Carbon nanofiber frameworks for Li metal batteries: the synergistic effect of conductivity and lithiophilic-sites

Zhen Yang; Zheng Li; Yanqi Liu; Zhaohai Song; Zhihao Gao; Jianmin Zhang; Zongmin Zheng

doi:10.1088/1361-6528/ac73a5

1. Introduction

Advanced energy storage is an intrinsic driving force for modern life. Lithium-ion batteries have achieved great applications in recent years as they can be utilized conveniently [1–3]. To satisfy the rapid development of electric vehicles and portable electronic devices, Li metal, Li–S and Li–O₂ batteries are regarded as the most promising battery systems because of the lowest negative potential (–3.04 V) and the highest theoretical capacity (3860 mAh g^–1) of Li metal [3–6]. However, the intrinsic high activity of Li metal and the formation of Li dendrites will lead to the consumption of a large amount of electrolyte and the penetration of the separator, which are the main cause of poor cyclic stability, low coulomb efficiency and serious safety problems [7–10]. In order to solve the uneven Li plating/stripping, kinds of strategies such as solid electrolyte [11], electrolyte additives [12–14], artificial lithiophilic surface [9, 15], functional interlayers [16–18] and 3D current collectors [19–21] have been used to guide metallic Li nucleation. Among these strategies, great attentions have been paid to effectively suppress the formation of Li dendrites due to the fact that lithiophilic sites can guide the uniform Li nucleation and growth. For example, Fu et al [9] found that sodium formate (SF)-treated Cu current collectors can enhance the lithiophilicity and inoxidizability of Cu current collectors. To further promote the energy density of Li metal batteries, the lithiophilic modifying strategy changes from traditional metal current collectors to lighter collectors, even collector-free. Feng et al [15] reported N-doped carbon bowls as the lithiophilic scaffold host to guide the Li plating process, presenting a low lithium metal nucleation overpotential of 18 mV, high CE of 98%, and stable cycling without obvious voltage fluctuation for over 600 cycles at a current density of 1 mA·cm⁻². High conductivity, highly porous structure, and lightweight make carbon frameworks the most concerning materials for anode current collectors. Indeed, carbon interlayers also can effectively reduce the charge density distribution and thus slow down the dendrite growth. Thus, large quantities of hollow carbon nanospheres [22], carbon nanofibers [18, 23] and nanosheets [24–26] have been used as interlayers between Li metal and the separator for Li metal batteries. However, due to the poor affinity between lithium and pure carbon, uneven deposition of metallic Li still happens on the interlayer surface, resulting in the poor durability of the long cycle performance of the battery. Thus, lithiophilic-sites such as electronegative heteroatoms are introduced in non-polar carbon frameworks. These lithiophilic-sites act as 'seeds' to guide the metallic Li nucleation [25, 27]. And here we can see that the conductivity and lithiophilic-sites are both essential for uniform Li plating/stripping. But what is the relationship between these two contributing factors? The more lithiophilic-sites the better for the reversibility? Little work has been taken into consideration for these questions.

With regard to three-dimensional and conductive structure, carbon nanofiber framework (CNF) prepared by electrospinning and carbonization of polyacrylonitrile was studied as the interlayer for Li metal protection. The amounts of N and O elements in CNF can be easily regulated by adjusting the conditions of heat-treatment. With the optimized CNF membrane as the interlayer, the voltage profiles reveal a low and stable nucleation overpotential for Li deposition. In the Li metal and Li–S full cells, the optimized CNF performs excellent cyclic stability and rate properties.

2. Experimental section

2.1. Chemicals

Polyacrylonitrile (PAN, average M.W. 150 000) was purchased from J&K Scientific Co., Ltd. S powders and N,N dimethyformamide (DMF, 99.9%) was purchased from Sinopharm Chemical Reagent Co., Ltd. Li foils were supplied by China Energy Lithium Co., Ltd. LiFePO₄ (LFP) powders and the electrolyte were purchased from Suzhou Duoduo Chemical Technology Co., Ltd. All chemicals were used as received.

2.2. Preparation of N, O-doped CNF membrane

CNF interlayer membranes were prepared by a simple electrospinning method followed by pre-oxidation and carbonization. In a typical procedure, 0.7 g of PAN was added to 5 ml of DMF. The mixture was vigorously stirred for 5 h and then ejected from a stainless-steel capillary with a voltage of 14 kV. The flooding rate was set to 2 μl min⁻¹. The distance between the capillary and the collector was 15 cm. The precursor membrane was collected on a roller coated with Al foil (20 × 20 cm). As-collected membrane was firstly treated at 280 °C for 2 h in air with a heating rate of 2 °C min⁻¹. Then, the pre-oxidated membranes were calcined under Ar atmosphere at 400 °C, 500 °C, 600 °C, 700 °C and 800 °C, respectively. The heating rate was 5 °C min⁻¹. The CNF membranes with different thicknesses (20–92 μm) were collected by controlling the injection volume (1.7–6.7 ml) of the precursor solution respectively.

2.3. Structural and morphological characterizations

The morphology and structure of the membranes and Li metals were characterized by scanning electron microscopy (SEM, Zeiss SUPRA55 SAPPHIRE). The elements and structural analysis were characterized with the energy dispersive spectrometer (EDS) and x-ray photoelectron spectroscopy (XPS, PHI 5000 Versaprobe III). Before the SEM analysis of cycled membranes and Li foils, the samples were rinsed with DME(1,2-dimethoxyethane)-DOL(1,3-dioxolane) mixture to remove the residual electrolytes.

2.4. Electrochemical measurements

The CNF membranes were cut into circle pieces (16.5 mm in diameter). The Li plating/stripping behaviors on CNF membranes with different carbonization temperatures and thicknesses were investigated using symmetric CNF@Li∣CNF@Li cells and asymmetrical CNF@Li∣Cu foil half-cells (CR2032 type). The LiFePO₄ (LFP) cathode and S cathode were prepared according to the previous works [28, 29]. The LFP cathodes were made by casting slurries composed of LFP (80 wt%), Super P (10 wt%) and polyvinylidene difluoride (10 wt%) in N-Methyl pyrrolidone (NMP) solvent onto Al foils by a spreader. After being dried at 110 °C in a vacuum drying oven for 12 h, the film was cut into circle discs (12 mm in diameter). The areal mass loading of LFP active materials was controlled at around 3.7 mg cm⁻². For the S cathode preparation, a homogeneous slurry of C/S material (75 wt% of S and 25 wt% of Super P), Super P and LA132 (the weight ratio is 8:1:1) was coated onto Al foils by a spreader. After being dried at 60 °C in a vacuum drying oven for 12 h, the film was cut into circle discs (12 mm in diameter). The weight of S in the cathode was controlled to ∼1.3 mg. To investigate the practical applications of the optimized CNF membranes, CNF@Li∣LFP and CNF@Li∣C/S full cells were also evaluated. The separator and the electrolyte were the commercial Celgard-2500 and 1.0 M LiTFSI/DME-DOL (V:V, 1:1) with 1 wt% LiNO₃, respectively. The charging–discharging tests were taken with Neware battery test system (BTS-4000), respectively. All the cells were assembled and disassembled in an Ar-filled glovebox (<0.1 ppm of O₂). All the electrochemical tests were performed at 25 °C.

3. Results and discussion

Carbon nanofibers prepared by pyrolytic electrospinning PAN is a usual method and the chemical and morphological characteristics are different at different cracking temperatures [30, 31]. Panels a and b of figure 1 show that the carbon nanofibers with a uniform diameter of 400 nm are interlaced to form a network. The good mechanical property maintains the integrity of CNF membrane by confining the volume change in long-range order during cycling. As the temperature rises from 400 °C to 800 °C, the average diameter decreases from 510 to 380 nm (figures S1, S2 (available online at stacks.iop.org/NANO/33/355402/mmedia)) and the average Ohmic resistance decreases from 110 k to 450 Ω (figure S3). According to the EDS results in figure S1, the total atom ratio of N and O elements decreases from 36.4% to 13.4% and the bulk N/O atomic ratio is for 2.4, 3.1, 6.6, 4.8 and 1.1, respectively. These results are common with many carbonized PAN fibers [32]. To investigate the surface chemical structure of the different membranes, the XPS spectrums were performed. As shown in figure S4, the full survey XPS spectrums indicate that C, N, and O elements are present and the total atom ratio of N and O elements decreases from 29.1% to 13.4%, matching the EDS results in figure S1. As the temperature rises from 400 °C to 800 °C, the surface N/O ratio is 1.4, 1.2, 0.4, 1.5 and 0.58, respectively. The high-resolution XPS of C 1s, N 1s and O 1s shown in figures 1(c)–(e) all can be deconvoluted into kinds of bond structures, indicating that the heteroatoms are doped in the carbon skeleton [25, 33]. The C 1s XPS spectrums in figure 1(c) show that, when the carbonization temperature is below 600 °C, the ratio of conductive carbon units (C=C bonds, 284.8 eV) is less than 50% and cannot form long conductive channels. This result helps to explain the poor conductivity of CNF-400 and CNF-500 samples. The N 1s XPS spectrums in figure 1(d) show that, as the temperature rises from 600 °C to 800 °C, the ratio of graphitic-N increases while the pyrrolic-N decreases. The O 1s XPS spectrums in figure 1(e) show that, the O element resulting from the pre-oxidation process gradually transform into the O–C (≈530.8 eV) and O=C(≈532.0 eV) bonds on the surface of CNF. Based on the above analysis, it can be concluded that the high carbonization temperature of CNF film brings low amounts of heteroatoms and good conductivity. However, the N/O ratio displays differently in the bulk and surface, which may affect the bonding environments resulting in different lithophilic regions on different CNF samples.

Figure 1. Refer to the following caption and surrounding text. — **Figure 1.** (a), (b) The SEM images of CNF membrane with inset digital photo; (c) the high-resolution C 1s, (d) N 1s and (e) O 1s XPS spectrums of different CNF membranes.
Download figure:
Standard image High-resolution image

The symmetric cells are used to evaluate the voltage profiles of different CNF membranes. Figure 2(a) shows the voltage profiles of CNF membranes carbonized at different temperatures at 5 mA cm⁻² and 2 mAh cm⁻². As the temperature rises from 400 °C to 600 °C, the overpotential of Li plating/stripping is greatly reduced. As a result of enhanced conductivity, the CNF film acts as a dense conductive network to equalize the current density and guide the uniform Li deposition. But when the CNF membrane is carbonized at 800 °C, the overpotential gets bigger, which is resulted from the decreased N and O lithiophilic sites and uneven deposition of Li. Furthermore, with the increase of the cyclic number, membranes carbonized at low temperatures exacerbating the uneven deposition of lithium, resulting in dendrite generation and even short circuits. It is essential for the CNF membrane with stable protection at over 600 °C. Other symmetric cells with the interlayers of CNF-600 and CNF-700 at a high current density of 10 mA cm⁻². The optimized CNF-700 membrane displays stable overpotential at about 15 and 30 mV over 1200 h at 5 and 10 mA cm⁻², respectively (figure S5). Meanwhile, the voltage profiles of CNF membranes for Li deposition plating/stripping are also evaluated and the results are shown in figure 2(c). It is observed that the thin membrane can't play very well in protecting anode as the repeated deposition and stripping of Li metal. However, the too thick membrane is unfavorable for the fast transition of Li ions, resulting in a large overpotential. The reasons can be well explained by the maximum capacity of CNF membranes with different thicknesses for Li deposition (figure 2(d)). The thick CNF membrane is necessary for providing enough surfaces for Li deposition. Combining the analysis of conductivity with the amount of N and O elements, the good electronic conductivity is more useful to balance the charge density between Li plating/stripping, and the heteroatoms lithiophilic-sites are favor for the uniform nucleation of metallic Li. The synergistic effects of the two factors contribute to the stable Li plating/stripping process.

Figure 2. Refer to the following caption and surrounding text. — **Figure 2.** (a) The voltage profiles of different CNF membranes at 5 mA cm⁻² and 2 mAh cm⁻²; (b) voltage profiles of CNF-600 and CNF-700 at 10 mA cm⁻² and 2 mAh cm⁻²; (c) voltage profiles of CNF membranes with different thickness at 5 mA cm⁻² and 2 mAh cm⁻² and (d) the capacities of CNF membranes with different thickness at 5 mA cm⁻².
Download figure:
Standard image High-resolution image

To further investigate the morphology of the electroplated Li metal on bare Li foil and CNF membrane, the symmetric cells after 10 and 100 cycles are disassembled and characterized, respectively. As shown in figure 3(a), obvious wormlike Li metal dendrites grow on the bare Li foils as the cycle increases, indicating an uncontrolled Li plating behavior. Plenty of mossy 'dead Li' cover on the surface. However, most of the surface of CNF-protected Li foil is smooth even after repeated Li plating/stripping for 100 times (figure 3(b)). The covered CNF membrane acts as the host for Li metal plating, even after 100 cycles of repeated Li plating/stripping, uniform Li agglomerations grow along the surface of carbon fibers (figures 3(c) and (d)). No obvious 'branch' or 'dead' Li formed on the separator side, which would greatly improve the cyclic performance of Li metal battery. It is observed that the lithiophobic surface and conductivity of CNF interlayer are favorable for guiding the Li nucleation.

Figure 3. Refer to the following caption and surrounding text. — **Figure 3.** The surface SEM images of (a) bare Li, (b) CNF membrane protected Li and (c), (d) CNF membrane, respectively, after 100 cycles with the capacity of 2 mAh cm⁻² at 5 mA cm⁻².
Download figure:
Standard image High-resolution image

A current of 1 mA cm⁻² was applied to the half-cell to electroplate 1 mAh cm⁻² of Li metal on the Cu foil electrodes with and without CNF protection. The Li electroplate was stripped by charging the cell to 1 V, and the CE (Coulombic Efficiency) was calculated by dividing the amount of stripped Li by that of plated Li, as shown in figure 4(a). For the bare Cu foil electrode, Li is notorious for its high reactivity and is prone to dendrite growth, resulting in low CE and short circuits upon repeated plating/stripping cycles. With the protection of CNF membrane, the CE of CNF@Cu half-cell could remain at 100% over 120 cycles. But it can't be ignored that the CE of the CNF@Cu half-cell is gradually raised from 85% to 100%. Many works reported the same results due to certain unreversible Li reactions within the interlayer and the consumption of SEI in the first several cycles. Figures 4(b) and (c) show the voltage profiles of Li electroplating/stripping at different cycles. The CNF-protected electrode displays a higher initial nucleation overpotential than the bare Cu electrode, which is caused by the lower conductivity of the CNF membrane compared with Cu metal. When the surface of the CNF membrane reaches a stable state, the overpotential of CNF@Cu for Li electroplate/stripping is around 20 mV, which is lower than the bare Cu foil (23 mV).

Figure 4. Refer to the following caption and surrounding text. — **Figure 4.** (a) The CEs of CNF@Cu and bare Cu foil electrodes upon repeated Li plating/stripping cycles and the corresponding voltage profiles: (b) CNF@Cu and (c) bare Cu foil.
Download figure:
Standard image High-resolution image

The protecting effect of the CNF membrane is evaluated by the charging–discharging measurements of the Li–LFP coin cells (the mass loading of LFP is ∼3 mg cm^–2). As shown in figure 5(a), the CNF@Li∣LFP cell obtains an initial reversible capacity of 137 mAh g^–1 at a high-rate of 1 C (1 C = 170 mA g^–1), which is higher than the Li foil∣LFP cell. After 1000 cycles, the CNF@Li∣LFP cell can still work with a capacity of 101 mAh g^–1, and the average decay rate of specific capacity per-cycle is only 0.026%. Surprisingly, the CE of the CNF@Li∣LFP cell keep around 99.8% after 1000 cycles (figure 5(b)). Nevertheless, a serious fluctuation of CE happens in Li foil∣LFP cell after 700 cycles, probably resulting from the formation of 'dead Li'. Rate-properties are also measured for the two cells and the specific capacities and the charging–discharging curves are shown in figures 5(c)–(e). It can be seen that the reversible specific capacities of CNF@Li∣LFP cell at 0.2, 0.5, 1 and 2 C are 163, 156, 147 and 134 mAh g⁻¹, respectively, which are significantly higher than that of Li foil∣LFP cell. Furthermore, the small polarization of CNF@Li∣LFP cell at high current density indicates that the CNF membrane can enhance the transport of both electrons and lithium ions.

Figure 5. Refer to the following caption and surrounding text. — **Figure 5.** (a) Cycling performance of the Li foil∣LFP and CNF@Li∣LFP cells at 1 C and (b) the corresponding CEs; (c) rate capabilities of the CNF@Li∣LFP and Li foil ∣ LFP cells and the corresponding charging–discharging curves: (d) CNF@Li∣LFP and (e) Li foil ∣ LFP, respectively.
Download figure:
Standard image High-resolution image

Given the excellent properties of CNF@Li∣LFP cell, Li–S cells using commercial C/S cathodes (Super P with 75 wt% S) are also investigated. In addition to the similar plating/stripping problems of Li metal Li–LFP cells, serious adverse reactions exist on the anode side of Li–S cell caused by the 'shuttle effect' [5, 34]. Figure 6(a) shows the charging–discharging specific capacities of Li–S cells at 0.2 C (1 C = 1675 mA g^–1). Compared to the Li foil∣C/S cell (781 mAh g^–1), CNF@Li∣C/S obtains a higher initial discharging specific capacity of 1072 mAh g^–1. After several cycles, the capacity can remain stable and the reversible capacity of 839 mAh g^–1 can be obtained after 100 cycles with stable CE over 99%. The charging–discharging curves of these two Li–S cells are also shown in figures 6(b) and 6(c). It is indicated that CNF@Li∣C/S performs a longer platform at low voltage (liquid–solid formation regime) of the entire discharging time, indicating the little capacity loss by the 'shuttle' of polysulfide [35]. The results show that the CNF membrane can not only guide the nucleation of Li metal, but also reduce the direct contact between polysulfide and the surface of Li anode.

Figure 6. Refer to the following caption and surrounding text. — **Figure 6.** (a) Cycling performance of the CNF@Li∣C/S and Li foil∣C/S cells at 0.2 C and the corresponding charging–discharging curves at different cycles: (b) CNF@Li∣C/S and (c) Li foil∣C/S, respectively.
Download figure:
Standard image High-resolution image

4. Conclusions

In this work, the CNF membrane prepared by electrospinning and carbonization of polyacrylonitrile was studied as the interlayer for Li metal protection. The good electronic conductivity is useful to balance the charge density between Li plating/stripping, while the surface heteroatoms lithiophilic-sites are favored for the uniform nucleation of metallic Li. The thickness of CNF membrane determines the maximum capacity for Li deposition. With the optimized CNF membrane as the interlayer, the CNF@Li∣LFP cell can still work with a capacity of 101 mAh g^–1 after 1000 cycles at 1 C, and the average decay rate of specific capacity per-cycle is only 0.026%. Meanwhile, the CNF@Li∣C/S cell delivers a reversible capacity of 839 mAh g^–1 after 100 cycles at 0.2 C. The synergistic effects of the conductivity and lithiophilic-sites in CNF make it a potential candidate for Li plating/stripping in various Li metal batteries.

Acknowledgments

The authors thank the financial support from the National Natural Science Foundation of China (21805146), Postdoctoral Science Foundation of China (2018M630751). The authors also would like to thank Dr F Pei for the SEM measurements and San Zhang from Shiyanjia Lab (www.shiyanjia.com) for the XPS analysis.

Data availability statement

All data that support the findings of this study are included within the article (and any supplementary files).

Notes

The authors declare no competing financial interest.

Author e-mails

Author affiliations

ORCID iDs

Dates

Peer review information