1 Introduction

The continuously increasing demands for sustainable energy and severe environmental crisis have boosted the development of various advanced energy technologies around the world, with the purpose of efficient utilization and storage of renewable energy [1, 2]. High energy density and economical rechargeable batteries are the key components of these advanced energy technologies [3,4,5]. Operated based on lithium ion (Li-ion) intercalation chemistry, Li-ion batteries have enjoyed great success in powering commercial portable electronics and electric vehicles [6]. However, the limited capacity of electrode materials and their high cost hinder the penetration of traditional Li-ion batteries in large-scale emerging fields. Therefore, it is increasingly important to develop electrochemical energy storage devices with higher energy density and lower cost [7,8,9].

Lithium–sulfur (Li–S) batteries are considered one of the most promising energy storage systems beyond Li-ion batteries due to their high energy density and low cost [10]. Typically, Li–S batteries consist of elemental sulfur (S8) cathodes and Li anodes, as shown in Fig. 1a. Based on the multi-electron conversion mechanism between S8 and Li metals (S8 + 16Li ↔ 8Li2S) [11, 12], Li–S batteries deliver high theoretical specific capacity of 1675 mAh g−1 and specific energy of 2,600 Wh kg−1, which is 2–5 times that of Li-ion batteries [13]. The widely accepted reaction mechanism of Li–S batteries is shown in Fig. 1c. During the discharge process, solid S8 is firstly reduced to soluble lithium polysulfides (LiPSs, usually denoted as Li2Sn, 2 < n ≤ 8) in a first discharge plateau at around 2.35 V and then continues to be reduced to solid lithium sulfide (Li2S) in a second discharge plateau at around 2.1 V. Due to the involved solid–solid conversion between Li2S2 and Li2S, the corresponding reaction kinetics performs sluggish. During the subsequent charge process, Li2S is reconverted to LiPSs and finally to S8, forming a reversible cycle [44,45,46].

In this review, we present recent advances in CxNy-based materials applied in Li–S batteries, including the optimized g-C3N4, g-C3N4-based composite materials, and other novel CxNy materials. We systematically summarized their synthetic methods, structures, properties, and effects on Li–S batteries, with a focus on the structure–activity relationship. Based on an extensive analysis of literature, we identified the limitations of existing CxNy-based materials and provided our perspective on the rational design of advanced CxNy-based materials for high-performance Li–S batteries.

2 Basics of Representative Carbon Nitride: g-C3N4

CxNy material, represented by g-C3N4, was firstly reported in the nineteenth century [37]. Due to its unique structure and properties, g-C3N4 has been widely applied in various fields ever since, such as photocatalysis [47], carbon dioxide capture [43] and pyrolysis temperature [61] on the properties of g-C3N4. Among the g-C3N4 materials prepared with different precursors, including urea, melamine, thiourea, and dicyandiamide, urea-based g-C3N4 shows the highest specific surface area (~ 93 m2 g−1). For the pyrolysis temperature, g-C3N4 synthesized at 550 °C shows the highest specific surface area with a rich mesoporous lamellar structure. These results were later reconfirmed by Versaci et al. [44]. In addition, they further proved that g-C3N4 prepared at 550 °C with urea had a high content of -NH2 group, which was conducive to the immobilization of soluble LiPSs.

Based on the direct condensation mechanism, the template method is introduced to fabricate g-C3N4 with 3D secondary structures such as hollow or core–shell structures, which could not only provide high surface area but also accommodate the volume change during cycling. The template and reaction temperature selected are important, which affect the structure and properties of the product. Silica is a common template applied. A hollow g-C3N4 material was prepared using mesoporous silica as a template and further constructed into an S@C3N4 composite cathode with a core–shell structure [62]. In addition, Han et al. used silica microspheres as a hard template and investigated the effect of calcination temperature on the structure of the synthesized hollow g-C3N4 microspheres [45]. With calcination temperatures set as 600, 700, and 800 °C, the synthesis procedure is shown in Fig. 3e. According to the SEM images of as-prepared g-C3N4 (Fig. 3f-h), the thickness of the shell decreases, and the microsphere structure collapses as the temperature increases from 600 to 800 °C. This could be related to the excessively decompose of the precursors at high temperatures. As a result, the cell with g-C3N4 prepared in 600 °C as the S host exhibited a low capacity fading rate of 0.076% per cycle after 500 cycles at 0.5C.

Generally, g-C3N4 is applied as an additive to the S cathode. Li et al. fabricated composite cathodes with g-C3N4 and S, which exhibited a high capacity of 1200 mAh g−1 at 0.2C and maintained a high capacity of 800 mAh g−1 after 100 cycles with the coulombic efficiency above 99.5% [60]. Yet some studies also use g-C3N4 to construct multifunctional layers on the cathode or separator to limit the diffusion of LiPSs. Li et al. coated a layer of g-C3N4 nanosheets on the surface of the S cathode (S-C3N4) by the spraying method [63]. This unique design has the following advantages: (1) the g-C3N4 layer has a strong chemical adsorption capability for LiPSs, which can limit LiPSs shuttling and alleviate the self-discharge phenomenon; (2) spraying technology ensures the uniformity of the coating, and it is easy to large-scale production with the controlled thickness. Therefore, the cell with an S-C3N4 composite cathode displayed a high capacity of 630 mAh g−1 at 5C. Similarly, **e et al. coated ultra-thin g-C3N4 nanosheets on the commercial polypropylene (PP) separator (g-C3N4 separator) by using the vacuum filtration technology [64], which effectively prevents LiPSs from diffusing across the separator but allows lithium ions to pass freely (Fig. 3i-j). Moreover, the LiPSs permeation test also showed the strong restriction effect of g-C3N4 for LiPSs diffusion (Fig. 3k-l). Thus, the cell with a g-C3N4 separator performed a high capacity of 829 mAh g−1 after 200 cycles at 0.2C.

3 Optimization of g-C3N4

With various synthesis methods, g-C3N4 could perform different microstructures with enhanced specific surface area. Beyond this, the LiPSs absorption capability, catalytic activity, and electron conductivity of g-C3N4 could be further improved via defect engineering and heteroatom do**. Defect engineering plays an important role in adjusting the atomic distribution and surface properties of nanomaterials and has widespread application in various fields including hydrogen evolution reaction [65], oxygen evolution reaction [66], and carbon dioxide reduction reaction [67]. Heteroatom do** is also an effective method to regulate the polarity of carbon materials, and various heteroatoms including nonmetal atoms and metal-single atoms have been studied extensively [68,69,70]. In particular, the introduction of metal single atoms with unsaturated coordination environments, unique electronic structures, and high surface free energy could significantly enhance the catalytic activity of the materials [71,72,73]. In recent years, defect engineering and heteroatom do** have attracted more and more attention in Li–S systems due to their significant potential in inhibiting LiPSs shuttling and promoting the redox chemistry [74,75,76,77].

3.1 Defect Engineering

With a certain proportion of N defects, g-C3N4 materials show enhanced adsorption and catalytic performance of LiPSs. Huang et al. prepared ultrafine spindle g-C3N4 (sCN) with N defects by K treatment (Fig. 4a) [58]. Compared with the original g-C3N4, the sCN performs spindle-like morphology (Fig. 4b) and an obvious different molecular structure with a large number of defects manifested as N vacancies or cyano groups (Fig. 4c-d). The introduction of N defects increases the polarity of sCN, which leads to 2–3 times increased LiPSs binding energy compared with that of the original g-C3N4. Therefore, the Li–S cell with sCN modified separator delivered a high initial capacity of 637 mAh g−1 at 5C and a low capacity fading rate of 0.05% per cycle after 500 cycles. Besides, various g-C3N4 materials with different defect structures, concentrations, and preparation methods have been reported, which obviously improve the performance of Li–S batteries [78, 79]. However, excessive N defects could destroy the structure of g-C3N4 and thus decrease its electron transport and LiPSs adsorption capability. According to Du et al. [46] (Fig. 4e), as the N content decreases from the original 60% (GCN-60%N) to 6% (GCN-6%N), the content of defect increases, which leads to an increased LiPSs adsorption capability of the material. It is worth noting that when the nitrogen content continues to drop below 6%, the adsorption capability of the material (GCN-2%N) for LiPSs begins to decrease. This could be related to the destruction of the material structure, which is also reflected in the performance of Li–S batteries. The cell with optimized GCN-6%N/S composite cathode displayed a high initial capacity of 852.2 mAh g−1 at 0.5C and retained a reversible capacity of 532.4 mAh g−1 after 300 cycles (Fig. 4f).

Fig. 4
figure 4

Copyright 2021, Wiley–VCH. e UV–Vis spectra of a bare Li2S8 solution and Li2S8 solutions with different GCN materials after aging for 24 h; the inset is the corresponding optical photograph. f Cycling performance of cells with different GCN/S cathodes at 0.5C [46]. Copyright 2020, American Chemical Society

a Schematic illustration of the preparation process of sCN/PDA/PP separator. b TEM images of sCN. c Adsorption energies for Li2S, Li2S2, Li2S4, Li2S6 and Li2S8 on pristine CN, the cyano group of sCN and the N vacancy of sCN. d The stable adsorption models of Li2S8 or Li2S4 on the cyano group of sCN and the N vacancy of sCN [58].

Full size image

3.2 Heteroatoms Do**

Heteroatom-doped g-C3N4, including nonmetal atom- and metal single-atom-doped materials, are reported to be applied in Li–S batteries with enhanced cycling performance and distinct working mechanisms.

Nonmetal atom do**, such as N, S, O, P, and B, enhances the electronic conductivity and the LiPSs absorption capability of g-C3N4. Liu et al. [80] prepared O-doped g-C3N4 nanosheets (OCN) by one-step self-supported solid-state pyrolysis (OSSP) technique with urea as the precursor and glucose as the oxygen source (Fig. 5a). The introduction of O atoms into g-C3N4 promotes the chemical interactions with LiPSs by forming Li–O bonds. Thus, the cell with OCN/S composite cathode performed a high capacity of 447.3 mAh g−1 after 500 cycles at 0.5C with the capacity fading rate of 0.1% per cycle (Fig. 5b). Zhang et al. prepared P-doped g-C3N4 (PCN), which was used as the S host to enhance the performance of Li–S batteries [83]. According to density functional theory (DFT) calculation results [83, 84], both OCN and PCN have higher conductivity and stronger adsorption capability for LiPSs compared with original g-C3N4, which is conducive to improving the S utilization efficiency. In addition to O and P, B-doped g-C3N4 nanosheets (BCN) were prepared by a one-pot thermal condensation method and used as functional separator coating for Li–S batteries [81]. As shown in Fig. 5c, in the heat treatment process, the g-C3N4 bulk was exfoliated to g-C3N4 nanosheets due to the blowing erosion caused by the decomposition of ammonium chloride. At the same time, B atoms were successfully doped into g-C3N4 matrix with N-B-N bonds. The TEM images in Fig. 5d show that BCN performs the wrinkled and irregular lamellar structure. The low polarization overpotential and high capacity of Li–S cells with BCN-coated separator at 0.5C (Fig. 5e) suggest the improvement of S utilization efficiency and redox kinetic.

Fig. 5
figure 5

Copyright 2015, American Chemical Society. c Schematic illustration of the preparation process and structure of BCN. d STEM image of BCN. e Charge–discharge voltage profiles of cells with different modified separators [81]. Copyright 2020, Elsevier. f Adsorption energies for Li2S4, Li2S6 and Li2S8 on g-C3N4, B-g-C3N4, O-g-C3N4, P-g-C3N4 and S-g-C3N4. g Differential charge densities of Li2S4 adsorbed on g-C3N4, B-g-C3N4, O-g-C3N4, P-g-C3N4 and S-g-C3N4 [82]. Copyright 2021, Elsevier

a Schematic illustration of the preparation process and structure of OCN. b Cycling performance of cells with g-C3N4 and OCN-900 as sulfur hosts at 0.5 C [80].

Full size image

Yanmsang et al. investigated the adsorption capabilities and mechanisms of g-C3N4-doped with different heteroatoms (B, O, P and S) for LiPSs on the molecular level by DFT calculations [82]. As shown in Fig. 5f-g, the B-doped g-C3N4 (B-g-C3N4) shows the strongest adsorption capability for LiPSs among investigated g-C3N4 materials. The result can be attributed to the lower capability of B atoms to attract electrons than C atoms, resulting in more negative charge accumulation around pyridinic N atoms and thus facilitating charge transfer between g-C3N4 and LiPSs.

Metal single-atom do**, such as Fe, Co, and Ni, also improves the adsorption and electrical conductivity of g-C3N4 [85,86,87]. In addition, the metal atoms as electropositive active sites could directly interact with LiPSs, which largely improves the catalytic activity of doped g-C3N4 for LiPSs redox reactions [86].

Fe atom-doped g-C3N4 (Fe-N2/CN) material with a hierarchical porous lamellar structure was successfully prepared by Qiu et al. (Fig. 6a-b) [85]. The uniform pyridinic N sites of g-C3N4 control the coordination structure of Fe-NC. As shown in Fig. 6c, a large number of independent Fe atoms with the size of about 2 Å are evenly distributed in obtained g-C3N4. According to the X-ray absorption near-side structure spectra (XANES) and Fourier transform of Fe K-edge extended X-ray absorption fine structure (EXAFS) spectra of Fe-N2/CN (Fig. 6d-g), Fe atoms on g-C3N4 are positively charged and coordinated with two N atoms through N–Fe–N bond. These Fe-N2 unsaturated sites show not only stronger LiPSs adsorption capability with higher binding energy (Fig. 6h-i) but also higher catalytic activity for Li2S decomposition with a lower energy barrier (Fig. 6j). Therefore, the cell with Fe-N2/CN@S composite cathode exhibited a low capacity fading rate of only 0.011% per cycle after 2000 cycles at 2C (Fig. 6k). Co atom-doped g-C3N4 (Co@C3N4) material was also reported with a similar working mechanism in Li–S batteries [87]. The formation of Co-S bonds effectively immobilizes LiPSs.

Fig. 6
figure 6

Copyright 2020, American Chemical Society. l Differential charge densities of Li2S6 adsorbed on C3N4, Fe-C3N4, Co-C3N4 and Ni-C3N4. m CV curves of symmetric cells with g-C3N4/C, Fe-C3N4/C, Co-C3N4/C and Ni-C3N4/C modified separators. n Decomposition barriers of Li2S on C3N4, Fe-C3N4, Co-C3N4 and Ni-C3N4. o Cycling performance of cells with g-C3N4/C, Fe-C3N4/C, Co-C3N4/C and Ni-C3N4/C modified separators at 0.5 A g−1 [86]. Copyright 2019, Elsevier

a TEM image and b corresponding EDS map**s and c HAADF-STEM image of Fe-N2/CN. d Fe K-edge XANES spectra and e Fourier transformation of Fe K-edge EXAFS spectra of Fe-N2/CN, Fe foil, and Fe2O3. f Fe k-space EXAFS curve and corresponding fitting curve, and g Fe r-space EXAFS curve and corresponding fitting curve of Fe-N2/CN. h Optimized structures and binding energies of Li2S4 adsorbed on Fe-N2/CN and CN surfaces. i UV–vis spectra of Li2S4 solution with CN and Fe-N2/CN. j Decomposition energy barriers of Li2S on Fe-N2/CN and CN surfaces. k Cycling performance of cells with CN@S and Fe-N2/CN@S cathodes at 2C [85].

Full size image

Chen et al. compared the adsorption and electrocatalytic capability of single-metal-atom-doped g-C3N4 (M-C3N4, where M = Fe, Co, or Ni) for LiPSs [86]. According to DFT calculation results, the metal atoms’ do** can enhance the conductivity of g-C3N4, among which Fe-C3N4 and Co-C3N4 show semi-metallic properties, while Ni-C3N4 exhibits metallic properties. Moreover, as shown in Fig. 6l, Ni-C3N4 shows the strongest interaction with Li2S6 and leads to the largest response current when applied in Li–S batteries (Fig. 6m). In addition, Li2S decomposition presents the lowest energy barrier on the surface of the Ni-C3N4 substrate (Fig. 6n), suggesting that Ni-C3N4 can promote solid–liquid conversion. Therefore, the cell with a Ni-C3N4/C-modified separator performed a high capacity of 893 mAh g−1 after 300 cycles at 0.5 A g−1 with a capacity retention of 89.4%, showing good cycling stability and high S utilization efficiency (Fig. 6o).

To summarize, defect engineering and heteroatom do** are effective methods to regulate the adsorption capability and catalytic activity of g-C3N4. The performances of Li–S cells with optimized g-C3N4 materials are compared and listed in Table 1. However, excessive defects and heteroatoms could destroy the structure of g-C3N4. An in-depth understanding of the do**- and defect-structure–activity relationship still remains a grand challenge.

Table 1 Comparison of performances of Li–S cells with optimized g-C3N4 materials
Full size table

4 Design of g-C3N4-Based Composites

Although g-C3N4 has made remarkable progress in Li–S systems, it is difficult for pristine g-C3N4 to enable practical performance in Li–S batteries, owing to its intrinsic properties, including poor conductivity and low electrocatalytic activity. To explore novel g-C3N4-based materials with satisfying physical/chemical properties, g-C3N4 has been incorporated with other functional materials, such as conductive carbon materials, metal nanoparticles, and polar compounds. And the final g-C3N4-based composites exhibit various advantages, such as strong LiPSs immobilization, rapid Li-ion transfer, and accelerated conversion of S species.

4.1 Conductive Carbon/g-C3N4 Composites

Various conductive carbon materials were applied in g-C3N4-based composite materials, including carbon nanotubes (CNTs), porous carbon material, graphene, and carbon cloth. The combination of g-C3N4 with conductive carbon materials could effectively improve the electronic conductivity of composite materials, which is critical for its application in Li–S batteries. In addition to this, different carbon materials could also provide various other benefits due to their specific structures and properties. For example, a large specific surface area of carbon materials could suppress LiPSs diffusion.

4.1.1 CNTs Constructed Conductive Networks

With the characteristic 1D structure and excellent electronic conductivity, CNTs could construct a conductive network and achieve fast electron conduction. By the high-temperature-assisted self-assembly method, Wang et al. directly synthesized g-C3N4 on the CNTs (Fig. 7a) [33]. Through hydrogen bonds, cyanic acid and melamine not only construct the triazine structure of g-C3N4 but also connect together with the CNTs. After heating treatment, the supramolecular structure can be further transformed into the final g-C3N4/CNTs composite. The obtained g-C3N4/CNTs composite shows a network structure with uniform coverage of the g-C3N4 layer. With largely improved conductivity and high LiPSs adsorption capability, the Li–S cell with g-C3N4/CNTs/S composite cathode displayed a high-capacity retention of 77.1% after 200 cycles at 1C at a 5 mg cm−2. Chen et al. [89] and Yao et al. [90] prepared g-C3N4/CNTs composite-based membranes and separately applied them as shielding layer and self-supported cathode. Profited from the strong LiPSs adsorption capability of g-C3N4 and good conductivity network of CNTs, the shuttle effect is largely inhibited, and S utilization efficiency is obviously improved.

Fig. 7
figure 7

Copyright 2019, American Chemical Society. b Schematic illustration of the preparation process of NG-CN/CMC-CA/S composite. c Charge–discharge voltage profiles of cells using NG-CN/CMC-CA/S cathodes with different sulfur loadings at 0.5 mA cm−2 [42]. Copyright 2016, Wiley–VCH. d Schematic illustration of the preparation process of S/GCN hybrid sponge. e Optical images of G-sponge (left) and S/GCN hybrid sponge (right). f TEM image of S/GCN. g Cycling performance of 3D S/GCN, GCN-Li2Sn and 2D S-GCN cathodes at 0.5 A g−1 [32]. Copyright 2018, Wiley–VCH. h Schematic illustration of the preparation process of the S@C3N4/C microsphere. TEM images of i PC-C3N4/C and j S@C3N4/C [

  • 1.

    As an inactive material, CxNy should be added at low amounts to increase the overall energy density of the Li–S cell. According to some reported works [35, 45], g-C3N4 added as S host accounts for about 15% of the total cathode weight. The reduction of CxNy content requires an increase in their specific surface area, which means an increased in the adsorption capability, and therefore, a similar function could be realized with a smaller amount of CxNy materials. Different preparation methods, such as the hard template method and strip**-assisted method, can be used to obtain the g-C3N4 with controllable morphology and high specific surface area;

  • 2.

    The electronic conductivity of g-C3N4, which limits the rate performance of Li–S batteries, should be increased. Although improved by numerous reported methods, the conductivity of g-C3N4 is still unsatisfactory. The effect of heteroatom do** and defect treatment on the conductivity of g-C3N4 appears limited; the additional carbon material introduced in the form of g-C3N4/carbon composite increases the amount of inactive materials. Besides these methods, the electronic conductivity of the materials could also be increased by regulating the C/N ratios. Synthesizing novel CxNy materials with higher electron conductivity and studying their adsorption and electrocatalytic effect on LiPSs could be prospective directions in Li–S systems.

  • 3.

    The catalytic mechanism of CxNy-based materials should be systematically studied. Although most of the CxNy-based materials are proposed with a working mechanism, the reaction mechanism of S8 with CxNy-based materials is still unclear. Since the conversion process of LiPSs is quite complex, it is necessary to combine advanced in situ characterization techniques such as cryo-electron microscopy, in situ Raman, and XRD to monitor the electronic structure and morphology changes of intermediates on CxNy-based materials under various conditions in real time. By comparing series of materials in parallel, the structure–activity relationship can be revealed, which is critical for further material design.

  • 4.

    The enhancement of Li anode stability with CxNy-based materials worths more attention. The growth of dendrite limits the cycling life of Li anode. Besides, the soluble LiPSs lead to the passivation of Li anode in Li–S batteries. g-C3N4 has high shear modulus and good affinity with Li ions, and thus, it is promising in promoting the uniform deposition of Li ions and inhibiting the growth of Li dendrites [39, 40]. However, relevant studies are very limited. With more effort on this topic, an enhancement of Li anode stability with CxNy-based material is expected.

    • Looking into the future, there are infinite opportunities and challenges for the vigorous development of CxNy-based materials. With further efforts, it is expected that CxNy-based materials will promote the practical application of high energy density and long-life Li–S batteries.