High-capacity, low-tortuosity, and channel-guidedlithium metal anodeYing Zhanga,1, Wei Luoa,1, Chengwei Wanga, Yiju Lia, Chaoji Chena, Jianwei Songa, Jiaqi Daia, Emily M. Hitza,Shaomao Xua, Chunpeng Yanga, Yanbin Wanga, and Liangbing Hua,2

aDepartment of Materials Science and Engineering, University of Maryland, College Park, MD 20742

Edited by Thomas E. Mallouk, The Pennsylvania State University, University Park, PA, and approved February 17, 2017 (received for review November 15, 2016)

Lithium metal anode with the highest capacity and lowest anodepotential is extremely attractive to battery technologies, but infinitevolume change during the Li stripping/plating process results incracks and fractures of the solid electrolyte interphase, low Coulom-bic efficiency, and dendritic growth of Li. Here, we use a carbonizedwood (C-wood) as a 3D, highly porous (73% porosity) conductiveframework with well-aligned channels as Li host material. We discov-ered that molten Li metal can infuse into the straight channels ofC-wood to form a Li/C-wood electrode after surface treatment. TheC-wood channels function as excellent guides in which the Li strip-ping/plating process can take place and effectively confine the vol-ume change that occurs. Moreover, the local current density can beminimized due to the 3D C-wood framework. Therefore, in symmetriccells, the as-prepared Li/C-wood electrode presents a lower overpo-tential (90 mV at 3 mA·cm−2), more-stable stripping/plating profiles,and better cycling performance (∼150 h at 3 mA·cm−2) comparedwith bare Li metal electrode. Our findings may open up a solutionfor fabricating stable Li metal anode, which further facilitates futureapplication of high-energy-density Li metal batteries.

lithium metal batteries | wood channels | low tortuosity | stable cycling |high capacity

The ever-increasing demand from portable devices, electricvehicles, and clean energy has stimulated intensive research

on new battery technologies (1–6). Among various batteries,Li−S (7–15) and Li−O2 (16–20) batteries have shown greatpromise due to their high energy densities. Li metal has longbeen regarded as the ultimate anode for Li-ion based batteriesdue to its highest specific capacity (3,860 mAh·g−1), lowestelectrochemical potential (−3.040 V vs. standard hydrogenelectrode), and light weight (0.53 g·cm−3) (21–23), which is alsoindispensable for these desirable Li−S and Li−O2 batteries.However, compared with graphite (24–26) or Li4Ti5O12 (27),which exhibits small/zero volume change during Li ionintercalation/deintercalation, infinite volume change occurs inLi metal anode upon repeated stripping/plating cycles (21, 28).Such a huge volume change inevitably destroys the solid elec-trolyte interphase (SEI), making fresh Li metal exposed toelectrolyte and leading to excessive growth of SEI as a result(22). All these issues will result in rapid consumption of elec-trolyte, low Coulombic efficiency, and high resistance when Limetal is used as an anode. Furthermore, Li dendritic growth isanother critical issue associated with Li metal anode, wheresafety concerns about internal short circuit and battery explo-sion have been raised (29, 30). Hence, how to develop a stableand safe Li metal anode becomes a pivotal and challengingquestion for chemists and materials scientists to answer.In the past several decades, considerable efforts have been

devoted to solve the above-mentioned issues of Li metal anode.For reinforcing the SEI layer, one approach is to use additivesin liquid electrolyte, such as adding Cs+ and Rb+ ions (31, 32),LiF (33), Cu(CH3COO)2 (34), Li2S8, and LiNO3 (35). Introducingan artificial SEI has also been intensively investigated, whererepeated deposition/stripping of Li beneath the artificial SEIexhibited stable Coulombic efficiency (3, 36–38). Unfortunately,

huge volume change upon stripping/plating of Li metal still occursand leads to the cracking of SEI. To solve this problem, 3D open-structured designs for Li metal have been demonstrated (39).Taking the 3D porous Cu current collector as an example, Limetal is restricted in the 3D structure without conspicuous growthof Li dendrites, thus presenting a stable cycling performance (upto 560 h at 1 mA·cm−2) and low voltage hysteresis without shortcircuit (40, 41). Meanwhile, the local current density plays a largerole in the Li dendrite formation and growth, where a high localcurrent density normally accelerates the growth of Li dendritesand leads to an unstable cycling performance. By minimizing thelocal current density, Zhang et al. (42) adopted an unstackedgraphene framework with 3D drums as a host for deposited Limetal, which achieved a high Coulombic efficiency and effectivelysuppressed Li dendrites. However, it is still a challenge to preloadLi metal into these 3D current collectors. Most recently, a newconcept for improving the Li wettability of 3D current collectorhas been demonstrated, which resulted in significant progress (28,43, 44).Natural wood has a unique channel structure along the

growth direction, to transport water, ions, and substances (45).As a functional material, natural wood recently has garneredgreat interest due to its low tortuosity, high porosity, and eco-friendliness. Here we introduce a design of Li metal/carbonizedwood (Li/C-wood) composite by preinfusing metallic Li into thechannels of C-wood. Due to the unique channel structure ofLi/C-wood composite, the electrochemical Li stripping/platingprocess is expected to occur in the channels, and the hugevolume change can be well confined. Moreover, the local cur-rent density has been significantly decreased due to the 3D

Significance

Li metal is considered as the “Holy Grail” anode for Li batteriesdue to its highest theoretical capacity and lowest electrochemicalpotential. However, the infinite volume change during the Listripping/plating process would lead to issues like solid electro-lyte interphase cracks and Li dendrites. This work describes ahigh-capacity and low-tortuosity Li metal anode, which wasprepared by infusing molten Li into carbonized wood channels.The straight channels of carbonized wood acting as an ideal hostcan effectively accommodate the Li volume change, which de-livered a lower overpotential and better cycling performancecompared with bare Li metal. This work demonstrated the im-portance of structure design, especially low-tortuosity Li metalstructure, for enabling Li metal anode in high-energy batteries.

Author contributions: Y.Z., W.L., C.W., and L.H. designed research; Y.Z. and W.L. performedresearch; Y.Z., W.L., Y.L., C.C., J.S., J.D., and Y.W. contributed new reagents/analytic tools;Y.Z., W.L., S.X., C.Y., and L.H. analyzed data; and Y.Z., W.L., E.M.H., and L.H. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1Y.Z. and W.L. contributed equally to this work.2To whom correspondence should be addressed. Email: binghu@umd.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1618871114/-/DCSupplemental.

3584–3589 | PNAS | April 4, 2017 | vol. 114 | no. 14 www.pnas.org/cgi/doi/10.1073/pnas.1618871114

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 18

, 202

0

structure of the C-wood channels to obtain uniform Li nucleationand growth while alleviating dendritic Li growth. Owing to theabove merits, the Li/C-wood electrodes enable a much bettercycling performance than bare Li metal anode. The essentialprinciples of using C-wood to guide the Li stripping/plating

process and accommodating the volume change are illustrated inFig. 1B. On the surface of planar bare Li metal, the volumechange during repeated stripping/plating cycles will lead to thefailure of SEI and growth of dendrites (Fig. 1A). In contrast, Livolume expansion/contraction can be effectively accommodated

Fig. 1. Schematic diagrams of Li stripping/plating behavior for bare Li metal electrodes and Li/C-wood electrodes with well-aligned channels. (A) Volumechange during Li stripping/plating leads to the failure of SEI and growth of Li dendrites. (B) Li metal was accommodated in aligned C-wood channels duringstripping/plating process, which enables long-term cycling stability without dendrite formation. Note that the carbon walls are thin in the C-wood, leading toa low tortuosity and a high porosity (73%) for hosting Li metal.

Fig. 2. Fabrication process of the Li/C-wood composite. (A) A schematic of the material design and the subsequent synthesis from C-wood (Left), to ZnO-coated C-wood (Middle), and, finally, to Li/C-wood composite (Right). (B) The corresponding digital images of C-wood, ZnO-coated wood, and Li/C-wood (sizeis 5 mm × 7 mm), showing the successful infusion of Li metal into C-wood. (C−E) The corresponding SEM images of (C) C-wood, (D) ZnO-coated C-wood, and(E) Li/C-wood composite, which indicates Li infused into the C-wood channels completely. (F−H) Digital images present the fast infusion of molten Li metalinto ZnO-coated C-wood. The whole infusion takes only 1 s, which demonstrates the fast infusion process due to the reaction between ZnO and molten Li. SeeMovie S1 for the whole process.

Zhang et al. PNAS | April 4, 2017 | vol. 114 | no. 14 | 3585

ENGINEE

RING

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 18

, 202

0

within the natural channels of C-wood, thus enabling a long-termcycling stability (Fig. 1B).

Results and DiscussionDesign and Fabrication of Li/C-Wood Composites. Fig. 2A illustratesthe typical fabrication process of the Li/C-wood composite. Bass-wood is adopted in this study, which possesses channel structureswith sizes ranging from 15 μm to 100 μm (SI Appendix, Fig. S1).After carbonization, the C-wood shows clear texture patternsinherited from the pristine wood, as shown in the left digital imageof Fig. 2B. Detailed scanning electron microscopy (SEM) images ofpristine wood and C-wood in SI Appendix, Fig. S1 and Fig. 2Crevealed that the unique channel structure was well preserved. Tofabricate the Li/C-wood composite, the wettability between C-woodand Li metal is critical. As previously reported by Liu et al. (43), athin layer of ZnO coated by atomic layer deposition technology canchange the polymer substrate from lithiophobe to lithiophile. Herewe used a low-cost and simple solution-process method to coat thechannel walls of C-wood with a thin layer of ZnO (see more detailsin Methods). Fig. 2D and SI Appendix, Fig. S2 show that ZnOparticles formed inside the channels and dispersed uniformly. Asexpected, molten Li was fast flooded into channels of ZnO-coatedC-wood, which formed the shiny Li/C-wood composite (Fig. 2 Band E). Fig. 2 F−H shows the infusion process of molten Li metalinto C-wood, where the whole process occurred in less than 1 s (seeMovie S1). Such a fast process is mainly attributed to the favorablereaction between Li metal and ZnO and the capillary force onlithiophilic surface as driving force.

Morphology Characterizations of the Li/C-Wood Composite. To ex-amine the morphology evolution of C-wood, ZnO-coated C-wood,and Li/C-wood composite, SEM characterization was performed.Fig. 3A shows a top-view SEM image of C-wood, which exhibits itstypical channel structure. The channel size ranges from 5 μm to60 μm (see SI Appendix, Fig. S3), and the thickness of channel wallis about 1 μm, which can provide a porous conductive framework(porosity of 73%) for accommodating Li metal. As shown in Fig.3B, several stripes on the channel wall can be observed by cross-sectional SEM image, which is consistent with the microstructureof pristine wood (see SI Appendix, Fig. S1) and confirms that thecarbonization process can maintain the intrinsic microstructure ofwood. The intensity ratio of I(D)/I(G) from Raman spectra is 1.26,indicating the amorphous feature of the C-wood (46) (see SIAppendix, Fig. S4). After soaking the C-wood in ZnO precursorsolution followed by a thermal treatment, ZnO particles weregenerated on the top of C-wood (Fig. 3C). More importantly,ZnO particles also grew evenly on the inner channel walls, whichcan form reaction sites between C-wood and molten Li (Fig. 3D).By controlling the concentration of ZnO precursor solution andthe applied solution volume, about 11 wt% ZnO was loaded ontoC-wood. After Li metal was successfully infused into the ZnO-coated C-wood, the morphology of Li/C-wood composite wasfurther investigated. Fig. 3E shows the cross-section view ofLi/C-wood composite. It appears that Li was drawn from thebottom side upward along the channels and occupied the wholeinner void space of wood channels without destroying the channelstructures (Fig. 3F), to deliver a high areal mass loading of28 mg·cm−2. Energy dispersive X-ray spectroscopy was used tofurther confirm the location of Li metal in the aligned woodchannels (see SI Appendix, Fig. S5−S7). The overlapped mappingimage and the individual elemental map corresponding to carbonshow a clearly defined carbonized framework of different-sizedwood channels. However, we did not detect carbon in the regionwithin the channels, indicating that these void spaces were occu-pied by Li metal. Interestingly, the top surface of Li metal in thesechannels exhibits a concave shape, which was attributed to thevolume contraction upon cooling, as displayed in Fig. 3 G−H.

Electrochemical Stability Comparison of Li Metal and Li/C-WoodElectrodes. To evaluate the electrochemical performance of theLi/C-wood composite, symmetric cells with two identical Li/C-wood electrodes were fabricated using 1.0 M LiPF6 in ethylenecarbonate/diethyl carbonate (EC:DEC = 1:1 by volume) withoutadditives as the electrolyte (see details in Methods). Meanwhile,symmetric cells with two bare Li metal electrodes were assembledas a control. Fig. 4 A−C shows the long-term cycling profiles of thesymmetrical cells under different current densities of 0.5 mA·cm−2,1 mA·cm−2, and 3 mA·cm−2, respectively, following the samestripping/plating capacity (1 mAh·cm−2). As shown in Fig. 4A, thecell with Li/C-wood electrodes exhibited a stable voltage profileover long-term cycling with small hysteresis compared with the cellusing bare Li metal electrodes. In particular, Li/C-wood electrodesdelivered a low initial stripping/plating overpotential of 20 mV at acurrent density of 0.5 mA·cm−2 that increased to 65 mV after213 cycles (Fig. 4A, Insets). In sharp contrast, the overpotential ofthe bare Li metal electrodes increased from 38 mV to 178 mVunder the same conditions, where the increase in overpotential ofthe bare Li metal cell is threefold larger than that of the Li/C-woodcell. When the current density was increased to 1 mA·cm−2 or3 mA·cm−2, Li/C-wood cells could be cycled for a long time(165 cycles at 1 mA·cm−2 and 225 cycles at 3 mA·cm−2) with stablestripping/plating platforms and low overpotential (43 mV in the75th cycle at 1 mA·cm−2 and 100 mV in the 101th cycle at3 mA·cm−2), indicative of dendrite-free electrodes (Fig. 4 B andC). Contrastingly, for bare Li metal cells, overpotential increasedgradually upon cycling, and a random voltage oscillation canbe observed. Specifically, an abrupt voltage drop occurred

Fig. 3. Morphology study of C-wood, ZnO-coated C-wood, and Li/C-woodcomposite. (A and B) Top-view and cross-sectional SEM images of C-woodpresent typical channel structure. (C and D) Top-view and cross-sectionalSEM images of ZnO-coated C-wood. Insets in A and B are correspondingzoomed-in SEM images. (E and F) Cross-sectional SEM images of Li infusionin the channel void space. (G and H) Top-view SEM images of Li/C-woodcomposite exhibit a concave morphology of Li in the different sized chan-nels, revealing that Li metal is encapsulated inside the natural channels.

3586 | www.pnas.org/cgi/doi/10.1073/pnas.1618871114 Zhang et al.

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 18

, 202

0

(Fig. 4B, Inset) after dramatic voltage increase, which can beascribed to a short circuit of the cell induced by the formationof dendritic Li (see SI Appendix, Fig. S8). Fig. 4D and SI Ap-pendix, Table S1 compared the detailed voltage profiles of bareLi metal cells and Li/C-wood cells at these three current den-sities. Clearly, Li/C-wood cells delivered more-stable stripping/plating curves with smaller overpotential, which indicates im-proved cycling stability in a full cell without voltage swings. Itshould be noted that a dual function offered by the 3D porousC-wood structure makes this Li/C-wood anode novel and sta-ble. A homogeneous contact with Li+ flux enabling the uniformLi nucleation and growth during the Li plating process wasachievable due to the well-aligned and electrically conductivewood channels, which should account for the improved stability

of Li/C-wood electrodes. The larger overpotential and voltagefluctuation in bare Li metal cells can be attributed to the in-stability of the SEI layer, thus leading to the large impedance inthe following cycles and finally resulting in a short cycling life(28, 42). On the other hand, Li metal can be confined in thechannel structure of C-wood and the local current density canbe further minimized due to the high porosity, which promotedthe improved performance of the C-wood cells (21). To betterunderstand the mechanism, electrochemical impedance spectra(EIS) were analyzed before cycling and after three cycles (see SIAppendix, Fig. S9). The semicircle of the EIS curve represents thecharge-transfer resistance (Rct) associated with the Li/Li+ re-action. The Rct values for bare Li and Li/C-wood cells are 461 Ωand 60 Ω, respectively. After three cycles, compared with the bareLi metal cell (Rct 190 Ω), the Li/C-wood cell gave a much smallerRct of 40 Ω, indicating a fast charge transfer at the interfacebetween electrode and electrolyte and fast kinetics of theelectrochemical reaction. It can be inferred that the Li/C-woodwill display an excellent cycling performance with a smalloverpotential under even large current densities, in accordancewith the results shown in Fig. 4 A−C.To elucidate the structural evolution of Li/C-wood electrodes

upon stripping/plating, ex situ SEM images were taken. As shown

Fig. 4. Electrochemical stability comparison of symmetric cells using Limetal and Li/C-wood electrodes. (A−C) Voltage profiles of Li/C-wood cell(black) and bare Li metal cell (red) at current densities of 0.5 mA·cm−2,1 mA·cm−2, and 3 mA·cm−2 with a cycling capacity of 1 mAh·cm−2. Insets inA−C are zoomed-in profiles of typical cycles. Compared with bare Li metalcell, Li/C-wood cells exhibited smaller voltage hysteresis and long-term cy-cling stability. (D) Voltage profiles of bare Li metal cells (Left) and Li/C-woodcells (Right) at different current densities; notably, Li/C-wood cells showstable voltage profiles with smaller hysteresis.

Fig. 5. Structural evolution of Li/C-wood electrodes upon electrochemicalstripping/plating process. (A) A typical Li stripping curve of the Li/C-woodelectrode at a current density of 1 mA·cm−2, which delivers a specific capacityof 2,650 mAh·g−1. (B and C) The corresponding morphology of Li/C-woodelectrodes when stripping one third of Li (∼875 mAh g−1, marked as “B” inA) and all of the Li (2,650 mAh·g−1, marked as “C” in A) out. (D) A Li stripping/plating cycle with a specific cycling capacity of 3 mAh·cm−2 at 1 mA·cm−2. Cross-sectional SEM images of the Li/C-wood after (E and F) stripping 3 mAh·cm−2 Liout and then (G) plating 3 mAh·cm−2 Li back.

Zhang et al. PNAS | April 4, 2017 | vol. 114 | no. 14 | 3587

ENGINEE

RING

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 18

, 202

0

in Fig. 5A, a typical Li/C-wood electrode can deliver a highcapacity of 2,650 mAh·g−1 (75 mAh·cm−2). After stripping875 mAh·g−1 (about one third of total capacity), the top layer ofLi/C-wood composite became empty, whereas Li metal can stillbe observed in the bottom of the channels, as shown in SIAppendix, Fig. S10 and Fig. 5B. When all of the Li was strippedout from the Li/C-wood electrode, the hollowed channelstructure of C-wood was returned, which revealed that C-woodwith low tortuosity and high porosity can serve as a stableframework for Li metal electrodes (Fig. 5C and SI Appendix,Fig. S11). Benefiting from the porous C-wood structure, a ho-mogenous Li+ flux distribution is provided to enable a uniformLi nucleation and growth on this electrically conductive frame-work. Furthermore, after a full stripping/plating cycle with acapacity of 3 mAh·cm−2 at 1 mA·cm−2 (Fig. 5D), the top-viewSEM observations can clearly show the removal and depositionof Li from/into the channels of Li/C-wood electrode (see SIAppendix, Fig. S12 A and B), revealing an effective accommo-dation of Li volume change. For comparison, bare Li metalelectrode displayed a mossy surface with dendrites (see SIAppendix, Fig. S12 C and D).Unlike the completely hollow structure of the Li/C-wood

electrode that stripped all Li out, the Li/C-wood electrode thatstripped 3 mAh·cm−2 of Li exhibited a hollow structure of theupper channels with residual Li grains on the channel walls (inFig. 5 E and F). After plating Li back (Fig. 5G), the channelswere refilled with Li metal, which demonstrated good guidanceof Li into/out of wood channels and, furthermore, enabledlong-term cycling stability.

ConclusionIn summary, we designed and fabricated a highly porous frame-work with well-aligned channels for a stable Li stripping/platingprocess. The electrode with an aligned Li metal structure wasfabricated by infusing metallic Li into the channel structure ofC-wood. The wettability between C-wood and Li has been enabledby ZnO coating via a facile solution method. Due to its 3D porousconductive framework structure, C-wood is an ideal host for Limetal, helping to accommodate the volume change during thelong-term cycling of Li metal. Moreover, the Li stripping/platingprocess mainly occurs in the channels of Li/C-wood electrodes, sothat the channel structure effectively guides the electrochemicalprocess. Compared with the conventional planar Li foils, the

well-aligned and electrically conductive carbonized wood channelscan offer a homogeneous contact with Li+ flux, which enables auniform Li nucleation and growth during the Li plating process.As a result, a long-term cycling stability with small hysteresis ofLi/C-wood electrodes was achieved even under a high currentdensity of 3 mA·cm−2 in the commercially available carbonate-based electrolyte. Because Li metal anodes are extremely prom-ising for high-energy-density Li batteries, the strategy using naturalchannels from wood to guide the stripping/plating process andconfine volume change can shed light on further research inframework design for high-capacity, dendrite-free Li metal anodes.

MethodsFabrication of Li/C-Wood Composite. Basswood was purchased from WalnutHollow Company. A piece of basswood was calcined at 260 °C for 6 h in air andat 1,000 °C for 6 h in Ar to give the C-wood product. Zn(NO3)2·6H2O (>98%;Sigma-Aldrich) was dissolved in ethanol with a concentration of 50 mg·mL−1.Then, a piece of C-wood was immersed into the Zn(NO3)2 solution, followed bydrying and calcining at 400 °C for 10 min in air. To infuse Li metal into channels,ZnO-coated C-wood was dipped into molten Li, as shown in Movie S1.

Material Characterizations and Electrochemical Measurements. The morphol-ogies were studied using Hitachi SU-70 field emission SEM. Raman microscope(Yvon Jobin LabRam ARAMIS) was applied to measure the structure of theC-wood with a helium/neon 10-mW laser excitation source (632.8 Raman).Symmetric cells were used to study the Li stripping/plating behaviors by as-sembling two identical Li/C-wood electrodes into 2,032-type coin cells. LiPF6, ECand DEC in battery grade were commercially available from BASF Inc. BatteryMaterials; 1.0 M LiPF6 in EC/DEC (volume ratio 1:1) solution without any ad-ditives is used as the electrolyte. The galvanostatic cycling performance wasconducted on an Arbin BT2000 system. In control experiments, bare Li metalfoils (99.9%, 0.75 mm in thickness, 19 mm in width; Sigma-Aldrich) were usedas electrodes for assembling symmetric cells. EIS measurements were per-formed by a Biologic VMP3 multichannel system. For morphology observationafter electrochemical cycling, the electrodes were washed with dimethyl car-bonate and dried in the glove box before SEM characterization.

ACKNOWLEDGMENTS. The authors acknowledge support from US Depart-ment of Energy, Advanced Research Projects Agency–Energy Project (DE-AR0000726). This work was also supported as part of the Nanostructures forElectrical Energy Storage, an Energy Frontier Research Center funded by theUS Department of Energy, Office of Science, Basic Energy Sciences underAward DESC0001160. Y.Z. acknowledges the China Scholarship Council(CSC 201506680044) for financial support. The authors also thank the KeyLaboratory of Superlight Materials and Surface Technology of Ministry ofEducation, College of Materials Science and Chemical Engineering, HarbinEngineering University for support.

1. Armand M, Tarascon J-M (2008) Building better batteries. Nature 451(7179):652–657.2. Li Y, et al. (2016) Mastering the interface for advanced all-solid-state lithium re-

chargeable batteries. Proc Natl Acad Sci USA 113(47):13313–13317.3. Huang C, et al. (2014) Manipulating surface reactions in lithium-sulphur batteries

using hybrid anode structures. Nat Commun 5:3015.4. Long JW, Dunn B, Rolison DR, White HS (2004) Three-dimensional battery architec-

tures. Chem Rev 104(10):4463–4492.5. Guo YG, Hu JS, Wan LJ (2008) Nanostructured materials for electrochemical energy

conversion and storage devices. Adv Mater 20(15):2878–2887.6. El-Kady MF, et al. (2015) Engineering three-dimensional hybrid supercapacitors and

microsupercapacitors for high-performance integrated energy storage. Proc Natl

Acad Sci USA 112(14):4233–4238.7. Villaluenga I, et al. (2016) Compliant glass−polymer hybrid single ion-conducting

electrolytes for lithium batteries. Proc Natl Acad Sci USA 113(1):52–57.8. Bruce PG, Freunberger SA, Hardwick LJ, Tarascon J-M (2011) Li-O2 and Li-S batteries

with high energy storage. Nat Mater 11(1):19–29.9. Ji X, Lee KT, Nazar LF (2009) A highly ordered nanostructured carbon-sulphur cathode

for lithium-sulphur batteries. Nat Mater 8(6):500–506.10. Yang Y, Zheng G, Cui Y (2013) Nanostructured sulfur cathodes. Chem Soc Rev 42(7):

3018–3032.11. Li Z, et al. (2014) A highly ordered meso@microporous carbon-supported sulfur@smaller

sulfur core-shell structured cathode for Li-S batteries. ACS Nano 8(9):9295–9303.12. Ji X, Nazar LF (2010) Advances in Li–S batteries. J Mater Chem 20(44):9821–9826.13. Wei Seh Z, et al. (2013) Sulphur-TiO2 yolk-shell nanoarchitecture with internal void

space for long-cycle lithium-sulphur batteries. Nat Commun 4:1331.14. Chen S, et al. (2016) Functional organosulfide electrolyte promotes an alternate reaction

pathway to achieve high performance in lithium-sulfur batteries. Angew Chem Int Ed

Engl 55(13):4231–4235.

15. Huang J-Q, et al. (2014) Ionic shield for polysulfides towards highly-stable lithium–

sulfur batteries. Energy Environ Sci 7(1):347–353.16. Xiao J, et al. (2011) Hierarchically porous graphene as a lithium-air battery electrode.

Nano Lett 11(11):5071–5078.17. Shao Y, et al. (2013) Making Li‐air batteries rechargeable: Material challenges. Adv

Funct Mater 23(8):987–1004.18. Lee JS, et al. (2011) Metal–air batteries with high energy density: Li–air versus Zn–air.

Adv Energy Mater 1(1):34–50.19. Xiao J, et al. (2010) Optimization of air electrode for Li/air batteries. J Electrochem Soc

157(4):A487–A492.20. Trahey L, et al. (2013) Synthesis, characterization, and structural modeling of high‐

capacity, dual functioning MnO2 electrode/electrocatalysts for Li‐O2 cells. Adv EnergyMater 3(1):75–84.

21. Xu W, et al. (2014) Lithium metal anodes for rechargeable batteries. Energy EnvironSci 7(2):513–537.

22. Kim H, et al. (2013) Metallic anodes for next generation secondary batteries. ChemSoc Rev 42(23):9011–9034.

23. Tarascon J-M, Armand M (2001) Issues and challenges facing rechargeable lithiumbatteries. Nature 414(6861):359–367.

24. Chen Z, Christensen L, Dahn JR (2003) Large-volume-change electrodes for Li-ionbatteries of amorphous alloy particles held by elastomeric tethers. ElectrochemCommun 5(11):919–923.

25. Cohn AP, et al. (2014) Assessing the improved performance of freestanding, flexiblegraphene and carbon nanotube hybrid foams for lithium ion battery anodes.Nanoscale 6(9):4669–4675.

26. Allen MJ, Tung VC, Kaner RB (2010) Honeycomb carbon: A review of graphene. ChemRev 110(1):132–145.

27. Borghols WJH, Wagemaker M, Lafont U, Kelder EM, Mulder FM (2009) Size effects inthe Li4+xTi5O12 spinel. J Am Chem Soc 131(49):17786–17792.

3588 | www.pnas.org/cgi/doi/10.1073/pnas.1618871114 Zhang et al.

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 18

, 202

0

28. Liang Z, et al. (2016) Composite lithiummetal anode by melt infusion of lithium into a 3Dconducting scaffold with lithiophilic coating. Proc Natl Acad Sci USA 113(11):2862–2867.

29. Zhang K, Lee GH, Park M, Li W, Kang YM (2016) Recent developments of the lithiummetal anode for rechargeable non‐aqueous batteries. Adv Energy Mater 6(20):1600811.

30. Ilott AJ, Mohammadi M, Chang HJ, Grey CP, Jerschow A (2016) Real-time 3D imagingof microstructure growth in battery cells using indirect MRI. Proc Natl Acad Sci USA113(39):10779–10784.

31. Ding F, et al. (2013) Dendrite-free lithium deposition via self-healing electrostaticshield mechanism. J Am Chem Soc 135(11):4450–4456.

32. Zhang Y, et al. (2014) Dendrite-free lithium deposition with self-aligned nanorodstructure. Nano Lett 14(12):6889–6896.

33. Lu Y, Tu Z, Archer LA (2014) Stable lithium electrodeposition in liquid and nano-porous solid electrolytes. Nat Mater 13(10):961–969.

34. Zu C, Manthiram A (2014) Stabilized lithium-metal surface in a polysulfide-rich en-vironment of lithium-sulfur batteries. J Phys Chem Lett 5(15):2522–2527.

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

36. Zheng G, et al. (2014) Interconnected hollow carbon nanospheres for stable lithiummetal anodes. Nat Nanotechnol 9(8):618–623.

37. Yan K, et al. (2014) Ultrathin two-dimensional atomic crystals as stable interfaciallayer for improvement of lithium metal anode. Nano Lett 14(10):6016–6022.

38. Choudhury S, Mangal R, Agrawal A, Archer LA (2015) A highly reversible room-

temperature lithium metal battery based on crosslinked hairy nanoparticles. Nat

Commun 6:10101.39. Ji X, et al. (2012) Spatially heterogeneous carbon-fiber papers as surface dendrite-free

current collectors for lithium deposition. Nano Today 7(1):10–20.40. Yang C-PP, Yin Y-XX, Zhang S-FF, Li N-WW, Guo Y-GG (2015) Accommodating lithium

into 3D current collectors with a submicron skeleton towards long-life lithium metal

anodes. Nat Commun 6:8058.41. Lu L-L, et al. (2016) Free-standing copper nanowire network current collector for

improving lithium anode performance. Nano Lett 16(7):4431–4437.42. Zhang R, et al. (2016) Conductive nanostructured scaffolds render low local current

density to inhibit lithium dendrite growth. Adv Mater 28(11):2155–2162.43. Liu Y, et al. (2016) Lithium-coated polymeric matrix as a minimum volume-change and

dendrite-free lithium metal anode. Nat Commun 7:10992.44. Lin D, et al. (2016) Layered reduced graphene oxide with nanoscale interlayer gaps as

a stable host for lithium metal anodes. Nat Nanotechnol 11(7):626–632.45. Zhu H, et al. (2016) Wood-derived materials for green electronics, biological devices,

and energy applications. Chem Rev 116(16):9305–9374.46. Ferrari A, Robertson J (2001) Resonant Raman spectroscopy of disordered, amor-

phous, and diamondlike carbon. Phys Rev B 64(7):075414.

Zhang et al. PNAS | April 4, 2017 | vol. 114 | no. 14 | 3589

ENGINEE

RING

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 18

, 202

0