SEARCH

SEARCH BY CITATION

Keywords:

  • dendrites;
  • electrolytes;
  • lithium batteries;
  • nanoporous materials;
  • separators

Advances in materials that enable high-energy and safe electrochemical storage are understood to be a critical next step for portable electronic devices and for electric vehicles. Progress in both fields requires high-density, reliable and safe storage of electrical energy. Rechargeable lithium ion batteries (LIB), due to their high energy density, low internal resistance and minimal memory effects, are currently the most attractive storage technology;[1-4] they are expected to dominate the marketplace for the foreseeable future. Two well-known drawbacks of current LIB technology stem from the carbonaceous material used to host lithium in the anode. First, the 6/1 C/Li molar ratio in the anode lowers the anode specific capacity by more than one order of magnitude: from 3860 mAh g−1 to around 360 mAh g−1.[5] It also limits the choice of cathode to relatively low capacity, lithiated compounds, such as lithiated metal oxides, phosphates, and silicates, and presents a barrier to usage of novel, high-storage capacity cathode chemistries based on un-lithiated materials including oxygen, sulfur, and carbon dioxide.[6, 7] A second, less appreciated drawback arises from the small difference in potential that separates lithium insertion into the host and lithium plating onto the host.[8] Thus, either an overcharged or too quickly charged lithium ion battery can become a lithium metal battery (LMB), wherein the deposited metallic lithium provides the primary storage material in the anode. This means that notorious safety issues associated with non-uniform electrodeposition on metallic lithium anodes, dendrite formation and potential for catastrophic cell failure by internal short circuits,[8] which are normally associated with LMBs, are also an important concern for LIBs.

Development of electrolyte and separator platforms that permit safe and reliable cycling of lithium batteries that utilize metallic lithium anodes provide a potential solution to both of these problems, and is thus an important scientific undertaking. A key requirement of such an electrolyte/separator would be the ability to suppress or eliminate uneven electrodeposition of Li and/or to retard subsequent dendrite formation and proliferation during repeated charge-discharge cycles.[9] Several strategies have been proposed in the older literature for suppressing/managing lithium dendrite growth in LMBs. The list includes electrode coatings that prevent dendrite-induced short-circuits;[10] introducing so-called solid electrolyte interface (SEI) additives into the electrolyte, which facilitate electrodeposition of Li,[11] application of external pressure on the lithium metal electrode to “flatten” the electrode/electrolyte interface.[12] Mechanical blocking using solid or solid-like electrolytes with sufficiently high modulus to prevent growth of any formed dendrites[13, 14] has good theoretical support from the model of Monroe and Newman, which shows that an electrolyte with shear modulus around two times that of lithium metal can prevent dendrite growth in a LMB.[15] Unfortunately, the most commonly used electrolyte materials that present sufficiently high mechanical moduli, including polymers and ceramics, are insufficiently conductive and/or too brittle to be used in practical room-temperature LMBs.

Recently, new approaches have become available to retard lithium dendrite growth and proliferation in secondary batteries. Self-suspended suspensions of polyethylene glycol (PEG) functionalized nanoparticles have for example been shown to undergo a jamming transition, leading to formation of a nanoporous network of a lithium conductive PEG phase, mechanically reinforced by a silica nanoparticle network.[16, 17] While the confinement of PEG in the nanopores leads to moderate enhancements in Li transference number, the room temperature ionic conductivities of the jammed materials are not high. By blending similar silica nanoparticles tethered with an ionic liquid (IL) in a conventional propylene carbonate (PC)–lithium bis(trifluoromethanesulfonyl) imide (LiTFSI) electrolyte, Lu et al. showed that high room-temperature ionic conductivities can be achieved in such hybrids. And, when used as the electrolytes in LMBs, the IL-nanoparticle hybrid electrolytes lead to as much as a ten-fold increase in cell lifetime.[18] The authors explained their observation in terms of the space charge mechanism proposed by Chazaviel.[19] Specifically, it was argued that because the anion (TFSI) is the same for both the nanoparticle-tethered IL and for the electrolyte salt, the IL-tethered particles provide a reservoir of TFSI throughout the electrolyte that neutralizes development of the space charge. More recently an elegant, self-healing approach has been proposed for dendrite prevention through mixing small amounts of salts based on cesium ions with a lithium salt-based electrolyte.[20] This method has so far shown promising ability to prevent lithium dendrite formation in post-mortem studies of lithium electrodes, but it has been insufficiently studied to determine its applicability to LMBs cycled at moderate and high current densities.

Here, we report a model nanoporous separator/electrolyte configuration that facilitates both high mechanical moduli and facile ion transport at room temperature. The separator is created by laminating a nanoporous γ-Al2O3 sheet with a high pore density between macroporous poly(vinylidene fluoride-co-hexafluoropropylene) (PVDF-HFP) polymer layers to create a sandwich-type composite structure (Figure 1). The obtained composite separator exhibits remarkably improved toughness, in comparison to nanoporous alumina, and readily imbibes an electrolyte based on 1 m LiTFSI in PC to produce an electrolyte/separator material with room-temperature ionic conductivity above 1 mS cm−1 and mechanical modulus of at least 0.5 GPa at room temperature.[21] To the best of our knowledge the reported materials are among the first to exhibit this attractive combination of mechanical and ion transport properties at room temperature.

image

Figure 1. Middle and left: Schematics of the structure and preparation method of PVDF-HFP/Al2O3 separator. SEM images of the PVDF-HFP/Al2O3 with 100 nm nanopores: top, cross-section of the composite; right, cross-section of the internal alumina layer; bottom, boundary between alumina and polymer.

Download figure to PowerPoint

The nanoporous γ-Al2O3 is prepared by voltage-controlled anodic oxidation of metallic aluminum,[22, 23] and it is possible to precisely manipulate the pore dimensions through the processing voltage. PVDF-HFP/γ-Al2O3 separators were prepared using a phase separation method described in the Experimental Section. The resultant composite films were immersed in a 1 m LiTFSI/PC solution to form the electrolyte-separator used for electrochemical studies. The scanning electron microscopy (SEM) image (Figure 1) shows the material possesses a tri-layer, laminated structure in which the top and the bottom layers are PVDF-HFP and the middle layer is alumina comprised of a dense, uniform distribution of nanometer-sized pores. Before compositing with PVDF-HFP, the porous alumina is extremely brittle, which makes it difficult to handle. The macroporous PVDF-HFP coating produced by the phase separation procedure provides exceptionally high levels of mechanical reinforcement for the alumina, without infiltrating its pores. This configuration results in a composite film with dramatically higher mechanical flexibility, high mechanical modulus, and room temperature conductivity approaching that of the liquid LiTFSI/PC electrolyte hosted in the open pores of the PVDF-HFP/Al2O3 laminate.

The shear mechanical modulus is both intuitively and based on theory considered an important physical property for assessing the ability of an electrolyte/separator to impede lithium dendrite growth in a LMB. Because of its brittleness, the mechanical modulus of the unlaminated nanoporous alumina cannot be characterized using normal mechanical testing methods. We instead employ an atomic force microscopy (AFM) approach to first obtain a load-displacement curve (see Supporting Information Figure S3), and by applying Oliver and Pharr's method,[24] subsequently deduce the reduced elastic modulus to be around 500 MPa. This value is substantially lower than the theoretical modulus for bulk Al2O3 and somewhat lower than expected even if one factors in the nanoporous nature of the material.[25] It suggests that even the very small strains applied in the AFM measurement may cause some amount of brittle failure of the unlaminated material. This situation is quite different for the PVDF-HFP/Al2O3 composite film, which can be subjected to orders of magnitude larger mechanical deformations without showing any evidence of mechanical failure. Figure 2a reports the elastic/storage modulus of PVDF-HFP/Al2O3 separator film measured using dynamical mechanical analysis of a bulk specimen over a broad range of temperature (–130 °C to 150 °C). The pore size of the Al2O3 is varied from 20 nm to 200 nm and the figure also presents data for the macroporous PVDF-HFP copolymer film without the nanoporous Al2O3 interlayer, for comparison. It is apparent that irrespective of the measurement temperature, the elastic modulus for the PVDF-HFP/Al2O3 is about one order of magnitude higher than that of the PVDF-HFP and exhibits at most a weak dependence on the nanopore dimensions. At room temperature, the elastic modulus is close to 0.4 GPa for the composite material with the largest Al2O3 pore dimension. In every case, the elastic modulus decreases gradually with increasing temperature, which is attributed to the broad glass transition region for the highly random PVDF-HFP copolymer.[26] After soaking the PVDF-HFP/Al2O3 separators in a 1 m LiTFSI/PC electrolyte the materials become even tougher, but slippery, which makes it difficult to measure their mechanical modulus. Based on several repeat experiments we conclude that the storage modulus of a composite film based on Al2O3 is at least 0.15 GPa (see Supporting Information Figure S4).

image

Figure 2. a) Storage modulus of PVDF-HFP/Al2O3 separators with various pore sizes and for PVDF-HFP as a function of temperature. b) DC ionic conductivity of PVDF-HFP/Al2O3 separators with various pore sizes after immersing in LiTFSI/PC and separators with 100 nm pores after immersion in LiTFSI/PEG. c) Impedance spectra of PVDF-HFP/100 nm Al2O3/LiTFSI/PC against temperature. d) Bulk resistance and interfacial resistance of the composite in (c) against temperature, analyzed using the equivalent circuit shown in the inset.

Download figure to PowerPoint

Figure 2b reports the conductivity of PVDF-HFP/Al2O3/LiTFSI/PC electrolyte-separator with different pore sizes for temperatures ranging from –5 °C to 100 °C. The DC conductivity is determined from AC measurements (Supporting Information Figure S6) by the method proposed by Jonscher.[27] A typical AC conductivity vs. frequency profile is provided in the Supporting Information. The conductivity of PVDF-HFP/Al2O3/LiTFSI/PC electrolytes with 100 and 200 nm pore sizes are over 1 × 10−3 S cm−1 at room temperature, which is close to that of the LiTFSI/PC liquid electrolyte. Such high conductivity is attributed to the unrestricted movement of LiTFSI in liquid PC hosted in the pores of the separator. Materials based on Al2O3 with smaller, 20 nm pores shows slightly lower conductivity (8 × 10−4 S cm−1) at room temperature, which may be evidence that the decreased pore size to some extent hinders ion transport. The solid lines through the data are fitted using the Arrhenius equation inline image (see Supporting Information Figure S5). The deviations between theory and data in the low temperature region are attributed to the broad glass transition temperature of PVDF-HFP. The activation energy obtained from the fits is provided in Table S1 in the Supporting Information. The activation energy values are evidently all close to consensus values for LiTFSI/PC, implying that for the pore dimensions studied, the pores are large enough that the PVDF-HFP/Al2O3 serves essentially as a host for the liquid LiTFSI/PC electrolyte.

In addition to facilitating good ion transport in bulk, a suitable electrolyte for a LMB must also present low barriers for injection and removal of Li ions at the electrode/electrolyte interface. Figure 2c reports impedance spectra of a PVDF-HFP/Al2O3/LiTFSI/PC material based on Al2O3 with 100 nm nanopores measured in a symmetric lithium coin cell as a function of temperature. By fitting the results to the equivalent circuit model depicted in the inset to Figure 2d, both the bulk and interfacial resistance can be obtained (Supporting Information Table S2). It is apparent that both the interfacial and bulk resistances at 25 °C are low (48.8 Ω and 31.9 Ω, respectively) and as expected decrease with increasing temperature. The higher interfacial resistance compared with the bulk resistance shows that the main obstacle for ion conduction is the interfacial diffusion. The corresponding room-temperature bulk and interfacial resistances for the electrolyte-separators based on Al2O3 with 20 nm and 200 nm nanopores are, respectively, 9.5 Ω, 51.2 Ω, 60.6 Ω and 82.0 Ω. (see Supporting Information Figure S7 and Table S2). It means that for the range of pore dimensions studied, the materials are good candidates for application in batteries.

To assess the stability of our PVDF-HFP/Al2O3/LiTFSI/PC electrolyte-separators in batteries employing metallic lithium anodes, we performed electrochemical cycling of a Li/Li4Ti5O12(LTO) cell utilizing a laminated material based on Al2O3 with 100 nm pores as both the separator and electrolyte. This cell configuration was chosen because of the well-known, stable electrochemical cycling of Li/LTO cells in conventional electrolytes at both low and high rates. It therefore allows the new separator/electrolyte materials to be evaluated at high current densities and over large numbers of charge-discharge cycles to establish their performance limits.[28]

Figure 3a,b report the capacity and Coulombic efficiency as a function of cycle number at a fixed current density of 0.315 mA cm−2 (1C) and 1.575 mA cm−2 (5C). The experiment is also taken at 0.630 mA cm−2 (2C) (Supporting Information Figure S9). It is apparent from this figure that apart from a small amount of capacity fading over the first few cycles, the cells exhibit stable, high-efficiency cycling over at least 1100 charge/discharge cycles, with no evidence of short circuiting and with a capacity approaching the theoretical maximum (175 mAh g−1) for LTO. Scrutiny of Figure 3a,b shows that the initial capacity fading is accompanied by as rapid a decrease of the Coulombic efficiency from above 100% to stable values close to 100% for more than 1000 cycles. We therefore attribute the initial fading and the excess discharge to electrolyte decomposition and formation of the solid electrolyte interface during the first few cycles, as well as to unwanted side reactions, which further cause electrolyte decomposition.[29] Figure 3c,d report the galvanostatic charge-discharge profiles after the 1st, 100th, and 1000th cycles at a fixed current density of 0.315 mA cm−2 (1C) and 1.575 mA cm−2 (5C). It is apparent that the voltage plateau, round-trip efficiency, and capacity retention are all quite high.

image

Figure 3. Galvanostatic charging/discharging plots for Li/(PVDF-HFP/100 nm Al2O3/LiTFSI/PC)/LTO cells under: a) 0.315 mA cm−2 (1C) and b) 1.575 mA cm−2 (5C). Discharge and charge voltage profiles versus capacity for 1st, 100th, and 1000th cycle for c) 0.315 mA cm−2 (1C) and d) 1.575 mA cm−2 (5C).

Download figure to PowerPoint

To further evaluate the performance of the PVDF-HFP/Al2O3/LiTFSI/PC electrolyte in LMBs, we employed a cyclic lithium plate/strip electrochemical procedure in a symmetric lithium cell to characterize performance over extended periods of time. Because the capacity of the cathode/anode is not limited by the finite capacity of the LTO host used for the experiments reported in Figure 3, much larger amounts of lithium can be moved between electrodes within each cycle, which increases the chances of cell failure by dendrite-induced short circuits. In the present experiments a fixed protocol was used wherein cells were periodically charged for 3 h and discharged for 3 h at a range of current densities. Formation of a short circuit in this configuration produces an internal path for current flow in the electrolyte, which lowers the internal resistance and causes the measured voltage to drop. Thus, by monitoring the voltage versus time during these strip/plate cyclic experiments, it is possible to identify the onset of short-circuiting from the voltage drop. Figure 4a depicts the time-dependent voltage profile for a cell based on the electrolyte-separators containing Al2O3 with 100 nm pores, cycled for up to 1000 h under a constant current density of 0.2 mA cm−2. Figure 4b reports results from similar measurements, except without the nanoporous Al2O3 layer in the separator. The sudden drop observed in the peak-to-peak voltage amplitude in Figure 4b is attributed to cell failure by dendrite-induced short circuits. A variety of hypotheses have been presented in the literature for why these dendrites form,[19, 30-33] and proliferate to the point that they lead to cell failure.[15, 19] Although these results shed no new light on the dendrite nucleation processes, they clearly show that the pore configuration and mechanics of the electrolytes-separators employed are important impediments to dendrite proliferation. In particular, it is seen that while cells based on the PVDF-HFP/LiTFSI/PC fail by short circuiting after as little as 60 h of operation at 0.2 mA cm−2, cells based on the PVDF-HFP/Al2O3/LiTFSI/PC electrolyte exhibit stable voltage profiles even after 1000 h of operation. Considering the modulus of PVDF-HFP/Al2O3 is almost ten times higher than that of PVDF-HFP itself, this result is consistent with Monroe and Newman's analysis.[15] Results from similar measurements using PVDF-HFP/Al2O3/LiTFSI/PC electrolyte-separators at both lower and higher current densities are provided in the supporting materials section (Supporting Information Figure S10). It is apparent from these experiments that the materials are quite effective in stabilizing the cells against failure by short circuiting.

image

Figure 4. Voltage profiles for lithium plating/striping experiment as a function of time for symmetric lithium coin cells cycled at a fixed current density of 0.2 mA cm−2 using: a) PVDF-HFP/100 nm Al2O3/LITFSI/PC and b) PVDF-HFP/LiTFSI/PC as separators/electrolytes. The thickness for both separator samples is approximately 0.8 mm.

Download figure to PowerPoint

Post mortem studies of the laminated separator (see Supporting Information Figure S11 and S12) taken after 1000 h of operation show that mossy deposits form in small patches on the surface and edges of the electrolyte-separators, meaning that the PVDF-HFP/Al2O3/LiTFSI/PC materials does not stop uneven electrodeposition, but are apparently able to contain the non-uniform deposits within a small region near the electrode/electrolyte interface. As a comparison with earlier studies, the cell lifetimes achieved with the present PVDF-HFP/Al2O3/LiTFSI/PC materials are a substantial improvement over recent reports for a block copolymer electrolyte with storage modulus of 0.1 GPa.[33] Significantly, because the PVDF-HFP/Al2O3-based materials are more conductive, the present results are obtained from room temperature measurements, which represents an additional improvement over copolymer based electrolytes, which currently require battery operating temperatures around 90 °C.

In summary, we report a nanoporous electrolyte-separator comprised of laminated PVDF-HFP/Al2O3 composite membranes infused with a conventional low-volatility, liquid electrolyte. The materials are shown to be stable against metallic lithium and exhibit good toughness, high mechanical modulus, high ionic conductivity, and low interfacial impedances at room temperature. Using Li/LTO cells we show that the materials allow for exceptional, stable cycling performances for more than 1000 charge/discharge cycles. In symmetric lithium/lithium cells, the PVDF-HFP/Al2O3 membranes exhibit more than 1000 h of stable operation at current densities ranging from 0.02 to 0.2 mA cm−2. These last results are substantial improvements over symmetric cells based on PVDF-HFP without the Al2O3 inter-layer, but soaked in the same liquid electrolyte; these cells fail in as little as 60 h when cycled at 0.2 mA cm−2. The porous media electrolyte concept explored in this study provides a powerful, new approach for creating high-modulus electrolytes with acceptable room-temperature ionic conductivity. It is straightforward to extend it to other nanoporous ceramics, solid polymers, and their laminates in which the areal density of pores is high.

Experimental Section

  1. Top of page
  2. Experimental Section
  3. Acknowledgements
  4. Supporting Information

Polyvinylidene fluoride hexafluropropylene (PVDF-HFP, supplied by Sigma Aldrich.) was dissolved in N,N-dimethylformamide (DMF, supplied by Sigma Aldrich) at 10 wt% concentration. The viscous solution was poured onto a clean glass plate, covered by nanoporous alumina membrane (Whatman Anodisc 25 with 20 nm, 100 nm, and 200 nm pore sizes, supplied by Fisher). The surface flatness and overall laminate membrane thickness (around 0.8 mm) were controlled using a doctor blade technique. The formed solid composite separator was completely dehydrated in vacuum. To prepare the electrolyte separator, the composite separator was soaked in 1 m lithium bis(trifluoromethanesulfonyl)imide (LiTFSI)/propylene carbonate (PC) solution for at least 24 h. The symmetric lithium coin cells and the Li/LTO coin cells (both 2032 type) were prepared under argon protection (glove box, MBraun. Labmaster). The symmetric lithium/lithium coin cells had Li/(PVDF-HFP/Al2O3/LiTFSI/PC)/Li structure, while the Li/LTO coin cells have Li/(PVDF-HFP/Al2O3/LiTFSI/PC)/LTO structure. The LTO electrode was composed of 10% PVDF binder, 10% carbon black, and 80% LTO. A small amount of N-methylpyrrolidone (NMP) was used as solvent for homogenizing all components. The resultant slurry was coated on a copper plate and rigorously dried. Because the laminated PVDF-HFP/Al2O3/LiTFSI/PC electrolyte-separators hosted large amounts of liquid electrolyte that wets LTO well, it was assumed that activation of the LTO electrode occured almost immediately after contact with the electrolyte-separator and none of the usual electrochemical activation processes were used in the current experiments.

A thermogravimetric analyzer (TGA) was used to study the thermal stability of PVDF-HFP/100 nm Al2O3/LiTFSI/PC and PVDF-HFP/100 nm Al2O3/LiTFSI/PEG (Supporting Information Figure S8). Scanning electron microscopy (LEO-1550-FESEM) was used to characterize the laminated structure in the composite separator. The separator was cut in liquid nitrogen to achieve clean edges. The sample was placed vertically on a SEM stub for cross-section observation. The SEM images were obtained under 3 kV voltages with aperture size of 30 μm. Mechanical properties of the separator/electrolyte materials were characterized using dynamic mechanical analysis (DMA-Q800) in the temperature range from –130 °C to 150 °C. A heating rate of 10 °C min−1 and frequency of 1 Hz were employed for these measurements. Atomic force microscopy (AFM, Asylum-MFP-3D-Bio-AFM) was used to indirectly measure the modulus of the unlaminated Al2O3 film. The force mode was chosen to obtain the force plot against the indent depth. Conductivity and impedance were measured against frequency using a Novocontrol N40 broadband dielectric spectroscopy at different temperature from –5 °C to 100 °C. The lithium plate/strip experiment and galvanostatic charge/discharge experiment were performed on a Neware CT-3008 battery tester. The plate-strip experiment was performed with symmetric lithium coin cells under different current density (0.02, 0.05, 0.1, and 0.2 mA cm−2). The coin cells harvested after the plate-strip experiment were taken apart in a glove box and the separator/electrolytes dried in the vacuum chamber of the glove box and stored for SEM analysis. The galvanostatic experiment was performed under different charging/discharging rate (0.315, 0.630, and 1.575 mA cm−2, which corresponds to 1C, 2C and 5C, respectively).

Acknowledgements

  1. Top of page
  2. Experimental Section
  3. Acknowledgements
  4. Supporting Information

This material is based on work supported as part of the Energy Materials Center at Cornell, an Energy Frontier Research Center funded by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences under Award Number DESC0001086. This work made use of the electrochemical characterization facilities of the KAUST-CU Center for Energy and Sustainability, which is supported by the King Abdullah University of Science and Technology (KAUST) through Award # KUS-C1–018–02. Electron microscopy facilities at the Cornell Center for Materials Research (CCMR), an NSF supported MRSEC through Grant DMR-1120296, were also used for the study.

Supporting Information

  1. Top of page
  2. Experimental Section
  3. Acknowledgements
  4. Supporting Information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re-organized for online delivery, but are not copy-edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

FilenameFormatSizeDescription
aenm201300654-sup-0001-S1.pdf1832KSupplementary

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.