Rational integration of spatial confinement and polysulfide conversion catalysts for high sulfur loading lithium–sulfur batteries

Abstract

Spatial confinement is a desirable successful strategy to trap sulfur within its porous host and has been widely applied in lithium–sulfur (Li–S) batteries. However, physical confinement alone is currently not enough to reduce the lithium polysulfide (Li₂S_n, 4 ≤ n ≤ 8, LIPSs) shuttle effect with sluggish LIPS-dissolving kinetics. In this work, we have integrated spatial confinement with a polar catalyst, and designed a three-dimensional (3D) interconnected, Co decorated and N doped porous carbon nanofiber (Co/N-PCNF) network. This Co/N-PCNF film serves as a freestanding host for sulfur trapping, which could effectively facilitate the infiltration of electrolyte and electron transport. In addition, the polar Co species possess strong chemisorption with LIPSs, catalyzing their reaction kinetics as well. As a result of this rational design and integration, the Co/N-PCNF@S cathode with a sulfur loading of 2 mg cm⁻² exhibits a high initial discharge capacity of 878 mA h g⁻¹ at 1C, and maintains a discharge capacity of 728 mA h g⁻¹ after 200 cycles. Even with high sulfur loading of 9.33 mg cm⁻², the cathode still keeps a stable areal capacity of 7.16 mA h cm⁻² at 0.2C after 100 cycles, which is much higher than the current areal capacity (4 mA h cm⁻²) of commercialized lithium-ion batteries (LIBs). This rational design may provide a new approach for future development of high-density Li–S batteries with high sulfur loading.

---

New concepts

It is noteworthy that extensive research efforts have focused on designing conductive porous carbon as a sulfur host material to both improve the conductivity of the electrodes, and entrap sulfur/lithium polysulfides (LIPSs) within confined spaces in lithium–sulfur (Li–S) batteries. However, the physical constraint of ionic LIPSs is limited because the interaction between polar LIPSs and nonpolar carbon is weak. Herein, we propose a new concept to overcome the above issues that involves rational integration of spatial confinement with a polar catalyst, where the Co/N-PCNF film serves as a freestanding host for the sulfur trapping, and which could effectively facilitate the infiltration of electrolyte and electron transport. In addition, the polar Co species possess strong chemisorption with LIPSs, catalyzing their reaction kinetics as well. This rational design may provide a new approach for future development of high-energy density Li–S batteries with high sulfur loading.

Introduction

With the rapidly increasing demand for electric vehicles and consumer electronics, rechargeable battery systems with improved energy density are expected to expand the traditional family of lithium-ion batteries (LIBs). Among the proposed candidates, lithium–sulfur (Li–S) batteries have become a promising next-generation energy storage option due to their high theoretical specific capacity (1675 mA h g⁻¹), high theoretical energy density (2567 Wh kg⁻¹) and low material cost.^1–8 Nevertheless, the insulating nature of both sulfur and its discharging products (Li₂S₂, Li₂S) causes a slow dynamic process, and hence poor rate capabilities.^9,10 More challenging, the excellent solubility of ionic lithium polysulfides (Li₂S_n, 4 ≤ n ≤ 8, LIPSs) in the electrolyte results in the loss of electrochemically active sulfur and generates the “shuttle effect”.^11–14 To address these issues, extensive research efforts have focused on designing conductive porous carbon as a sulfur host material to both improve the conductivity of the electrodes, and entrap sulfur/LIPSs within confined spaces.^15–18 Porous carbon structures possess large specific surface area, providing sufficient contact between sulfur and the electronic percolation, and numerous pores further imprison LIPSs through physical adsorption. Both functions could improve the utilization of the active materials.^19,20 However, the physical constraint of ionic LIPSs is limited because the interaction between polar LIPSs and nonpolar carbon is weak. While the slow conversion of dissolved LIPSs into insoluble Li₂S₂/Li₂S only further worsens the “shuttle effect” in long-term cycling.^21–23

Some recent research has proved that polar materials can greatly improve the cycling stability of sulfur cathodes by anchoring LIPSs via strong chemical bonding.^24–29 For example, Zhang et al. reported a composite of carbon nanotubes and sulfur decorated with polar TiN nanoparticles which delivered a high capacity of 1269 mA h g⁻¹ at 0.05C and showed long cycling stability with 1C for 400 cycles.³⁰ Polar materials not only realize the uniform adsorption of LIPSs, but can also be used as a catalyst to accelerate the conversion of LIPSs into discharge products (Li₂S/Li₂S₂).^31–34 Nazar et al. reported that polar MnO₂ served as a highly efficient polysulfide mediator binding LIPSs and accelerating their conversion.³¹ More recently, Lee et al. elucidated the catalytic activity of polar WO_3−x and MoS_2−x/rGO, which significantly enhanced the LIPS conversion kinetics in Li–S batteries.^33,34 These results suggest that the introduction of a catalyst to accelerate the conversion of dissolved LIPSs into Li₂S/Li₂S₂ presents a highly efficient method for improving electrochemical performance in Li–S batteries. Ideally, rational integration of a LIPS conversion catalyst with the spatial confinement effect of host materials appears interesting and promising; however, this idea has not yet been explored or reported. In addition, it is still difficult to realize the theoretical energy-density in practical applications of Li–S batteries as high areal sulfur loading remains an enormous challenge.^35,36 Freestanding cathodes, eliminating non-energy carrier materials such as the metal collector, conductive agents or polymer binders, have natural advantages in increasing sulfur loading compared with cathodes prepared from the conventional slurry-coating method. Therefore, freestanding 3D current collectors have been used to fabricate high-energy sulfur cathodes.

In this work, we developed a freestanding three-dimensional (3D) network consisting of polar Co species and nitrogen dopant porous carbon nanofibers (Co/N-PCNF) as a sulfur host via electrospinning. The fabricated Co/N-PCNF@S cathode possesses several advantages. Firstly, due to the enormous specific surface area from the 3D conductive network, Co/N-PCNF@S is easily infiltrated by electrolyte and provides abundant active sites. Secondly, the engineered structural pores could not only provide space for the volume changes of sulfur, but also provide physical confinement for LIPSs. Meanwhile the polar Co species and nitrogen dopant will generate strong chemical interactions with LIPSs. Most importantly, the polar Co species could effectively promote the redox reaction of LIPSs and reduce the “shuttle effect”. Finally, a high sulfur-loading cathode was obtained by a facile stacking layer-by-layer of freestanding Co/N-PCNF film strategy for developing high-energy batteries. With these advantages, the freestanding Co/N-PCNF@S realized high capacity and excellent rate capability in Li–S batteries, delivering a remarkably high capacity of 926 mA h g⁻¹ at 0.2C with a high sulfur loading of 9.33 mg cm⁻², corresponding to an areal capacity of 7.16 mA h cm⁻², which is almost two times that of the commercial LIBs.

Results and discussion

The Co/N-PCNF film was prepared with a simple electrospinning procedure as schematically drawn in Fig. 1. Firstly, monodispersed ZnCo-ZIF nanoparticles (Fig. S1a, ESI†) were mixed with an N,N-dimethylformamide (DMF) solution of polyacrylonitrile (PAN) to form a homogeneous turbid liquid. Then, this bluish liquid mixture was electrospun into ZnCo-ZIFs@PAN nanofibers (Fig. S1c, ESI†) that assembled into a purple film (Fig. S1d, the size of 9 × 15 cm, ESI†). Next, ZnCo-ZIFs@PAN nanofibers would be converted into a black Co/N-PCNF film (Fig. S2a, ESI†) after a carbonizing treatment. The N doped porous carbon nanofiber (N-PCNF) network was prepared as a control sample by using Zn-ZIF as a self-sacrificing template. Upon 180° bending (Fig. S2b, ESI†), the Co/N-PCNF film shows no breakage, revealing good mechanical strength and flexibility. Finally, sulfur was infiltrated into the carbonized Co/N-PCNF. The whole structure was termed as Co/N-PCNF@S and directly used as a cathode.

Fig. 1 Schematic fabrication process of the Co/N-PCNF@S hybrid composite.

The micro-structure and elemental mappings of the Co/N-PCNF film were characterized with scanning electron microscopy (SEM) and transmission electron microscopy (TEM). Carbon fibers in random directions interweave with each other forming a 3D interconnected network structure (Fig. 2a). The spaces allow good contacts with the electrolyte and enhance the rapid ionic transfer. The diameter of Co/N-PCNF fibers is about 500–800 nm, and the length reaches several hundreds of micrometers. This high length-to-diameter ratio endows the interweaving film with robust texture and excellent conductivity. In the SEM and TEM images (Fig. 2b–d), hollow nanocages are clearly visible over the fibers. It is clear that the ZnCo-ZIF nanoparticles dispersed in the fibers and acted as sacrificial templates. During the carbonizing treatment, organics of ZIF and PAN decomposed to a N doped carbon substrate. The ionic Zn in ZIFs was converted by carbon to metallic Zn and subsequently evaporated from the fibers at high temperature,³⁷ forming void space, and pores. Around the cages, carbon shells converted from PAN prevented the collapse of the ZIF polyhedron and some solid particles exist in hollow nanocages or voids (Fig. 2d). The HRTEM image (Fig. 2e) confirms the existence of graphitic carbon, as the annular inter-stripe distance of 0.34 nm is the typical distance between graphitic carbon (002) crystal planes.³⁸ Indeed in Fig. 2f, solid particles were identified as metallic Co given the lattice fringes, as the space between the particles was measured as 0.177 nm, agreeing with the (200) crystal plane of metallic Co.³⁹ High-angle annular dark field scanning transmission electron microscopy (HAADF-STEM) further confirms the porous structure of the Co/N-PCNF fibers (Fig. 2g), and the black dots on that image correspond to Co nanoparticles, which is further well consistent with the high light dots in Co elemental mapping (Fig. 2 Co). The signals of the C map and N map are coincident, revealing that N homogeneously distributed in the C fiber. In addition, the dispersive signal of the Co map indicated that there is considerable Co dispersed in the N doped carbon substrate in a smaller form. The morphology of N-PCNF is similar to Co/N-PCNF (Fig. S3–S5 ESI†) except that no Co nanoparticles exist in the hollow nanocages, which illustrates that introducing Co will not change the porous structure of the nanofibers.

Fig. 2 (a–c) SEM image of Co/N-PCNF; (d) TEM image of Co/N-PCNF; (e and f) HRTEM image of Co/N-PCNF; (g) HAADF-STEM image of a single Co/N-PCNF fiber and the elemental mappings of C, Co, and N; the corresponding colours are cyan, red and purple, respectively.

X-ray photoelectron spectroscopy (XPS) (Fig. S6 and S7 ESI†) was used to explore the element state of Co/N-PCNF and N-PCNF. From the Zn2p spectrum in Fig. S6d and S7b (ESI†) we found that no characteristic peaks of Zn appeared in the positions of 1022 eV (Zn2p_3/2) and 1045 eV (Zn2p_1/2).⁴⁰ The residual Zn content in the Co/N-PCNF and N-PCNF samples is infinitesimal and difficult to be detected, therefore, we believe that the influence of Zn on this system is very small and didn't take Zn into consideration. Signals for N1s have been detected around 400 eV. N is doped in the form of pyridinic N (398.5 eV), graphitic N (401 eV), pyrrolic N (400 eV) and in both Co/N-PCNF and N-PCNF. Meanwhile, the Co–N (399.2 eV) bond only appeared in Co/N-PCNF, and the corresponding peak can be found in the Co2p spectra. The Co–N bond indicates the existence of Co–N_x moieties, which were formed in the thermal decomposition of ZnCo-ZIF.

The N₂ adsorption/desorption measurement was carried out to investigate the porosity of the Co/N-PCNF and N-PCNF. The N₂ adsorption/desorption isotherms (Fig. S8a, ESI†) are typical H₄-type hysteresis loops, indicating the existence of micropores and mesopores in both Co/N-PCNF and N-PCNF.⁴¹ Co/N-PCNF and N-PCNF carry high Brunner–Emmett–Teller (BET) specific surface areas of 684 m² g⁻¹ and 662 m² g⁻¹, respectively. There is no big different between the two samples because of the same porous structure. The density functional theory (DFT) method was employed to analyze the pore size distribution. Results from Fig. S8b (ESI†) also proved the presence of mesopores and micropores. The high specific surface areas grant the materials with good physical adsorption for sulfur species and large interfaces for sulfur conversion, which could enhance the sulfur utilization. Fig. S9 (ESI†) is the Raman spectrum of Co/N-PCNF. The peak located around 1348 cm⁻¹ (D band) is associated with the disorder-induced asymmetric vibration mode and the one at 1598 cm⁻¹ (G band) is related to the in-plane stretching motion of sp2 carbons. The value of I_G/I_D both samples is less than 1, meaning that rich defects exist in the carbon matrix.⁴² A higher I_G/I_D is observed for Co/N-PCNF(0.963) than N-PCNF(0.864) resulting from the graphitization catalytic effect of Co nanoparticles.⁴³

To get Co/N-PCNF@S cathodes, the host material Co/N-PCNF films were cast-dropped with S/CS₂ solution and subjected to a melt-diffusion process at 155 °C. Upon diffusion, Xray diffraction (XRD) was used to determine the structures of both Co/N-PCNF and Co/N-PCNF@S. Before S-loading, Co/N-PCNF (Fig. 3a) showed a broad peak at around 23.5°, which was characteristic for amorphous carbon.⁴² Two weak peaks at 26° and 43° confirm that some of the carbon has been graphitized.⁴¹ While in the Co/N-PCNF@S spectrum, peaks at 23.3°, 26.0°, 26.8°, 27.9°, 28.9°, and 31.6° corresponding to elemental sulfur (PDF 42-1278) were all detected, indicating that sulfur had been successfully loaded into the porous carbon fibers.⁴⁴ According to the TGA analysis curves in Fig. 3b, the content of sulfur in the Co/N-PCNF@S is 62.2% and in N-PCNF@S it is 59.6%, which is consistent with the quantity we introduced in the Experiment section. When the temperature increases above 165 °C, sulfur begins to sublime and causes weight loss of Co/N-PCNF@S. Between 200 °C and 265 °C, the Co/N-PCNF@S suffers a quick weight loss of 58.6%. Then there is a gentle slope of 3.6% weight loss between 265 °C and 320 °C, because S embedded in the micropores of Co/N-PCNF is more stable and needs a higher temperature to evaporate away.

Fig. 3 (a) XRD patterns of Co/N-PCNF and Co/N-PCNFS; (b) TGA curves of Co/N-PCNF@S and N-PCNF@S; (c) and (d) SEM images of the Co/N-PCNF@S film; element mappings of the Co/N-PCNF@S; the corresponding colors of C, Co, N, and S element are cyan, red, purple and yellow, respectively.

After the sulfur loading, the Co/N-PCNF@S electrode still shows good flexibility and the freestanding nature (Fig. S10a and b, ESI†). The cross section of Co/N-PCNF@S shown in Fig. S10c (ESI†), has a thickness around 200 μm, with uniform and long fibers intertwined together suggesting that the fiber structures are well conserved. The corresponding elemental mapping analysis of Co/N-PCNF@S reveals the uniform distribution of S as shown in Fig. 3. More importantly, there is no obvious sulfur aggregation within the 3D network or on the surface of the fibers (Fig. 3c and d). The abundant pores in Co/N-PCNF provided space and large interface area, and sulfur can diffuse into the inside hollow carbon nanocages at 155 °C. The morphology of the N-PCNF@S electrode is also shown in Fig. S11 and S12 (ESI†). The distribution of sulfur in N-PCNF@S exhibits no difference to Co/N-PCNF@S, as N-PCNF has the same porous structure.

The Co/N-PCNF@S film was directly used as a freestanding cathode with lithium foil acting as an anode. The N-PCNF@S freestanding film cathode was also prepared as the contrast sample and tested under the same conditions. The cyclic voltammetry (CV) curves of the two cathodes, obtained within the voltage window 1.7–2.8 V at a scan rate of 0.1 mV s⁻¹, are shown in Fig. 4a. The curve shows two pairs of redox peaks. The cathodic peaks appear at about 2.00 and 2.19 V corresponding to the conversion of element S to the soluble long-chain LIPSs and the following transition of LIPSs to the solid-state Li₂S₂/Li₂S, respectively. Meanwhile, the two oxidative peaks at 2.32 and 2.40 V associated with the reverse reactions.²¹ The nearly overlapped CV curves in the initial 5 cycles (Fig. S13, ESI†) reveal that the electrochemical process of Co/N-PCNF@S is well reversible. Compared with the contrast sample (N-PCNF@S), the Co/N-PCNF@S cathode shows a smaller gap between the cathodic and anodic peaks in its curve, demonstrating a less severe potential polarization which suggests a lower energy barrier on the electrochemical conversion. Furthermore, as shown in Fig. 4b and Table S1, (ESI†), the Co/N-PCNF@S experiences a smaller charge-transfer resistance measured from electrochemical impedance spectroscopy³¹ than the N-PCNF@S, confirming the faster transformation of LIPSs. Improved sulfur transition kinetics suggests better rate capacities.

Fig. 4 (a) CV curves of the Co/N-PCNF@S and N-PCNF@S cathodes at 0.1 mV s⁻¹ between 1.7 and 2.8 V; (b) EIS spectra of the Co/N-PCNF@S and N-PCNF@S cathodes; (c) rate performance of the Co/N-PCNF@S and N-PCNF@S cathodes; (d) charge and discharge curves of the Co/N-PCNF@S cathode; (e) cycling performance of Co/N-PCNF@S and N-PCNF@S cathodes at 1C rate after two cycles activation at 0.05C and 0.2C, respectively.

The rate capabilities of Co/N-PCNF@S and N-PCNF@S cathodes, each with a sulfur loading of 2 mg cm⁻², are compared at various discharge/charge current densities (Fig. 4c). The electrodes were used for 10 cycles at each current density between 1.8 and 2.8 V. The Co/N-PCNF@S electrode gives the discharge capacities of 1048, 909, 850, 780, and 672 mA h g⁻¹ at different rates of 0.2C, 0.5C, 1C, 2C and 3C, respectively. When the current density is reduced back to 0.2C, it exhibits a great reversibility, and the discharge capacity still reaches 952 mA h g⁻¹. In comparison, the N-PCNF@S electrode shows much lower discharge capacities of 877, 746, 663, 558, and 512 mA h g⁻¹, respectively. The Co/N-PCNF@S cathode owes its better rate performance to the Co species that greatly improved the redox kinetics of LIPSs. Fig. 4d shows typical galvanostatic discharge–charge curves of the Co/N-PCNF@S cathode at the tested current densities. At all current densities, both the discharge curves and charge curves have two plateaus that are associated with the electrochemical conversion of sulfur, and agree with the positions of the redox peaks in the cathode's CV profiles. The cycling performances of Co/N-PCNF@S and N-PCNF@S are also compared (Fig. 4e). Before cycling at a high rate of 1C, the electrodes were activated for 2 cycles at 0.05C and 0.2C, respectively. The Co/N-PCNF@S cathode shows a high discharge capacity 878 mA h g⁻¹ at the fifth cycle, and gradually decreases to 728 mA h g⁻¹ after 200 cycles. The capacity retention is 83%; this means that the decay ratio per cycle is 0.07% on average. While the N-PCNF@S cathode delivers an inferior performance achieving capacity retention of 74% after 200 cycles.

Higher areal loading of sulfur is a straightforward way to achieve high energy density, yet still a critical challenge due to the irreversible aggregation of sulfur. Herein, by facilely stacking layer-by-layer the freestanding Co/N-PCNF film (each film about 2 mg cm⁻² of sulfur), we also increased sulfur loading and tested the cycling performances of the Co/N-PCNF@S cathodes with different sulfur loadings of (two-layer film) 4.67, (three-layer film) 6.41, and (four-layer film) 9.33 mg cm⁻², respectively. As shown in Fig. 5a, these Co/N-PCNF@S cathodes can deliver high areal capacities of 4.38, 6.84 and 8.35 mA h cm⁻², respectively, at a current density of 0.2C. The high areal capacities are considerably better than that of commercial LIBs (4 mA h cm⁻²). We also compared the performance of the Co,N-PCNF/S cathode with other freestanding sulfur cathodes. It can be seen from the Table S3 (ESI†) that the sulfur loading of Co,N-PCNF/S is not the highest, but it delivers the biggest areal capacities among the several carbon nanofiber-based cathodes which were prepared by the electrospinning route. This means that the areal energy density of Co,N-PCNF/S is superior.

Fig. 5 (a) Areal capacities of the N-PCNF@S cathode with 4.67, 6.41, and 9.33 mg cm⁻² of sulfur at 0.2C. (b) Schematic of the 3D conductive network and the catalysis function of Co species on LIPS conversion.

The superior performance is due to the integration of LIPS conversion catalysts with the spatial confinement effect due to the rational design of the 3D Co/N-PCNF network host materials as depicted in Fig. 5b. Firstly, the interlinked carbonaceous nanofibers act not only as an efficient current collector, but also as a wonderful sulfur host. After carbonization, the polar carbon substrate derived from Zn/Co-ZIF exists in the internal surface of the in situ formed hollow carbon nanocages. Sulfur could diffuse into the inside hollow carbon nanocages at 155 °C through the abundant pores and have a good contact with the polar N dopants and Co species. The 3D Co/N-PCNF network with a tremendous number of micro- and mesopores promotes the infiltration of electrolyte, and speeds up the transportation of electrons and Li⁺, which eventually guarantees improved performances even at high sulfur loadings and large charge/discharge rates. Then, the efficient confinement and the good catalytic activity of Co species promote the redox kinetics of soluble intermediate LIPSs. These benefits would not only guarantee a reduced loss of active material, but also alleviate the “shuttle effect” as the life of soluble LIPSs is shortened.

Further experiments and theoretical calculation have confirmed the above points. The ability to trap soluble LIPSs was examined with simple liquid phase adsorption experiments. Li₂S₆ (2 mmol L⁻¹) in 1,2-dimethoxyethane (DME) and 1,3-dioxolane (DOL) solvents (1 : 1 volume) was used as the representative LIPS. The same amount of Co/N-PCNF and N-PCNF was immersed in the Li₂S₆ solution for two days at room temperature, individually. The color changes of the Li₂S₆ solution after exposure to blank, Co/N-PCNF and N-PCNF composites are shown in the inset of Fig. 6a. The blank Li₂S₆ was barely faded after exposure for two days, and the one with N-PCNF still appears slightly yellowish. Meanwhile, a colorless Li₂S₆ solution was observed for that exposed to Co/N-PCNF, suggesting a significantly enhanced LIPS affinity with Co/N-PCNF. Fig. 6a shows the strong capability of Co/N-PCNF to adsorb Li₂S₆, which was further demonstrated by UV-vis spectroscopy, with a reduced LIPS concentration for the strongest absorbance in the 400–500 nm region for the polysulfide solution.

Fig. 6 (a) UV-Vis spectra of the Li₂S₆ solution after exposure to blank, Co/N-PCNF and N-PCNF; (inset is the optical photograph of Li₂S₆ solution after addition of samples); (b) CV of symmetric cells of N-PCNF and Co/N-PCNF electrodes with Li; the DFT calculation results of interaction between Li₂S₆ and different substrates: (c) N-graphene, (d) Co@N-graphene.

We have also studied the adsorption energy (E_ad) of Li₂S₆ on different substrates, which was calculated based on density functional theory (DFT). Since adsorption is an important step of catalysis, the adsorption energy can also reflect the catalyst affinity between different catalytic sites and the polysulfide intermediates. It can be found from Fig. 2g that the amount of Co particles is limited. Generally, one hollow carbon nanocage only has one Co particle. Meanwhile, Co–N_x species are numerous and uniformly disperse on the internal surface of the hollow carbon nanocages, which have adequate contact with the sulfur species. Therefore, we take the Co–N₄ mode in our calculation. The Co–N-graphene delivers an E_ad of −1.18 eV (Fig. 6d) to Li₂S₆ which is more negative than that of N-graphene (−0.98 eV, Fig. 6c). The above results confirm that the introduction of Co species can enhance the interaction between N-doping graphene and LIPSs, also indicating that the Co–N-graphene is more active for the catalysis conversion of Li₂S₆. As the “shuttle effect” is mainly caused by the soluble long chain polysulfides (Li₂S_x, 4 ≤ x ≤ 8), the DFT calculations of the interactions between Co–N-graphene and Li₂S₈ and Li₂S₄, respectively, were also performed. The results are shown in Fig. S14 (ESI†). The Co–N-G substrate displays a similar adsorption energy (−1.17 eV and −1.15 eV, respectively) of Li₂S₆ for both Li₂S₈ and Li₂S₄, which indicates that Co–N-graphene could fix soluble polysulfides effectively through strong bonding.

The cycled coin cells were disassembled to characterize the structure of the electrode after cycling and evaluate the migration of LIPSs from the sulfur cathodes to the Li anodes. In Fig. S15 (ESI†), the continuity of the fibers and the porous structure Co/N-PCNF and N-PCNF are well maintained after 300 cycles, which reveals the good structural stability of the films when used as a freestanding cathode on Li–S batteries. SEM images, elemental mapping of S (Fig. S16, ESI†) and energy dispersive spectroscopy (EDs) (Table S2, ESI†) of the Li anodes after 200 cycles offer more evidence. Polysulfides dissolved in the organic electrolytes would be transported to the anode side and corrode the surface of lithium, which will destroy the SEI protective layer, resulting in the terrible situation of a lithium anode. The morphology of the lithium anode surface paired with the Co/N-PCNF@S cathode is flat and lithium dendrites can’t be observed. The S content is only 5.65% (wt%) and it mainly exists in the crannies corroded by polysulfides. Meanwhile, lithium dendrites and more S content, 16.65% (wt%), nearly distribute in the whole surface of the lithium anode which was paired with the N-PCNF@S cathode. With the Co/N-PCNF@S cathode, even during long cycles much less LIPSs were dissolved in the organic electrolytes transported to the anode (Li foil) side and causing loss of active materials.⁴² These results reveal that the high specific surface area of the porous carbon nanofiber can effectively trap Li₂S₆ through physical absorption. The introduction of Co species can greatly strengthen the adsorption effect through chemical bonding.

To confirm the electrocatalysis of the LIPSs, we studied the kinetics reactions of LIPSs in the liquid phase on N-PCNF and Co/N-PCNF. The symmetric cells were assembled with Co/N-PCNF (or N-PCNF) disks as both working and counter electrodes in Li₂S₆ (0.2 M) electrolytes (Fig. 6b). In the blank electrolyte without Li₂S₆ there is only very low current occurring in the symmetric cell and no apparent peak is detected.^45,46 The Co/N-PCNF system displays larger current peaks than the N-PCNF electrode, suggesting the boosted redox reactions and accelerated conversion of LIPSs for the former.

The CV curves under different scan rates are described in Fig. S17 (ESI†), and one can find that as the scan rates increase, both the Co/N-PCNF@S curves and N-PCNF@S curves show stronger redox peaks, but the redox peaks in the Co/N-PCNF@S curves are higher and more distinct, which further proves the superiority of Co/N-PCNF@S. Furthermore, previous reports have suggested that there is a linear relationship between the current of the redox peaks and the square root of the scan rate, which could reflect the diffusion process of lithium ions. The lithium ion diffusion process follows the Randles–Sevcik equation:^45,46I_p = (2.69 × 10⁵)n^1.5AD^0.5Cv^0.5 (25 °C).

In this equation I_p is the peak current (A), n is the number of charge transfers, A is the active electrode area (cm²), D is the lithium ion diffusion coefficient (cm² s⁻¹), C is the concentration of Li ions (mol cm⁻³), and v is the scan rate (V s⁻¹). As n, A, and C in the cell are kept constant, the slope of the plot (I_p/v^0.5) is directly proportion to the diffusion rate of Li⁺D. Therefore, we investigated the behavior of the Co/N-PCNF@S and N-PCNF@S electrodes at different scanning rates to determine the Li⁺ diffusion rates in the two structures. For all scanning rates tested, the redox peaks for the Co/N-PCNF@S cathode are higher and more distinct from each other. The peak currents were plotted versus the square roots of the scan rates and obviously, the slopes for Co/N-PCNF@S are larger than those for N-PCNF@S in all three major redox peaks (Fig. S18, ESI†). The comparison indicates a faster diffusion of Li⁺ occurring in the Co/N-PCNF@S electrode. During the electrochemical cycling, the release of LIPSs into the electrolyte will increase its viscosity, resulting in restricted Li⁺ diffusion. Therefore, the faster diffusion, important for the rapid conversion of LIPSs, also implies the excellent LIPSs catalyzing capability and strong adsorption for LIPSs near the Co/N-PCNF@S surface.

To understand the reaction mechanisms of Co or N and LIPSs, the chemistry structure of Co/N-PCNF after the adsorption of Li₂S₆ was clarified with XPS. In the survey spectra (Fig. S19, ESI†), the characteristic peaks at around 164, 285, 400, 532, and 779 eV confirm the existence of S, C, N, O and Co elements. The detailed chemical states of C, Co, N, and S are described in the high-resolution XPS spectra. Fig. 7a shows that the C1s spectrum can be divided into three peaks, with the ones at 284.7 eV and 285.6 eV corresponding to the C=C bond and C=N bond, respectively. The peaks confirm the N heteroatoms in the carbon matrix. The high-resolution XPS spectrum of N1s (Fig. 7b) could be divided into four peaks corresponding to pyridinic N (398.4 eV), graphitic N (401 eV), pyrrolic N (400 eV) and Co–N (399.2 eV), and further prove the doping of N.^47,48 According to previous reports, pyridinic N and pyrrolic N enhance the polarity of the carbon matrix through forming strong chemical affinity with LIPSs. The four distinct peaks in the high-resolution Co2p spectrum can be assigned to metallic Co, Co–N, Co²⁺ and satellites as demonstrated in Fig. 7c. It has been verified that Co–N_x moieties can effectively catalyze reduction and oxidation in Li–S batteries.^41,48–50Fig. 7d possesses two dominating peaks in the S2p spectrum at 163.9 and 165.1 eV, corresponding to the terminal S and bridging S. Moreover, the peak located at 162.6 eV is assigned to the Co–S bonding, further confirming the chemical bonding between polysulfides and Co/N-PCNF.^41,42

Fig. 7 High-resolution XPS spectra of Co/N-PCNF after the adsorption of Li₂S₆: (a) C1s spectrum, (b) N1s spectrum, (c) Co2p_3/2 spectrum and (d) S2p spectrum.

Conclusions

In this work, we developed a freestanding network of Co decorated and N doped porous carbon nanofibers as advanced host materials for sulfur. It is confirmed by a series of experiments that the 3D structure network and high porosity of Co/N-PCNF can effectively trap Li₂S₆ through physical confinement. In particular, the introduction of Co nanoparticles can greatly strengthen the adsorption effect with LIPSs through chemical bonding, but also catalyze and promote the redox kinetics of LIPSs. Furthermore, the Co not only shows a strong interaction to LIPSs but also promoted redox kinetics of LIPSs. The results suggest that Co/N-PCNF can confine and catalyze the soluble LIPSs, leading to enhanced cycling stability of Li–S batteries. The Co/N-PCNF@S cathode can deliver a high areal capacity of 7.16 mA h cm⁻² at 0.2C for 100 cycles even when the sulfur loading is increased to 9.33 mg cm⁻². The rational integration paves the way for high energy density Li–S batteries with high sulfur loading.

Experimental section

Synthesis of materials

Preparation of metal zeolitic imidazolate frameworks (ZnCo-ZIFs). Nano-sized ZnCo-ZIFs were synthesized in a methanol solution according to previous reports⁵¹ Typically, 2.4 g Zn(NO₃)₂·6H₂O (99%; Macklin) and 0.6 g Co(NO₃)₂·6H₂O (99%; Macklin) were dissolved in 150 mL of methanol (99%; Guanghua Guangdong) to form a clear solution A. 4.0 g 2-methylimidazole (C₄H₆N₂, 98%, Macklin) was dissolved in another 150 mL of methanol to form solution B. Then, solution B was quickly poured into solution A under vigorous stirring. After the reaction, the mixture was aged for 3 h at room temperature and a purple dispersion of colloidal particles was observed. Next, the prepared particles were collected via centrifugation and washed with ethanol 3 times. After drying at 60 °C in a vacuum, the finely dispersed ZnCo-ZIFs nanoparticles were obtained. For comparison, Nano-sized ZIF8 nanoparticles were also prepared using the same method by reacting only Zn(NO₃)₂·6H₂O.

Preparation of Co/N-PCNF and N-PCNF. First, the ZnCo-ZIFs@PAN nanofibers were prepared with the electrospinning technique. Briefly, 0.48 g ZnCo-ZIFs and 0.48 g polyacrylonitrile (PAN, average MW 150 000, Sigma-Aldrich) were added into 6 mL of N,N-dimethylformamide (DMF), followed by magnetically stirring overnight at room temperature then sonication for 1 h. Then, the homogeneous liquid mixture was loaded into a syringe with a steel needle and electrospun onto a steel roller receiver wrapped with a flow rate of 0.3 mL h⁻¹. The distance between the needle and the receiver is 15 cm and a working voltage of 12 kV was provided with a high voltage DC power supply. The as-woven ZnCo-ZIFs@PAN nanofibers were pre-oxidized in air in a quartz tube heated to 220 °C at the rate of 1 °C min⁻¹ and kept at that temperature for 2 h. Then, these pre-oxidation products were heated to 800 °C at a heating rate of 2 °C min⁻¹ and calcined for 4 h under an Argon atmosphere. For comparison, plain N-PCNF was also synthesized from a similar process employing ZIF8 as the precursor.

Preparation of Co/N-PCNF@S and N-PCNF@S. Sulfur powder was slowly added to liquid carbon disulfide (CS₂, Macklin) to form an S/CS₂ solution of 50 mg mL⁻¹. For facilitated sulfur loading, the Co/N-PCNF film was first cut into circular disks with a diameter of 12 mm. Quantitative sulfur was added to CS₂ to form S/CS2 solution (50 mg mL⁻¹). Then the S/CS₂ solution was cast-dropped on the disks and excess CS₂ was evaporated at 30 °C. The disks loaded with sulfur were further sealed in a glass bottle and kept at 155 °C for 12 h to ensure adequate and uniform diffusion of sulfur into the structural pores. The sulfur mass content was about 65 wt% and the areal sulfur loading in the Co/N-PCNF@S composite disks was set at 2.0–3.0 mg cm⁻². The N-PCNF@S composite disks were prepared using the same procedure.

Electrochemical measurements

CR2032-type coin cells were assembled in an M Braun glove box (MB-10-G-V2A, H₂O and O₂ < 0.5 ppm) filled with argon. The Co/N-PCNF@S or N-PCNF@S disks were used as cathodes directly. The anodes were lithium chips with a thickness and diameter of 0.45 mm and 16 mm, respectively. The electrolyte consisted of LiTFSI (1 mol L⁻¹) dissolved in DME/DOL (1 : 1 vol) with LiNO₃ (2 wt%) additive, and the ratio of electrolyte to sulfur was about 20 μL mg⁻¹. The separator was a Celgard 2400 polypropylene membrane. A NEWARE BTS battery charger system (Shenzhen, China) was used to test the cells at a voltage range of 2.8–1.7 V. For the coin cells of symmetric electrochemical measurements, 50 μL of LiTFSI [1 mol L⁻¹ in DME/DOL (1 : 1 vol)] and Li₂S₆ [0.2 mol L⁻¹ in DME/DOL (1 : 1 vol)] in a 1 : 1 (v/v) ratio was used as the electrolyte and the Co/N-PCNF or N-PCNF disks were used as both the working and counter electrodes. Cyclic voltammetry (CV) were performed with a CHI660D electrochemical workstation (CH Instruments Co., Ltd, Shanghai, China) at the scanning rate of 0.1 mV s⁻¹. The electrochemical impedance spectroscopy³¹ was measured with the same electrochemical workstation for a frequency range of 100 kHz to 0.1 Hz.

Characterization of materials

The structural information was measured by X-ray diffraction (XRD, Rikagu Ultima IV, Cu Kα radiation, 40 kV, 40 mA). The morphology and microstructure of the samples were collected by field-emission scanning electron microscopy (FESEM, SUPRA-55, ZEISS, Germany) and field emission transmission electron microscopy (FETEM, TECNAI-F30, Philips-FEI, Netherlands). The high-resolution transmission electron microscopy (HRTEM) and elemental mapping were carried out with a F30 at a working voltage of 200 kV. Raman spectra were recorded using a micro-Raman 2000 system (Renishaw, Britain). The specific surface area measurements were performed with 3H-2000PM2 (BeiShiDe, China) equipment with N₂ as the adsorption–desorption gas. And thermogravimetric analysis (TGA, SDTQ600) was applied to analyse the sulfur contents in the composites within a temperature range of 25–700 °C with a heating rate of 10 °C min⁻¹ in nitrogen. X-ray photoelectron spectroscopy (XPS) analysis was carried out on a PHI QUANTUM 2000 (monochromatic Al K X-ray source).

First-principles calculations

The calculations were performed via the density functional theory (DFT) method by using the Vienna ab initio simulation package (VASP).^52,53 To describe electron–ion interactions, we employed the projector augmented wave (PAW) pseudopotentials⁵⁴ and used the gradient-corrected Perdew–Burke–Ernzerhof (GGA-PBE) functional to calculate electron exchange – correlation effects.⁵⁵ A supercell containing 6 × 6 unit graphene cells with a vacuum spacing of 16 Å was used to mimic the N-graphene and Co@N-graphene. A cutoff energy of 400 eV for the plane-wave basis set was used; moreover, the total energy convergence and the force convergence for geometric optimization were set to be 1 × 10⁻⁵ eV and 0.02 eV Å⁻¹, respectively. The Brillouin zone was sampled by k-point grids of 5 × 5 × 1 for geometric optimization. The adsorption energy (E_ad) of Li₂S₆ on the substrate is defined as:

E_ad = E_Li2S6+Sub − E_Li2S6 − E_Sub

where E_Li2S6+Sub is the total energy of Li₂S₆ adsorbed on the substrate, the E_Li2S6 and E_Sub are the total energy of isolated Li₂S₆ and substrate, respectively. The vdW-DF functional was adopted to describe the van der Waals interactions.⁵⁶

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work was financially supported by the National Natural Science Foundation of China (Grant No. 51901031 and 51931006), NSF of Fujian Province, China (No. 2017J01103) and the National Key R&D Program of China (No. 2016YFA0202602). This work was also supported by Fundamental Research Funds for the Central Universities of China (Xiamen University: No. 20720190013), the Guangdong Basic and Applied Basic Research Foundation (No. 2019A1515011070) and the “Double-First Class” Foundation of Materials and Intelligent Manufacturing Discipline of Xiamen University. The State Key Laboratory for Physical Chemistry of Solid Surfaces at the Xiamen University is acknowledged for the supercomputer resources.

Notes and references

1. M. Armand and J.-M. Tarascon, Nature, 2008, 451, 652.
2. Z. W. Seh, Y. Sun, Q. Zhang and Y. Cui, Chem. Soc. Rev., 2016, 45, 5605–5634.
3. Y. X. Yin, S. Xin, Y. G. Guo and L. J. Wan, Angew. Chem., Int. Ed., 2013, 52, 13186–13200.
4. P. G. Bruce, S. A. Freunberger, L. J. Hardwick and J. M. Tarascon, Nat. Mater., 2011, 11, 19–29.
5. Q. Pang, X. Liang, C. Y. Kwok and L. F. Nazar, Nat. Energy, 2016, 1, 16132.
6. J. Zhang, H. Huang, J. Bae, S.-H. Chung, W. Zhang, A. Manthiram and G. Yu, Small Methods, 2018, 2, 1700279.
7. L. Qie and A. Manthiram, Adv. Mater., 2015, 27, 1694–1700.
8. Z.-W. Zhang, H.-J. Peng, M. Zhao and J.-Q. Huang, Adv. Funct. Mater., 2018, 28, 1707536.
9. J. He, Y. Chen, W. Lv, K. Wen, C. Xu, W. Zhang, Y. Li, W. Qin and W. He, ACS Nano, 2016, 10, 10981–10987.
10. J. Zhou, X. Liu, L. Zhu, J. Zhou, Y. Guan, L. Chen, S. Niu, J. Cai, D. Sun, Y. Zhu, J. Du, G. Wang and Y. Qian, Joule, 2018, 2, 2681–2693.
11. M. Wild, L. O'Neill, T. Zhang, R. Purkayastha, G. Minton, M. Marinescu and G. J. Offer, Energy Environ. Sci., 2015, 8, 3477–3494.
12. F. Wu, H. Lv, S. Chen, S. Lorger, V. Srot, M. Oschatz, P. A. van Aken, X. Wu, J. Maier and Y. Yu, Adv. Funct. Mater., 2019, 29, 1902820.
13. S. Evers, T. Yim and L. F. Nazar, J. Phys. Chem. C, 2012, 116, 19653–19658.
14. L. Zhang, X. Chen, F. Wan, Z. Niu, Y. Wang, Q. Zhang and J. Chen, ACS Nano, 2018, 12, 9578–9586.
15. X. Ji, K. T. Lee and L. F. Nazar, Nat. Mater., 2009, 8, 500–506.
16. F. Pei, T. An, J. Zang, X. Zhao, X. Fang, M. Zheng, Q. Dong and N. Zheng, Adv. Energy Mater., 2016, 6, 1502539.
17. F. Wu, E. Zhao, D. Gordon, Y. Xiao, C. Hu and G. Yushin, Adv. Mater., 2016, 28, 6365–6371.
18. Z. Qiao, F. Zhou, Q. Zhang, F. Pei, H. Zheng, W. Xu, P. Liu, Y. Ma, Q. Xie, L. Wang, X. Fang and D.-L. Peng, Energy Storage Mater., 2019, 23, 62–71.
19. G. Li, X. Wang, M. H. Seo, M. Li, L. Ma, Y. Yuan, T. Wu, A. Yu, S. Wang, J. Lu and Z. Chen, Nat. Commun., 2018, 9, 705.
20. Y. Tao, Y. Wei, Y. Liu, J. Wang, W. Qiao, L. Ling and D. Long, Energy Environ. Sci., 2016, 9, 3230–3239.
21. H. Al Salem, G. Babu, C. V. Rao and L. M. Arava, J. Am. Chem. Soc., 2015, 137, 11542–11545.
22. N. Xu, T. Qian, X. Liu, J. Liu, Y. Chen and C. Yan, Nano Lett., 2017, 17, 538–543.
23. C. Zhang, J. J. Biendicho, T. Zhang, R. Du, J. Li, X. Yang, J. Arbiol, Y. Zhou, J. R. Morante and A. Cabot, Adv. Funct. Mater., 2019, 29, 1903842.
24. Y. Li, J. Fan, J. Zhang, J. Yang, R. Yuan, J. Chang, M. Zheng and Q. Dong, ACS Nano, 2017, 11, 11417–11424.
25. X. Lu, Q. Zhang, J. Wang, S. Chen, J. Ge, Z. Liu, L. Wang, H. Ding, D. Gong, H. Yang, X. Yu, J. Zhu and B. Lu, Chem. Eng. J., 2019, 358, 955–961.
26. Y. Wu, X. Zhu, P. Li, T. Zhang, M. Li, J. Deng, Y. Huang, P. Ding, S. Wang, R. Zhang, J. Lu, G. Lu, Y. Li and Y. Li, Nano Energy, 2019, 59, 636–643.
27. T. Chen, L. Ma, B. Cheng, R. Chen, Y. Hu, G. Zhu, Y. Wang, J. Liang, Z. Tie, J. Liu and Z. Jin, Nano Energy, 2017, 38, 239–248.
28. W. Chen, T. Qian, J. Xiong, N. Xu, X. Liu, J. Liu, J. Zhou, X. Shen, T. Yang, Y. Chen and C. Yan, Adv. Mater., 2017, 29, 1605160.
29. H. Li, S. Ma, H. Cai, H. Zhou, Z. Huang, Z. Hou, J. Wu, W. Yang, H. Yi, C. Fu and Y. Kuang, Energy Storage Mater., 2019, 18, 338–348.
30. J. Zhang, C. You, W. Zhang, J. Wang, S. Guo, R. Yang and Y. Xu, Electrochim. Acta, 2017, 250, 159–166.
31. X. Liang, C. Hart, Q. Pang, A. Garsuch, T. Weiss and L. F. Nazar, Nat. Commun., 2015, 6, 5682.
32. W. G. Lim, S. Kim, C. Jo and J. Lee, Angew. Chem., Int. Ed., 2019, 58, 18746–18757.
33. H. Lin, S. Zhang, T. Zhang, H. Ye, Q. Yao, G. W. Zheng and J. Y. Lee, Adv. Energy Mater., 2018, 8, 1801868.
34. H. Lin, L. Yang, X. Jiang, G. Li, T. Zhang, Q. Yao, G. W. Zheng and J. Y. Lee, Energy Environ. Sci., 2017, 10, 1476–1486.
35. L. Qie, C. Zu and A. Manthiram, Adv. Energy Mater., 2016, 6, 1502459.
36. W. Xue, Z. Shi, L. Suo, C. Wang, Z. Wang, H. Wang, K. P. So, A. Maurano, D. Yu, Y. Chen, L. Qie, Z. Zhu, G. Xu, J. Kong and J. Li, Nat. Energy, 2019, 4, 374–382.
37. S.-K. Park, J.-S. Park and Y. C. Kang, J. Mater. Chem. A, 2018, 6, 1028–1036.
38. S. Liu, J. Li, X. Yan, Q. Su, Y. Lu, J. Qiu, Z. Wang, X. Lin, J. Huang, R. Liu, B. Zheng, L. Chen, R. Fu and D. Wu, Adv. Mater., 2018, 30, e1706895.
39. Q. Bai, F. C. Shen, S. L. Li, J. Liu, L. Z. Dong, Z. M. Wang and Y. Q. Lan, Small Methods, 2018, 2, 1800049.
40. S. Wang, F. Gao, R. Ma, A. Du, T. Tan, M. Du, X. Zhao, Y. Fan and M. Wen, Metals, 2018, 8, 755.
41. D. Fang, Y. Wang, C. Qian, X. Liu, X. Wang, S. Chen and S. Zhang, Adv. Funct. Mater., 2019, 29, 1900875.
42. W. Cai, G. Li, D. Luo, G. Xiao, S. Zhu, Y. Zhao, Z. Chen, Y. Zhu and Y. Qian, Adv. Energy Mater., 2018, 8, 1802561.
43. J. Meng, C. Niu, L. Xu, J. Li, X. Liu, X. Wang, Y. Wu, X. Xu, W. Chen, Q. Li, Z. Zhu, D. Zhao and L. Mai, J. Am. Chem. Soc., 2017, 139, 8212–8221.
44. R. Liu, Z. Liu, W. Liu, Y. Liu, X. Lin, Y. Li, P. Li, Z. Huang, X. Feng, L. Yu, D. Wang, Y. Ma and W. Huang, Small, 2019, 15, 1804533.
45. Y.-T. Liu, D.-D. Han, L. Wang, G.-R. Li, S. Liu and X.-P. Gao, Adv. Energy Mater., 2019, 9, 1803477.
46. G. Zhou, H. Tian, Y. Jin, X. Tao, B. Liu, R. Zhang, Z. W. Seh, D. Zhuo, Y. Liu and J. Sun, Proc. Natl. Acad. Sci. U. S. A., 2017, 114, 840.
47. J. Xu, W. Zhang, Y. Chen, H. Fan, D. Su and G. Wang, J. Mater. Chem. A, 2018, 6, 2797–2807.
48. S.-K. Park, J.-K. Lee and Y. C. Kang, Adv. Funct. Mater., 2018, 28, 1705264.
49. Y.-J. Li, J.-M. Fan, M.-S. Zheng and Q.-F. Dong, Energy Environ. Sci., 2016, 9, 1998–2004.
50. P. Wang, Z. Zhang, X. Yan, M. Xu, Y. Chen, J. Li, J. Li, K. Zhang and Y. Lai, J. Mater. Chem. A, 2018, 6, 14178–14187.
51. W. Zhang, X. Jiang, X. Wang, Y. V. Kaneti, Y. Chen, J. Liu, J. S. Jiang, Y. Yamauchi and M. Hu, Angew. Chem., Int. Ed., 2017, 56, 8435–8440.
52. G. Kresse and J. Furthmüller, Phys. Rev. B: Condens. Matter Mater. Phys., 1996, 54, 11169.
53. G. Kresse and J. Furthmüller, Comput. Mater. Sci., 1996, 6, 15–50.
54. P. E. Blöchl, Phys. Rev. B: Condens. Matter Mater. Phys., 1994, 50, 17953.
55. J. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett., 1996, 77, 3865–3868.
56. M. Dion, H. Rydberg, E. Schröder, D. C. Langreth and B. I. Lundqvist, Phys. Rev. Lett., 2004, 92, 246401.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c9nh00663j

---