Improved performances of lithium-ion batteries with a separator based on inorganic fibers

Abstract

Battery separators made of inorganic powders and polymeric substrates have attracted intensive attention from both industry and academia. However, existing technology suffers from an inherent limitation in that the aggregation of powders during manufacturing and the escape of powders in use are inevitable. In this work, we report for the first time, a novel separator prepared by blending inorganic ZrO₂ staple fibers with poly(vinylidenefluoride-co-hexafluoropropylene) (PVDF-HFP). Through a unique phase inversion process, the separator shows a skinless surface, a highly porous internal structure and a uniform distribution of pore size. Systematical comparison of a fiber-based separator and a powder-based separator demonstrates that the reinforcement from the fiber skeleton can provide the separator with desired porosity, sufficient mechanical strength, superior thermal stability and improved electrochemical performance. We anticipate that the concept of combining inorganic fiber substrates with polymeric media will offer a platform strategy towards the fabrication of high-safety separators for lithium-ion batteries and other energy storage systems.

---

Introduction

As the most popular energy storage system, lithium-ion batteries have received extensive attention for portable electronic devices including laptops, digital cameras and cell phones with the worldwide market valued at ten billion dollars.^1–3 A complete lithium battery system is composed of an anode, a cathode and a separator.⁴ Among them, the role of a separator is to serve as the medium for the transfer of lithium ions between electrodes, as well as to control the number and mobility of the ions.^5–9

Currently, polyolefin microporous membranes (Celgard) are widely used as separators for lithium-ion batteries because of their suitable chemical stability, thickness, pore size and mechanical strength.^10–12 However, these polyolefin membranes have been proven to cause safety issues at elevated temperature, because of their low thermal stability (i.e., melting temperature of ∼130 °C for polyethylene (PE) and ∼165 °C for polypropylene (PP)).^13–15 Even though polyolefin membranes have the function of thermal shut-down at a specific temperature, they cannot withstand high temperature for a long time.¹⁶ What is worse, when the temperature exceeds the melting point of the polymer, the melted separator will easily cause short circuits between the electrodes.^17–20 The poor wettability to the conventional liquid electrolyte of polyolefin separators has also brought serious disadvantages in battery performances.^21–23

A considerable amount of effort has been devoted to solving these problems. A popular strategy is to incorporate inorganic particles into organic macromolecules. For instance, Choi et al. developed SiO₂-coated polyester nonwoven separators for lithium-ion batteries and provided a new route to tune the porous structure by exploiting particles with various sizes.²⁴ Feng et al. used a joint method of electrospinning and electrospraying to prepare separators based on polymeric nanofibers and SiO₂ powders.²⁵ Kang et al. adopted a biomimetic method to coat SiO₂ on PE membranes, and the adhesion of SiO₂ layer was inspired by mussels.²⁶ Liao et al. acquired a gel polymer electrolyte (GPE) by adding Al₂O₃ nanoparticles on poly(ethylene oxide) and PVDF-HFP blends.²⁷ Although the embedded inorganic particles could effectively regulate the pore size and improve the thermal stability, many small pores were inevitably blocked in such a process, leading to the decrease of overall porosity. Moreover, when soaked in the electrolyte for a long period, the particles, especially those on the surface of the separator, may easily fall out from their organic host, and thus limits the service life of the battery.

Different from previous studies that all involve inorganic particles, herein we report a novel method that applies inorganic staple fibers as the substrate to fabricate separators for lithium-ion batteries. In a typical procedure (Fig. 1A), ZrO₂ staple fibers with various concentrations were dispersed in PVDF-HFP solution using 1-methyl-2-pyrrolidone (NMP) and ammonia water as solvents. The fiber suspension was then casted on a smooth glass plate to form a film with controllable thickness. Through simple immersion in deionized water, a modified phase inversion process was triggered to generate tiny and numerous pores throughout the film, after which the film was peeled off from the glass. The resulting separator inherits the intrinsic thermal and chemical stability of raw ZrO₂ fibers, and the superior electrolyte affinity of PVDF-HFP, and hence exhibits better electrochemical performance and significantly higher thermal stability than the commercial PP separator, Celgard 2400. More importantly, the ZrO₂ fiber skeleton improves the mechanical strength and successfully prevents the deformation of the separator from intensive combustion. The concept of combining an inorganic fiber substrate with porous organic macromolecules does not require sophisticated procedures, harsh conditions or expensive reagents, making it possible to become a general strategy for the manufacture of battery separators.

Fig. 1 (A) Schematic diagram of the preparation process of ZrO₂/PVDF-HFP membranes. (B) Surface and cross-sectional SEM micrographs of the membranes containing 0, 25%, 50% and 75% ZrO₂ fiber. Scale bar = 20 μm for the large images and 5 μm for the inset.

Experimental

Materials

PVDF-HFP powder (PVDF solef 21216, M_n = 60 000, Solvay) was purchased from Jinhua Co. Ltd., China. ZrO₂ fiber was prepared by using the following protocol.²⁸ Briefly, 15.2 g of nitric acid was added into 100 g of liquid zirconium acetate under vigorous stirring. The precursor solution was heated at 75 °C until the viscosity increased to 25 Pa s, then the temperature was set at 45 °C and the solution was further condensed to 100 Pa s. The acquired solution was fed into a self-developed, high-speed centrifugal spinning nozzle to spin gel fibers. The fibers were sintered in a furnace at the heating rate of 2 °C min⁻¹. At 610 °C the temperature was maintained constant for 1 h, and subsequently increased to 1050 °C for another 1 h (heating rate = 8 °C min⁻¹). Once taken out, the fibers were quenched to room temperature so that the ZrO₂ fibers were prepared.

1-Methyl-2-pyrrolidone (NMP) and ammonia water were purchased from Shanghai Lingfeng Chemical Reagents Co., Ltd., China. Nano-zirconia powders (average particle size = 30 nm) were obtained from Nanjing Mingshan Advanced Materials Co., Ltd. Commercial PP membranes (Celgard 2400, Celgard) with a thickness of about 25 μm were selected for comparison. Electrolyte solution (1 M LiPF₆ solution in dimethyl carbonate (DMC) : ethylmethyl carbonate (DEC) : ethylene carbonate (EC) = 1 : 1 : 1 (w/w/w)) was purchased from Kunlun Xianghe Chemical Co., Ltd.

Membrane preparation

The ZrO₂/PVDF-HFP separator was prepared by dissolving PVDF powders in NMP at a concentration of 5% (w/v). When a transparent solution was formed, 200 μL of ammonia water was added into the solution under continuous stirring. Different proportions of ZrO₂ fibers were then added into the solution and kept stirring for 12 h to form a uniform solution. The solution was casted on a smooth glass plate by using a scraper coating machine (casting speed = 2.5 cm s⁻¹). After being exposed in open air for 30 min, the glass plate was immersed in water for 10 min to achieve a phase inversion process, during which NMP was exchanged with water. The acquired separator was peeled off from the glass plate and dried in a 70 °C vacuum oven for 12 h.

Characterization

The samples were observed by using a scanning electron microscope (SEM, Hitachi TM3000). Cross-sections of the samples were prepared by dipping the membrane into liquid nitrogen, and then broken into two parts.²⁹ Fourier transform infrared spectroscopy (FT-IR) was conducted using a Nicolet 6700 infrared spectrometer. Wide angle X-ray diffraction was performed by using a D/Max-2550 PC X-ray diffractometer (Rigaku Corporation, Japan). Energy dispersive X-ray spectroscopy (EDX) analysis was conducted utilizing a field emission scanning electron microscope (FESEM, S-4800, Japan) instrument to confirm the elemental composition of the sample. Mechanical properties were tested by using a universal tensile tester (H5 K-S, Hounsfield, UK). Thermal gravimetric (TG) analysis and differential scanning calorimetry (DSC) of the membranes were conducted by using a TGA4000/DSC4000 (Perkin-Elmer). The wettability of the electrolyte was tested by using an optical contact angle meter (OCA15EC, Dataphysics, Germany). Electrolyte uptake was measured by immersing the membrane into the liquid electrolyte solution at 25 °C for 2 h. The membrane was then taken out and weighed immediately on a microbalance after removing the excrescent surface solution with wipes. The amount of liquid electrolyte uptake (E) was calculated using the equation:
where E is the imbibition rate, W is the dry weight of the sample (g), and W₁ is the wet weight of the sample (g).

The pore size was measured by using a PMI aperture tester (CFP-1100-AI). Porosity was measured by weighing the samples before and after being soaked in methyl silicone oil, and then calculated as follows:

where P is the porosity, W is the dry weight of the sample (g), W₁ is the wet weight of the sample (g), P is the density of methyl silicone oil (0.96 g cm⁻³), and V is the apparent volume of samples (cm³).

The areal density was calculated by weighing the samples of a fixed size (3 × 3 cm). A densometer (4110 N, Thwing-Albert, USA) was used to measure the Gurley value, which is defined as the time for nitrogen to pass through a fixed volume (30 cm⁻³) under a given pressure (0.02 MPa). For the evaluation of thermal stability, each separator was positioned between slide glasses and subjected to tensile shear stress. The temperature was increased to 150 °C and 180 °C at 5 °C min⁻¹, respectively.³⁰ Each sample was heated for 12 h at each temperature and the contraction rate was compared.

To evaluate the electrochemical performance, 2025 type lithium coin cells were assembled by employing separators containing 25%, 50% and 75% ZrO₂ fiber, 75% ZrO₂ powder and Celgard 2400, respectively. Detailed procedures are shown in Fig. S1.† The AC impedance (frequency range: 0.1 Hz to 100 kHz and AC amplitude of 10 mV) was measured by using an electrochemical workstation (Autolab 302 N, Switzerland). Cyclic voltammetry (3.0–4.3 V) at various scan rates (0.1, 0.2, 0.5 and 1.0 mV s⁻¹) was used to investigate the rate capability and kinetics of the cell reactions in the electrodes based on Li(Ni_1/3Co_1/3Mn_1/3)O₂. The Li⁺ diffusion coefficient (D_Li⁺) can be calculated by using the Randles–Sevcik equation:³¹

where I_p represents the peak current (A), n is the number of electrons per molecular reaction, A is the surface area of the cathode (cm²), C is the lithium shuttle concentration (mol cm⁻³) and n is the scanning rate (V s⁻¹).

The ionic conductivity was calculated using the following equation:

where d is the thickness of the membrane, R is the bulk resistance and S is the surface area of the separators.

A LAND CT2001 battery testing system was used to test the electrochemical properties of the assembled lithium ion battery at both room temperature and elevated temperature (60 °C).³² The charge and discharge cut-off voltage of the battery is set between 3 and 4.3 V. When testing the cycle performance of the battery, the battery was charged and discharged at the rate of 0.2C/0.2C at all times. For the rate performance test, the battery was charged and discharged at the rate of 0.2C for 5 cycles in order to form a good solid electrolyte interface (SEI) film on the electrode. For the follow-up rate performance test, the battery was charged at a permanent rate of 0.5C and discharged at 0.5C, 1C, 2C, 4C and 8C, respectively (n = 5). For the calculation of energy density, each voltage is multiplied by the values of discharge current and the time before the voltage jumps to the next stage. The results are summarized to acquire the total delivered discharge energy. For discharge power, the voltage is multiplied by the discharge current and the results are summed and averaged over the number of points.³³ The change of open circuit voltage (OCV) was monitored at various time spots (0–210 min). Before being placed into an oven at 140 °C, the cells were fully charged to 4.3 V at the rate of 0.5C under room temperature.³⁴

Results and discussion

Surface and cross-sectional SEM images in Fig. 1B indicate that the thickness of the membranes is ∼30 μm, and ZrO₂ fibers are uniformly distributed in both horizontal and vertical directions.

Increasing the proportion of ZrO₂ leads to the increase of fiber density and membrane weight (Fig. S2†), but has no interference on fiber distribution. The large gaps between ZrO₂ fibers are filled with PVDF-HFP porous media having an average pore size of ∼600 nm. The porous structure of PVDF-HFP is highly comparable to that of Celgard 2400 (Fig. S3†). Formation of these pores can be attributed to the exchange of water with NMP, and subsequent transformation of water-rich region into pores.³⁵ It is worth noting that, unlike a common phase inversion process, the addition of ammonia in PVDF-HFP/NMP solution plays a key role in inducing localized microphase separation, which generates a skinless surface. For comparison, PVDF-HFP membranes containing 75% ZrO₂ powders are also prepared using the same process (Fig. S3†), although the particles seem to be well scattered in all directions, it is difficult to prevent them from aggregation with each other, which increases the particle dimension from tens of nanometers to several micron meters.

The presence and distribution of ZrO₂ in the membranes are further confirmed by FT-IR and XRD analyses. The absorption peaks at 1403, 1184 and 880 cm⁻¹ (Fig. 2A) can be attributed to the CH₂, CF₃ and CF groups of PVDF-HFP, respectively.³⁶ The peak at 645 cm⁻¹, which corresponds to the ZrO group of ZrO₂,³⁷ becomes stronger when increasing the ratio of ZrO₂. XRD patterns in Fig. 2B suggest the same trend. The peaks at 29.74°, 34.52°, 49.82° and 59.26°, which correspond to the tetragonal phases (101), (110), (200) and crystal planes (211),³⁸ become more apparent with the increase of ZrO₂ ratio. For samples containing 75% ZrO₂ powders and 75% ZrO₂ fibers, the existence of C, O, F and Zr was further confirmed by EDX (Fig. S4†), demonstrating that the ZrO₂ contents, whether in powder or fibrous forms, are well distributed among the membranes.

Fig. 2 (A) FT-IR spectra and (B) XRD patterns of membranes containing 0, 25%, 50% and 75% ZrO₂ fiber and 75% ZrO₂ powder; (C) dry and (D) wet stress–strain curves of membranes containing 0, 25%, 50% and 75% ZrO₂ fiber and 75% ZrO₂ powder.

During the assembly and use of batteries, the separators are required to withstand certain tension. Therefore, mechanical properties in both dry and wet states are tested (Fig. 2C and D). In general, the incorporation of ZrO₂ fiber causes a slight decrease in elongation, but a significant increase in strength. When the fiber ratio increases to 75%, the stress of the membrane exceeds 5 MPa, making the membrane robust enough to be used in battery systems. In comparison, membranes composed of 75% ZrO₂ powder show a stress of 0.6 MPa in the dry state and 1.9 MPa in the wet state, both are dramatically lower than those of the fiber membranes.

An ideal separator should absorb and retain a significant amount of liquid electrolyte to achieve low internal resistance and high ionic conductivity.^39,40 The absorption capacities of different samples are visualized by electrolyte immersion-heights. The membrane composed of pure PVDF-HFP shows the largest height (Fig. 3A), which can be attributed to the combined effect of high porosity, affinity to electrolyte and capillary action. More ZrO₂ fibers result in the decrease of immersion-height, but the decreasing trend is inhibited when the fiber ratio exceeds 50%. Samples made of 75% ZrO₂ fiber and 75% ZrO₂ powder have similar immersion-heights that are both higher than that of Celgard 2400. Such a difference is in accordance with the difference in electrolyte spread on the sample surface. As shown in Fig. S5,† an instant spread of electrolyte droplet is observed on membranes containing PVDF-HFP. However, on celgard 2400 the droplet remains spherical, due to the poor wettability of PP. We further compared the structural stability of wet membranes having the same ratio of ZrO₂. When pressing the 75% powder sample by fingers, a proportion of powders are found to transfer from the membrane surface to finger, whereas the same phenomenon does not appear on the 75% fiber sample (see ESI video 1†). More interestingly, after soaking the samples in the electrolyte, a clear shrinkage is observed from membranes having 75% ZrO₂ powder or no ZrO₂, but membranes having ZrO₂ fiber are able to remain flat (Fig. 3B). These results clearly prove that the structural support from ZrO₂ fibers can effectively prevent the deformation of wet membranes.

Fig. 3 (A) Electrolyte immersion-height and (B) appearance change of Celgard 2400 and membranes containing various ratios of ZrO₂. (C) Porosity of membranes containing various ratios of ZrO₂ fiber and electrolyte uptake of Celgard 2400 and membranes containing 75% ZrO₂ fiber and 75% ZrO₂ powder. (D) Pore size of membranes containing various concentrations of ZrO₂ fibers and pore size distribution of Celgard 2400, membranes containing 75% ZrO₂ fiber and 75% ZrO₂ powder.

As can be seen from Fig. 3C, the porosity, electrolyte uptake and blending ratio of ZrO₂/PVDF-HFP are positively correlated to each other. With the decrease of polymer ratio, the porosity gradually decreased from 87.53% to 78.38%, leading to the decrease of electrolyte uptake from 351.2% to 252.5%. Fast and similar uptaking processes of 75% fiber sample and 75% powder sample (Fig. 3C, inset) suggest that the shape of ZrO₂ is not relevant to electrolyte uptake. Because of the hydrophobic nature of PP, Celgard 2400 shows a lower electrolyte uptake than any sample that contains PVDF-HFP.

As a result of the heterogeneous structure formed by organic/inorganic blends, the mean and largest pore sizes of the membranes slightly increase with the increase of ZrO₂ fiber ratio (Fig. 3D). Compared to Celgard 2400, membranes having 75% ZrO₂ fiber or powder have larger average pore diameters, but the largest pore is still confined within 1 μm. Statistical analysis in the inset of Fig. 3D proved the narrow pore size distribution of the membranes. The pore structure is further reflected by the Gurley value (Fig. S6†). In general, the Gurley value increases with the increase of ZrO₂ ratio. The 75% fiber membrane and 75% powder membrane exhibit higher values (243 s 100 cm⁻³ and 261 s 100 cm⁻³) than that of Celgard 2400 (178 s 100 cm⁻³), mainly because of the longer tortuous path of the pores, which is regarded as conducive to prevent the internal short circuit and alleviate the occurrence of self-discharge.¹⁴ These results are of importance because separators with immoderately large-sized pores and a non-uniform pore size distribution tend to inadequately inhibit the leakage current between electrodes.

The above findings lead to the conclusion that membranes having a high ratio of ZrO₂ fibers exhibit good shape retention, appropriate electrolyte absorption and suitable porosity. Therefore, we selected 75% ZrO₂ fiber membranes for further characterization. Celgard 2400 and 75% ZrO₂ powder membranes were also subjected to the same tests for comparison. Thermal stabilities of the membranes are displayed in Fig. 4. At the ZrO₂ ratio of 75%, the corresponding TG curves show little change before 450 °C, while Celgard 2400 decomposes rapidly from ∼400 °C (Fig. 4A). A further increase of temperature causes the complete decomposition of Celgard 2400 at 495 °C, but the ZrO₂ membrane is found to maintain more than 80% of the initial weight. The results imply that the weight loss of the membrane is mainly prevented by the ZrO₂ component. As the decomposition point of ZrO₂ is as high as 2500 °C, elevated temperature may melt down the polymeric part, but leaves the inorganic skeleton intact. More evident improvement from the presence of ZrO₂ is shown by the DSC curves in Fig. 4B. Celgard 2400 has an endothermic peak at 167 °C, which corresponds to the melting point of PP. At this temperature, the micropores completely shut down and the separator consequently transforms to a nonporous sheet. Although this behaviour is a crucial safety feature to prevent a thermal runaway situation, the nonporous sheet may experience severe shrinkage and even break down if the temperature continues to increase.¹² For ZrO₂/PVDF-HFP membranes, no obvious peaks were observed below 300 °C, except for a small peak at 127 °C (melting point of PVDF-HFP). The main reason is that PVDF-HFP is stable until 370 °C, while PP can react with oxygen and triggers the decomposition of Celgard 2400.

Fig. 4 (A) TG and (B) DSC curves of Celgard 2400, 75% ZrO₂ fiber separator and 75% ZrO₂ powder separator. (C) Thermal contraction behavior and (D) combustion behavior of Celgard 2400, 75% ZrO₂ fiber separator and 75% ZrO₂ powder separator. The blue arrows represent the machine direction and the red arrows represent the cross-direction of the separators. (E) Schematic diagram of the difference in combustion behavior between 75% fiber and 75% powder separators.

Apart from TG and DSC curves, heat resistance of the separators was visualized and compared through thermal contraction tests. To simulate the tensile shear of the positive and negative poles in the battery, each separator was positioned between glass slides under tensile stress and then heated for at least 2 hours. As shown in Fig. 4C, ZrO₂ samples are immune to long-term high temperature with almost no shrinkage being found in both machine direction and cross-direction. On the contrary, an early shrinkage of Celgard 2400 along the cross-direction starts at 160 °C and becomes more severe at 170 °C.

Most polymeric separators and electrolyte in lithium-ion batteries are highly flammable and extremely dangerous if ignited by short circuit or traffic accidents. On May 26^th 2012, the explosion of an electric taxi in Shenzhen, China, caused 3 deaths and 3 injuries. Recently, one million Samsung mobile phones have been recalled due to the potential risk of battery explosion. Although grave concerns have been raised to the safety of these batteries, similar incidents occur from year to year. The superior flame-retardancy of ZrO₂ makes us believe that the ZrO₂/PVDF-HFP membrane may inherit this advantage and have the potential of reducing the risk of fire and blast. To validate this consumption, the combustion behavior of the membranes was examined. As can be seen from Fig. 4D, once getting close to fire, Celgard 2400 immediately curls up into a ball-like structure (see also ESI video 2†). In contrast, the ZrO₂ separators show extraordinary flame-retardancy and cannot be ignited throughout the experiment. These separators undergo 2 discernable stages during combustion: fast carbonization of PVDF-HFP and exposure of ZrO₂. The carbonization of PVDF-HFP turns the yellowish separators to black, after which the polymer is completely burned out so that the white ZrO₂ component is exposed. Notably, the combustion behavior of the fiber-based sample differs significantly from that of the powder-based sample. The powder-based sample loses its structural integrity within 16 s (ESI video 3†), while the fiber-based sample is highly stable against continuous combustion for a period as long as 240 s (ESI video 4†). The feature is explained by the schematic illustration in Fig. 4E. Before being placed in flame, PVDF-HFP is critical to the basic strength of 75% ZrO₂ powder separator because it acts as the binder to bind the powders together. After the burning out of the polymer, instantaneous scattering of ZrO₂ powders occurs due to the loss of contact between neighbouring powders. In a ZrO₂ fiber separator, however, the structural integrity is reinforced by the organic binder as well as inorganic fibers. These fibers are tightly connected in a nonwoven structure, in which the friction and cohesion among fibers efficiently prevent the deformation of the membrane. Combining these results with mechanical properties, it is reasonable to conclude that the employment of ZrO₂ fibers is essential to enhance both heat-resistance and mechanical strength of the separators.

Electrochemical stability, an essential factor in characterizing battery separators, was examined by linear-sweep voltammetry (LSV) on a stainless steel electrode as a working electrode with lithium as a reference electrode.⁴¹ When the voltage is below 1 V, no current is detected among the cells fabricated with Celgard 2400, ZrO₂ powder and ZrO₂ fiber separators (Fig. 5A). The finding demonstrates the excellent anodic stability of electrolytes from ZrO₂/PVDF separators, which can be ascribed to the good electrolyte retention of porous PVDF and superior interfacial compatibility of both PVDF and ZrO₂. A further increase of the voltage results in a tiny current generated at 4.5 V, indicating that all the separators have large electrochemical windows in the liquid electrolyte, and hence are suitable to be used in lithium ion batteries with a high voltage. Fig. 5B shows the self-discharge detection curves of the three types of cells. As expected, no obvious difference in open circuit voltage was observed among the cells, regardless of the working time. Since the open circuit voltage of lithium-ion batteries is almost linear to their amount of charge, the small and synchronous variation suggests that ZrO₂-based separators are qualified for commercial application. In addition, the calculated Li⁺ diffusion coefficients of the cells are similar to each other (4.42 × 10⁻¹², 4.35 × 10⁻¹² and 4.49 × 10⁻¹² cm s⁻¹ for 75% fiber group, 75% powder group and Celgard 2400, respectively, Table S1†), which confirms that the high ratio of ZrO₂ has no impact on the diffusion of active materials on the surface of the positive electrode.

Fig. 5 Electrical properties and electrochemical performance of the cells composed of Celgard 2400, 75% ZrO₂ powder and 75% ZrO₂ fiber separators. (A) LSV curves, (B) self-discharge curves, (C) EIS profiles, (D) first charge/discharge curves, (E) cycle performance in the cut-off voltage range of 3.0 to 4.3 V and a current rate of 0.2C and (F) rate capability under charge/discharge rates from 0.2 to 8C.

As one of the most important indexes to evaluate the electrochemical performance of cells, ionic conductivities of the separators are determined by electrochemical impedance spectroscopy (EIS). According to the interfacial resistance shown in Fig. 5C, the calculated ionic conductivity is 0.320 mS cm⁻¹ for ZrO₂ fiber separator and 0.299 mS cm⁻¹ for ZrO₂ powder separator, each is ∼60% higher than that of Celgard 2400 (0.190 mS cm⁻¹). The ZrO₂ fiber separator exhibits the highest ionic conductivity, which is consistent with the porosity and pore size results. The increase of porosity leads to a higher electrolyte uptake and consequently enhances the ionic conductivity, while a mixture of large and small pores can synergistically improve the battery performances by generating a uniform ion transport pathway.^42–44

The influence of porous structure on cell performance is further verified by charging and discharging the cells at a constant current of 0.2C in the voltage range of 3.0–4.3 V. The first cycle indicates no abnormal charge/discharge behavior from the three groups of cells (Fig. 5D). The discharge capacity of the ZrO₂ fiber group (165.7 mA h g⁻¹) is slightly higher than those of Celgard 2400 (160.8 mA h g⁻¹) and ZrO₂ powder group (150.2 mA h g⁻¹), and the relatively mild discharging process of the ZrO₂ fiber group is attributed to the low interfacial resistance. The cycling performances of the three groups of cells are evaluated by charging and discharging the cells at a rate of 0.2C/0.2C for 50 cycles. As depicted in Fig. 5E, the ZrO₂ fiber group delivers the highest retention rate of ∼93%, which is 3% higher than that of Celgard 2400 and 45% higher than that of the ZrO₂ powder group. A similar trend is also found in the inset of Fig. 5E, where the coulombic efficiency (calculated from the discharge/charge capacity) of cells using the ZrO₂ fiber separator and Celgard 2400 remains at ∼100% after 50 cycles. The desirable cycling performance of the ZrO₂ fiber group can be explained by the rapid transmission of lithium ions. It also implies that with the mechanical support from ZrO₂ fibers, the separator is robust enough to withstand any stress change, or any possible expansion/contraction from positive and negative active substances during the continuous charge/discharge cycles. By contrast, the coulombic efficiency of cell using the ZrO₂ powder separator gradually decreases with the increase of cycling number, and is ∼93.2% at the end of the process. Due to the fact that chemical composition, blending ratio, porosity and ionic conductivity of ZrO₂ powder separator and ZrO₂ fiber separator are highly comparable, the lower coulombic efficiency of the powder group is most likely caused by the insufficient wet strength of the separator. With such a low strength, the structural integrity of the separator might be easily damaged during cycling, resulting in the generation of unwanted electric leakage.

The rate capacities of separators under various charge/discharge rates are measured to validate the reliability of the batteries. The first 5 cycles are operated at a 0.2C/0.2C rate to form a solid electrolyte membrane on the surface of the electrode. When the discharge rate is below 2C, the specific capacities of the cells decrease slightly (Fig. 5F). However, a notable fading in capacity is observed when the discharge rate exceeds 4C, probably due to the internal polarization of the battery. The capacity of the ZrO₂ powder group is significantly lower than those of Celgard 2400 and the ZrO₂ fiber group at all discharge rates. This trend becomes more severe when the discharge rate increased to 8C, as quantitative results in the inset of Fig. 5F suggest that the capacity retention rates of the ZrO₂ powder group, ZrO₂ fiber group and Celgard 2400 are 24%, 38% and 35%, respectively. Under high current density, although the initial pore size of the ZrO₂ powder separator is ideal, its poor mechanical properties fail to prevent the shape and size change of the pores. The energy and power densities of the cells were computed from the discharge data. As illustrated in the Ragone plots (Fig. S7†), despite the high areal density caused by the addition of ZrO₂ fibers, the highest energy density is found in cells fabricated by 75% fiber separators, which once again reflects the advantage of using fibers as the reinforcement for separators.

For comparison, cycle stability and rate capacities under different charge/discharge rates were measured at 60 °C. The moderate increase of temperature shows no interference on the three types of cells, as the overall electrochemical performances of the cells are similar to those at room temperature (Fig. S8A–C†). Since further increasing of temperature may cause the decomposition of the electrolyte, we used OCV to monitor the thermal stability of the cells (Fig. S8D†). The OCV of Celgard 2400 group decreases to 2.39 V in less than 10 min, and is only 0.26 V after 210 min, probably due to the internal short circuits caused by the thermal shrinkage of PP. On the contrary, cells composed of 75% ZrO₂ fiber separator and 75% ZrO₂ powder separator are able to maintain the original OCV throughout the testing time. In order to find the optimal fiber ratio, electrochemical properties of cells based on 25% and 50% fiber separators were also assessed (Fig. S9†). Along with the increase of cycle number and charge/discharge rate, the retention rates and interfacial resistance of these cells become lower than those of the 75% fiber group, proving that a higher concentration of ZrO₂ fibers can improve the thermal stability without any sacrifice of the electrochemical performances.

The electrochemical results indicate that the 75% ZrO₂ fiber separator is suitable for long-term battery systems. When it comes to practical applications, it is essential to scale up the production and reduce the cost. Our approach does not involve sophisticated procedures or delicate equipment. More attractively, the phase inversion and the blending of organic/inorganic materials can be achieved simultaneously, making us believe the feasibility of manufacturing battery separators by using inorganic fiber substrates. Future study will try to reduce the cost by replacing ZrO₂ fibers with cheaper alternatives (i.e., glass fibers).

Conclusions

In summary, we have proposed and demonstrated the concept of using inorganic fibers as substrates to fabricate separators for lithium-ion batteries. The uniform distribution of inorganic fibers among porous organic macromolecules, along with the friction and cohesion between fibers, provides good mechanical properties (under both dry and wet conditions), appropriate pore size, high electrolyte affinity and extraordinary thermal stability (against long-term heating and combustion). Compared to the separator composed of inorganic powders, the fiber-based separator exhibits superior structural integrity, and therefore enhanced electrochemical performances. Taken together, we believe that the simple but effective method developed in this study may become a platform strategy for the manufacture of battery separators.

Acknowledgements

The authors wish to acknowledge financial support from the National Natural Science Foundation of China (No. 51403033), “Chen Guang” Project from Shanghai Municipal Education Commission and Shanghai Education Development Foundation (No. 14CG34) and the Fundamental Research Funds for the Central Universities (No. 2232014D3-15).

Notes and references

1. B. L. Ellis, K. T. Lee and L. F. Nazar, Chem. Mater., 2010, 22, 691–714.
2. P. Zhang, L. C. Yang, L. L. Li, M. L. Ding, Y. P. Wu and R. Holze, J. Membr. Sci., 2011, 379, 80–85.
3. Y. S. Zhu, S. Y. Xiao, Y. Shi, Y. Q. Yang and Y. P. Wu, J. Mater. Chem. A, 2013, 1, 7790–7797.
4. H. Y. Liao, H. Q. Hong, H. Y. Zhang and Z. H. Li, J. Membr. Sci., 2016, 498, 147–157.
5. J. H. Kim, J. H. Kim, K. H. Choi, H. K. Yu, J. H. Kim, J. S. Lee and S. Y. Lee, Nano Lett., 2014, 14, 4438–4448.
6. J. L. Hao, G. T. Lei, Z. H. Li, L. J. Wu, Q. Xiao and W. Li, J. Membr. Sci., 2013, 428, 11–16.
7. T. H. Cho, M. Tanaka, H. Ohnishi, Y. Kondo, M. Yoshikazu, T. Nakamura and T. Sakai, J. Power Sources, 2010, 195, 4272–4277.
8. T. Ma, Z. Cui, Y. Wu, S. Qin, H. Wang, F. Yan, N. Han and J. Li, J. Membr. Sci., 2013, 444, 213–222.
9. W. Xiao, L. N. Zhao, Y. Q. Gong, J. G. Liu and C. W. Yan, J. Membr. Sci., 2015, 487, 221–228.
10. Y. E. Miao, G. N. Zhu, H. Hou, Y. Y. Xia and T. Liu, J. Power Sources, 2013, 226, 82–86.
11. K. Fu, Y. H. Gong, J. Q. Dai, A. Gong, X. G. Han, Y. G. Yao, C. W. Wang, Y. B. Wang, Y. N. Chen, C. Y. Yan, Y. J. Li, E. D. Wachsman and L. B. Hu, Proc. Natl. Acad. Sci. U. S. A., 2016, 113, 7094–7099.
12. S. Y. Hu, S. D. Lin, Y. Y. Tu, J. W. Hu, Y. Wu, G. J. Liu, F. Li, F. M. Yu and T. T. Jiang, J. Mater. Chem. A, 2016, 4, 3513–3526.
13. Q. S. Wang, P. Ping, X. Y. Zhao, G. Q. Chu, J. H. Sun and C. H. Chen, J. Power Sources, 2012, 208, 210–224.
14. J. L. Shi, Y. G. Xia, Z. Z. Yuan, H. S. Hu, X. F. Li, H. M. Zhang and Z. P. Liu, Sci. Rep., 2014, 5, 8255.
15. T. Lee, W. K. Kim, Y. Lee, M. H. Ryou and M. L. Yong, Macromol. Res., 2014, 22, 1190–1195.
16. W. Jiang, Z. H. Liu, Q. S. Kong, J. H. Yao, C. J. Zhang, P. X. Han and G. L. Cui, Solid State Ionics, 2013, 232, 44–48.
17. B. Scrosati, J. Hassoun and Y. K. Sun, Energy Environ. Sci., 2011, 4, 3287–3295.
18. Y. S. Jung, A. S. Cavanagh, L. Gedvilas, N. E. Widjonarko, I. D. Scott, S. H. Lee, G. H. Kim, S. M. George and A. C. Dillon, Adv. Energy Mater., 2012, 2, 1022–1027.
19. C. J. Orendorff, T. N. Lambert, C. A. Chavez, M. Bencomo and K. R. Fenton, Adv. Energy Mater., 2013, 3, 314–320.
20. X. Luo, W. Pan, H. G. Liu, J. H. Gong and H. Wu, Ionics, 2015, 21, 3135–3139.
21. C. Y. Cao, L. Tan, W. W. Liu, J. Q. Ma and L. Li, J. Power Sources, 2014, 248, 224–229.
22. Y. Song, Y. Shen, H. Y. Liu, Y. H. Lin, M. Li and C. W. Nan, J. Mater. Chem., 2012, 22, 16491–16498.
23. J. J. Zhang, L. P. Yue, Q. S. Kong, Z. H. Liu, X. H. Zhou, C. J. Zhang, Q. Xu, B. Zhang, G. L. Ding and B. S. Qin, Sci. Rep., 2014, 4, 3935.
24. E.-S. Choi and S.-Y. Lee, J. Mater. Chem., 2011, 21, 14747–14754.
25. Z. Feng, X. Ma, C. B. Cao, J. L. H. Li and Y. Q. Zhu, J. Power Sources, 2014, 251, 423–431.
26. S. M. Kang, M. H. Ryou, J. W. Choi and H. Lee, Chem. Mater., 2012, 24, 3481–3485.
27. Y. H. Liao, X. P. Li, C. H. Fu, R. Xu, L. Zhou, C. L. Tan, S. J. Hu and W. S. Li, J. Power Sources, 2011, 196, 2115–2121.
28. K. C. Ren, Y. H. Xu, D. R. Chen, C. X. Wu, Y. Q. Ren, and X. P. Zhou, China Pat., CN 103757749 A, 2014.
29. Y. S. Zhu, F. X. Wang, L. L. Liu, S. Y. Xiao, Z. Chang and Y. P. Wu, Energy Environ. Sci., 2013, 6, 618–624.
30. Y. J. Kim, S. M. Lee, S. H. Kim and H. S. Kim, J. Power Sources, 2011, 16, 6997–7001.
31. Z. Z. Zhang, H. M. Xu, Z. L. Cui, P. Hu, J. C. Chai, H. P. Du, J. J. He, J. J. Zhang, X. H. Zhou and P. X. Han, J. Mater. Chem. A, 2016, 4, 2277–2285.
32. L. H. Yu, Y. Jin and Y. S. Lin, RSC Adv., 2016, 6, 40002–40009.
33. G. Nagasubramanian and R. G. Jungst, J. Power Sources, 1998, 72, 189–193.
34. J. C. Song, M. H. Ryou, B. Son, J. N. Lee, J. L. Dong, M. L. Yong, J. W. Choi and J. K. Park, Electrochim. Acta, 2012, 85, 524–530.
35. W. B. Zhang, Z. Shi, Z. Feng, L. Xia, J. Jian and J. Lei, Adv. Mater., 2013, 25, 2071–2076.
36. M. M. Noor, M. A. Careem, S. R. Majid and A. K. Arof, Mater. Res. Innovations, 2013, 15, 157–160.
37. M. M. Mikhailov and V. A. Vlasov, Radiat. Meas., 2016, 91, 15–20.
38. M. Siddiqui, A. Ibrahimalotaibi, A. Mohmmedmahfouz and R. Mohamad, Mater. Res., 2012, 15, 986–989.
39. Q. J. Wang, Z. X. Jian, W. L. Song, S. C. Zhang and L. Z. Fan, Electrochim. Acta, 2014, 149, 176–185.
40. H. Lee, M. Yanilmaz, O. Toprakci, K. Fu and X. Zhang, Energy Environ. Sci., 2014, 7, 3857–3886.
41. J. J. Zhang, Z. H. Liu, Q. S. Kong, C. J. Zhang, S. P. Pang, L. P. Yue, X. J. Wang, J. H. Yao and G. L. Cui, ACS Appl. Mater. Interfaces, 2013, 5, 128–134.
42. J. Heo, Y. Choi, K. Y. Chung and J. H. Park, J. Mater. Chem. A, 2016, 4, 9496–9501.
43. M. Kim and J. H. Park, Adv. Energy Mater., 2013, 3, 1417–1420.
44. Y. B. Kim, T. Tranphu, M. Kim, D. W. Jung, G. R. Yi and J. H. Park, ACS Appl. Mater. Interfaces, 2015, 7, 4511–4517.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ta08404d

---