ARTICLESPUBLISHED: 31 JULY 2017 | VOLUME: 2 | ARTICLE NUMBER: 17119

A facile surface chemistry route to a stabilizedlithiummetal anodeXiao Liang1†, Quan Pang1, Ivan R. Kochetkov1, Marina Safont Sempere2, He Huang1, Xiaoqi Sun1

and Linda F. Nazar1*

Lithium metal is a highly desirable anode for lithium rechargeable batteries, having the highest theoretical specific capacityand lowest electrochemical potential of all material candidates. Its most notable problem is dendritic growth upon Li plating,which is a major safety concern and exacerbates reactivity with the electrolyte. Here we report that Li-rich composite alloyfilms synthesized in situ on lithium by a simple and low-cost methodology e�ectively prevent dendrite growth. This isattributed to the synergy of fast lithium ion migration through Li-rich ion conductive alloys coupled with an electronicallyinsulating surface component. The protected lithium is stabilized to sustain electrodeposition over 700 cycles (1,400h) ofrepeated plating/stripping at a practical current density of 2mAcm−2 and a 1,500 cycle-life is realized for a cell paired with aLi4Ti5O12 positive electrode. These findings open up a promising avenue to stabilize lithium metal with surface layers havingtargeted properties.

Over the past decades, rechargeable batteries based on Li-ionintercalation positive and negative electrode materials havebeen widely used as ‘cathodes’ and ‘anodes’, respectively.

However, cells based on intercalation chemistry can provide onlylimited energy density, which is problematic in light of growingdemands for large-scale storage1. Alternative approaches based onconversion chemistry such as Li/O2 and Li/S batteries have attractedincreasing attention2–5. The advantages of such potentially highspecific energy systems arise from the combination of a very highcapacity cathodewith a lithiummetal anodewhich exhibits an orderof magnitude higher specific capacity than graphite (3,860 versus370mAh g−1; ref. 6). Although significant effort has been put intothe cathode side of these batteries, their eventual success relies on therealization of a stable and safe lithiummetal anode, and a favourableelectrode–electrolyte interface7. If stabilized lithium anodes couldbe employed for Li-ion batteries to replace graphitic carbons,significantly higher energy densities could also be achieved8.

A major problem associated with lithium metal is the formationof Li dendrites on cycling9. These result in cell short circuits, andaccelerate detrimental reactions with the electrolyte6, which notonly corrodes the Li metal anode but also consumes the electrolyte,leading to cell ‘dry-out’ and early failure10. Much effort has beendevoted to tackling this problem. However, there have been fewsignificant breakthroughs to date that enable long-term stablecycling at practical current densities greater than 1mA cm−2, withsome noteworthy exceptions based on electrolyte developments11,12.Electrolytes with low Cs+ concentrations can eliminate dendritegrowth at low currents13. Other approaches, including in situgeneration of a solid electrolyte interphase (SEI) on lithium usingelectrolyte additives14–16, high lithium salt concentrations17, andex situ embellishment of lithiumwith artificial protection layers18–20,or pre-treatment methodologies21, can stabilize the lithium surface.However, owing to the low electron/ion conductivity and poormechanical stability of most protection layers, they are ofteneffective only for a few hundred cycles at relatively low current

densities (that is, <0.5mA cm−2). Suppressing dendrite formationby reducing the local current density via high-surface-area currentcollectors was reported recently22–24. Although this approach isefficient, because the plated lithium duplicates the porous structure,electrolyte reactivity could be aggravated. Physically blockingdendrite growth by solid electrolytes25,26 and polymer electrolytes27with a high shear modulus can prevent dendrites from piercingthe separator. However, it does not change the fundamental, self-amplifying behaviour of dendrite growth6. Application of inorganicsolid electrolytes is also challenged by their poor interface withlithium28, and difficulty in fabricating thin membranes. Althoughsome sulfide-based and garnet-type solid electrolytes exhibitcomparable ionic conductivity to that of liquid electrolytes—andare relatively stable toward Li—dendrite growth through the grainboundaries is nonetheless a major issue29.

Here we report a highly effective approach to prevent Li dendriteformation by in situ formed surface films comprised of lithium-based compounds (for example, Li13In3, LiZn, Li3Bi, or Li3As),which exhibit fast lithium diffusion in the bulk. Our approach wasrealized by direct reduction of themetal chlorides by Li at room tem-perature. The resultant metal reacts with the underlying Li to give afilm comprised of the respective lithium-richest LixM alloy (M canbe In, Zn, Bi or As), and electronically insulating LiCl is formed as aby-product of the initial reaction. This alloy remains composition-ally invariant on cycling—providing a conduit for Li transport tothe underlying lithium—and the LiCl bestows insulating qualities tothe film, preventing reduction of the Li+ on the surface. The alloy-protected Limetal delivers an extended cycling life of 1,400 h of con-tinuous plating/stripping in symmetric cells at a high current densityof 2mA cm−2. Furthermore, cells with protected lithiumpairedwitha Li4Ti5O12 (LTO) electrode were cycled up to 1,500 cycles at a 5Crate (1C = 1 Li/h).Operando imaging of the lithium surface in liquidelectrolyte cells during cycling confirms that dendrite formation issuppressed by the surface layer, and ex situ SEMmeasurements showthat Li electrodeposition occurs under the alloy layer.

1Department of Chemistry, University of Waterloo, Waterloo, Ontario N2L 3G1, Canada. 2BASF SE, Ludwigshafen 67056, Germany.†Present address: College of Chemistry and Chemical Engineering, Hunan University, 410006 Changsha, China. *e-mail: lfnazar@uwaterloo.ca

NATURE ENERGY 2, 17119 (2017) | DOI: 10.1038/nenergy.2017.119 | www.nature.com/natureenergy 1

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

ARTICLES NATURE ENERGY

∗∗∗

20 40 60 802θ

Li13In3|Li

Li3As|Li

Li3Bi|Li

LiZn|Li

∗Li13In3 Li3As Li3Bi LiZn

Li

In

Li

Cl

Inte

nsity

(a.u

.)

a b

c

e

f

g

h

d

Figure 1 | Characterization of alloy-protected lithium foil. a, X-ray di�raction patterns of lithium metal and alloy-protected lithium metal (reference datafrom the corresponding JCPDS files: 00-033-0615 for Li13In3, 01-074-1158 for Li3As, 00-027-0427 for Li3Bi and 98-064-2412 for LiZn, respectively). Afterthe Li foil was reacted with metal chlorides it was aged for two days in vacuum. b, SEM image of ‘fresh’ lithium. c,d, EDS mapping of In and Cl, respectively,of the cross-section of Li13In3|Li. e,g, SEM images of the top alloy-covered surface of Li13In3|Li (e) and LiZn|Li (g). f,h, Cross-sectional SEM images ofLi13In3|Li (f) and LiZn|Li (h). The scale bars are 20 µm.

Fabrication of alloy-protected lithiummetal foilsThe facile, scalable procedure to fabricate thin alloy layers chemi-cally bound onto Li metal foil starts with in situ reduction of metalchlorides by Li:

xLi+MClx→M+xLiCl (M= In,Zn,Bi,As) (1)

The metal layer immediately undergoes reaction with the underly-ing lithium, proceeding until a single phase composition is achieved:

yLi+ zM→LiyMz (2)

The kinetics of equation (1) are very fast. The colour of the lithiummetal changes within seconds upon dipping it in the chloridesolutions (Supplementary Fig. 1). The equilibrated LiM protectionlayer was characterized by X-ray diffraction (XRD) (Fig. 1a),showing that the lithium-rich phases Li13In3, LiZn, Li3Bi, and Li3Asare the only crystalline phases evident on the respective lithiumfoils. Crystalline defects are undoubtedly present in the films, owingto the room-temperature growth conditions: significant XRD line-broadening was observed for Li3Bi, for example (Fig. 1a). Scanningelectron microscopy (SEM) images of a lithium surface (Fig. 1b)compared to that reacted with InCl3 and ZnCl2 as representatives(Fig. 1e,g) show the Li surface is uniformly covered with the alloyfilms. Energy dispersive spectroscopy (EDS) mapping of the In filmshows a uniform distribution of In and Cl in the layer (Fig. 1c,d),where the latter is present as LiCl (see equation (1)). It accounts fora weight ratio of 14.5% in the layer (EDS spectrum, SupplementaryFig. 2). The protected lithium foils are denoted henceforth asLiyMz|Li, where LiCl is omitted for clarity. The composite alloylayer is less than 10 µm thick as measured by cross-sectional SEM(Fig. 1f,h).

Importantly, the layers function quite differently than alloy anodematerials (Li–Mg, Li–Al, Li–In, and so on) that have been employedas sole lithium sources. Cycling the alloys through a range of Licompositions results in deleterious volume expansion/contractionand invokes rate-limiting two-phase transitions between differentline phases30,31. Our approach benefits from the fast Li conductionof the LixM alloys while the underlying Li foil provides thesource of lithium. Bulk chemical diffusion coefficients (DLi) forhigh-temperature synthesized phases are 4.7 × 10−8 cm2 s−1 for

LiZn at 300K (ref. 32), 3× 10−6 cm2 s−1 for Li3Bi at 320K (ref. 33),and 10−8–10−6 cm2 s−1 for the Li–In phase containing 47–63 at%Li (ref. 34). Assuming a 200 µm lithium foil, the protecting alloylayer accounts only for <5% volume. Because the alloys are linephases—not solid solutions—they remain as the Li-rich phase oncycling by virtue of contact with the lithium foil (see SupplementaryFig. 3, Li13In3|Li). The bulk alloy phases are likely to be electronicconductors (either metals or small band-gap semiconductors, assuggested by our band structure calculations, Supplementary Fig. 4)but the fortuitous co-production of insulating LiCl with the alloyrenders the composite film resistive. Such resistance, as exhibited bysome amorphous carbons and graphenes which have been exploredas protective layers35–37, is sufficient to establish an electric fieldacross the film, and provides a driving force for lithium to migratethrough the layer, as we will show in the next sections.

Physical characterization of the alloy protection layersThe composition of the protection layer was determined by X-rayphotoelectron spectroscopy (XPS) analysis, using In-protected Lias the example. The In 3d spectrum of a reference material,bulk Li13In3, exhibits a main peak at 442.2 eV (Fig. 2a, top),assigned to Inδ− in Li13In3 (ref. 38). The same Li13In3 alloyphase is clearly formed on the surface of the InCl3-treatedlithium. Some In (443.2 eV) (and its oxide, (444.5 eV; ref. 39))are also present (Fig. 2a, centre), but their fraction dramaticallydecreased after Ar sputtering, indicating their existence only onthe upper 10 nm (Fig. 2a, bottom). This means that the self-alloying process (equation (2)) is a kinetically limited process andresults in a gradient of alloy concentration. The presence of LiClin the composite layer is evidenced by its characteristic signal at199.8 eV in the Cl 2p spectrum (Fig. 2b, top; see survey spectrum,Supplementary Fig. 5a). The Li fraction (as LiCl) is about 16 at% inthe uppermost portion of the layer that is probed by XPS; that is,a weight fraction of 19% (Li 1s XPS, Supplementary Fig. 5b). TheLi–Zn protection layer contains 12wt% LiCl. LiCl is also evidentin the Li–In film after Ar sputtering for 10min (Fig. 2b, bottom)—indicating that it penetrates well below the upper surface, in accordwith the bulk probe afforded by the EDS (Fig. 1d) that shows anoverall 14.5 wt% Li(Cl) content in the layer.

This insulating halide component results in resistive protectingfilms, as indicated by direct current–voltage measurements of the

2

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

NATURE ENERGY 2, 17119 (2017) | DOI: 10.1038/nenergy.2017.119 | www.nature.com/natureenergy

NATURE ENERGY ARTICLES

455 450Binding energy (eV) Binding energy (eV)

445 440

455 450Binding energy (eV) Binding energy (eV)

445 440

Inte

nsity

(a.u

.)

Inte

nsity

(a.u

.)In

tens

ity (a

.u.)

In2O3In0

In0

Li13In3

Li13In3

Li13In3

10 min sputter

10 min sputter

5 min sputter

5 min sputter

10 min sputter

10 min sputter

a

60 58 56 54 52 50

Li

Li2O

Li3CO2/LiOH

LiCl

d

205 200 195 190

b

c

1

Cl 2p

Li 1s

In 3d

In 3d

Figure 2 | XPS analysis of the film-protected lithiummetal before and after Li plating. Data are shown as open circle markers, and the fits in solid colour.a, High-resolution In 3d spectra of cold-pressed bulk Li–In alloy (as a reference, top), film-protected Li before (centre) and after 10 min Ar sputtering(bottom), respectively. b, Cl 2p spectra of the Li–In film-protected Li before (top) and after (bottom) 10 min sputtering. c,d, In 3d and Li 1s spectra,respectively, of protected Li with 2 mAh cm−2 Li plating, before sputtering (top), after 5 min sputtering (centre), and after 10 min sputtering (bottom).The chemical environments of the individual peaks are as shown. Details of the fitting procedure can be found in the Methods.

In- and Bi-based films as representative examples (Fig. 3). Theelectronic resistivity of the Li13In3 and Li3Bi protection layers is(0.5± 0.2)× 104� cm and (2.8 ± 0.9) × 104� cm, respectively(that is, σ e = 2 × 10−4 and 3.6 × 10−5 S cm−1; see SupplementaryFig. 6 for details). The resistance values are two–three orders ofmagnitude higher than those of amorphous carbon nanospherefilms (1.3 × 101� cm) that induce Li plating underneath thefilm35. The polarization I–V curves of symmetric cells also showthat the Li|Li13In3 electrode exhibits a much smaller I–V slopethan the Li electrode (20± 2mAV−1 versus 4.7 ± 0.1mAV−1,Supplementary Fig. 6a). This supports electron transfer impedancefrom the protection layer, consistent with the electrochemicalimpedance measurements (Supplementary Fig. 6b).

Probing lithium deposition under the alloy layerWe employed a combination of SEM and optical microscopy toinvestigate Li electrodeposition with the alloy protective layers(Fig. 4). Li13In3|Li and LiZn|Li were used as the representativefoils. Figure 4a shows a top-surface SEM view of the dendriticmorphology of the electrodeposited lithium on Li metal. Thesehigh-surface-area dendrites exacerbate reaction with the electrolyte,

0 20 40 60 80

0.0

0.2

0.4OCV 5 mA

Vol

tage

(V)

Time (s)

Li

Li3Bi/Li/Li3Bi

Li13In3/Li/Li13In3

Figure 3 | Measurements of d.c. conductivity of lithiummetal protectedwith the alloy-composite films using blocking electrodes. Voltageresponse of pristine lithium (red curve), Li3Bi|Li|Li3Bi (blue curve) andLi13In3|Li|Li13In3 (olive curve) electrodes to an applied current of 5 mA, fromwhich the resistance of the films are calculated (see Supplementary Fig. 6).

leading to lithium pulverization after 100 cycles (200 h) of repeatedplating/stripping (Supplementary Fig. 7a)10. In contrast, thecomposite alloy-protected lithium anodes show more featureless,

NATURE ENERGY 2, 17119 (2017) | DOI: 10.1038/nenergy.2017.119 | www.nature.com/natureenergy

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

3

ARTICLES NATURE ENERGY

Before cycling 71 cycles 117 cycles 131 cycles 185 cycles 220 cycles

ElectrolyteElectrolyte

Li metal

f

g

a b

c

e

d

2 µm

Li foil

10 µm

Li foil

Li foil

Deposited Li

Alloy composite layer

Alloy composite layer

20 µm

20 µm

100 µm

Li13 In3 |Li

Alloy composite layer

2 µm

Figure 4 | Scanning electron microscopy and optical microscopy study of the alloy-protected lithiummetal. a,b, Surface view of fresh lithium metal (a)and the composite Li13In3|Li anode plated with 2 mAh cm−2 of Li (b). c–e, Cross-sectional image of a composite Li13In3|Li foil before Li plating (c),cross-sectional image for composite Li13In3|Li plated with 2 mAh cm−2 of Li (d) and the corresponding image in backscattered electron mode (e). The reddotted line in d,e outlines the electrodeposited Li. f,g, Operando optical microscopy images of the front surfaces of two electrodes, fresh lithium (f) andcomposite Li13In3|Li (g) in a symmetric transparent cell, recorded at the specified number of plating/stripping cycles. The cross-section edge is slightlylifted in places due to stresses induced in the cutting procedure. The scale bar in f–g is the same, as indicated.

non-dendritic lithium deposition (Fig. 4b). Corrosion of the surfaceis suppressed, as seen in the cross-sectional view (SupplementaryFig. 7b); similar data are shown for LiZn|Li in SupplementaryFig. 7c,d. Because the alloy protection layers are compositionallyinvariant on cycling, they do not undergo any deleterious volumechanges. The alloy layers are also expected to be less reactive withthe electrolyte than Li because of their lower chemical potential.

The cross-sectional view of the Li13In3|Li electrode before(Fig. 4c) and after electrodeposition with 2mAh cm−2 of lithium(Fig. 4d,e) shows that the lithium is plated under the composite alloylayer. The plated lithium layer appears bright in secondary electrondetection mode (Fig. 4d), but dark in backscattered mode (Fig. 4e),allowing it to be distinguished from the alloy. The thickness of thedeposited metal—about 8–10 µm—is in accord with that expectedfor the quantity of Li deposited.

To further illustrate the suppression of dendrite growth by thecomposite film, we monitored Li plating/stripping in a sealedtransparent cell with an optical microscope equipped with a digitalcamera (details shown in Supplementary Fig. 8a,b). Using Li13In3|Lias an example, Li foil as the counter-electrode, and an electrolyte-soaked separator, an alternating current (4mA cm−2, 10min) wasapplied to the electrodes. Figure 4f,g shows photos taken at

different plating/stripping cycles. Before cycling, the surfaces ofboth electrodes were smooth. After about 100 cycles, protrusionsstart to appear along the edge of the Li electrode: evidence ofinhomogeneous Li deposition (Fig. 4f). These protrusions nucleate,and aggregate to form high-surface-area dendrites that are verypronounced at 220 cycles, as also demonstrated by the porousmorphology of the lithium electrode (Supplementary Fig. 7a). Incontrast, the alloy-protected Li exhibits smooth lithium depositionwith no sign of dendrites or pulverization on cycling (Fig. 4g).

Figure 2c,d reports the XPS depth profiling of the Li13In3|Lielectrode immediately after being plated with 2mAh cm−2 of Li.The absence of an In signal (Fig. 2c, top) and the presence oflithium components (Fig. 2d, top) implies that the very uppermostlayer is covered by an interphase consisting of a LiOH/Li2CO3(55.0 eV)/Li2O (53.7 eV) passivation film on electrodeposited Li(52.8 eV; ref. 40) (see Supplementary Fig. 9a,b). However, afteronly 5min of Ar-ion sputtering, the In 3d spectrum shows thepredominance of Li13In3 with aminor contribution from In0 (Fig. 2c,centre); furthermore, no interphase composition is now apparentin the Li 1s spectrum (Fig. 2d, centre). Thus the interphase is onlyabout 10 nm in depth (the amount removed based on a sputteringrate of 1–2 nmmin−1; ref. 40), which is far less than 10 µm, the

4

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

NATURE ENERGY 2, 17119 (2017) | DOI: 10.1038/nenergy.2017.119 | www.nature.com/natureenergy

NATURE ENERGY ARTICLESLi metal

Li+ ion

a b Alloy/LiCl-protected Li

ΔE

Figure 5 | Schematic depicting the function of the alloy-protected lithium foil. a, Unprotected lithium foil is subject to dendrite growth. b, Alloy/LiCl-protected Li foil allows for a significant potential gradient across the protection layer (1E), and good Li+ ion di�usion through the∼10 µm film whichenhances charge transfer at the interface. The two factors act in concert to inhibit dendrite growth. The shaded olive region in b is the interface; the alloyphase and the amorphous/nanocrystalline LiCl phase are represented in orange and light blue, respectively.

approximate thickness of lithium deposited at 2mAh cm−2. Itconfirms that the vast majority of the plated lithium atoms havepassed into or through the protection layer, as demonstrated inFig. 4d. The corresponding Li 1s spectrum (Fig. 2d, centre, andSupplementary Fig. 9c) clearly shows the Li13In3 phase at 55.4 eV,along with LiCl. After a 10min sputter, the ratio of Li13In3/LiClincreases, indicating the alloy composite becomes richer in alloywith increasing depth (Fig. 2d, bottom). The XPS spectra also showthe fraction of residual In0 in the protection layer is greatly reducedafter lithium plating (Fig. 2c, centre) compared to pristine Li13In3|Li(Fig. 2a, centre). This demonstrates that the In at the surface alloyswith the plated lithium on deposition until the composition Li13In3is reached.

Advantages of the alloy-composite protective layersThe uniqueness of the protection layers that allow electrodepositionof Li underneath the film relies on two factors, as summarized inFig. 5. First, the increase in resistance conveyed by the insulatingLiCl component is critical to establishing the necessary potentialgradient across the film to drive Li+ diffusion through the layer.Although LiCl is not a good ionic conductor, Li migration takesplace through the alloy phase. This is vital to plate Li underneath thefilm. The strategy differs fromutilizing a high-surface-area substrateon which non-dendritic Li plates at low local currents22–24, where itscontact with the electrolyte leads to potential reactivity. Secondly,while the insulating component establishes the potential gradient,the high diffusion coefficients of the alloy phases enables lithiumdiffusion. DLi is in the vicinity of 10−8–10−6 cm2 s−1 (refs 32–34)in bulk phases—which far exceeds that of Li in bulk lithiummetal (5.69× 10−11 cm2 s−1; ref. 41) or most other conventionalLi-ion electrode materials (typically <10−10 cm2 s−1; ref. 42). Infact, the impact of fast lithium diffusion on alleviating dendriteformation was highlighted by Huggins almost three decades ago43.We anticipate DLi to be even higher in our room-temperature-formed alloy phases, which are likely to contain significantLi vacancies.

Stripping/plating measurements were carried out in symmetricLi–Li cells to validate the role of the composite alloy layer for allmaterials. A practical capacity of lithium at a high current density(2mAh cm−2 at a current density of 2mA cm−2) was repeatedlydeposited/dissolved in each discharge/charge cycle. Figure 6a–cshows the voltage profiles as a function of time. Lithium metalshort circuits due to dendrites within 200 h (100 cycles), signalledby voltage fluctuations (Fig. 6a). In contrast, the alloy-protectedlithium foils show a very stable voltage profile (Fig. 6b,c). Both

Li13In3|Li and LiZn|Li cells cycle for more than 1,000 h withoutany sign of a short circuit. Cells comprised of Li3Bi|Li and Li3As|Lishow similar performance: they can be repeatedly plated/strippedfor over 1,200 and 1,400 h, respectively (Supplementary Fig. 10). Theover fivefold improvement in cycling life proves that the protectionlayer is effective in stabilizing lithium electrodeposition. Thesecomposite alloy-protected lithium anodes also exhibit stabilizedperformance in a classic Li-ion electrolyte (1M LiPF6 in EC-DMC;Supplementary Fig. 11). To our knowledge, this is the bestperformance of Li plating/stripping in symmetric cells at a highcurrent density/surface capacity compared to other studies thathave employed electrolyte additives13,15, artificial SEIs18,19 or solidelectrolytes25,26 for protection.

The composite alloy-protected lithium exhibits a lower averageoverpotential compared to lithium itself (65 versus 90mV,Fig. 6a–c), indicating better charge transport through the protectivelayer compared to the native SEI formed on lithium metal. Asour impedance measurements demonstrate, lithium undergoesside reactions with the electrolyte44, resulting in the build-up ofa poorly ionically conductive SEI/passivation layer that leads toever-increasing charge transfer resistance (Supplementary Fig. 12a).The protected lithium foil is much more inert to reaction withthe electrolyte: the resistance is much lower, and remains almostconstant (Supplementary Fig. 12b). This implies favourable chargetransfer/atom diffusivity in the layer, consistent with the lowerpolarization observed in Fig. 6.

The performance of the alloy-protected lithium as an anode wasfurther investigated in a full cell using zero-strain Li4Ti5O12 (LTO)45as the cathode (Fig. 6d–f). Cells were examined at a high rate of 5C,corresponding to a current density of 2.5mA cm−2. In the cell witha non-protected lithium anode, the overall overpotential becamesignificant after 100 cycles (Fig. 6e), attributed to increasingly severepassivation/pulverization of the lithium surface, as shown in Fig. 4.The discharge capacity also gradually decreased, and the cell failedafter 600 cycles. In sharp contrast, cells with the alloy-protectedlithium anode show excellent cycling life, delivering stable capacityfor 1,500 cycles without significant degradation. No significantoverpotential increase was observed, as illustrated by the Li13In3|Lianode (Fig. 6f).

ConclusionsIn summary, we have demonstrated that Li-based surface compositealloy films (Li13In3, LiZn, Li3Bi, or Li3As)/LiCl, generated by asimple surface chemistry in solution, provide a stable interface forlithium electrodeposition. These alloys have much higher lithium

NATURE ENERGY 2, 17119 (2017) | DOI: 10.1038/nenergy.2017.119 | www.nature.com/natureenergy

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

5

ARTICLES NATURE ENERGY

0 50 100 150 200 250

−0.4

−0.2

0.0

0.2

0.4

Vol

tage

(V)

Vol

tage

(V)

Vol

tage

(V)

Vol

tage

(V)

Vol

tage

(V)

Vol

tage

(V)

a

0 200 400 600 800 1,000 1,200−0.2

0.0

0.2

0.4

0 5−0.2

0.0

0.2

−0.2

0.0

0.2

−0.2

0.0

0.2

600 605 1,000 1,005

0 5 600 605 1,000 1,005

b

0 200 400 600 800 1,000−0.2

0.0

0.2

0.4

Time (h)

c

0 300 600 900 1,200 1,5000

40

80

120

160

200d

1.0

1.5

2.0

2.5

646−650396−40096−10046−501−5

e

1.0

1.5

2.0

2.5

Cycle number

1,496−1,500996−1,000496−50096−1001−5

f

LiZn|Li

Li13In3|LiLi3As|Li

Fresh Li

Li3Bi|LiV

olta

ge (V

)

−0.2

0.0

0.2

−0.2

0.0

0.2

−0.2

0.0

0.2

Capa

city

(mA

h g−1

)

Figure 6 | Electrochemical performance of protected lithiummetal. Lithium stripping/plating in symmetric cells at a current density of 2 mA cm−2 for 1 hin 1M LiTFSI 1:1 DOL:DME electrolyte. a–c, Voltage profile of fresh lithium metal anode (a), Li13In3|Li anode (b) and LiZn|Li anode (c). The red dotted circlein a shows the instability due to the onset of dendrite growth. The inserts in b and c are the high-resolution voltage profiles at specific times. d, Cyclingperformance of LTO electrodes paired with all four alloy films as indicated. e,f, Voltage profiles of the LTO cells during cycling, with a Li metal anode (e) andwith a representative Li13In3|Li anode (f). The dotted red lines in e highlight the increased polarization in the cell with unprotected Li. The areal loading ofthe LTO electrodes was about 3 mg cm−2, and cells were examined at a 5C rate.

diffusion coefficients than lithium metal. In addition, the resistivenature of the film provided by the insulating LiCl component inthe alloy layer inhibits reduction of lithium ions on the surface,and generates a driving force for deposition of lithium under theprotective alloy surface layer. The alloys were prepared by a facilemetathesis reaction in solution followed by self-alloying, and thefilms enabled suppression of dendrite growth on repeated lithiumstripping/plating deposition in a symmetric cell over 1,400 h ata practical current density of 2mA cm−2. Ultra-long cycling lifewas realized by such protected lithium metal anodes paired witha Li4Ti5O12 cathode. Our finding is significant since it providesa simple and inexpensive strategy to stabilize lithium metal,which fundamentally changes its electrodeposition behaviour fromdendritic to non-dendritic, and opens up a new and promisingresearch direction. Given the existence of numerous availablelithium alloys along with other materials and their rich surfacescience chemistry, explorations combined with computationalapproaches may yield more practical lithium metal batteries.

MethodsPreparation of the protected lithium electrode. Electrode preparation wascarried out in an argon-filled glove box with <1 ppm oxygen and moisture.Lithium metal foil (99.9%, Aldrich) was polished until the surface wasextremely shiny. After polishing, the lithium foil was immersed in a 0.167MMClx solution in tetrahydrofuran (THF) for 20 s (M = As, In, Zn or Bi). Uponremoval from the THF solution, the excess liquid on the treated lithium foil wascarefully removed. The foil was then rinsed with THF, dried under vacuum fortwo days at room temperature and cut into 11mm circles for investigation inelectrochemical cells.

Preparation of the Li13In3 bulk phase. Indium foil and lithium foil were polishedprior to use. The desired weight of foils was measured and stacked together. Thestacked foil was pressed with 4 tons of pressure for 30min in a 12-mm-diameterdie. All steps were conducted in an Ar-filled glovebox.

Materials characterization. XRD measurements were carried out on a BrukerD8-Advance powder X-ray diffractometer operating at 40 kV and 30mA, usingCu Kα radiation (λ = 0.15405 nm). SEM studies were carried out on a Zeiss Ultrafield-emission SEM instrument. Electrodes were gently washed with THF toremove the electrolyte salt and dried under vacuum prior to the SEMcharacterization. The XPS samples were sealed in a vial before being quicklytransferred to the chamber of an ultrahigh vacuum Imaging XPS Microprobesystem for analysis (Thermo VG Scientific ESCALab 250). All spectra were fittedwith Gaussian–Lorentzian functions and a Shirley-type background usingCasaXPS software. The binding energies were calibrated using the C 1s peakat 285.0 eV.

Electrochemical measurements. Electrochemical studies were performed in 2325coin cells. For the impedance and lithium plating/stripping studies, symmetriccells (fresh lithium on each side or protected lithium foil on each side) wereassembled with 40 µl of 1M lithium bis(trifluoromethane)sulfonimide (LiTFSI) indioxolane/dimethoxyethane (DOL/DME) (1:1 vol) as the electrolyte. We used aprotocol of 1 h of stripping followed by 1 h of plating with a current density of2mA cm−2. To investigate the performance of the protection layer compared tothe lithium metal anode, cells were fabricated with Li4Ti5O12 (LTO) as thecathode. The LTO electrodes were prepared by casting a dimethylformamide(DMF) slurry containing Li4Ti5O12 (Sigma-Aldrich), Super P andpoly(vinylidene difluoride) (PVDF) in a weight ratio of 8:1:1 onto carbon-coatedAl foil. The cathodes were cut into disks with a diameter of 11mm and dried at60 ◦C prior to use. The areal loading of LTO was about 3mg cm−2. Approximately40 µl of 1M LiTFSI in DOL/DME (1:1 vol) was used as the electrolyte.Electrochemical impedance measurements were conducted at room temperatureusing a VMP-3 with a frequency range of 0.1Hz to 100 kHz. The cycling of theLTO cells was conducted on an Arbin cycler with a voltage window between 1and 2.5V. The transparent cell was monitored using a Leica DM 2700Mmicroscope. Electronic resistivity of the protection layers was measured byexamining the voltage response to a direct current on the cells with the pristineor protected Li foils sandwiched between two stainless steel blocking electrodes(see further details in the Supplementary Information). Polarization curves andelectrochemical impedance of pristine Li and Li|Li13In3 electrodes were recordedin three-electrode symmetric Swagelok cells using lithium metal referenceelectrodes, in an electrolyte comprised of 1M LiTFSI in DOL/DME.

6

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

NATURE ENERGY 2, 17119 (2017) | DOI: 10.1038/nenergy.2017.119 | www.nature.com/natureenergy

NATURE ENERGY ARTICLESThe electrolyte volume was 200 µl. Measurements were conducted immediatelyafter cell assembly, to avoid extensive SEI formation.

Data availability. The data that support the plots within this paper and otherfindings of this study are available from the corresponding author uponreasonable request.

Received 27 April 2017; accepted 18 June 2017;published 31 July 2017

References1. Armand, M. & Tarascon, J. M. Building better batteries. Nature 451,

652–657 (2008).2. Bruce, P. G., Freunberger, S. A., Hardwick, L. J. & Tarascon, J. M. Li–O2 and

Li–S batteries with high energy storage. Nat. Mater. 11, 19–29 (2012).3. Chen, Y., Freunberger, S. A., Peng, Z., Fontaine, O. & Bruce, P. G. Charging a

Li–O2 battery using a redox mediator. Nat. Chem. 5, 489–494 (2013).4. Lu, Y. C. et al. Lithium–oxygen batteries: bridging mechanistic understanding

and battery performance. Energy Environ. Sci. 6, 750–768 (2013).5. Pang, Q., Liang, X., Kwok, C. Y. & Nazar, L. F. Advances in lithium–sulfur

batteries based on multifunctional cathodes and electrolytes. Nat. Energy 1,16132 (2016).

6. Xu, W. et al. Lithium metal anodes for rechargeable batteries. Energy Environ.Sci. 7, 513–537 (2014).

7. Busche, M. R. et al. Dynamic formation of a solid–liquid electrolyteinterphase and its consequences for hybrid-battery concepts. Nat. Chem. 8,426–434 (2016).

8. Gallagher, K. G. et al. Quantifying the promise of lithium-air batteries forelectric vehicles. Energy Environ. Sci. 7, 1555–1563 (2014).

9. Harry, K. J., Hallinan, D. T., Parkinson, D. Y., MacDowell, A. A. & Balsara, N. P.Detection of subsurface structures underneath dendrites formed on cycledlithium metal electrodes. Nat. Mater. 13, 69–73 (2014).

10. Aurbach, D. et al. Attempts to improve the behavior of Li electrodes inrechargeable lithium batteries. J. Electrochem. Soc. 149,A1267–A1277 (2002).

11. Qian, J. et al.High rate and stable cycling of lithium metal anode. Nat.Commun. 6, 6362 (2015).

12. Wang, H. et al. A reversible dendrite-free high-areal-capacity lithium metalelectrode. Nat. Commun. 8, 15106 (2017).

13. Ding, F. et al. Dendrite-free lithium deposition via self-healing electrostaticshield mechanism. J. Am. Chem. Soc. 135, 4450–4456 (2013).

14. Gofer, Y., Ben-Zion, M. & Aurbach, D. Solutions of LiAsF in 1,3-dioxolane forsecondary lithium batteries. J. Power Sources 39, 163–178 (1992).

15. Lu, Y., Tu, Z. & Archer, L. A. Stable lithium electrodeposition in liquid andnanoporous solid electrolytes. Nat. Mater. 13, 961–969 (2014).

16. Li, W. et al. The synergetic effect of lithium polysulfide and lithium nitrate toprevent lithium dendrite growth. Nat. Commun. 6, 7436 (2015).

17. Suo, L., Hu, Y. S., Li, H., Armand, M. & Chen, L. A new class of solvent-in-saltelectrolyte for high-energy rechargeable metallic lithium batteries.Nat. Commun. 4, 2513 (2013).

18. Zheng, G. et al. Interconnected hollow carbon nanospheres for stable lithiummetal anodes. Nat. Nanotech. 9, 618–623 (2014).

19. Kozen, A. C. et al. Next-generation lithium metal anode engineering via atomiclayer deposition. ACS Nano 9, 5884–5892 (2015).

20. Bucur, C. B., Lita, A., Osada, N. & Muldoon, J. A. Soft, multilayeredlithium–electrolyte interface. Energy Environ. Sci. 9, 112–118 (2016).

21. Umeda, G. et al. Protection of lithium metal surfaces using tetraethoxysilane.J. Mater. Chem. 21, 1593–1599 (2011).

22. Yang, C., Yin, Y. X., Zhang, S. F., Liu, N. W. & Guo, Y. Accommodating lithiuminto 3D current collectors with a submicron skeleton towards long-life lithiummetal anodes. Nat. Commun. 6, 8058 (2015).

23. Zhang, R. et al. Conductive nanostructured scaffolds render low localcurrent density to inhibit lithium dendrite growth. Adv. Mater. 28,2155–2162 (2015).

24. Lin, D. et al. Layered reduced graphene oxide with nanoscale interlayer gaps asa stable host for lithium metal anodes. Nat. Nanotech. 11, 626–632 (2016).

25. Armand, M. B., Duclot, M. J. & Rigaud, P. Polymer solid electrolytes: stabilitydomain. Solid State Ion. 3–4, 429–430 (1981).

26. Zhou, W. et al. Plating a dendrite-free lithium anode with apolymer/ceramic/polymer sandwich electrolyte. J. Am. Chem. Soc. 138,9385–9388 (2016).

27. Khurana, R., Schaefer, J. L., Archer, L. A. & Coates, G. W. Suppression oflithium dendrite growth using cross-linked polyethylene/poly(ethylene oxide)electrolytes: a new approach for practical lithium-metal polymer batteries.J. Am. Chem. Soc. 136, 7395–7402 (2014).

28. Wenzel, S. et al. Direct observation of the interfacial instability of the fastionic conductor Li10GeP2S12 at the lithium metal anode. Nat. Chem. 28,2400–2407 (2016).

29. Ren, Y., Shen, Y., Lin, Y. & Nan, C. Direct observation of lithium dendritesinside garnet-type lithium-ion solid electrolyte. Electrochem. Commun. 57,27–30 (2015).

30. Richardson, T. J. & Chen, G. Solid solution lithium alloy cermet anodes.J. Power Sources 174, 810–812 (2007).

31. Stark, J. K., Ding, Y. & Kohl, P. A. Dendrite-free electrodeposition andreoxidation of lithium–sodium alloy for metal-anode battery. J. Electrochem.Soc. 158, A1100–A1105 (2011).

32. Shi, Z., Liu, M. & Gole, J. L. Electrochemical properties of Li–Zn alloyelectrodes prepared by kinetically controlled vapor deposition for lithiumbatteries. Electrochem. Sol. State Lett. 3, 312–315 (2000).

33. Hiratani, M., Miyauchi, K. & Kudo, T. Effect of a lithium alloy layer insertedbetween a lithium anode and a solid electrolyte. Solid State Ion. 28–30,1406–1410 (1988).

34. Hiratani, M. et al. Solid state lithium battery. US patent 4,645,726 (1987).35. Zheng, G. et al. Interconnected hollow carbon nanospheres for stable lithium

metal anodes. Nat. Nanotech. 9, 618–623 (2014).36. Kim, J. S., Kim, D. W., Jung, H. T. & Choi, J. W. Controlled lithium dendrite

growth by a synergistic effect of multilayered graphene coating and anelectrolyte additive. Chem. Mater. 27, 2780–2787 (2015).

37. Yan, K. et al. Ultrathin two-dimensional atomic crystals as stableinterfacial layer for improvement of lithium metal anode. Nano Lett. 14,6016–6022 (2014).

38. Web, S. A., Baggetto, L., Bridges, C. A. & Veith, G. M. The electrochemicalreactions of pure indium with Li and Na: anomalous electrolyte decomposition,benefits of FEC additive, phase transitions and electrode performance. J. PowerSources 248, 1105–1117 (2014).

39. Hewitt, R. W. &Winograd, N. Oxidation of polycrystalline indium studied byX-ray photoelectron spectroscopy and static secondary ion mass spectroscopy.J. Appl. Phys. 51, 2620–2624 (1980).

40. Kanamura, K., Tamura, H., Shiraishi, S. & Takehara, Z. XPS analysis of lithiumsurfaces following immersion in various solvents containing LiBF4.J. Electrochem. Soc. 142, 340–347 (1995).

41. Dologlou, E. Self diffusion in solid lithium. Glass Phys. Chem. 36,570–574 (2010).

42. Park, M., Zhang, X., Chung, M., Less, G. B. & Sastry, A. M. A reviewof conduction phenomena in Li-ion batteries. J. Power Sources 195,7904–7929 (2010).

43. Huggins, R. A. Polyphase alloys as rechargeable electrodes in advanced batterysystems. J. Power Sources 22, 341–350 (1988).

44. Aurbach, D., Zinigrad, E., Cohen, Y. & Teller, H. A short review of failuremechanisms of lithium metal and lithiated graphite anodes in liquid electrolytesolutions. Solid State Ion. 148, 405–416 (2002).

45. Ohzuku, T., Ueda, A. & Yamamoto, N. Zero-strain insertion material ofLi[Li1/3Ti5/3]O4 for rechargeable lithium cells. J. Electrochem. Soc. 142,1431–1435 (1995).

AcknowledgementsThis research was supported by the BASF International Scientific Network forElectrochemistry and Batteries. L.F.N. also thanks NSERC for generous support via theirCanada Research Chair, and Discovery Grant programs. We greatly appreciate helpfuldiscussions with K. Zavadil, P. Bruce and J. Janek.

Author contributionsX.L. and L.F.N. designed the experimental work. X.L. performed all the physicalmeasurements on the films and the electrochemistry on symmetric and full cells. I.R.K.and Q.P. carried out the resistivity measurements of the protective layers. M.S.S., H.H.and X.S. participated in the discussion of the data. X.L., I.R.K. and L.F.N. wrote themanuscript. L.F.N. directed the work.

Additional informationSupplementary information is available for this paper.Reprints and permissions information is available at www.nature.com/reprints.Correspondence and requests for materials should be addressed to L.F.N.How to cite this article: Liang, X. et al. A facile surface chemistry route to a stabilizedlithium metal anode. Nat. Energy 2, 17119 (2017).Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims inpublished maps and institutional affiliations.

Competing interestsThe authors declare no competing financial interests.

NATURE ENERGY 2, 17119 (2017) | DOI: 10.1038/nenergy.2017.119 | www.nature.com/natureenergy

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

7