Flexible, solid-state, ion-conducting membrane with 3Dgarnet nanofiber networks for lithium batteriesKun (Kelvin) Fua,b,1, Yunhui Gonga,1, Jiaqi Daib, Amy Gonga,b, Xiaogang Hana,b, Yonggang Yaob, Chengwei Wanga,b,Yibo Wangb, Yanan Chenb, Chaoyi Yanb, Yiju Lib, Eric D. Wachsmana,b,2, and Liangbing Hua,b,2

aUniversity of Maryland Energy Research Center, University of Maryland, College Park, MD 20742; and bDepartment of Materials Science and Engineering,University of Maryland, College Park, MD 20742

Edited by Yi Cui, Stanford University, Stanford, CA, and accepted by Editorial Board Member Tobin J. Marks May 4, 2016 (received for review January 10, 2016)

Beyond state-of-the-art lithium-ion battery (LIB) technology withmetallic lithium anodes to replace conventional ion intercalationanode materials is highly desirable because of lithium’s highestspecific capacity (3,860 mA/g) and lowest negative electrochem-ical potential (∼3.040 V vs. the standard hydrogen electrode). Inthis work, we report for the first time, to our knowledge, a 3Dlithium-ion–conducting ceramic network based on garnet-typeLi6.4La3Zr2Al0.2O12 (LLZO) lithium-ion conductor to provide con-tinuous Li+ transfer channels in a polyethylene oxide (PEO)-basedcomposite. This composite structure further provides structural re-inforcement to enhance the mechanical properties of the polymermatrix. The flexible solid-state electrolyte composite membraneexhibited an ionic conductivity of 2.5 × 10−4 S/cm at room temper-ature. The membrane can effectively block dendrites in a symmetricLi j electrolyte j Li cell during repeated lithium stripping/plating atroom temperature, with a current density of 0.2 mA/cm2 for around500 h and a current density of 0.5 mA/cm2 for over 300 h. Theseresults provide an all solid ion-conducting membrane that can beapplied to flexible LIBs and other electrochemical energy storagesystems, such as lithium–sulfur batteries.

solid-state electrolyte | 3D garnet nanofibers | polyethylene oxide |ionic conductor | flexible membrane

High capacity, high safety, and long lifespan are three of themost important key factors to developing rechargeable

lithium batteries for applications in portable electronics, trans-portation (e.g., electrical vehicles), and large-scale energy stor-age systems (1–5). Based on state-of-the-art lithium-ion battery(LIB) technology, metallic lithium anode is preferable to replaceconventional ion intercalation anode materials because of thehighest specific capacity (3,860 mAh/g) of lithium and the lowestnegative electrochemical potential (∼3.040 V vs. the standardhydrogen electrode), which can maximize the capacity densityand voltage window for increased battery energy density (1).Moreover, the success of beyond LIBs, such as lithium–sulfurand lithium–oxygen, will strongly rely on lithium metal anodedesigns with good stability to achieve their targeted goals of highenergy density and long cycle life.Using lithium metal in organic liquid electrolyte systems faces

many challenges in terms of battery performance and safety. Forexample, lithium–sulfur batteries suffer from the dissolution ofintermediate polysulfides in the organic electrolyte that causessevere parasitic reactions on lithium metal surfaces, leading tolithium metal degradation and low lithium cycling efficiency (6).Lithium–oxygen batteries have the challenge of chemically in-stable liquid electrolytes on the oxygen electrode that causelimited battery cycling (7). All of these challenges are associatedwith the use of lithium metal in liquid electrolyte battery systems.Another major associated challenge is lithium dendrite growthon lithium metal anodes, which causes internal short circuitsafter lithium dendrites penetrate through the separator andtouch the cathode. In addition, solid–electrolyte interphase(SEI) formation during the uneven lithium deposition will con-tinuously consume Li metal and dry up the electrolyte, leading to

an increase of cell resistance and decrease of cell Coulombicefficiency (1, 8). Although extensive studies have been per-formed to address these challenges, Li dendrite and SEI forma-tion are inevitable and mainly caused by the intrinsic problems of thethermodynamically unstable Li with low-molecular weight organicsolvents and the poor strength of formed SEI layers (1).A fundamental strategy to address Li dendrite penetration

and SEI formation is to develop a solid-state electrolyte tomechanically suppress the lithium dendrite and intrinsicallyeliminate SEI formation (9–14). Among the different types ofsolid-state electrolytes (inorganic oxides/nonoxides and Lisalt-contained polymers), solid-state polymer electrolyteshave been the most extensively studied (15–21). Polyethyleneoxide (PEO)-based composite electrolytes have attracted themost interest (22, 23). In PEO-based composite, powders areincorporated into a host PEO polymer matrix to influence therecrystallization kinetics of the PEO polymer chains to pro-mote local amorphous regions, thereby increasing the Li salt–polymer system’s ionic conductivity (15). The addition ofpowders will also improve the electrochemical stability and en-hance the mechanical strength. As studied in previous work, thefillers can be either non–Li+-conductive nanoparticles, such asAl2O3 (15), SiO2 (24), TiO2 (25), ZrO2 (26), and organic polymerspheres (23), or Li+-conductive nanoparticles, such as Li0.33La0.557-TiO3 (22), tetragonal Li7La3Zr2O12 (27), and Li1.3Al0.3Ti1.7(PO4)3(28). Developing nanostructured fillers is an essential approach toincrease the ionic conductivity of polymer composite electrolytesbecause of the increased surface area of the amorphous region andimproved interface between fillers and polymers. Typically, 1D

Significance

This work describes a flexible, solid-state, lithium-ion–conductingmembrane based on a 3D ion-conducting network and polymerelectrolyte for lithium batteries. The 3D ion-conducting network isbased on percolative garnet-type Li6.4La3Zr2Al0.2O12 solid-stateelectrolyte nanofibers, which enhance the ionic conductivity ofthe solid-state electrolyte membrane at room temperature andimprove the mechanical strength of the polymer electrolyte. Themembrane has shown superior electrochemical stability to highvoltage and high mechanical stability to effectively block lithiumdendrites. This work represents a significant breakthrough toenable high performance of lithium batteries.

Author contributions: K.(K.)F., Y.G., E.D.W., and L.H. designed research; K.(K.)F., Y.G., J.D.,A.G., and Y.W. performed research; K.(K.)F., Y.G., X.H., Y.Y., C.W., Y.C., C.Y., Y.L., E.D.W.,and L.H. analyzed data; and K.(K.)F. and Y.G. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission. Y.C. is a guest editor invited by the EditorialBoard.1K.(K.)F. and Y.G. contributed equally to this work.2To whom correspondence may be addressed. Email: ewach@umd.edu or binghu@umd.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1600422113/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1600422113 PNAS Early Edition | 1 of 6

ENGINEE

RING

Dow

nloa

ded

by g

uest

on

Dec

embe

r 13

, 202

0

nanowire fillers, based on perovskite-type lithium-ion–conductingLi0.33La0.557TiO3 material, were shown by Cui and coworkers (22)to enhance the ionic conductivity of the polymer composite elec-trolyte. This enhanced ionic conductivity was because the nano-wire fillers provide extended ionic transport pathways in thepolymer matrix instead of an isolated distribution of nanoparticlefillers in the polymer electrolyte (22). However, the agglomerationof ceramic fillers may remain, and it will become a challenge for itsmixing with polymer to fabricate uniform solid polymer electrolyteon a large scale. To solve this challenge, in situ synthesis of ce-ramic filler particles with high monodispersity in polymer elec-trolyte was recently reported (29). By in situ synthesizingnanosized SiO2 particles into PEO–Li salt polymer, the reportedsolid polymer electrolyte exhibited an ionic conductivity of4.4 × 10−5 S/cm at 30 °C, which needs additional improvementto achieve a higher ionic conductivity at room temperature.Based on our understanding, therefore, creating a continuousnanosized network with interconnected long-range ion transportand controlling a minimum/nonfiller agglomeration are the maindirections to design high ionic-conductive polymer compositeelectrolytes.In this work, we have successfully developed a 3D ceramic

network based on garnet-type Li6.4La3Zr2Al0.2O12 (LLZO) nano-fibers to provide continuous Li+ transfer channels in PEO-basedcomposite electrolytes as all solid ion-conducting membranes forlithium batteries. Here, we select garnet-type lithium-ion–conductingceramic as the inorganic component because of several desiredphysical and chemical properties, including (i) high ionic con-ductivity approaching 10−3 S/cm at room temperature withoptimized element substitution, (ii) good chemical stabilityagainst lithium metal, and (iii) good chemical stability againstair and moisture (11, 30, 31). Fig. 1 shows the schematic structureof the 3D LLZO–polymer composite membrane. The LLZO po-rous structure consists of randomly distributed and interconnectednanofibers, creating a continuous lithium-ion–conducting network.The Li salt–PEO polymer is then filled into the porous 3D ceramicnetworks, forming the 3D garnet–polymer composite membrane.Different from conventional methods to prepare polymer electro-lytes, the 3D garnet–polymer composite membrane does not need tomechanically mix fillers with polymers; instead, we can directly soaka preformed 3D ceramic structure into Li salt–polymer solutions toget the desired polymer composite electrolyte hybrid structure, thus

simplifying fabrication process and avoiding the agglomerationof fillers.

Results and DiscussionFig. 2 schematically shows the procedure to synthesize flexiblesolid-state garnet LLZO nanofiber-reinforced polymer compos-ite electrolytes. As shown in Fig. 2A, garnet LLZO nanofiberswere prepared by electrospinning of polyvinylpyrrolidone (PVP)polymer mixed with relevant garnet LLZO salts followed by thecalcination of the as-prepared nanofibers at 800 °C in air for 2 h.On the drum collector of the electrospinning setup, a thin non-woven fabric was covered to collect the nanofibers.The schematic fabrication of fiber-reinforced polymer com-

posite (FRPC) lithium-ion–conducting membrane using the 3Dporous garnet nanofiber network is shown in Fig. 2B. A PEOpolymer mixture with Li salt, such as bis(trifluoromethane)sulfo-nimide lithium salt (LiTFSI), is prepared. Then, the Li salt–PEOpolymer is reinforced by the 3D nanofibers to form a compositeelectrolyte, which can be called FRPC electrolyte membrane.Compared with filler-containing polymer electrolyte, the FRPCelectrolyte membrane maintains the framework of 3D garnetnanofiber networks and is believed to have a better mechanicalproperty because of the continuous nanofiber structure thatenhances the integrity of polymer electrolyte.Morphologies of the as-spun PVP–garnet salt nanofibers and

calcinated garnet nanofibers were characterized by SEM asshown in Fig. 2 C and E. Before calcination, the PVP–garnet saltnanofibers have smooth surfaces, and nanofibers have a di-ameter of 256 nm on average. The corresponding diameterdistribution is shown in Fig. 2D. After the calcination at 800 °C inair, PVP polymers were removed, and garnet LLZO nanofiberswere obtained. The average diameter of the nanofibers de-creased to 138 nm. Their diameter distribution is given in Fig.2F. We can see that, after annealing, garnet nanofibers were“interwelded” with each other, forming cross-linked 3D garnetnanofiber networks. The large volume of interspace betweennanofibers can facilitate Li salt–polymer infiltration to form thecomposite membrane. The flexibility of the membrane is shownin Fig. 2G. The bendable electrolyte membrane can then be usedto construct flexible solid-state lithium batteries. Note that thedesign of flexible 3D ion-conducting networks mainly dependson ceramic garnet nanofibers, which require a thin and mechan-ically stable structure for good ionic conductivity and feasiblebattery fabrication. To achieve a thinner polymer compositeelectrolyte while maintaining a good mechanical stability, somekey parameters need to be considered, which include electro-spinning process (collecting time, drum rotating speed, andsyringe moving speed), precursor solution preparation (garnetsalt concentration, polymer concentration, polymer molecularweight, and solvent selection), and thermal annealing optimi-zation (heating rate, temperature, time, and cooling rate).Therefore, we believe that additional development of 3D ce-ramic nanofiber networks as well as polymer–salt optimizationbecome important and necessary in the future understanding ofmechanical properties and electrochemical performance of the3D ion-conducting network-based solid-state electrolyte forlithium batteries and beyond.Fig. 3 shows the morphological characterization of the garnet

nanofibers and resulting FRPC electrolyte. As shown in Fig. 3A,garnet nanofibers were bonded together at their intersectionpoints, forming a cross-linked network. These interconnectedgarnet nanofibers offer a continuous ion-conducting pathway be-cause of the extended long-range lithium transport channels, whichshould be superior to the isolated particle fillers that are dis-tributed in typical polymer matrixes (22). Fig. 3 B and C showsthe transmission electron microscopy (TEM) images of thegarnet nanofibers. The garnet nanofiber has a polycrystallinestructure consisting of interconnected small crystallites to form

Fig. 1. Schematic of the hybrid solid-state composite electrolyte, whereceramic garnet nanofibers function as the reinforcement and lithium-ion–conducting polymer functions as the matrix. The interwelded garnet nano-fiber network provides a continuous ion-conducting pathway in the elec-trolyte membrane.

2 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1600422113 Fu et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 13

, 202

0

the long, continuous nanofiber (Fig. 3B). Fig. S1 shows themagnified TEM image of a garnet nanofiber with an averagegrain size of 20 nm in diameter. Fig. 3C indicates the highlycrystalized structure of the garnet grain.The morphologies of FRPC electrolyte were examined by

SEM (Fig. 3 D–F). The FRPC electrolyte exhibited a smoothsurface, which came from the PEO–LiTFSI polymer (Fig. 3D).Inside of the FRPC electrolyte, we can see that the 3D porousgarnet nanofiber network supported the main structure of thecomposite and that the PEO–LiTFSI polymer was infiltratedinto the porous garnet membrane and filled the interspace be-tween garnet nanofibers. The cross-section image of the FRPCelectrolyte showed a thickness of 40–50 μm (Fig. 3E). To in-crease interphase contact between garnet nanofibers and PEO–

LiTFSI polymer, the FRPC electrolyte was thermally treated at60 °C, which is slightly above the polymer melting temperature

(Tm), to enable the melted PEO–LiTFSI polymer to fully in-filtrate the 3D porous garnet nanofiber network. As shown inFig. 3F, after thermal treatment, PEO–LiTFSI polymer was fullyembedded with garnet nanofibers. We can see that garnetnanofibers increased to an average diameter of 500 nm becauseof the PEO–LiTFSI polymer coating. The interconnected poreswere filled with polymer to maintain good lithium-ion transfer.The FRPC electrolyte membrane is proposed to have three ion-conducting pathways: the first one is the interwelded ceramicgarnet nanofiber network, the second one is the continuous garnetfiber–polymer interface, and the third one is the Li salt-containingpolymer matrix. Because of the higher ionic conductivity of gar-net-type electrolytes than that of Li salt-containing polymer elec-trolyte, we believe that the former two ion-conducting pathwaysare the dominant factors to provide improved ionic conductivity tothe electrolyte membrane.

Fig. 2. Fabrication of the flexible solid-state FRPC electrolyte. (A) Schematic setup of electrospinning garnet–PVP nanofibers. (B) Schematic procedure tofabricate the FRPC lithium-ion–conducting membrane. (C) SEM image of the as-spun nanofiber network. (D) Diameter distribution of the as-spun nanofibers.(E) SEM image of the garnet nanofiber network. (F) Diameter distribution of the garnet nanofibers. (G) Photo image to show the flexible and bendable FRPClithium-ion–conducting membrane.

Fig. 3. Morphological characterizations of garnet nanofiber reinforcement and the solid-state FRPC electrolyte. (A) SEM image showing the interweldedgarnet nanofibers. (B) TEM image of polycrystalline garnet nanofiber. (C) High-resolution TEM image of an individual garnet nanofiber. (D) SEM image ofFRPC electrolyte membrane surface. (E) Cross-sectional SEM image of the membrane. (F) Magnified SEM image of the cross-section morphology. The freespace of garnet 3D porous structure was filled with polymer.

Fu et al. PNAS Early Edition | 3 of 6

ENGINEE

RING

Dow

nloa

ded

by g

uest

on

Dec

embe

r 13

, 202

0

Thermogravimetric analysis (TGA) was used to study thegarnet nanofiber formation during the calcination process. TheTGA was carried out under airflow with a rapid heating rate of10 °C/min. Fig. 4A shows the TGA profile of the as-spun nano-fibers containing PVP polymer and garnet precursor. The resultshows that, above 750 °C, the weight became stable, indicatingthat stable garnet nanofibers were formed. Fig. 4B compares theTGA profiles of the PEO–LiTFSI and the FRPC electrolyte.Both electrolytes were thermally stable to around 200 °C. Inthe rapid heating process, polymers began to decompose above200 °C and showed a significant weight loss at around 400 °Cbecause of the almost complete decomposition of the polymer.The slope at 400 °C was the decomposition of LiTFSI. For theFRPC electrolyte, the weight was stable at 500 °C, and theremaining was the garnet nanofiber membrane caused bythe superior stability of garnet material in air. For the polymerelectrolyte, the weight was stable at 650 °C, leaving with decom-posed LiTFSI salt.Thermal stability is an important consideration for using solid-

state electrolytes, especially polymer electrolyte. Traditionalliquid electrolytes, such as carbonate electrolytes, tend to causethermal runaway when batteries are under extreme conditions ofshort circuits, overcharge, and high temperature (19). Because ofits relatively high thermal stability, polymer electrolyte becomesa safer choice compared with liquid electrolyte. Because tradi-tional polymer electrolytes are built on their own polymerstructure and fillers cannot offer sufficient mechanical supportfor the electrolyte, the polymer electrolyte inevitably melts andshrinks at high temperature, especially above the polymer ther-mal decomposition temperature, which may cause direct contactbetween cathode and anode and is a significant safety concern.The FRPC electrolyte is able to address this concern, becausethe garnet nanofiber membrane within the polymer electrolyteprovides a ceramic barrier to physically block cathode and anodecontact, even after loss of the polymer.Fig. 4 C and D compares the combustion tests of a traditional

polymer electrolyte and the FRPC electrolyte developed in this

work. The traditional polymer electrolyte was prepared using thesame recipe used to prepare the PEO–LTFSI polymer but usinggarnet nanopowders (vs. the 3D garnet network) as fillers. Themass ratio of polymer and filler was controlled at 4:1. In Fig. 4C,the polymer electrolyte caught fire instantly when it came closeto the ignited lighter and was quickly burned off into ashes. Thishigh flammability indicates poor thermal stability of the polymerelectrolyte. In comparison, the FRPC electrolyte exhibited anoutstanding thermal stability; although the polymer componentwas gone, the garnet nanofiber membrane still retained its3structure (Fig. 4D). This low-flammability FRPC electrolyte canprovide enhanced safety for all lithium metal and LIBs.Powder X-ray diffraction (XRD) patterns of LLZO garnet

nanofibers that were calcined at 800 °C for 2 h are shown in Fig.5A. Almost all of the diffraction peaks match very well with thoseof cubic-phase garnet Li5La3Nb2O12 (Joint Committee on Pow-der Diffraction Standards card 80-0457). Li5La3M2O5 (M = Nb,Ta) is the first example, to our knowledge, of a fast lithium-ion–conductive processing garnet-like structure, which is the typicalstructure that has been widely used as a model to study the garnetstructure of LLZOmaterial. Here, we use the standard Li5La3Nb2O12XRD profile to identify the synthesized garnet nanofiber structure.A small amount of La2Zr2O7 was identified, but other impuritieswere below detection limit. According to the thermogravimetricresults, decomposition of precursors to oxide was completed at∼750 °C. Additional heating at 800 °C resulted in reaction of theoxides and formation of cubic-phase LLZO garnet structure.However, the small amount of La2Zr2O7 phase could also beformed by lithium loss at elevated temperature.The total lithium-ion conductivity of FRPC electrolyte was

characterized by electrochemical impedance spectroscopy (EIS).Fig. 5B shows the typical Nyquist plots of FRPC electrolytesandwiched between stainless steel blocking electrodes in thefrequency range from 1 Hz to 1MHz. Each impedance profileshows a real axis intercept at high frequency, a semicircle at in-termediate frequency, and an inclined straight tail at low fre-quency. The intercept of the extended semicircle on the real axisand the semicircle in the high- and intermediate-frequency rangerepresent the bulk relaxation of FRPC electrolyte. The low-frequency

Fig. 4. Thermal properties and flammability tests of the solid-state FRPCelectrolyte. (A) TGA curve of the as-spun nanofibers. (B) TGA curves of Lisalt–PEO polymer and FRPC electrolyte membrane. (C) Flammability test of Lisalt–PEO polymer mixed with garnet nanoparticles. (D) Flammability test ofFRPC electrolyte membrane.

Fig. 5. Phase structure of garnet fiber and electrical properties of solid-state FRPC electrolyte. (A) XRD pattern of the garnet nanofibers and thepowder diffraction file (PDF) of Li5La2Nb2O12. (B) EIS profiles of the FRPCelectrolyte membrane at different temperatures (25 °C, 40 °C, and 90 °C). (C)Arrhenius plot of the FRPC electrolyte membrane at elevated temperatures(from 20 °C to 90 °C and record every 10 °C increase). (D) LSV curve of theFRPC electrolyte membrane to show the electrochemical stability window inthe range of 0–6 V. OCV, open-circuit voltage.

4 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1600422113 Fu et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 13

, 202

0

tail is caused by the migration of lithium ions and the surfaceinhomogeneity of the blocking electrodes. Fig. 5C shows theArrhenius plot of the FRPC electrolyte. Lithium-ion conductivitywas calculated based on the thickness of FRPC electrolyte anddiameter of stainless electrodes. As reported, lithium-ion con-ductivity of the cubic-phase LLZO garnet pellet would reach ashigh as 10−3 S/cm, whereas lithium salt-stuffed PEO is generallyon the order of 10−6–10−9 S/cm at room temperature (15, 22).Our FRPC electrolyte combining conductive cubic LLZO garnetand lithium–PEO could exhibit reasonably high ionic conduc-tivity of 2.5 × 10−4 S/cm at room temperature.A large electrochemical window is another key factor to deter-

mine the polymer electrolyte application for high-voltage lithiumbatteries. Fig. 5D shows the result of the linear sweep voltammetry(LSV) profile of the FRPC electrolyte using lithium metal as thecounter and reference electrode and stainless steel as the workingelectrode. The FRPC electrolyte exhibits a stable voltage windowup to 6.0 V vs. Li/Li+, indicating that this ion-conducting membranecan satisfy the requirement of most high-voltage lithium batteries.The mechanical stability of the FRPC electrolyte membrane

against Li dendrites was evaluated by using a symmetric Li j FRPCelectrolyte j Li cell. During charge and discharge processes at aconstant current, lithium ions are plating/stripping the lithium

metal electrode to mimic the operation of charging and dis-charging lithium metal batteries. Fig. 6A represents the sche-matic of the symmetric cell setup. The FRPC electrolyte membranewas sandwiched between two lithium metal foils and sealed incoin cell. Fig. 6B shows the time-dependent voltage profile of thecell with FRPC electrolyte membrane cycled over 230 h at aconstant current density of 0.2 mA/cm2 and a temperature of15 °C. The symmetric cell was periodically charged and dis-charged for 0.5 h. The positive voltage is the Li stripping, and thenegative voltage value refers to the Li plating process. In the first70 h, the cell’s voltage slightly increased from 0.3 to 0.4 V andthen, stabilized at 0.4 V.When the testing temperature increased to 25 °C, the voltage

dropped to 0.3 V because of the improved ionic conductivity atelevated temperature as shown in Fig. 6C. In the following long-time cycles, the voltage kept decreasing to 0.2 V with increasingcycle time to 700 h (Fig. S2). The fluctuation of voltage wascaused by the surrounding environmental temperature change.Two voltage profiles of the symmetric cell at two differentstripping/plating process times were compared as shown in Fig.S3. The voltage hysteresis apparently decreased with increase ofcycle time. This decrease in voltage is quite different from theliquid electrolyte system, in which the voltage normally increaseswith the increase of time and is mainly ascribed to the non-uniform Li deposition and severe electrolyte decomposition thatcause impedance increase (32). Similar voltage decrease hasbeen observed in recent polymer electrolyte studies, but thereason why voltage keeps decreasing with the increasing cycletime has not yet been explained (23, 33). Based on our un-derstanding, the decrease in voltage might be because of theimproved interface between the electrolyte membrane and lith-ium metal during the repeated Li electrodeposition, which isconfirmed by the EIS spectra of the symmetric cell measured at300, 500, and 700 h (Fig. 6D). The depressed semicircles at lowerfrequency indicate decreased interfacial impedance betweenelectrolyte membrane and lithium metal during cycling. At highfrequency (Fig. 6E), the semicircle also decreased with the in-creased cycle time, indicating the decreased bulk impedance ofthe electrolyte membrane. When the current density increased to0.5 mA/cm2, the voltage increased to 0.3 V, and the cell alsoexhibited slight decrease in voltage with increasing time to 1,000 h(Fig. 6F), showing good cycling stability with long cycle life.

ConclusionIn conclusion, all solid ion-conducting membranes of 3D garnet–polymer composite were synthesized for lithium batteries. 3Dgarnet nanofiber networks were prepared by electrospinning andhigh-temperature annealing. The garnet nanofibers constructedan interwelded 3D structure that provides long-range lithium-iontransfer pathways and further provides structural reinforcementto enhance the polymer matrix. This flexible solid-state electro-lyte composite membrane exhibited an ionic conductivity of 2.5 ×10−4 S/cm at room temperature. The membrane can effectivelyblock dendrites in a symmetric Li j electrolyte j Li cell duringrepeated lithium stripping/plating at room temperature, with acurrent density of 0.2 mA/cm2 around 500 h and a current densityof 0.5 mA/cm2 over 300 h. The decrease of voltage with increasingcycle time is observed for the symmetric cell, which is possiblybecause of the improved interfaces during repeated lithium elec-trodeposition. Our work is the first report, to our knowledge, of thedevelopment of 3D lithium-ion–conducting ceramic materials insolid-state electrolytes, which can be potentially applied to flexibleLIBs and other electrochemical energy storage systems, such aslithium–sulfur batteries.

ACKNOWLEDGMENTS. This work was supported by EERE, which is funded bythe US Department of Energy. We acknowledge the support of the Mary-land NanoCenter and its FabLab and NispLab.

Fig. 6. Electrochemical performance of the FRPC electrolyte membranemeasured in the symmetric Li j FRPC electrolyte j Li cell. (A) Schematic of thesymmetric cell for the lithium plating/stripping experiment. (B) Voltageprofile of the lithium plating/striping cycling with a current density of0.2 mA/cm2 at 15 °C. (C) Voltage profile of the continued lithium plating/stripping cycling with a current density of 0.2 mA/cm2 at 25 °C. (D) The im-pedance spectra of the symmetric cell measured at different cycle times (300,500, and 700 h). (E) Magnified EIS spectra in the high-frequency region.(F) Voltage profile of the continued lithium plating/stripping cycling with acurrent density of 0.5 mA/cm2 at 25 °C.

Fu et al. PNAS Early Edition | 5 of 6

ENGINEE

RING

Dow

nloa

ded

by g

uest

on

Dec

embe

r 13

, 202

0

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

2. Xu K (2014) Electrolytes and interphases in Li-ion batteries and beyond. Chem Rev114(23):11503–11618.

3. Bruce PG, Freunberger SA, Hardwick LJ, Tarascon JM (2011) Li-O2 and Li-S batterieswith high energy storage. Nat Mater 11(1):19–29.

4. Yin Y-X, Xin S, Guo Y-G, Wan L-J (2013) Lithium-sulfur batteries: Electrochemistry,materials, and prospects. Angew Chem Int Ed Engl 52(50):13186–13200.

5. Dunn B, Kamath H, Tarascon J-M (2011) Electrical energy storage for the grid: Abattery of choices. Science 334(6058):928–935.

6. Manthiram A, Fu Y, Chung SH, Zu C, Su YS (2014) Rechargeable lithium-sulfur bat-teries. Chem Rev 114(23):11751–11787.

7. Jung H-G, Hassoun J, Park J-B, Sun Y-K, Scrosati B (2012) An improved high-perfor-mance lithium-air battery. Nat Chem 4(7):579–585.

8. Cheng X-B, et al. (2015) A review of solid electrolyte interphases on lithium metalanode. Adv Sci 3(3):1500213.

9. Quartarone E, Mustarelli P (2011) Electrolytes for solid-state lithium rechargeablebatteries: Recent advances and perspectives. Chem Soc Rev 40(5):2525–2540.

10. Takada K (2013) Progress and prospective of solid-state lithium batteries. Acta Mater61(3):759–770.

11. Thangadurai V, Narayanan S, Pinzaru D (2014) Garnet-type solid-state fast Li ionconductors for Li batteries: Critical review. Chem Soc Rev 43(13):4714–4727.

12. Luntz AC, Voss J, Reuter K (2015) Interfacial challenges in solid-state Li ion batteries.J Phys Chem Lett 6(22):4599–4604.

13. Wang Y, et al. (2015) Design principles for solid-state lithium superionic conductors.Nat Mater 14(10):1026–1031.

14. Tan G, et al. (2016) Solid-state li-ion batteries using fast, stable, glassy nanocompositeelectrolytes for good safety and long cycle-life. Nano Lett 16(3):1960–1968.

15. Croce F, Appetecchi GB, Persi L, Scrosati B (1998) Nanocomposite polymer electrolytesfor lithium batteries. Nature 394(6692):456–458.

16. Gadjourova Z, Andreev YG, Tunstall DP, Bruce PG (2001) Ionic conductivity in crys-talline polymer electrolytes. Nature 412(6846):520–523.

17. Christie AM, Lilley SJ, Staunton E, Andreev YG, Bruce PG (2005) Increasing the con-ductivity of crystalline polymer electrolytes. Nature 433(7021):50–53.

18. Aetukuri NB, et al. (2015) Flexible ion-conducting composite membranes for lithiumbatteries. Adv Energy Mater 5(14):1500265.

19. Zhou D, et al. (2015) In situ synthesis of a hierarchical all-solid-state electrolyte basedon nitrile materials for high-performance lithium-ion batteries. Adv Energy Mater5(15):1500353.

20. Pan Q, Smith DM, Qi H, Wang S, Li CY (2015) Hybrid electrolytes with controllednetwork structures for lithium metal batteries. Adv Mater 27(39):5995–6001.

21. Zhang J, et al. (2015) Safety-reinforced poly(propylene carbonate)-based all-solid-state polymer electrolyte for ambient-temperature solid polymer lithium batteries.Adv Energy Mater 5(24):1501082.

22. Liu W, et al. (2015) Ionic conductivity enhancement of polymer electrolytes with ce-ramic nanowire fillers. Nano Lett 15(4):2740–2745.

23. ZhouW, Gao H, Goodenough JB (2016) Low-cost hollow mesoporous polymer spheresand all-solid-state lithium, sodium batteries. Adv Energy Mater 6(1):1501802.

24. Liu S, et al. (2010) Effect of nano-silica filler in polymer electrolyte on Li dendriteformation in Li/poly(ethylene oxide)–Li(CF3SO2)2N/Li. J Power Sources 195(19):6847–6853.

25. Gurevitch I, et al. (2013) Nanocomposites of titanium dioxide and polystyrene-poly(ethylene oxide) block copolymer as solid-state electrolytes for lithium metalbatteries. J Electrochem Soc 160(9):A1611–A1617.

26. Damen L, Hassoun J, Mastragostino M, Scrosati B (2010) Solid-state, rechargeable Li/LiFePO4 polymer battery for electric vehicle application. J Power Sources 195(19):6902–6904.

27. Choi JH, Lee CH, Yu JH, Doh CH, Lee SM (2015) Enhancement of ionic conductivity ofcomposite membranes for all-solid-state lithium rechargeable batteries incorporatingtetragonal Li7La3Zr2O12 into a polyethylene oxide matrix. J Power Sources 274:458–463.

28. Wang Y-J, Pan Y, Kim D (2006) Conductivity studies on ceramic Li1.3Al0.3Ti1.7(PO4)3-filled PEO-based solid composite polymer electrolytes. J Power Sources 159(1):690–701.

29. Lin D, et al. (2016) High ionic conductivity of composite solid polymer electrolyte viain situ synthesis of monodispersed SiO2 nanospheres in poly(ethylene oxide). NanoLett 16(1):459–465.

30. Teng S, Tan J, Tiwari A (2014) Recent developments in garnet based solid stateelectrolytes for thin film batteries. Curr Opin Solid State Mater Sci 18(1):29–38.

31. Du F, et al. (2015) All solid state lithium batteries based on lamellar garnet-type ce-ramic electrolytes. J Power Sources 300:24–28.

32. Liang Z, et al. (2015) Polymer nanofiber-guided uniform lithium deposition for bat-tery electrodes. Nano Lett 15(5):2910–2916.

33. Choudhury S, Mangal R, Agrawal A, Archer LA (2015) A highly reversible room-temperature lithium metal battery based on crosslinked hairy nanoparticles. NatCommun 6:10101.

34. Rangasamya E, Wolfenstineb J, Sakamoto J (2012) The role of Al and Li concentrationon the formation of cubic garnet solid electrolyte of nominal composition Li7La3Zr2O12.Solid State Ionics 206:28–32.

6 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1600422113 Fu et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 13

, 202

0