Hybrid Li Ion Conducting Membrane as Protection for the Li Anode in

May 8, 2017 - Search; Citation; Subject .... *E-mail: [email protected]. This article is part of the Fundamental Interfacial Science for Energy...
0 downloads 0 Views 2MB Size
Article pubs.acs.org/Langmuir

Hybrid Li Ion Conducting Membrane as Protection for the Li Anode in an Aqueous Li−Air Battery: Coupling Sol−Gel Chemistry and Electrospinning Gilles Lancel,†,‡ Philippe Stevens,‡ Gwenael̈ le Toussaint,‡ Manuel Maréchal,§ Natacha Krins,†,# Damien Bregiroux,†,# and Christel Laberty-Robert*,†,# †

Sorbonne Universités, UPMC Univ Paris 06, CNRS, Collège de France, Laboratoire de Chimie de la Matière Condensée de Paris, 4 place Jussieu, 75005 Paris, France ‡ EDF R&D, 77818 Moret Sur Loing, Cedex, France § Univ. Grenoble Alpes, CNRS, CEA, INAC, SYMMES, F-38000 Grenoble, France S Supporting Information *

ABSTRACT: Aqueous lithium−air batteries have very high theoretical energy densities, which potentially makes this technology very interesting for energy storage in electric mobility applications. However, the aqueous electrolyte requires the use of a watertight layer to protect the lithium metal, typically a thick NASICON glass−ceramic layer, which adds ohmic resistance and penalizes performance. This article deals with the replacement of this ceramic electrolyte by a hybrid organic−inorganic membrane. This new membrane combines an ionically conducting inorganic phase for Li ion transport (Li1.3Al0.3Ti1.7(PO4)3 (LATP) and a poly(vinylidene fluoride-co-hexafluoropropylene) (PVDF−HFP) polymer for water tightness and mechanical properties. The Li ion transport through the membrane is ensured by an interconnected 3-D network of crystalline LATP fibers obtained by coupling an electrospinning process with the sol−gel synthesis followed by thermal treatment. After an impregnation step with PVDF−HFP, hybrid membranes with different volumetric fractions of PVDF−HFP were synthesized. These membranes are watertight and have Li ion conductivities ranging from 10−5 to 10−4 mS/ cm. The conductivity depends on the PVDF−HFP volume fraction and the fibers’ alignment in the membrane thickness, which in turn can be tuned by adjusting the water content in the electrospinning chamber during the process. The alignment of fibers parallel to the membrane surface is conductive to poor conductivity values whereas a disordered fiber mat leads to interesting conductivity values (1 × 10−4 mS/cm) at ambient temperature.



negative electrode and discharge product weight2. However, the power densities of the battery are limited by the ohmic loss essentially caused by the electrical resistance of the solid electrolyte.3 All things taken into account, an energy density of 500 W h/kg is considered to be a more realistic first objective (but not a limit) that has already been achieved on the laboratory scale.6 The first rechargeable lithium−air battery was made of a Li metal foil as the anode, an organic polymer membrane as the solid electrolyte, and a thin carbon composite as the oxygen electrode.4 The limitation of Li−air batteries using an organic solvent is both the formation of an insulating layer of Li2O2

INTRODUCTION One of the main limitations in the development of electrical vehicles (EV) is their relatively short driving range (150−400 km) compared to that of internal combustion engine (ICE)powered vehicles. The driving range is mainly determined by the battery energy density used to power the EV. A breakthrough in battery technology is needed to be able to extend this range beyond 400 km. Lithium−air batteries, which are based on the Li/O2 electrochemical couple, is the most attractive in terms of energy density.1 The lithium−air cell configuration shown in Figure 1 is based on the use of lithium metal as the negative electrode in combination with an oxygen positive electrode.2 This electrode has infinite capacity because the active material, oxygen from the air, is not stored in the battery. Only the lithium negative electrode, the electrolyte, and the product of the reaction are stored in the battery, which are the limiting factor affecting the energy density. Among the metal−air batteries, the lithium−air battery has the highest theoretical energy density of 3582 W h/kg based on the © XXXX American Chemical Society

Special Issue: Fundamental Interfacial Science for Energy Applications Received: February 28, 2017 Revised: May 3, 2017 Published: May 8, 2017 A

DOI: 10.1021/acs.langmuir.7b00675 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

Figure 1. (a) Aqueous lithium−air battery proposed by Stevens et al.6 (b) Schematic of the hybrid membrane (details in Figure S1).

LiPON protective layer of the LISICON membrane.12 Lithium metal was then electrochemically deposited between the LiPON thin film and the stainless steel current collector during the first charge process. This also has the advantage of being able to fabricate the battery without any lithium metal, thus reducing the risks and constraints associated with handling lithium metal.12 Previous studies on aqueous lithium−air batteries highlighted that the thickness of a glass−ceramic LISICON membrane coated with a thin protective LiPON film was still too high; a resistance of 67 Ω has been determined for a 55-μm-thick membrane of 1 cm2 area.14 Such a resistance causes an important ohmic loss that limits the battery’s power. Small cracks can also form with very thin ceramic electrolytes that will affect its watertight protection. In this study, a different approach was investigated to answer these issues. A hybrid organic−inorganic membrane combining two components was developed and manufactured with the objective of replacing the LISICON glass−ceramic (Figure 1a). The advantage of this approach is that thin hybrid membranes of ∼20 μm can be easily manufactured. These hybrid membranes present characteristics from both the organic and ceramic components, combining high shear modulus, flexibility, and transport properties. Hybrid organic−inorganic membranes have been investigated in the past to improve the Li ion conductivity in lithium−ion batteries but also in hightemperature fuel cells to improve proton conductivity at both relatively high humidity (30% of RH) and high temperature (100 °C).15−17 In the latter, the strategy was to decouple the transport of protons from the mechanical properties of the membrane. The inorganic component was responsible for the proton conduction whereas the polymer introduced flexibility into the membrane. This approach enables materials with conductivities comparable to the state of the art by controlling the membrane structure at different scales: going from the molecular scale to the macroscale. The aim of this article is to apply this approach to the aqueous lithium−air device and to design a thin (≤30 μm) lithium ion-conducting hybrid membrane. In this membrane, the mechanical properties, the inertness, and the water tightness are brought about by the polymer, whereas the ion-

inside the pores of the oxygen electrode and the decomposition of the organic solvent resulting from the formation of a superoxide radical.5 Aqueous lithium−air batteries using aqueous electrolytes such as a LiOH solution do not have these problems because the product of the reaction (LiOH· H2O) is stored as a solid in the electrolyte and not in the electrode.6 In addition, the inclusion of a polymeric anionexchange membrane at the air electrode/electrolyte interface prevents both LiOH precipitation inside the electrode and the formation of Li2CO3, increasing the lifetime of the air electrode from 10 to 1000 h.6,7 This modification enables the Li−air batteries to work in untreated air. Another strong limitation to the lifetime of this battery is the poor reversibility of the air electrode. Toussaint et al. designed an oxygen bielectrode where a stainless steel electrode is placed in the aqueous electrolyte to shunt the porous air electrode during charge,8 increasing the durability of the electrode by up to more than 1000 cycles. This additional electrode plays the role of an oxygen evolution electrode. In this aqueous lithium−air battery design, the lithium metal electrode needs to be protected to prevent its oxidation by water. A lithium ion conducting membrane separating the metal from the aqueous environment is placed between them. This protective barrier needs to be watertight and stable to both the aqueous electrolyte and Li metal. Water-stable NaSICON-like electrolyte Li1.3Al0.3Ti1.7(PO4)3 (LATP) is a good candidate with a conductivity of 7 × 10−1 mS/cm which has a theoretical ohmic resistance of 14 Ω·cm2 for a 100-μm-thick electrolyte.9 Comparable conductivities have been found for a composite membrane, including LATP containing a small amount of AlPO4.10 To stabilize these materials versus lithium metal, a thin layer of lithium phosphorus oxynitride glass (LiPON) is used to protect the solid electrolyte.11−13 Alternatively, an interfacial layer between the LISICON membrane and the lithium metal can be used. This interfacial layer can be, for example, a liquid organic electrolyte such as 1 M LiClO4 in ethylene carbonate/dimethyl carbonate, but this strategy does not prevent lithium dendrites from coming into contact with the ceramic electrolyte, leading to its degradation over cycling. A very good electrode/electrolyte interface has been achieved by depositing a stainless steel thin film (0.3 μm) onto the B

DOI: 10.1021/acs.langmuir.7b00675 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

until complete dissolution was achieved. After homogenization, 2 mL of the solution was electrospun at 20 kV and with a flow rate ranging from 10 to 100 μL/min, in air with a relative humidity ranging from 4 to 10 g/m3 at ambient temperature (∼25 °C). The counter electrode was a rotating aluminum roll. The distance between the spinneret and the counter electrode was varied from 10 to 15 cm. A white, flexible membrane was deposited onto the aluminum foil that plays the role of counter electrode (placed on the rotating electrode). Finally, the electrospun membranes were dried for 24 h at 70 °C. The electrospun membrane was then detached from the support, and a homogeneous self-standing membrane was obtained. The electrospun membrane was then calcined in air at various temperatures, from 800 to 950 °C, to produce crystalline LATP fibers. Synthesis of the Hybrid LATP Membranes. The LATP fiber mat was impregnated with a solution with 3 wt % PVDF−HFP in DMF. (Note that in this configuration, PVDF−HFP is used as a polymer to bring about the watertightness of the membrane.) To achieve homogeneous hybrid membranes, two successive impregnations were necessary. The solvent was then evaporated in air at 70 °C for 24 h. The thickness of the membranes obtained varied between ∼20 and 50 μm. During the impregnation step, the mass of the solution was weighed. By knowing the mass concentration of the solution and the mass of the membrane, it was then possible to determine the volume fraction of PVDF−HFP in the membrane. Spark Plasma Sintering. Dense ceramic pellets of LATP were formed by spark plasma sintering. The sintering was performed at 900 °C for 1 min under 0.1 mbar vacuum by a spark plasma sintering technique (Dr. Sinter 515S Syntex machine). A graphite die with an inner diameter of 8 mm was filled with the LATP powder. The temperature was monitored by using a thermocouple. The sample shrinkage was followed by using the displacement of the lower punch. The heating rate, cooling rate, and applied pressure were set to 100 °C/min, 50 °C/min, and 100 MPa, respectively. Pressure was applied gradually from room temperature to 500 °C and maintained at its maximum value until the beginning of the cooling step, and the pressure was released over 2 min. The sintered sample was then annealed under flowing air at 800 °C for 12 h in order to remove carbon contamination from the graphite die and oxygen vacancies generated by the reducing environment during sintering. Scanning Electron Microscopy. Membranes morphologies were observed by SEM with a Hitachi S-3400N while element mapping was performed with an Oxford Xmax energy-dispersive X-ray (EDX) detector. Prior to the observation, a thin layer (∼10 nm) of carbon or gold was deposited onto the membranes or the fiber mats to facilitate the observation. FE−SEM analyses were performed using a Hitachi SU-70 microscope. The 3D reconstruction was obtained using Fiji software. The set of images was first converted in a stack and then segmented. The stack was then reframed using the Crop3D plugin. Viewing in 3D was carried out with the plugin VolumeViewer developed by Shmid et al.19 The estimation of the diameter of the fibers from the SEM images was performed with Fiji software using the DiameterJ plugin. The standard deviation was determined using 100 measurements made on several images. EDX analyses performed on different batches of LATP pellets indicated that the synthesized powders have the following composition: Li1.4Al0.4Ti1.6(PO4)3. For the sake of simplicity, the powders will be called LATP in the text. High-Resolution Transmission Electron Microscopy. HRTEM imaging was performed on a FEI Tecan Spirit 120 operating at 200 kV equipped with a camera CCD Gatan Orius 1000. The fibers or powders were first dispersed in an ethanolic solution and then transferred onto a copper grid covered with a carbon holey membrane. X-ray Analyses. X-ray diffraction analyses were performed using a D8 diffractometer operating in Bragg−Brentano mode and equipped with a LinxEye detector with a variable slit and Ni filter. For the Lebail mode, a fixed slit of 0.5 mm was used. SAXS Measurements. SAXS measurements were performed at the ESRF (European Synchrotron Radiation Facility, Grenoble, France) on the BM02-D2AM beamline. The incident photon energy

conducting ceramic ensures the transport of Li ions throughout the membrane (Figure 1b).18 LATP ceramic is used because it is made of abundant elements, has good stability in alkaline electrolytes, and has high Li+ ionic conductivity. The polymer was poly(vinylidene fluoride-co-hexafluoropropylene (PVDF− HFP) because of its processability, watertightness, and stability toward the alkaline aqueous electrolyte. Its stability under an alkaline aqueous electrolyte was tested before use. We have demonstrated that no modification of its chemical and microstructure occurred after immersing a membrane for 1 month in a 5 mol/L aqueous LiOH electrolyte (Figure SI1). A 3-D network of intermingled LATP fibers was f irst synthesized by coupling the electrospinning process and the sol−gel chemistry, followed by a thermal treatment. The solid felt of lithium ion-conducting LATP fibers was then impregnated with the polymer to give it watertight and gastight properties. By proceeding in this manner, the interconnected percolating network of the lithium ion conductor is preserved within the solid polymer membrane. The volume fraction of PVDF−HFP was tuned to achieve optimal watertightness. The electrical characterization of the hybrid membranes and its water tightness were evaluated as a function of the concentration of polymer in the membrane and the microstructure of the intermingled 3-D network of LATP fibers. Finally, the electrochemical properties of the best membrane were evaluated.



EXPERIMENTAL SECTION

Materials. The polymer used in this work was poly(vinylidene fluoride-co-hexafluoropropene) (PVDF−HFP) (Solvay (Solef 21216), Mn = 570−600). Titanium butoxide (Ti(OiPr)4, 98%, ABCR), acetylacetone (C5H8O2, 99%, Sigma-Aldrich), tetrahydrofuran (THF, C4H8O, VWR analaR normapur), dimethylformamide (DMF, C 3 H 7 NO, VWR analaR normapur), phenylphosphonic acid (PhPO3H2, 99%, Sigma-Aldrich), aluminum nitrate nonahydrate (Al(NO3)3·9H2O, 99.99%, Alfa Aesar), and lithium nitrate (LiNO3 99.99%, Alfa Aesar) were used for the synthesis of the sol. Prior to the synthesis, THF and DMF were dried using molecular sieves (3 Å, Sigma-Aldrich). All of the materials were kept in a desiccator in the presence of silica gel (Sigma-Aldrich) to avoid hydration. Synthesis of LATP Powder (Target Composition Li1.4Al0.4Ti1.6(PO4)3). The synthesis of LATP powder was first performed in order to determine the synthesis conditions necessary for the preparation of pure LATP. First, titanium butoxide was stabilized by mixing it with 2 equiv of acetylacetone. This mixture was then added to 100 mL of THF in a 250 mL flask. Phenylphosphonic acid, Al(NO3)3·9H2O, titanium butoxide, and Li(NO3) were added in the molar ratio 1.6:3:0.4:1.4 to the solution. The concentration of the precursor in the final solution was 1.67 M. The solution was stirred until complete dissolution was achieved. The solvent was then evaporated using a rotary evaporator until a gel formed. This gel was then calcined at 500 °C for 2 h in air to produce a powder that was ground with a mortar for several minutes. After grinding, the agglomerate size ranges from 1 to 50 μm. The resulting powder was then calcined under flowing air at different temperatures ranging from 700 to 950 °C for 2 h with a heating ramp of 5 °C/min. The final powder was ground with attrition milling using 2 mm zirconia balls. The agglomerate size decreased from 1 to 5 μm. Synthesis of the Electrospun LATP Fiber Mat. First, titanium butoxide was stabilized and mixed with 2 equiv of acetylacetone. Four hundred milligrams of PVDF−HFP (used here as the binder), 10 mL of tetrahydrofuran (THF), and 10 mL of dimethylformamide (DMF) were added to a 30 mL flask and stirred for 24 h until complete dissolution of the polymer. Then, the Ti/acetylacetone (molar ratio 1:2) mixture, phenylphosphonic acid, Al(NO3)3·9H2O, and Li(NO3) were added in a molar ratio of 1.6:3:0.4:1.4. The concentration of precursor in the final solution was 1.67 M. The solution was stirred C

DOI: 10.1021/acs.langmuir.7b00675 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

Figure 2. Characterization of the LATP powders obtained after calcining the gel at different temperatures for 2 h under air. (a, b) SEM images of powders (a) before and (b) after attrition. (c) Powder X-ray diffraction analyses of the powders heat-treated before attrition milling at (i) 700, (ii) 800, (iii) 900, and (iv) 950 °C and of the pellet after SPS treatment at (v) 900 °C, 1 min. (d) Experimental impedance diagram of the pellet and its fitting. The decades in frequency are mentioned. The diagram was fitted according to the equivalent circuit reported in Figure S4. was tuned to 11 keV, which corresponds to a wavelength of λ = 1.12 Å. A 2-D detector, a CDD camera developed by Princeton, presently Ropper Scientific, was used. The magnitude of the scattering vector is defined as q = (4π/λ)sin θ, where θ is half of the scattering angle and λ is the wavelength. The distance from the sample to the detector was 164 cm, which covers the q range from 0.06 to 1.56 nm−1. The corrections to the primary data were carried out using the Bm2Img software available on the beamline: (i) the dark current (i.e., nonilluminated camera), (ii) the flat-field response (i.e., a homogeneously illuminated camera), and (iii) the tapper distortion. The standard silver behenate was used for the q-range calibration. Twodimensional images were converted into radial averages over the image center to yield scattered intensity I vs scattering vector q. Conductivity. The through lithium conductivities of the electrospun LATP ceramics, fiber mats, and LATP/PVDF−HFP membranes were measured using a homemade two-electrode setup. The twoelectrode measurements were performed at ambient temperature in air and at various temperatures (from 25 to 100 °C) to estimate the activation energy. The dry dimensions of the membrane were used for conductivity calculations. Prior to the experiments, circular gold electrodes (thickness, 100 nm; area, 7 mm2) were deposited on each face of the membrane to ensure good contacts. The frequency range varied between 1 and 106 Hz with an amplitude of 100 mVrms at 0 V. Water Tightness. The watertightness experiments were performed in symmetrical cells composed of two compartments, containing either satured LiCl solution (100 mS/cm) or water obtained via osmosis (4 μS/cm), separated by the hybrid membrane. The conductivity of the compartment for water obtained via osmosis was measured after 24 h.

No change in conductivity indicates that ions do not diffuse across the membrane.



RESULTS AND DISCUSSION Synthesis and Characterization of LATP Powders from the Calcination of the Sol Precursor. The experimental conditions for the synthesis of a pure LATP phase were optimized by varying the heat-treatment temperature. The effect of the heating temperature on LATP crystallization was investigated by X-ray diffraction analysis, and the results are summarized in Figure 2c(i−iv). The peaks corresponding to the LATP phases appear at 700 °C beside impurities, which include TiO2, Li3PO4, and AlPO4. In contrast to the powders calcined at T ≥ 800 °C, here only the pure phase was obtained because only the peaks characteristic of the LATP phase are observed on the XRD pattern. These results agree with those in the literature, suggesting that high temperatures (T ≥ 800 °C) are needed to favor the substitution of Ti by Al.20,21 To gain further insight into whether Al is substituted, the X-ray diffraction patterns were refined using the Le Bail mode because the substitution of Ti by Al causes a decrease in the cell volume. This is due to the difference in ionic radius in 6-fold coordination between Al3+ (0.535 Å) and Ti4+ (0.605 Å) (Supporting Information, Table S1).22,23 From this modeling, parameters a, b (8.485 ± 0.001 Å), and c (20.797 ± 0.001 Å) were estimated. The obtained D

DOI: 10.1021/acs.langmuir.7b00675 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

Figure 3. Self-standing membrane of LATP fibers obtained by electrospinning after calcination. (a) Picture. (b) Cross-sectional SEM image. (c, d) TEM images of isolated fibers.

Impedance spectroscopy measurements were performed on the dense LATP pellets (Figure 2d). The frequency range used was 1 to 106 Hz. The electrical behavior was modeled using a series circuit consisting of a resistance Rmes with a resistance R in parallel with a constant phase element (CPE) for both the grain (Rg and CPEg) and the grain boundary response (Rjg, CPEjg), and a CPE to take into account the nonlinearity of the response (Figure S4). Using this model, a total conductivity of 5 × 10−1 mS/cm was obtained at room temperature for LATP ceramics with a relative density of 97%. This value agrees well with those found in the literature.26−28 By using the conductivity values measured at various temperatures, the activation energy was estimated to be 0.35 ± 0.03 eV. This value compares well with those reported for LATP.26,29 This set of experiments shows that powders synthesized from the sol− gel process exhibit the expected electrical behavior. Elaboration and Characterization of the Electrospun LATP Fiber Mat. The process leading to the synthesis of the electrospun LATP fibers is shown schematically in Figure S1. After homogenization, the precursor−PVDF−HFP solution was electrospun at ambient temperature (25 °C) in air under controlled humidity. After the deposition, drying, and calcination steps, self-standing white hybrid membranes made of LATP fibers were obtained (Figure 3). A photograph of the LATP fiber mat is reported in Figure 3a, showing the flexibility of the mat. A bending radius of about 1 cm could be used without breaking the fibers in the mat, as confirmed by postmortem analysis. The cross-sectional SEM image indicates that the fibers are well distributed and connected across the membrane (Figure 3b). Shrinkage of the fiber mat of 35% has been observed after heating the membrane to 950 °C. The thickness of the LATP fiber mat varies between 10 and 50 μm, depending on the volume of the precursor−PVDF−HFP solution. The Li ion conductivity of the electrospun LATP membranes is highly dependent on the microstructure of the LATP mats (i.e., the density and the organization of the fibers and the diameter of the fibers, their

values are smaller than the ones observed for LiTi2(PO4)3 (LTP), suggesting the substitution of Ti by Al.24 From these analyses, the crystallite domains were estimated to be around 100 nm. SEM imaging of the product calcined at 950 °C shows agglomerated powders with an agglomerate size ranging from 10 to 100 μm (Figure 2a,b). The specific surface area determined by the BET method is inferior to 1 m2/g, attesting to the presence of large agglomerates. Attrition milling was then used to decrease the agglomerate size in order to obtain dense pellets for the Li ion conductivity measurements. After attrition, the agglomerate size decreases to 2−5 μm, and the specific surface area increases to 7 m2/g. Ceramics were first synthesized using conventional sintering. Even after an optimization of the sintering parameters (temperature and dwelling time), the pellets remained porous (maximum relative density of 75%) and exhibited abnormal grain growth. To determine the conductivity of the powder synthesized within this study, spark plasma sintering (SPS) was used with the goal of obtaining a dense ceramic at lower sintering temperatures compared to conventional sintering techniques. This sintering process is very fast and encourages the densification of nanosize powders or nanostructures while limiting coarsening, which often accompanies standard densification routes. The optimization of the SPS sintering parameters is very important to achieving dense ceramics with reproducible conductivity values. The effects of the sintering temperature on the LATP pellet density were investigated. Different temperatures ranging from 850 to 950 °C were tested, and the results are reported in the Supporting Information (Figure S2). The optimal sintering temperature was found to be 900 °C for 1 min to produce an LATP ceramic with a relative density of 97%, which is consistent with the results found in the literature.25 The purity of the phase was then checked by X-ray diffraction analyses, and no impurity peaks appear after the SPS sintering (Figure 2c(v)). E

DOI: 10.1021/acs.langmuir.7b00675 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

Figure 4. SEM images showing the impact of water content on the electrospun LATP mat (20 kV, 10 μL/min). Mats obtained from low water content [1.5 g/m (a, c)] and from high water content [20 g/m (b, d)]. Mats before calcination (a, b) and mats calcined at 850 °C (c) and 900 °C (d).

decomposition of organic components and the LATP crystallization during the sintering processes, accompanied by a reduction of the diameter of LATP (Figure 4c,d). X-ray diffraction analyses confirmed the formation of LATP after sintering in air at 900 °C for 2 h (Figure S3). SEM observations on the samples reveal the impact of the electrospin conditions on the orientation of the fibers and, in particular, on the counter electrode rotating speed. The alignment of the fibers is linked to the increase in the rotational speed of the counter electrode when high water contents are used (Figure 4d). This was necessary to achieve a good deposition of the hybrid solution. The diameter of the crystalline sintered LATP fibers is 270 ± 60 nm when the green fibers are synthesized in air with a high water content while its size distribution is broadly between 100 and 200 nm when drier conditions are used. Note that the LATP fibers are made of monodisperse LATP nanoparticles of ∼20 nm that are well connected to each other, as can be seen in HR-TEM images (Figure 3c,d). The Li+ ion conductivity of these LATP mats was evaluated by impedance spectroscopy at ambient temperature, and an example of the Nyquist plot is shown in Figure S5. The Li ion conductivity was estimated to be 3 × 10−4 mS/cm; note that this value is not affected by the water content used in the electrospinning chamber and therefore by the microstructure of the LATP mat. This value is low compared to that observed on the dense ceramic; this is linked to the porosity of the LATP fiber mat. Taking into account that the porosity of the fiber mat is around 60% and using equation σmeasured 2 σcorrected = 3 1 − Por − 1

orientation, and their interconnectivity). Recently, it has been demonstrated that the water content of the air (grams of water/ m3 of air) during the electrospinning process of silica/PVDF− HFP solutions influences the microstructure of the electrospun mat.17 More specifically, it has been shown that it influences the diameter of sol−gel-based fibers because hydrolysis and condensation reactions of the silica precursors are modified. Large diameters are obtained for high humidity conditions because of the kinetics of hydrolysis, and the condensation of the silica network is accelerated. The diameter of the assynthesized fibers (called green LATP fibers) containing PVDF−HFP and the sol−gel LATP precursor will have a large impact on both the mechanical properties of the fibers, because the crystallization of LATP will occur within the fibers, and on Li ion transport. Large fibers will accommodate more crystals and therefore favor the transport of Li ions along the fibers if the ohmic resistance due to particle−particle contact is not too high. Here, we examine the influence of the water content during the electrospinning process on the diameter of both green and sintered LATP fibers. Precursor/PVDF−HFP solutions were electrospun in a chamber where the water content was controlled. The microstructure of the electrospun mat before and after heat treatment was studied by SEM analyses. We observed that working in air with a high water content requires a modification of the electrospinning experimental parameters such as the counter electrode rotational speed. The typical microstructure achieved for an electrospun LATP mat when the water content in the electrospun chamber ranges from 1.5 to 20 g/m3 is displayed in Figure 4. We observed a modification of the diameter of the green fibers as a function of water content, the fibers have a diameter of 460 ± 120 nm at high water contents (20 g/m3) compared to 120 ± 60 nm for “dry” conditions. (The water content in the electrospinning chamber was 1.5 g/m3). Sintering the green fibers up to 850 °C in air produces a decrease in the fiber diameter as a result of both the

(

dim

)

where σmeasured is the conductivity of the mat, Por is the volumic fraction of the pore in the mat, and Dim is the dimensionality of the porous network,30 the corrected conductivity values are ∼2 × 10−3 mS/cm.31 This value is 1 order of magnitude higher than the one achieved on ceramic with low relative density and F

DOI: 10.1021/acs.langmuir.7b00675 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

Figure 5. Hybrid LATP membranes. (a) Photograph of the hybrid LATP membranes. (b) Microscopy images of a cross-sectional membrane (66% by weight for PVDF−HFP, with the A and S sides being gold-plated). (c, d) SEM images of the surface of the LATP membrane of the S (c) and A sides (d). Note that A indicates the air interface and S indicates the surface interface.

is comparable to glass−ceramic.32 This good conductivity value shows that good particle−particle contact is achieved along the LATP fibers and that most of the LATP fibers are connected to each other to ensure a continuous Li ion pathway throughout the membrane. Elaboration and Characterization of Hybrid LATP Membranes. The objective was to synthesize hybrid membranes as a protection of the lithium metal from the aqueous electrolyte. The watertightness of the LATP fiber mat was obtained by impregnating it with a hydrophobic polymer, PVDF−HFP, according to the conditions described in the experimental section. The hybrid membranes thus obtained are white and flexible (Figure 5a). SEM images of both faces of the membrane indicate that the fibers are impregnated by the polymer (Figure 5c,d). Cross-sectional images confirm that the polymer is well distributed through the thickness of the membrane (Figures 5b and S5) and that the impregnation step does not disrupt the fiber network because no fiber breakage is observed. For a PVDF−HFP volume fraction equal to and superior to 65%, the presence of a PVDF−HFP overlayer has been observed (Figure SI2). Procedures were performed to selectively dissolve it, but this step was unsuccessful because the water tightness of the membrane suffered. Li ion conductivities measured by impedance spectroscopy are reported as a function of the volume faction of PVDF−HFP for LATP mat synthesized under dry or high water content (Figure 6). Note that in Figure 6 the LATP mat is the same; the only variable is the PVDF−HFP content. Note that for membrane synthesized under these two conditions and for a constant PVDF−HFP content, the difference in conductivity observed is mainly ascribed to the morphology of the membrane. The LATP fiber mat is more porous for membranes elaborated under high humidity conditions. These conditions affect the kinetics of the sol−gel reactions, accelerating the fiber solidification, leading to a more porous network. As a consequence, the LATP fiber mats can accommodate more PVDF−HFP without having the formation of an insulating PVDF−HFP overlayer. As previously mentioned, these conditions affect the conductivity values.

Figure 6. Li ion conductivity as a function of the volume fraction of PVDF−HFP for water content in the electrospinning chamber = 1.5 g/m3 (red) and 10 g/m3 (blue). Note that the limit of water tightness is reported on each figure by a red or blue line. The thicknesses of the membrane at low and high water contents are 50 and 20 μm, respectively.

For LATP fiber mats synthesized under dry conditions, the conductivity values measured on hybrid LATP fiber mats are the same order of magnitude as those measured on a pure LATP fiber mat, confirming that PVDF−HFP impregnation does not disrupt the LATP fiber mat and thereby the pathways for Li conduction along the 3D network of fibers. On the contrary, a difference is observed for LATP fiber mats synthesized with water in the electrospinning chamber, demonstrating a modification of the conductive pathways. Quantitatively, the conductivity values are lower by a factor of 10 for membranes elaborated from fibers synthesized in air with high water content. The difference observed is mainly ascribed to the morphology of the membrane: the LATP fibers here are mainly orientated in the plane, limiting the transport of Li ions through the plane. As a consequence, the Li ion conductivity is much lower under these conditions. The alignment is G

DOI: 10.1021/acs.langmuir.7b00675 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir confirmed by the SEM analyses that show the fiber orientation predominantly in the plane in these membranes (Figure 8). We

exhibits, at high q, a broad correlation peak (around q = 0.85 nm−1) related to the organization of the PVDF−HFP crystalline domains. This correlation length is attributed to the distance between the crystallites of the semicrystalline polymer phase considering a sufficient electronic contrast between the amorphous and crystalline phases.33 The semicrystalline polymer phase can be treated as lamellar systems. The lamellar period is the sum of the average widths of the lamellae and amorphous layers. In the present case, this value is 7.4 nm. The spectrum of pure PVDF−HFP fibers without LATP (a) is also represented as a reference. In the unsintered fiber mat (d), the peak is broader than the pure PVDF−HFP fibers (a), which would suggest a higher dispersion of the characteristic correlation lengths of PVDF−HFP crystallite domains. In all cases, at low q, the graphs [log(abs int) vs log(q)] are a straight line, which corresponds to q−x evolution (where x ≈ 4). This behavior tends to highlight, according to Porod’s law, the existence of a sharp and smooth interface between air and the material at the nanoscale. Please note that the evolution is not modified after the impregnation of the membrane, which suggests that the nanostructure is intact after calcination and impregnation. The watertightness of the hybrid membranes were then measured because these membranes will be used as separation layer for aqueous Li−air batteries. Watertightness was achieved for different volume fractions of PVDF−HFP in the membrane. A value of 65% is sufficient for unaligned fiber mats (lower humidity), whereas 82% is needed when the LATP fibers are aligned in the plane at higher humidity. This difference can be attributed to several factors, including the distribution of hydrophobic domains within the membrane. The ohmic resistance values were then estimated from the impedance spectra. Values of ∼35−40 kΩ·cm2 are observed for watertight membranes whereas values of ∼5−10 kΩ·cm2 are achieved for the others. These values are 1000 times superior to

Figure 7. SAXS absolute intensity as a function of the scattering-vector q modulus on fiber mats of (a) pure PVDF−HFP; (b, c) LATP (before calcination) obtained at high RH (b) and low RH (c); and (d) LATP/PVDF−HFP from low-RH conditions after impregnation.

note that the fibers are surrounded by PVDF−HFP, disrupting transversal Li ion pathways. On the contrary, the membranes elaborated at low humidity exhibit fibers connected to each other within the thickness of the membrane (Figure S6). A SAXS analysis, showing the homogeneity of the fiber mat before and after impregnation, is presented in Figure 7. The fiber mats (b, c) before impregnation give a featureless SAXS intensity, regardless of the humidity used during the electrospinning process. After impregnation (d), the spectrum

Figure 8. (a, c) FEG−SEM images of a cross section of a hybrid LATP fiber mat. The cross section was prepared by using a focused ion beam (b). Reconstruction of the LATP fiber mat in the hybrid membranes. H

DOI: 10.1021/acs.langmuir.7b00675 Langmuir XXXX, XXX, XXX−XXX

Langmuir



what has been observed for LIPON ceramic.34 Even though the ohmic resistance is high compared to dense LiPON ceramic, the LATP hybrid membranes have Li ion conductivities that compare well to that observed on dense ceramics corrected by the porosity. This result shows that these hybrid membranes are promising candidates as protective layers for lithium metal in an aqueous environment such as Li−air batteries.

Article

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.langmuir.7b00675. a, b, and c parameters as functions of Al content, synopsis of membrane synthesis, linear shrinkage during spark plasma sintering of attrition-milled LATP powder under pressure, XRD patterns of the calcined LATP mat, equivalent circuit used in this work, electrochemistry Nyquist plot of the LATP mat, and cross-sectional FESEM images of hybrid membrane synthesized under dry conditions (PDF) Results of stability tests and evolution of the surface of the hybrid membrane for different PVDF−HFP content (PDF)



CONCLUSIONS One of the main disadvantages of the electric vehicle is its short driving range, which is limited by the energy density of lithium ion batteries. There is therefore a need for batteries with higher energy densities. Among the various technologies existing on the market, the aqueous Li−air battery is a promising technology with energy densities potentially twice as high as for current batteries. Aqueous Li−air batteries today use a separator consisting of a Li ion conducting NaSICON-like ceramic electrolyte, itself protected from lithium metal by a thin LiPON layer. This ceramic separator needs to be sufficiently thick to ensure good mechanical properties, giving rise to a large ohmic resistance and therefore leading to batteries with poor power performances. By replacing the solid ceramic membrane with ceramic fibers, Li ion conduction is obtained in a nonbrittle, flexible material. Watertightness is obtained by filling the voids created by the fibers with an organic, hydrophobic, nonconducting polymer. The solid ceramic electrolyte is therefore replaced by a flexible hybrid organic− inorganic membrane, combining good Li ion conduction and watertightness. To achieve a good percolation of the inorganic Li ion conducting network, we propose to fabricate an interconnected 3D network of fibers obtained by combining sol−gel chemistry and electrospinning. The microstructure of the 3D fiber network was tuned through the electrospinning parameters, including the water content in air (or relative humidity (RH)), the speed of the rotating counter electrode, and the final sintering temperature. The water content of the surrounding air was found to have an impact on the fiber diameter whereas the rotating electrode speed impacts the alignment of the fibers produced: high speed and high humidity are conducive to aligned fibers with a large diameter. Sintering the fiber mat at high temperature (i.e., T ≥ 850 °C) produces crystalline fiber mats with good mechanical properties. The Li ion conductivity was estimated by impedance spectroscopy, and values ranging from 10−5 and 10−3 mS/cm have been measured depending on the microstructure of the fiber mat (i.e., fiber alignment). Taking into account a porosity of 30% in volume, the conductivity value is comparable to that measured on dense pellets under the same conditions, attesting to the good quality of the particle−particle contact along the LATP fibers. Impregnation of the crystalline fiber mat by a hydrophobic polymer (PVDF−HFP) is conducive to the formation of flexible hybrid membranes. The PVDF−HFP volume fraction was tuned from 40 to 80%, and the best results in terms of water tightness and conductivity values were obtained for membranes with a volume fraction of PVDF−HFP of 70% and a Li ion conductivity of 1 × 10−4 mS/cm for a thickness of 20 μm, corresponding to an ohmic resistance of 40 kΩ·cm2. Even if the membrane exhibits promising properties, further improvements in terms of the Li ion conductivity are needed to test it in a real Li−air battery.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Christel Laberty-Robert: 0000-0003-3230-3164 Author Contributions #

These authors contributed equally.

Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS The authors thank Stephan Borensztajn for the FIB and FESEM images and acknowledge Patrick Legriel for his help with the HR-TEM images and Cyrille Rochas as the local contact for the BM02-D2AM beamline at the ESRF. The authors also acknowledge Julie Kasparian, undergraduate student, for her invaluable help during her training period.



REFERENCES

(1) Imanishi, N.; Yamamoto, O. Rechargeable lithium−air batteries: characteristics and prospects. Mater. Today 2014, 17, 24−30. (2) Bruce, P. G.; Freunberger, S.; Hardwick, L. J.; Tarascon, J. M. LiO2 and Li-S batteries with high energy storage. Nat. Mater. 2012, 11, 19−29. (3) Nimon, Y.; De Jonghe, L. V.; Visco, S. Solid Electrolytes Based on Lithium Hafnium Phosphate for Active Metal Anode Protection. U.S. Patent US20060078790 A1, 2006. (4) Abraham, K. M.; Jiang, Z. A Polymer Electrolyte-Based Rechargeable Lithium/Oxygen Battery. J. Electrochem. Soc. 1996, 143 (1), 1−5. (5) Laoire, C. O.; Mukerjee, S.; Abraham, K. M.; Plichta, E. J.; Hendrickson, M. A. Elucidating the Mechanism of Oxygen Reduction for Lithium-Air Battery Applications. J. Phys. Chem. C 2009, 113 (46), 20127−20134. (6) Stevens, P.; Toussaint, G.; Caillon, G.; Viaud, P.; Vinatier, P.; Cantau, C.; Fichet, O.; Sarrazin, Ch.; Mallouki, M. Development of a Lithium Air Rechargeable Battery. ECS Trans. 2010, 28 (32), 1−12. (7) Stevens, P.; Toussaint, G. Air Electrodes for Aqueous Lithium Air Batteries. The Lithium Air Battery: Fundamentals; Springer, 2014. (8) Toussaint, G.; Stevens, P.; Akrour, L.; Rouget, R.; Fourgeot, F. Development of a rechargeable zinc-air battery. ECS Trans. 2010, 28 (32), 25−34. (9) Aono, H.; Sugimoto, E.; Sadaoka, Y.; Imanaka, N.; Adachi, G.-Y. Ionic Conductivity of the Lithium Titanium Phosphate (Li1+X MXTi2−X (PO4)3, M = Al, Sc, Y, and La) Systems. J. Electrochem. Soc. 1989, 136 (2), 590−591. I

DOI: 10.1021/acs.langmuir.7b00675 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir (10) Thokchom, J. S.; Kumar, B. Water Durable Lithium Ion Conducting Composite Membranes from the Li2O-Al2O3-TiO2-P2O5 Glass-Ceramic. J. Electrochem. Soc. 2007, 154 (4), A331−A336. (11) Bates, J. B.; Dudney, N. J.; Gruzalski, G. R.; Zuhr, R. A.; Choudhury, A.; Luck, C. F. Electrical properties of amorphous lithium electrolyte thin films. Solid State Ionics 1992, 53−56, 647−654. (12) Toussaint, G.; Stevens, P.; Caillon, G.; Viaud, P.; Cantau, C.; Vinatier, P. EDF-CNRS, French Patent 09 56772, 2009. (13) Wang, Y.; Zhou, H. A lithium-air battery with a potential to continuously reduce O2 from air for delivering energy. J. Power Sources 2010, 195 (1), 358−361. (14) Puech, L.; Cantau, C.; Vinatier, P.; Toussaint, G.; Stevens, P. Elaboration and characterization of a free standing LiSICON membrane for aqueous lithium-air battery. J. Power Sources 2012, 214, 330−336. (15) Laberty-Robert, C.; Vallé, K.; Pereira, F.; Sanchez, C. Design and properties of functional hybrid organic-inorganic membranes for fuel cells. Chem. Soc. Rev. 2011, 40, 961−1005. (16) Maneeratana, V.; Bass, J. D.; Azaïs, T.; Patissier, A.; Vallé, K.; Sanchez, C.; Laberty-Robert, C. Fractal Inorganic − Organic Interfaces in Hybrid Membranes for Effi cient Proton Transport. Adv. Funct. Mater. 2013, 23 (22), 2872−2880. (17) Dos Santos, L.; Rose, S.; Sel, O.; Maréchal, M.; Perrot, H. Electrospinning a versatile tool for designing hybrid proton conductive membrane. J. Membr. Sci. 2016, 513, 12−19. (18) Stevens, P.; Toussaint, G.; Lancel, G.; Laberty-Robert, C.; Bregiroux, D.; Sanchez, C. Procédé de fabrication d’une membrane flexible étanche à l’eau et conductrice ionique. French Patent WO2016097571 A1, 2015. (19) Schmid, B.; Schindelin, J.; Cardona, C.; Longair, M.; Heisenberg, M. A high-level 3D visualization API for Java and ImageJ.BMC Bioinform.2010, 274, doi: 10.1186/1471-2105-11-274 (20) Chang, C.-M.; Lee, Y. I.; Hong, S.-H.; Park, H.-M. Spark Plasma Sintering of LiTi2(PO4)3-Based Solid Electrolytes. J. Am. Ceram. Soc. 2005, 88, 1803−1807. (21) Key, B.; Schroeder, D. J.; Ingram, B. J.; Vaughey, J. T. SolutionBased Synthesis and Characterization of Lithium-Ion Conducting Phosphate Ceramics for Lithium Metal Batteries. Chem. Mater. 2012, 24, 287−293. (22) Shannon, R. D. Revised effective ionic radii and systematic studies of interatomic distances in halides and chalcogenides. Acta Crystallogr., Sect. A: Cryst. Phys., Diffr., Theor. Gen. Crystallogr. 1976, A32, 751−767. (23) Le Bail, A.; Duroy, H.; Fourquet, J. L. Ab-initio structure determination of LiSbWO6 by X-ray powder diffraction. Mater. Res. Bull. 1988, 23, 447−452. (24) Mariappan, C. R.; Galven, C.; Crosnier-Lopez, M.-P.; Le Berre, F.; Bohnke, O. Synthesis of nanostructured LiTi2(PO4)3 powder by a Pechini-type polymerizable complex method. J. Solid State Chem. 2006, 179, 450−456. (25) Pérez-Estébanez, M.; Isasi-Marin, J.; Rivera-Calzada, A.; León, C.; Nygren, M. Spark Plasma versus conventional sintering in the electrical properties of Nasicon-type materials. J. Alloys Compd. 2015, 651, 636−642. (26) Xu, X.; Wen, Z.; Wu, J.; Yang, X. Preparation and electrical properties of NASICON-type structured Li1.4Al0.4Ti1.6(PO4)3 glassceramics by the citric acid-assisted sol−gel method. Solid State Ionics 2007, 178 (1−2), 29−34. (27) Duluard, S.; Puech, L.; Vinatier, P.; Turq, V.; Rozier, P.; Lenormand, P.; Taberna, P.-L.; Simon, P.; Ansart, F. Lithium conducting solid electrolyte Li1.3Al0.3Ti1.7(PO4)3 obtained via solution chemistry. J. Eur. Ceram. Soc. 2013, 33 (6), 1145−1153. (28) Wu, X. M.; Li, X. H.; Zhang, Y. H.; Xu, M. F.; He, Z. Q. Synthesis of Li1.3Al0.3Ti1.7(PO4)3 by sol−gel technique. Mater. Lett. 2004, 58 (7−8), 1227−1230. (29) Imanaka, N.; Shimizu, T.; Adachi, G. Lithium conducting amorphous solid electrolytes obtained by explosion method. Solid State Ionics 1993, 62, 167−171.

(30) Bokstein, B. S.; Brose, H. D.; Trusov, L. I.; Khvostantseva, T. P. Diffusion in nanocrystalline nickel. Nanostruct. Mater. 1995, 6 (5−8), 873−876. (31) Muller, G.; Ringuede, A.; Laberty-Robert, C. Discussion on a Percolating Conducting Network of a Composite Thin-Film Electrode (≤ 1 μm) for Micro-Solid Oxide Fuel Cell Application. Langmuir 2014, 30 (29), 8889−8897. (32) Thokchom, J. S.; Kumar, B. The effects of crystallization parameters on the ionic conductivity of a lithium aluminum germanium phosphate glass−ceramic. J. Power Sources 2010, 195, 2870−2876. (33) Abbrent, S.; Plestil, J.; Hlavata, D.; Lindgren, J.; Tegenfeldt, J.; Wendsjo, A. Crystallinity and morphology of PVDF-HFP-based gel electrolytes. Polymer 2001, 42, 1407−1416. (34) Knauth, P. Inorganic solid Li ion conductors: An overview. Solid State Ionics 2009, 180, 911−916.

J

DOI: 10.1021/acs.langmuir.7b00675 Langmuir XXXX, XXX, XXX−XXX