction in the separator, leading to decep- per-cycle areal capacities and cumulative capacities approaching
tively ‘good’ cycling results. In many research publications with DOE goals; however, Fig.  2 shows that these efforts use far more
Li/Li symmetric cells, the cell potential does not fall to 0 V during lithium than acceptable for the targeted energy density. As one
extended cycling but rather stabilizes, often after a transition example of a technical challenge, cycling 15 μ​m (~3 mAh cm–2) of a
period of tens to hundreds of hours, which we believe could indi- 500 μ​m lithium foil results in far less volume change (on a fractional
cate soft shorts. What may be taking place in these cases is the basis) than cycling 15 μ​m out of 19 μ​m (the 80% target shown in
growth of lithium metal through the separator to contact the posi- Fig. 2). Coupling among the parameters shown in Fig. 2 is certain,
tive electrode and thus form a direct electronic pathway between and additional technical challenges may emerge as limited lithium
the electrodes. If this growth occurs slowly, with significant ionic is used more widely.
current taking place even as an electronic pathway is established, Third, the limited lithium paradigm allows the Coulombic
the cell potential may change only slowly, and never fall to zero as efficiency — another critical electrochemical parameter — to be
it would for a ‘hard’ short. Once an electronic pathway is estab- assessed readily through a limited plate and a full strip. While long-
lished, extremely stable cycling may be observed because lithium term cycling tests rarely use limited lithium, the use of a limited
is no longer being passed between the electrodes. The use of a plate and strip is well-established and practiced by some groups
limited amount of lithium metal can conclusively identify the today, with copper or stainless steel typically used as the working
presence of electronic shorting in the following way. When a electrode40–43. A full strip may be carried out on each cycle, or peri-
known amount of lithium is removed fully from one electrode, the odically, to assess the Coulombic efficiency. With 80% of the initial
potential should rise, thereby indicating full depletion; Faraday’s lithium passed per cycle (the goal), reaching 1,000 cycles requires a
law can be used to compare the charge passed with the amount of Coulombic efficiency of 99.98% — a significantly higher value than
lithium present. Lithium with a thickness of 20 μ​m, corresponding that reached by present approaches using liquid41,43 or solid44 elec-
to ~4  mAh  cm–2, can be fully stripped at a meaningful current trolytes. Notably, LiPON thin-film cells in which all of the lithium

18 Nature Energy | VOL 3 | JANUARY 2018 | 16–21 | www.nature.com/natureenergy

© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
Nature Energy Perspective
originates in the cathode have been built and cycled hundreds of but also to enable per-kilogram costs that reflect the current prices
times, thereby proving a Coulombic efficiency above 99.9% is pos- of battery-grade materials such as electrolytes (~20 US$ kg–1) and
sible in practice with lithium metal45. electrode materials (15–30  US$  kg–1). The role of density is also
evident, with significant density differences among leading Li+-
Hitting aggressive cost goals conducting materials46,47. Figure 3a shows that several commodity
In addition to meeting aggressive performance goals, lithium metal sheet materials (such as glass, Al, polyethylene and paper pulp) have
batteries for transportation or the grid must also meet challenging material costs under 0.10 US$ m–2, while more specialized materials
cost goals to achieve mass adoption. It is important to note that like CdTe and 8 mol% Y2O3-stabilized ZrO2 (8YSZ) have costs well
the ‘cost’ goal is really the price goal for cell buyers (for example, above 1 US$ m–2.
vehicle manufacturers) and includes the manufacturing cost, busi- Figure  3b plots the price of fabricated sheet products at the
ness overhead, warranty and profit. With the DOE cell goal of thickness at which they are made against their material cost;
100 US$ kWh–1, and assuming the cost breakdown reflects that of colours classify each product according to the manufacturing
an existing Li-ion energy cell (see Supplementary Table 2), the areal process by which they are made. Numerous material inputs and
cost goal for the cell repeat layers is 10–12 US$ m–2. With the cath- manufacturing processes result in sheet products at <​5 US$ m–2. At
ode and current collectors retaining a combined cost of ~7 US$ m–2, the low end, thin (<​10 μ​m) polyethylene sheet made with continu-
the cost goal for the separator and any lithium beyond that present ous, low-temperature (~150 ºC) melt processing results in a sheet
in the cathode is <​5 US$ m–2 (ref. 11). Lowering the cathode or cur- product at just cents per square metre; polymer processing more
rent collector costs (or using Li metal as the current collector rather generally has the clearest ability to meet the cost goal. The rolling
than Cu), or reducing manufacturing costs, would allow a higher of thin metal sheet is another low-temperature continuous process
cost for the separator plus lithium metal sheets. that results in costs of ~1 US$ m–2 or below. Float glass is a remark-
The critical question is how this challenging cost goal can be able inorganic example: despite starting with molten glass at
achieved. Each of the major contributions to cost depends on manu- >​1,000 °C and requiring an inert atmosphere (N2 and H2) to pro-
facturing volume, and each needs careful consideration. We analyse tect the molten tin bath, sheet product several millimetres thick is
the cost in the following three categories: (1) cost of the material sold at <​5 US$ m–2. Unfortunately, this float glass process is pres-
inputs in the appropriate purity and form; (2) cost of the processing ently limited to thicknesses of several hundred micrometres or
required to form the separator from the material inputs; and (3) higher. It is also notable that simple vapour deposition, for example
cost of integrating a separator, lithium metal (if needed) and other metallized plastic and the CdTe solar cell active layer, also achieves
cell components into a cell. Figure  3 addresses costs (1) and (2); product costs of <​~5  US$  m–2. The ability of high-temperature
cost (3) will be addressed at the end of this section. Figure 3a shows (>​1,000 ºC) sintering of ceramic materials to meet the cost target
material costs assuming full density and a 20  μ​m thickness as a will depend on the process temperature and duration; the process-
function of material cost (in US$ kg–1); diagonal lines correspond ing conditions used to densify the 8YSZ and other layers used in
to constant densities. Lithium metal, because of its low density and solid-oxide fuel cells is likely to be too expensive to meet the aggres-
molecular weight, costs ~1 US$ m–2 for 20 μ​m (corresponding to sive cost target defined here. Material and processing innovations
~4 mAh cm–2) despite its cost (as an ingot) of ~80 US$ kg–1. The may allow ceramic powders to be densified at lower cost48,49. While
lithium layer offers an opportunity for cost saving: if only a seed Fig.  3b shows there are several processing techniques consistent
layer is required (for example, 5 μ​m), with the cathode supplying with the aggressive cost goal, exceptional attention to material
the remaining lithium, material costs drop to ~0.25  US$  m–2. choice and process design will be required from the earliest stages
Presently, lithium foil at these costs and thicknesses is not available, of research to ensure the potential to reach the goal.
but remains of critical importance in the event of scale-up. Figure 3a Finally, it is also important to address cost area (3) listed above:
shows the importance of reaching a separator thickness of 20 μ​m, cell integration costs. Li metal forms a passivating surface film even
not only to ensure a high energy content for the repeating cell layers, in a dry room, which could lead to high interfacial resistance, and

a 100 b 100
8YSZ SOFC electrolyte layer (10 µm) Nafion (25 µm)
Material input cost for 20 µm sheet (US$ m–2)

CdTe Si solar cell (150 µm)


10 10 Approximate goal for
Sheet product prices (US$ m–2)

8YSZ
Approximate goal for separator and any Li metal ePTFE (25 µm)
separator and any Li metal Polysilicon Celgard (15 µm)
Cu Soda-lime float CdTe solar cell
1 1 glass (3 mm) active layer (3 µm)
Ti Nafion Tyvek (150 µm) Cu foil (10 µm)
Polyethylene Al-coated plastic (several µm)
Al foil (16 µm)
Paper pulp Li metal
Soda-lime Teflon (PTFE) Polymer melt processing
0.1 0.1 Paper (100 µm)
glass Float glass process
Al PE sheet (7.9 µm) Paper making
5 g cm–3 Vapour deposition
0.01 2.5 g cm–3 0.01 Metal rolling
0.5 g cm–3 Other

0.1 1 10 100 0.1 1 10 100


Cost of sheet materials (US$ kg–1, exludes processing) Cost of sheet materials (US$ kg–1, exludes processing)

Fig. 3 | Cost analysis for lithium metal batteries. a,b, Cost of producing a fully dense 20 μ​m sheet as a function of the cost of materials alone (diagonal
lines represent constant densities) (a), and approximate sheet product prices at the thickness at which they are sold, again as a function of the cost of the
material inputs (b). The goal of <​5 US$ m–2 is shown as a dashed horizontal line. Additional details on the points in Fig. 3 are provided in Supplementary
Table 3 (for a) and Supplementary Table 4 (for b).

Nature Energy | VOL 3 | JANUARY 2018 | 16–21 | www.nature.com/natureenergy 19


© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
Perspective Nature Energy

some of the separator technologies currently under development 12. Gallagher, K. G. et al. Quantifying the promise of lithium–air batteries for
(for example, rigid sheets of inorganic materials) cannot be wound. electric vehicles. Energ. Environ. Sci. 7, 1555–1563 (2014).
The use of an inert atmosphere, stacking rather than rolling, and unit 13. Mayer, S. W. & McKenzie, D. E. Lightweight secondary battery. US patent
3(185), 590 (1965).
operations such as vapour deposition or high-temperature sintering 14. Whittingham, M. S. Electrical energy storage and intercalation chemistry.
would be departures from existing roll-to-roll Li-ion processing. Science 192, 1126–1127 (1976).
Ultimately, achieving both cost and performance targets may 15. Tobias, C. W. & Jorne, J. Method of production of alkali metals and their
require the design of a new form of production line that integrates alloys. US Patent US3791945 A (1974).
novel unit operations with existing processing steps. 16. Hughes, M., Hampson, N. & Karunathilaka, S. A review of cells based on
lithium negative electrodes (anodes). J. Power Sources 12, 83–144 (1984).
17. Xu, W. et al. Lithium metal anodes for rechargeable batteries. Energ. Environ.
Outlook Sci. 7, 513–537 (2014).
The development of a reversible lithium metal electrode has eluded 18. Tikekar, M. D., Choudhury, S., Tu, Z. & Archer, L. A. Design principles for
researchers for decades, but remains of great interest due to its electrolytes and interfaces for stable lithium-metal batteries. Nat. Energy 1,
potential to reach performance and cost goals. In this Perspective we 16114 (2016).
19. Takeda, Y., Yamamoto, O. & Imanishi, N. Lithium dendrite formation on a
identified critical parameters against which to assess progress and lithium metal anode from liquid, polymer and solid electrolytes. Electrochem
provided a summary of published efforts. We also encouraged the 84, 210–218 (2016).
lithium metal battery community to use limited lithium (≤​30 μ​m) 20. Bates, J. B., Dudney, N. J., Neudecker, B., Ueda, A. & Evans, C. D.
rather than thick lithium foils for numerous reasons, including the Thin-film lithium and lithium-ion batteries. Solid State Ionics 135,
ability to detect soft shorts. Finally, we established a cost target for a 33–45 (2000).
21. Bates, J. B. et al. Fabrication and characterization of amorphous lithium
separator plus lithium sheet to enable a cell price of 100 US$ kWh–1, electrolyte thin films and rechargeable thin-film batteries. J. Power Sources 43,
and showed that there are several manufacturing and material 103–110 (1993).
classes that can reach or approach that target, although none to date 22. Hovington, P. et al. New lithium metal polymer solid state battery for an
with the performance properties required by a lithium metal battery. ultrahigh energy: nano C-LiFePO4 versus nano Li1.2V3O8. Nano Lett. 15,
If the lithium metal electrode can be proven to cycle in small 2671–2678 (2015).
23. Gross, J. & Eitouni, H. Drivers and technologies for Li-metal solid-state
research cells using the four parameters identified here, with mate- batteries. 16th Ann. Adv. Automot. Batt. Conf. (2016).
rial and processing costs consistent with the cost target at scale, 24. Monroe, C. & Newman, J. Dendrite growth in lithium/polymer systems.
additional challenges and opportunities will be evident. A dense J. Electrochem. Soc. 150, A1377–A1384 (2003).
lithium foil cycled with high per-cycle utilization is required and 25. Monroe, C. & Newman, J. The effect of interfacial deformation on
will result in significant volume changes in large-format cells; the electrodeposition kinetics. J. Electrochem. Soc. 151, A880–A886 (2004).
26. Monroe, C. & Newman, J. The impact of elastic deformation on deposition
resulting stresses and shape change may limit cycling and have del- kinetics at lithium/polymer interfaces. J. Electrochem. Soc. 152,
eterious side effects. Lithium metal electrodes with minimal volume A396–A404 (2005).
change, or novel types of large-format cells or packs containing 27. Tikekar, M. D., Archer, L. A. & Koch, D. L. Stabilizing electrodeposition
them, are possible solutions. Determining the safety and rate capa- in elastic solid electrolytes containing immobilized anions. Sci. Adv. 2,
bility of large-format cells, especially after extended cycling and e1600320 (2016).
28. Porz, L. et al. Mechanism of lithium metal penetration through inorganic
during charge at low temperatures, will be another priority. In terms solid electrolytes. Adv. Energ. Mater. https://doi.org/10.1002/
of manufacturing, lithium metal technologies that integrate easily aenm.201701003 (2017).
into existing lithium ion production lines will achieve the most 29. Aguesse, F. et al. Investigating the dendritic growth during full cell cycling of
rapid growth. The need to address these challenges is motivated by garnet electrolyte in direct contact with Li metal. ACS Appl. Mater. Interf.
the definitive shift towards an electrified future, and the fact that 9, 3808–3816 (2017).
30. Cheng, E. J., Sharafi, A. & Sakamoto, J. Intergranular Li metal propagation
high-energy rechargeable batteries are perhaps the single most through polycrystalline Li6.25Al0.25La3Zr2O12 ceramic electrolyte. Electrochim.
important enabler for that transformation to be widely successful. Acta. 223, 85–91 (2017).
31. Park, M. S. et al. A highly reversible lithium metal anode. Sci. Rep. 4,
Received: 2 August 2017; Accepted: 24 October 2017; 3815 (2014).
32. Tubandt, C. Handbuch der Experimentalphysik (Akad. Verlagsgesellschaft,
Published online: 18 December 2017
Leipzig, 1932).
33. Reddy, T. Linden’s Handbook of Batteries (eds Reddy, T. & Linden, D.) (2011).
References 34. Aurbach, D., Zinigrad, E., Cohen, Y. & Teller, H. A short review of failure
1. Pillot, C. Battery Market Development for Consumer Electronics, Automotive, mechanisms of lithium metal and lithiated graphite anodes in liquid
and Industrial (2014). electrolyte solutions. Solid State Ionics 148, 405–416 (2002).
2. Ryan, J. China is about to bury Elon Musk in batteries. Bloomberg  35. Liu, Y. et al. Lithium-coated polymeric matrix as a minimum
(28 June 2017). volume-change and dendrite-free lithium metal anode. Nat. Commun. 7,
3. Schmidt, O., Hawkes, A., Gambhir, A. & Staffell, I. The future cost of electrical 10992 (2016).
energy storage based on experience rates. Nat. Energy 6, 17110 (2017). 36. Wang, C. et al. Conformal, nanoscale ZnO surface modification of
4. Nykvist, B. & Nilsson, M. Rapidly falling costs of battery packs for electric garnet-based solid-state electrolyte for lithium metal anodes. Nano Lett. 17,
vehicles. Nat. Clim. Change 5, 329–332 (2015). 565–571 (2016).
5. USABC Goals for Advanced Batteries for EVs — CY 2020 37. Albertus, P. Integration and Optimization of Novel Ion-Conducting Solids
Commercialization (USABC, 2017). (IONICS) Funding Opportunity Announcement (2016); https://arpa-e-foa.
6. Choi, J. W. & Aurbach, D. Promise and reality of post-lithium-ion batteries energy.gov/FileContent.aspx?FileID=cfac9ce8-5a19-4623-b942-c3e65f3ccf77.
with high energy densities. Nat. Rev. Mater. 1, 16013 (2016). 38. Hernandez-Maya, R., Rosas, O., Saunders, J. & Castaneda, H. Dynamic
7. Thackeray, M. M., Wolverton, C. & Isaacs, E. D. Electrical energy storage for characterization of dendrite deposition and growth in Li-surface by
transportation — approaching the limits of, and going beyond, lithium-ion electrochemical impedance spectroscopy. J. Electrochem. Soc. 162,
batteries. Energ. Environ. Sci. 5, 7854–7863 (2012). A687–A696 (2015).
8. Goodenough, J. B. & Park, K.-S. The Li-ion rechargeable battery: a 39. Sharafi, A., Meyer, H. M., Nanda, J., Wolfenstine, J. & Sakamoto, J.
perspective. J. Am. Chem. Soc. 135, 1167–1176 (2013). Characterizing the Li–Li7La3Zr2O12 interface stability and kinetics as a
9. Christensen, J. et al. A critical review of Li/air batteries. J. Electrochem. Soc. function of temperature and current density. J. Power Sources 302,
159, R1–R30 (2012). 135–139 (2016).
10. Albertus, P., Lohmann, T. & Christensen, J. in The Li–Air Battery: 40. Aurbach, D., Gofer, Y. & Langzam, J. The correlation between surface
Fundamentals (eds Imanishi, N., Luntz, A. C. & Bruce, P.) 291–310 chemistry, surface morphology, and cycling efficiency of lithium
(Springer, 2014). electrodes in a few polar aprotic systems. J. Electrochem. Soc. 136,
11. McCloskey, B. D. Attainable gravimetric and volumetric energy density of 3198–3205 (1989).
Li–S and Li ion battery cells with solid separator-protected Li metal anodes. 41. Li, W. et al. The synergetic effect of lithium polysulfide and lithium nitrate to
J. Phys. Chem. Lett. 6, 4581–4588 (2015). prevent lithium dendrite growth. Nat. Commun. 6, 7436 (2015).

20 Nature Energy | VOL 3 | JANUARY 2018 | 16–21 | www.nature.com/natureenergy

© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
Nature Energy Perspective
42. Qian, J. et al. High rate and stable cycling of lithium metal anode. Acknowledgements
Nat. Commun. 6, 6362 (2015). The authors thank N. Dudney, S. Visco, L. Archer, K. Xu, T. Holme and P. Frischmann
43. Tu, Z. et al. Nanoporous hybrid electrolytes for high‐energy for helpful discussions.
batteries based on reactive metal anodes. Adv. Energ. Mater. 7,
1602367 (2017). Author contributions
P.A. conceived the figures, conducted the majority of the analysis and wrote most of
44. Kato, Y. et al. High-power all-solid-state batteries using sulfide superionic
the paper. S.B. and S.L. made significant contributions to the analysis and editing of the
conductors. Nat. Energy 1, 16030 (2016).
paper. A.N. contributed analysis and edited the paper.
45. Neudecker, B., Dudney, N. & Bates, J. ‘Lithium‐free’ thin‐film battery with in
situ plated Li anode. J. Electrochem. Soc. 147, 517–523 (2000). Competing interests
46. Murugan, R., Thangadurai, V. & Weppner, W. Fast lithium ion The authors declare no competing financial interests.
conduction in garnet‐type Li7La3Zr2O12. Angew. Chem. Int. Ed. 46,
7778–7781 (2007). Additional information
47. Bron, P. et al. Li10SnP2S12: an affordable lithium superionic conductor. Supplementary information is available for this paper at https://doi.org/10.1038/
J. Am. Chem. Soc. 135, 15694–15697 (2013). s41560-017-0047-2.
48. Guo, J. et al. Cold sintering: a paradigm shift for processing and integration
Reprints and permissions information is available at www.nature.com/reprints.
of ceramics. Angew. Chem. 128, 11629–11633 (2016).
49. Guo, J. et al. Cold sintering process of composites: bridging the processing Correspondence and requests for materials should be addressed to P.A.
temperature gap of ceramic and polymer materials. Adv. Func. Mater. 26, Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in
7115–7121 (2016). published maps and institutional affiliations.

Nature Energy | VOL 3 | JANUARY 2018 | 16–21 | www.nature.com/natureenergy 21


© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

You might also like