Elastic and Li-ion–percolating hybrid membranestabilizes Li metal platingQuan Panga,b, Laidong Zhoua,b, and Linda F. Nazara,b,1

aDepartment of Chemistry, University of Waterloo, Waterloo, ON N2L 3G1, Canada; and bWaterloo Institute for Nanotechnology, University of Waterloo,Waterloo, ON N2L 3G1, Canada

Edited by Galen D. Stucky, University of California, Santa Barbara, CA, and approved October 11, 2018 (received for review June 8, 2018)

Lithium metal batteries are capable of revolutionizing the batterymarketplace for electrical vehicles, owing to the high capacity andlow voltage offered by Li metal. Current exploitation of Li metalelectrodes, however, is plagued by their exhaustive parasiticreactions with liquid electrolytes and dendritic growth, which poseconcerns to both cell performance and safety. We demonstrate thata hybrid membrane, both elastic and Li+-ion percolating, can stabi-lize Li plating/strippingwith high Coulombic efficiency. The compactpacking of a Li+ solid electrolyte phase offers percolated Li+-con-ducting channels and the consequent infiltration of an elastic poly-mer endows membrane flexibility to accommodate volumechanges. The protected electrode allows Li plating with 95.8% effi-ciency for 200 cycles and stable operation of an LTOjLi cell for 2,000cycles. This rationally structured membrane represents an interfaceengineering approach toward stabilized Li metal electrodes.

Li-ion battery | lithium metal anode | Li metal protective flexiblemembrane | electrolyte-impermeable | Li-ion percolation membrane

The development of high-energy-density and low-cost batterysystems appears to be a major limiting factor toward the real-

world commercialization of electric cars (1). Traditional lithium-ion technology that relies on intercalation-type graphite anodesand metal oxide cathodes is approaching its theoretical limit inenergy density and cost (2, 3). New chemistry beyond in-tercalation is desirable. Lithium metal batteries, such as Li–O2and Li–S systems, promise lower-cost and higher-energy density(4, 5). This benefits from the coupling of high-capacity cathodes(O2/S) with a lithium metal anode, which exhibits an order ofmagnitude higher capacity than graphite (3,860 mAh g−1 vs.370 mAh g−1) (6, 7). In the past decade, much effort has beendevoted to extending the lifetime of cathodes; however, intrinsicchallenges with Li metal are often not apparent, owing to therelative low areal current density and excess electrolyte generallyused in such systems (8, 9).It is vital to stabilize the Li metal electrode to realize practical

lithium metal batteries. Lithium has the lowest reduction po-tential among alkali/alkali-earth metals and their alloys, which,although allowing a high-voltage full cell, causes fatal degrada-tion issues because Li reacts with almost all types of liquidelectrolytes (10, 11). Unlike the conventional intercalationgraphite anode that forms a stable solid electrolyte interphase(SEI) in carbonate electrolytes, plating of hostless Li results in adendritic morphology with a thick SEI (12, 13). The interplay ofparasitic reactions and dendritic growth gives rise to iterative SEIbreakdown, leading to “dead lithium” and potential cell failurebecause of electrolyte depletion (2, 3, 6). The exact mechanismof dendritic growth is still in debate, but two commonly acceptedtheories rely either on a modified Chazalviel model or on non-uniformity of the SEI. Chazalviel’s model states that an ionconcentration gradient exists over the thickness of the electrolyteunder polarization, which determines a limiting currentand Sand’s time, beyond which the dendrite forms (14). Thelow Li+-ion transference number of liquid electrolytes gives riseto anion depletion at the interface and establishes a strongelectric field that drives dendritic growth (15). The other

theory is complementary and emphasizes the nonuniformity ofthe as-formed SEI, which causes uneven Li+-ion flux and nu-cleation (6, 2).Based on understanding from these two theories, recent ef-

forts have made much progress in suppressing dendrites. Oneapproach is to tune the electrolyte compositions to create astable SEI. This has utilized stable glyme solvents/salts (16–18); aLiF additive (19); and a high salt concentration (20, 21). Effortsin reducing the establishment of an electric field are based onelectrolytes with high transference number (22–25). Accommo-dating Li plating into a porous host substrate represents astrategy to reduce electrolyte parasitic reactions (26, 27). An-other avenue resides in fabricating artificial SEI layers withmetal oxides (28), polymers (29–31), solid electrolytes (32–34),and alloys (35) to prevent direct contact with the electrolyte and/or to mechanically suppress Li dendrite formation. It remainschallenging to create a protective membrane with structuralcompatibility that can maintain a percolating Li+-conductingpathway and also low liquid electrolyte permeability. Further-more, it has been proposed that using a high-modulus solidelectrolyte can block the growth of Li dendrites to allow Li metalto be used as a negative electrode in all-solid-state batteries.However, there remain two challenges that prevent wide appli-cations of this approach: dendrite growth along the grain bound-aries and the formation of a nonconducting SEI layer by parasiticreactions (36). Herein we demonstrate that a hybrid membraneprepared by a rationally designed two-step process can overcomethese obstacles in liquid electrolytes (Fig. 1). First, the denselypacked β-Li3PS4 framework ensures a percolated Li+-conducting

Significance

Lithium metal batteries are capable of revolutionizing thebattery marketplace for electrical vehicles. They are, however,plagued by the reactive Li metal–electrolyte interface andgrowth of Li dendrites on cycling, which pose concerns to bothcell performance and safety. We demonstrate that a Li metalnegative electrode can be stabilized by a hybrid inorganic/organic membrane that is grown directly onto it. A solid elec-trolyte phase offers percolated Li+-ion–conducting channelsand the infiltration of an elastic polymer endows membraneflexibility to accommodate volume changes. The protectedelectrode allows Li deposition/stripping with 95.8% efficiencyand stable operation of an electrochemical cell for 2,000 cycles.This rationally structured membrane represents an interfaceengineering approach toward stabilized Li metal electrodes.

Author contributions: Q.P. and L.F.N. designed research; Q.P. and L.Z. performed research;Q.P. and L.Z. analyzed data; and Q.P. and L.F.N. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Published under the PNAS license.1To whom correspondence should be addressed. Email: lfnazar@uwaterloo.ca.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1809187115/-/DCSupplemental.

Published online November 19, 2018.

www.pnas.org/cgi/doi/10.1073/pnas.1809187115 PNAS | December 4, 2018 | vol. 115 | no. 49 | 12389–12394

CHEM

ISTR

Y

Dow

nloa

ded

by g

uest

on

Nov

embe

r 22

, 202

1

pathway without being blocked by the subsequently added polymercomponent, while exhibiting single-ion transport. Based on thespace-charge model established by Chazalviel (14), a transferencenumber of unity can eliminate the space charge in the vicinity of theLi metal surface, which is the driving force for dendrite growth.Second, the infiltrated siloxane polymer, being electrolyte-impermeable (in contrast to polyethylene oxide–Li salt electro-lytes), fills the gap between the percolated particles, preventingdirect electrolyte penetration and offering structure flexibility toaccommodate volume changes during Li plating/stripping (Fig. 1).

Results and DiscussionA Two-Step Strategy for Fabricating the Hybrid Membrane. A pro-tecting membrane for the Li metal electrode needs to be Li+-ionconducting, electron insulating, and electrolyte-proof (25). Aone-pot mixture of a solid electrolyte with a polymer filler cancreate an electrolyte-proof membrane, but the Li+ pathwaybetween particles may be blocked and prevent percolationthroughout the membrane. Therefore, we designed a two-stepfabrication process to resolve this dilemma. The thiophosphateβ-Li3PS4 was selected as the solid electrolyte because of its facilesynthesis in solution (37), high Li+ conductivity (measured to be1.54 × 10−4 S cm−1, SI Appendix, Fig. S1A) and good ductilitythat enables efficient room temperature densification. Poly-dimethylsiloxane (PDMS) is used as the polymer filler owing toits high elasticity, insolubility in carbonate/glyme electrolytes,and rapid solidification. Pure-phase β-Li3PS4, confirmed by its X-ray diffraction pattern, was synthesized following a previouslyreported solution-based procedure (SI Appendix, Fig. S1B) (37).We first coated a binder-free layer of β-Li3PS4 onto Cu foil bydoctor blading, followed by drying in vacuo. The membrane wasthen densified by cold-pressing at 1,000 psi (SI Appendix, Fig.S2A). A diluted PDMS solution along with a curing agent wasthen infiltrated into the layer by spin coating. As opposed to dipcoating, spin coating allows efficient infiltration of the polymerthrough the thickness of the membrane, without surface aggre-gation that would otherwise block the ion-conducting pathway.The cross-linking of PDMS occurred through “addition curing”by attaching the –Si-H group (from the curing agent) to –C=Cbonds (in PDMS). Cross-linking was realized by drying the

membrane at 130 °C. The reaction was confirmed by the signif-icantly reduced intensity of the Fourier transform infraredspectral bands assigned to -C=C and -Si-H bonds (SI Appendix,Fig. S3), compared with the non–cross-linked material. The as-obtained membrane (denoted as LPS-PDMS) was compact andcrack-free as shown in the scanning electron microscopy (SEM)images (Fig. 2A). The particle morphology is preserved free ofaggregated polymer on the surface, which only exhibits the rod-like crystallites of β-Li3PS4 (Fig. 2B). The polymer fills the voidsbetween the thiophosphate deep in the film as demonstrated byenergy-dispersive spectroscopy mapping of the cross-sectionalarea (SI Appendix, Fig. S4). A controllable membrane thick-ness around 2–4 μm was obtained (SI Appendix, Fig. S2B). Themembrane shows high flexibility that can accommodate volumeexpansion/extraction of the lithium (Fig. 1). The Li+ transferencenumber of the composite membrane is measured to be ∼1,confirming single-ion-conducting behavior (SI Appendix, Fig.S5). An optimum concentration of the PDMS solution for spin-coating infiltration (PDMS: toluene = 1:8, wt/vol) is critical toachieve good membrane morphology; a low (1:15) or a high (1:4)concentration leads to only partial infiltration or aggregatedpolymer on the surface (SI Appendix, Fig. S2 C and D).

Li Plates Under the Hybrid Membrane. The Li-plating morphologywith and without protection from the LPS-PDMS membrane wasinvestigated by depositing 4-mA h cm−2 Li on Cu foil from adimethoxyethane/dioxolane electrolyte at a current of 1 mA cm−2.The Li-plated electrodes were retrieved and washed with the samesolvent and analyzed by secondary electron and backscatteredSEM imaging. Secondary electron SEM images of lithium platedon Cu foil exhibit a classical needle-like wormy dendritic structure(Fig. 2C). In contrast, we did not observe any microstructured Lion the top of the LPS-PDMS membrane even after 10 cycles ofplating and stripping (ending with plating), only the membraneitself (Fig. 2D). We did observe some cracking of the hybridmembrane after 10 cycles, where electrolyte contact with thesubstrate can locally occur. However, as shown by surface SEI andelectrochemical studies below, there is only a very limited amountof electrolyte decomposition and the capability of Li to plateunderneath the protective membrane significantly reduces

Fig. 1. A schematic illustration of the designed structure of the hybrid membrane coated on Li or Cu foil. The densely packed β-Li3PS4 phase forms a per-colated Li+-conducting framework, which is then infiltrated with an elastic PDMS polymer. The photographic image showcases the flexibility of the mem-brane. The fast and single-ion-conducting channel allows nondendritic Li plating underneath the membrane.

12390 | www.pnas.org/cgi/doi/10.1073/pnas.1809187115 Pang et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 22

, 202

1

parasitic electrolyte reactions. Nevertheless, further optimizationof the polymer and infiltration process to maintain film integrityis necessary (30, 31). For better comparison, we plated Li on asingle Cu electrode having both a bare and coated surface in thesame cell. The phosphorus energy dispersive spectral mappingcorrelates with the backscattered SEM imaging (SI Appendix, Fig.S6); the brighter field is an LPS-PDMS–covered surface, in-dicating Li is plated under the membrane where it exists. Thecorresponding cross-sectional SEM images in Fig. 2 E and F lendfurther evidence of the role of the protective membrane after 10plate–strip–plate cycles. Backscattered SEM was used for imagingin this case as it highlights the contrast between the lightest-masselement (Li, which appears dark), the LPS-PDMS membrane(which appears brighter owing to its higher mass), and theunderlying Cu foil (which is very bright). Lithium plated on Cufoil shows a classic dendritic wormy morphology (Fig. 2E). Incontrast, plated Li (outlined by the dashed red line) is still de-posited underneath the LPS-PDMS membrane upon continuouscycles of stripping and plating, as distinguished by its dark contrastcompared with the light LPS-PDMS membrane that lies above(Fig. 2E).The electrochemical stability of the LPS-PDMS membrane to

Li plating was evaluated by cyclic voltammetry measurementsusing an LPS-PDMS Cu foil as the working electrode (SI

Appendix, Fig. S7). We did not observe any peak above 0 V (vs. Li+

/Li) that could be attributed to reduction of LPS or PDMS and Liplating occurs at an overpotential of ∼100 mV. X-ray photoelec-tron spectroscopy (XPS) was carried out to investigate the surfacecomposition before and after plating (Fig. 3). The pristine LPS-PDMS membrane exhibits one Li 1s component at 55.7 eV andone P 2p component at 132.8 eV (Fig. 3A) which are ascribed topristine β-Li3PS4 (34, 38). After plating 4-mAh cm−2 Li, the LPS-PDMS–coated electrode shows the same binding energies corre-sponding to β-Li3PS4 in both Li 1s and P 2p spectra (Fig. 3B). Thisagain confirms the reduction stability of the hybrid membrane andLi plating underneath the membrane. A small fraction of addi-tional components is observed, corresponding to the SEI com-ponents Li2CO3/LiOH and LiF, derived from the electrolytedecomposition (39). In contrast, the bare Cu electrode shows thepresence of a large fraction of SEI components, along with a lowerLi 1s binding-energy component that is ascribed to Li metal (Fig.3C) (40). The Si 2p spectra confirm the high chemical stability ofPDMS (SI Appendix, Fig. S8).

Li Plating Achieves Long-Term High Coulombic Efficiency. We eval-uated the Li plating/stripping Coulombic efficiency (CE) and itslong-term stability on the LPS-PDMS–coated Cu electrodes in aglyme-based electrolyte. A fixed amount of Li was plated on the

Fig. 2. Morphological characterization of the hybrid membrane before and after plating lithium. The representative (A) secondary electron SEM image and(B) magnified image of the surface of the fabricated hybrid LPS-PDMS membrane supported on Cu before Li plating. The secondary electron SEM images ofthe surface (C and D) and backscattered electron SEM images of the cross section (E and F) of 4-mA h cm−2 of Li plated on bare Cu (C and E) and onto LPS-PDMS–coated Cu (D and F) after 10 cycles (ending in plating, current density: 1 mA cm−2). The tiny particles evident in the dark lithium layer in F are presumedto be β-Li3PS4 crystallite fragments that were torn from the LPS-PDMS membrane during the sectioning procedure.

Pang et al. PNAS | December 4, 2018 | vol. 115 | no. 49 | 12391

CHEM

ISTR

Y

Dow

nloa

ded

by g

uest

on

Nov

embe

r 22

, 202

1

Cu, followed by stripping at the same current up to 1 V (vs. Li+/Li);the CE of each cycle was calculated by the amount of stripped Lidivided by the amount of plated Li. At a moderate current densityof 1 mA cm−2 and a capacity of 1 mA h cm−2, the bare Cu elec-trode shows a very low average CE of 75.3% (over 100 cycles) anddegrades significantly after 50 cycles. In contrast, the LPS-PDMS–

protected Cu electrode maintains a high average CE of 95.8% over200 cycles (Fig. 4A). An efficiency lower than 100% neverthelessindicates a marginal amount of parasitic reaction of Li with theelectrolyte. We thus measured the swelling effect of cross-linkedPDMS in the electrolyte, which shows that the PDMS absorbs theelectrolyte by about 9 wt % of its own weight (SI Appendix,Materialsand Methods). This implies the presence of some electrolyte contactwith Li metal. Nonetheless, the CE measurement confirms greatlyreduced Li corrosion and electrolyte decomposition by ∼80%compared with bare Cu. Furthermore, the voltage profiles of theLPS-PDMS electrode do not show a noticeable increase in po-larization compared with the bare Cu electrode throughout thecycling (Fig. 4B).To justify the impact of the two-step fabrication, it is necessary

to demonstrate the properties of each individual component(Li3PS4 and PDMS) on its own and to compare with a simplemixture of both. Therefore, Cu foils coated with PDMS only,β-Li3PS4 only, and a one-pot mixture of β-Li3PS4 with PDMS(noted as LPS/PDMS-mix) were fabricated. Their long-termcycling stability and average CE over 100 cycles are shown inSI Appendix, Fig. S9 and Fig. 4C, respectively. We found thatthe PDMS-coated electrode showed only a slight improvementover bare Cu electrodes (CE: 83.6% vs. 75.3%), although theβ-Li3PS4–protected electrode sustained relatively higher CE andlonger life (93.8%, 155 cycles). Nonetheless, the β-Li3PS4–coatedelectrode experienced a low CE (∼90%) conditioning periodduring the first 30 cycles, and unstable CE evolution (standarderror, SD of CE: ±5.9% vs. ±2.4% for the LPS-PDMS electrode,Fig. 4C and SI Appendix, Fig. S9A). In contrast, the LPS/PDMS-mix electrode showed an extremely unstable CE throughout thecycling period, at an average CE of 87.8% (SI Appendix, Fig. S9B).The voltage profile of the mixed-LPS/PDMS electrode also exhibiteda large overpotential, indicating inefficiency in Li+-ion conduc-tion across the membrane (SI Appendix, Fig. S9C). We concludefrom the above that the ion-conducting Li3PS4 component is

Fig. 3. Spectroscopic demonstration of Li-plating behavior with and with-out the membrane. (A) The Li 1s and P 2p XPS spectra of the LPS-PDMS–coated electrode in the pristine state and (B) after Li plating, along with(C) bare Cu foil after Li plating. Note that the P 2p region of the surveyspectrum was used for the Li-plated bare Cu electrode (C). The dotted andthe solid lines are the raw and fitted spectra, respectively. One minute ofsputtering was performed before XPS measurement to remove any super-ficial contaminants.

Fig. 4. The electrochemical behavior of Li plating on the hybrid-membrane–protected electrode with comparison with other systems. (A) The CE evolution ofLi plating/stripping on the bare Cu and LPS-PDMS–coated Cu (Cu–LPS-PDMS) over cycling (1 mA cm−2,1 mA h cm−2); (B) The representative voltage profiles atthe 2nd and 110th cycle. (C) The summary of the average CE over 100 plating/stripping cycles on different current collectors (200 cycles for the LPS-PDMS); theerror bars indicate the SD of the CE. (D) The Li-plating/stripping CE evolution at a higher current density and capacity (2 mA cm−2, 2 mA h cm−2).

12392 | www.pnas.org/cgi/doi/10.1073/pnas.1809187115 Pang et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 22

, 202

1

essential, and has to percolate through the membrane for effi-cient Li underplating; the electrolyte-proof and elastic polymerfiller is necessary to reduce parasitic reactions. We further showthat this membrane can withstand cycling at higher capacity andcurrent density. At a capacity of 2 mA h cm−2, we observe animpressive CE of 95.4% over 100 cycles, far superior to the bareCu electrode (SI Appendix, Fig. S10). At a current density of2 mA cm−2, the LPS-PDMS–protected electrode can sustain anaverage CE of 98.6% over 100 cycles, in comparison with the 50-cycle life of the bare Cu electrode with much lower CE of 75.7%(Fig. 4D). Even at a high current density of 4 mA cm−2, the LPS-PDMS electrode shows a high CE of up to 98%, in compari-son with about 60% for the bare Cu electrode (SI Appendix,Fig. S11).

Protected Li Electrode Enables Highly Stable Full Li Metal Cells. Thesignificance of the LPS-PDMS protective membrane was furtherdemonstrated by coating it onto Li foil (denoted as Li–LPS-PDMS). The same procedure was applied as for coating on Cufoil, except that the β-Li3PS4 layer was pressed at a lower pres-sure (200 psi) to avoid Li metal deformation. We showcaseperformance by assembling Li symmetric cells, and with full cellscoupled with Li4Ti5O12 (LTO) electrode. Here the comparison ismade to bare Li electrode, although a more practical and tech-nologically advanced Li metal electrode needs to be establishedas a benchmark across the community for future studies. Fig. 5Aexhibits the voltage profile of Li–LPS-PDMS symmetric cellsupon long-term cycling at 1 mA cm−2. We observe an extremelystable voltage evolution over 1,000 h of cycling, whereas the cellmade from nonprotected Li experiences intermittent voltagefluctuation indicating local structural changes, and fails eventually

due to a global short circuit (pure resistor behavior) after 270 h(Fig. 5A). It was recently reported that the voltage profile of aLijLi symmetric cell on the initial cycle is correlated to the growth/dissolution of microstructures (41). Such microstructure growthand dissolution give rise to a variation in electrode surface area,and thus a rise and decrease in cell voltage, as observed for thebare Li electrode (Fig. 5A, Inset). In contrast, the Li–LPS-PDMSelectrode neither shows such behavior in the first cycles northroughout cycling (Fig. 5A, Inset). The LTOjLi cells were cycledat a high rate of 5C (1C = 175 mA g−1LTO; areal current,2.1 mA cm−2). As the goal here is not to build a high-voltagebattery, using a zero-strain electrode such as LTO allows us toascribe any cell degradation and failure to the Li metal electrode.The capacity of the Li metal anode was about 20-fold excess inboth cells. The cell using the Li–LPS-PDMS electrode exhibitedvery stable capacity retention over 2,000 cycles, after a condi-tioning process over the first 30 cycles, whereas the bare Li cellshowed lower capacity and rapid fading after 175 cycles (Fig. 5B).The representative voltage profiles at the 2nd and 100th cyclesshowed about 2-fourfold-lower cell polarization for the LPS-PDMS–coated Li electrode (SI Appendix, Fig. S12 A and B). Moreover,the evolution of the round-trip energy efficiency of the whole cell,calculated by the discharge energy over the charge energy, clearlyshows superior efficiency of the LPS-PDMS–protected Li electrode(SI Appendix, Fig. S12C).

ConclusionsWe demonstrate a rationally structured hybrid membrane, in-corporating a single-ion-conducting component and a polymericfiller that stabilizes Li plating. While both components, β-Li3PS4and the PDMS polymer, are important to deliver high ion

Fig. 5. Long-term electrochemical performance of the hybrid-membrane–protected Li electrode. (A) The voltage evolution of LijLi symmetric cells using bareLi or LPS-PDMS–coated Li (Li–LPS-PDMS) (current: 1 mA cm−2, capacity: 1 mA h cm−2); (Insets) The voltage profiles at a specific cycling time. (B) The capacityretention and the Coulombic efficiency of LTO cathodes coupled with bare Li or Li–LPS-PDMS counter electrodes (rate: 5C, 2.1 mA cm−2).

Pang et al. PNAS | December 4, 2018 | vol. 115 | no. 49 | 12393

CHEM

ISTR

Y

Dow

nloa

ded

by g

uest

on

Nov

embe

r 22

, 202

1

conductivity/flexibility and be resistant to electrolyte permeation,we show that a two-step fabrication procedure is necessary tocreate a percolating ion pathway across the membrane. As op-posed to a protection layer prepared by simple mixture of thetwo components, the LPS-PDMS membrane allows stable plating/stripping at an average Coulombic efficiency of 95.8% for200 cycles. A long cycle life of 2,000 cycles was demonstrated foran LTOjLi cell using such protected electrode at high round-tripenergy efficiency. This concept of a hybrid protective membranecan extend to other sulfide- and oxide-based solid electrolytes,although efforts are needed to optimize the materials (e.g.,softness) to ensure efficient particle packing and a percolatingLi+ pathway. Further development of such hybrid membraneswill require improving the structural integrity using polymerswith better adhesion and fluidity (29, 30).

Materials and MethodsThe β-Li3PS4 was prepared following a solution-based approach. A stoi-chiometric amount of Li2S (Alfa Aesar) and P2S5 (Sigma-Aldrich) was addedto tetrahydrofuran (THF) and stirred for 12 h. The as-obtained white pre-cipitate was washed twice with THF before drying in vacuo at 130 °C. For

preparation of the LPS-PDMS hybrid membrane on Cu foil, the foil was firstcoated with a uniform slurry of β-Li3PS4 in THF using a doctor-blade in-strument with automatic thickness calibration. The coated foil was dried invacuo at room temperature overnight, followed by pressing at 1,000 psi for1 min. Subsequently, PDMS (SYLGARD) mixed with the curing agent (wt/wt:10:1) was diluted using toluene at the optimum ratio for spin coating. Theβ-Li3PS4–coated Cu foil was spin coated with the PDMS solution at a spinningspeed of 1,000 rpm (10 s) and 7,000 rpm (2 min). The final LPS-PDMS–pro-tected foil was obtained after drying at 130 °C in vacuo for 6 h. For LPS-PDMS–protected Li, the same procedure was applied except the foil waspressed at a lower pressure of 200 psi. Control samples of coating only onecomponent were prepared by single-step doctor blading of β-Li3PS4 or spincoating of PDMS solution on the Cu foil. The mixed-LPS/PDMS–coated Cu foilwas fabricated by doctor blading a solution of β-Li3PS4 and PDMS (wt/wt, 1:1,20 wt % in toluene) onto the foil. All procedures were conducted inside anargon-filled glovebox with very low levels of O2 (<1 ppm) and H2O (<1 ppm).More details of the methods are provided in SI Appendix.

ACKNOWLEDGMENTS. This research was supported by the BASF Interna-tional Scientific Network for Electrochemistry and Batteries. L.F.N. alsothanks NSERC for generous support via their Canada Research Chair, andDiscovery Grant programs.

1. Needell ZA, McNerney J, Chang MT, Trancik JE (2016) Potential for widespreadelectrification of personal vehicle travel in the United States. Nat Energy 1:16112.

2. Lin D, Liu Y, Cui Y (2017) Reviving the lithium metal anode for high-energy batteries.Nat Nanotechnol 12:194–206.

3. Cheng XB, Zhang R, Zhao CZ, Zhang Q (2017) Toward safe lithium metal anode inrechargeable batteries: A review. Chem Rev 117:10403–10473.

4. Bruce PG, Freunberger SA, Hardwick LJ, Tarascon JM (2011) Li-O2 and Li-S batterieswith high energy storage. Nat Mater 11:19–29.

5. Pang Q, Liang X, Kwok CY, Nazar LF (2016) Advances in lithium–sulfur batteries basedon multifunctional cathodes and electrolytes. Nat Energy 1:16132.

6. Aurbach D, Gofer Y (1989) The correlation between surface chemistry, surface mor-phology, and cycling efficiency of lithium electrodes in a few polar aprotic systems.J Electrochem Soc 136:3198–3205.

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

8. Urbonaite S, Poux T, Novak P (2015) Progress towards commercially viable Li–S batterycells. Adv Energy Mater 5:1500118.

9. Rosenman A, et al. (2015) Review on Li‐sulfur battery systems: An integral perspective.Adv Energy Mater 5:1500212.

10. Aurbach D, Zinigrad E, Cohen T, Teller H (2002) A short review of failure mechanismsof lithium metal and lithiated graphite anodes in liquid electrolyte solutions. SolidState Ion 148:405–416.

11. Cheng XB, et al. (2015) A review of solid electrolyte interphases on lithium metalanode. Adv Sci (Weinh) 3:1500213.

12. Tikekar MD, Choudhury S, Tu Z, Archer LA (2017) Design principles for electrolytesand interfaces for stable lithium-metal batteries. Nat Energy 1:16114.

13. Harry KJ, Hallinan DT, Parkinson DY, MacDowell AA, Balsara NP (2014) Detection ofsubsurface structures underneath dendrites formed on cycled lithium metal elec-trodes. Nat Mater 13:69–73.

14. Chazalviel J (1990) Electrochemical aspects of the generation of ramified metallicelectrodeposits. Phys Rev A 42:7355–7367.

15. Brissot C, Rosso M, Chazalviel JN, Lascaud S (1999) Dendritic growth mechanisms inlithium/polymer cells. J Power Sources 81–82:925–929.

16. Aurbach D, et al. (1995) Recent studies of the lithium-liquid electrolyte interface:Electrochemical, morphological and spectral studies of a few important systems.J Power Sources 54:76–84.

17. Ma Q, et al. (2016) Improved cycling stability of lithium-metal anode with concen-trated electrolytes based on lithium (fluorosulfonyl)(trifluoromethanesulfonyl)imide.ChemElectroChem 3:531–536.

18. Zheng J, et al. (2017) Electrolyte additive enabled fast charging and stable cyclinglithium metal batteries. Nat Energy 2:17012.

19. Lu Y, Tu Z, Archer LA (2014) Stable lithium electrodeposition in liquid and nano-porous solid electrolytes. Nat Mater 13:961–969.

20. Suo L, Hu YS, Li H, Armand M, Chen L (2013) A new class of solvent-in-salt electrolytefor high-energy rechargeable metallic lithium batteries. Nat Commun 4:1481.

21. Qian J, et al. (2015) High rate and stable cycling of lithium metal anode. Nat Commun6:6362.

22. Bouchet R, et al. (2013) Single-ion BAB triblock copolymers as highly efficient elec-trolytes for lithium-metal batteries. Nat Mater 12:452–457.

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

24. Villaluenga I, et al. (2016) Compliant glass-polymer hybrid single ion-conductingelectrolytes for lithium batteries. Proc Natl Acad Sci USA 113:52–57.

25. Zhao CZ, et al. (2017) An anion-immobilized composite electrolyte for dendrite-freelithium metal anodes. Proc Natl Acad Sci USA 114:11069–11074.

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

27. Zhang R, et al. (2017) Lithiophilic sites in doped graphene guide uniform lithiumnucleation for dendrite-free lithium metal anodes. Angew Chem Int Ed Engl 56:7764–7768.

28. Kazyak E, Wood KN, Dasgupta NP (2015) Improved cycle life and stability of lithiummetal anodes through ultrathin atomic layer deposition surface treatments. ChemMater 27:6457–6462.

29. Gao Y, et al. (2017) Interfacial chemistry regulation via a skin-grafting strategy en-ables high-performance lithium-metal batteries. J Am Chem Soc 139:15288–15291.

30. Liu K, et al. (2017) Lithium metal anodes with an adaptive “solid-liquid” interfacialprotective layer. J Am Chem Soc 139:4815–4820.

31. Zhu B, et al. (2017) Poly(dimethylsiloxane) thin film as a stable interfacial layer forhigh‐performance lithium‐metal battery anodes. Adv Mater 29:1603755.

32. Li NW, Yin YX, Yang CP, Guo YG (2016) An artificial solid electrolyte interphase layerfor stable lithium metal anodes. Adv Mater 28:1853–1858.

33. Liu Y, et al. (2017) An artificial solid electrolyte interphase with high Li-Ion conduc-tivity, mechanical strength, and flexibility for stable lithium metal anodes. Adv Mater29:16005531.

34. Pang Q, Liang X, Shyamsunder A, Nazar LF (2017) An in vivo formed solid electrolytesurface layer enables stable plating of Li meta. Joule 1:871–886.

35. Liang X, et al. (2017) A facile surface chemistry route to a stabilized lithium metalanode. Nat Energy 2:17119.

36. Shen F, Dixit MB, Xiao X, Hatzell KB (2018) Effect of pore connectivity on Li dendritepropagation within LLZO electrolytes observed with synchrotron X-ray tomography.ACS Energy Lett 3:1056–1061.

37. Liu Z, et al. (2013) Anomalous high ionic conductivity of nanoporous β-Li3PS4. J AmChem Soc 135:975–978.

38. Wenzel S, Leichtweiss T, Kruger D, Sann J, Janek J (2015) Interphase formation onlithium solid electrolytes—An in situ approach to study interfacial reactions by pho-toelectron spectroscopy. Solid State Ion 278:98–105.

39. Naoi K, Mori M, Naruoka Y, Lamanna W, Atanasosk R (1999) The surface film formed on alithium metal electrode in a new imide electrolyte, lithium bis(perfluoroethylsulfonylimide).J Electrochem Soc 146:462–469.

40. Contour JP, et al. (1979) Analysis by electron-microscopy and XPS of lithium surfacespolarized in anhydrous organic electrolytes. Microsc Spectrosc Electron 4:483.

41. Wood KN, et al. (2016) Dendrites and pits: Untangling the complex behavior oflithium metal anodes through operando video microscopy. ACS Cent Sci 2:790–801.

12394 | www.pnas.org/cgi/doi/10.1073/pnas.1809187115 Pang et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 22

, 202

1