Nitrogen Doping Improves the Immobilization and Catalytic Effects of Co9S8 in Li‐S Batteries

Several critical issues, such as the shuttling effect and the sluggish reaction kinetics, exist in the design of high‐performance lithium–sulfur (Li‐S) batteries. Here, it is reported that nitrogen doping can simultaneously and significantly improve both the immobilization and catalyzation effects of Co9S8 nanoparticles in Li‐S batteries. Combining the theoretical calculations with experimental investigations, it is revealed that nitrogen atoms can increase the binding energies between LiPSs and Co9S8, and as well as alleviate the sluggish kinetics of Li‐S chemistry in the Li2S6 cathode. The same effects are also observed when adding N‐Co9S8 nanoparticles into the commercial Li2S cathode (which has various intrinsic advantages, but unfortunately a high overpotential). A remarkable improvement in the battery performances in both cases is observed. The work brings heteroatom‐doped Co9S8 to the attention of designing high‐performance Li‐S batteries. A fundamental understanding of the inhibition of LiPSs shuttle and the catalytic effect of Li2S in the newly developed system may encourage more effort along this interesting direction.


Introduction
The increasing demand for energy storage devices with high energy/power density, long cycle life as well as the low material cost has boosted the development of battery technology in recent years. [1] Among the several most successful battery chemistries, Li-S battery has received significant attention due to its high theoretical specific capacity of 1672 mAh g −1 -about five times higher than the currently dominant lithium-ion battery. In addition, sulfur is naturally abundant, inexpensive, and environmentally friendly, making it an ideal choice as the battery electrode. [2][3][4][5] However, Li-S battery faces its own set of challenges. The most well-known issue is the uncontrolled dissolution of intermediate lithium polysulfides (LiPSs) into the electrolyte. This so-called "shuttling effect" results in fast capacity fading and low Coulombic efficiency (CE). [4,6] Another critical issue, which becomes a major focus in recent studies, [7,8] is the sluggish redox kinetics of the insulating Li 2 S during the electrochemical reactions. About 75% of the Li-S battery's theoretical capacity comes from the transformation of soluble Li 2 S 4 to solid Li 2 S. Therefore, the catalysis of the decomposition of Li 2 S and oxidization of Li 2 S to Li 2 S x , and finally to sulfur are also crucial steps to realizing high capacity and CE. [7] Therefore, innovative strategies need to be developed, which can take both issues into account.
Tremendous efforts have been devoted to improving the immobilization of polysulfide in Li-S batteries, mainly by physically constraining the sulfur within nanostructured carbon materials, [9][10][11][12][13] or adding carbon interlayer for inhibiting the LiPSs dissolution in the electrolyte. [14] But, the poor adsorption of the nonpolar carbon-based materials toward the LiPSs resulted in limited success. [11,15] There are also research works that use the smaller sulfur molecules (S 2-4 ) in Li-S batteries to avoid the shuttle problem, [16,17] however, the charge/discharge plateaus are relatively low in these works.
Recently, seminal results were reported on the introduction of heteroatoms into carbonaceous materials for the generation of polar functional groups to immobilize the LiPSs. [18][19][20][21] Besides, a wide variety of other materials (beyond carbon-based materials) also show great promise in trapping LiPSs due to the similar polar interactions with LiPSs or the Lewis acid-base interactions. For instance, metal oxide, [22,23] metal sulfides, [24][25][26] metal nitrides/carbides, [27,28] and indium tin oxide. [29] Very recently, single atoms are also reported as efficient catalysts for Li-S batteries. [8,30] Transition metal dichalcogenides (TMD), [7,[31][32][33] on the other hand, are interesting in Li-S batteries mainly due to that the metal d-orbitals and unsaturated heteroatoms (such as S) result in an effective d-band structure with catalytic characteristics. Their catalytic activities for the LiPSs are correlated to the exposed edge sites. [31,34] Although they can improve the Li−S battery performance to a certain extent (when compared with pure S or Li 2 S 6 ), the results are far from satisfactory.
In this work, we report that nitrogen (N) doping can improve, simultaneously, both the LiPSs immobilization and the redox catalyzation capabilities of Co 9 S 8 nanoparticles in Li-S batteries. This is because, on the one hand, N doping is an efficient approach to optimize the electrochemical performances in Li-S batteries, by reducing the intrinsic activation barriers during the catalytic reactions. [35][36][37] And on the other hand, Li-N bonds are more favorable than Li-S bonds when anchoring LiPSs in Li-S batteries. [38,39] Density functional theory (DFT) calculations provide an insight into the strong chemical bonding between the LiPSs and the surface of N-Co 9 S 8 nanoparticles. Further supported by experiments, we conclude that N atoms significantly increase the binding energies and hence help to prevent the LiPSs shuttling.
When compared with the undoped Co 9 S 8 (in Li 2 S 6 cathodes), the N-doped Co 9 S 8 shows significant improvements in the battery performances. The polarization is lower, the capacity is improved (1233 mAh g −1 at 0.2 A g −1 ), the redox reaction is faster (604 mAh g −1 at 20 A g −1 ), the capacity retention is better (decay of 0.037%/cycle over 1000 cycles), and the sulfur utilization is higher. The results at high mass loading (up to 5 mg cm −2 ) and high current densities (up to 20 A g −1 ) are also excellent.
Furthermore, N-Co 9 S 8 nanoparticles are also tested as an additive to the commercial Li 2 S in cathodes. We observe a much reduced potential barrier and a much larger specific capacity. This suggests a viable route to construct a Li 2 Sbased Li-S battery that has several intrinsic advantages. [40] For example, since the lithium is stored in Li 2 S, the lithium metal anode can be replaced by a high capacity silicon anode to avoid the dendrite problem and the low CE issue. Also, as the least-dense phase of the sulfur-containing species, no volume expansion is expected for the Li 2 S-based cathodes during the cell operations. Therefore, our work on the novel N-doped Co 9 S 8 material may open a new avenue for designing practical Li-S batteries and encourage more efforts along this interesting direction.

N-Co 9 S 8 Structural and Morphology Analysis
N-Co 9 S 8 nanoparticles were synthesized through the hydrothermal reactions (see Experimental Section). The as-fabricated N-Co 9 S 8 nanoparticles were first investigated by the X-ray diffraction (XRD) to determine its crystalline structure, and all the diffraction peaks are assignable to the face-centered cubic structure of Co 9 S 8 (JCPDS card number: 01-073-6395) without any noticeable impurity phase (Figure 1a). The detailed morphology, structure, and composition of the products were studied by the scanning electron microscopy (SEM) and the transmission electron microscopy (TEM). As shown in Figure 1b,c, all N-Co 9 S 8 nanoparticles show a well-defined spherical nanostructure with a diameter of around 40 nm, and therefore have large surface areas and abundant adsorption/catalytic active sites. The N-Co 9 S 8 nanoparticles, both after being coating on carbon paper and after the addition of Li 2 S 6 , show a uniform distribution, as shown in Figure S1, Supporting Information. The energy-dispersive X-ray spectrum (EDXS) elemental mappings (Figure 1d-g) show that the spatial distributions of the N, Co and S atoms are highly correlated over the entire nanoparticles.
X-ray photoelectron spectroscopy (XPS) measurements were performed in order to confirm the successful N-doping of the Co 9 S 8 nanoparticles (Figure 1h-j). The Co 2p signal consists of a Co 2p 3/2 and a Co 2p 1/2 component, each of which can be further deconvoluted into the Co 3+ (776.8 and 791.9 eV), Co 2+ (779.4 and 794.9 eV), and the satellite (784.1 and 801.0 eV) peaks. [36,41,42] The satellite peaks are due to the plasmon or shake-up losses, since their energy shifts are too large to be explained by a chemical shift. In particular, they cannot be due to bonds to O atoms as suggested in ref. [31]. Such peaks would overlap with the peaks resulting from the bonds to the S atoms. The S 2p spectrum shows a peak at 160.8 eV associated with S 2− , and a peak at 166.4 eV that is most likely due to the sulfite species. [43] Finally, the N 1s peak located at 398.5 eV confirms the presence of the N-Co species and the successful N-doping. [36] From the XPS analysis, the N concentration is approximately 5 at%. The undoped Co 9 S 8 nanoparticles show a similar structure and morphology to that of the N-Co 9 S 8 nanoparticles, as shown in Figure S2, Supporting Information.

LiPSs Adsorption
First, we have performed DFT calculations to investigate the interaction between the LiPSs and the surface of Co 9 S 8 /N-Co 9 S 8 nanoparticles. The goal is to evaluate the binding energies quantitatively between the two species. From the texture in XRD measurements, we know that the (311) and (440) surfaces are two main facets for both N-Co 9 S 8 and Co 9 S 8 nanoparticles. Therefore, the first-principles calculations were performed to study the adsorbed structures and adsorption energies of the two typical LiPSs (Li 2 S 2 and Li 2 S 4 , because of the capacity contribution in Li-S battery is mainly from this low-order LiPSs) on the (311) and (440) surfaces of Co 9 S 8 . From the adsorption structures (Figure 2a,b; Figure S3, Supporting Information), it is clear that the Li and S atoms of LiPSs chemically bonds with the S and Co atoms on the surfaces of Co 9 S 8 . Interestingly, after the surface doping with nitrogen atoms, the Li atoms tend to bond with the N atoms. This is confirmed by the formation of Li-N bonds in the XPS analysis ( Figure S4, Supporting Information).
The binding energy E ads between Co 9 S 8 and Li 2 S 2 on the (311) and (440)  . The E ads of LiPSs is increased (by more than three times in case of the (311) surface) after N-doping, indicating a more energetically favourable and stronger chemical anchoring of N-Co 9 S 8 nanoparticles to the LiPSs. We want to emphasize that this effect was not seen in the typical TMD structures, and it contributes to the greatly enhanced performances of the Li-S battery presented here.
The DFT calculation results ( Figure 2c) are strongly supported by the experimental results. Figure 2d shows the adsorption ability of N-Co 9 S 8 and Co 9 S 8 nanoparticles in the Li 2 S 6 solution. A significant color change was observed after adding N-Co 9 S 8 nanoparticles, indicating strong physical adsorption. The Co 9 S 8 nanoparticles, on the other hand, show lower adsorption capability and the Li 2 S 6 solution showed only a light yellow color after 6 h ( Figure 2d). Ultraviolet/visible spectroscopy was also performed, in order to demonstrate the concentration changes of Li 2 S 6 solutions after adding the N-Co 9 S 8 or Co 9 S 8 nanoparticles, respectively. Consistent with the physical adsorption results, the absorption peak of Li 2 S 6 is reduced remarkably after adding the N-Co 9 S 8 nanoparticles ( Figure S5, Supporting Information).
The DFT calculations, together with the LiPSs absorption experiments, prove that the notorious "shuttling effect" is inhibited by the N-doped Co 9 S 8 nanoparticles, which has not been reported by previous works. Now, the question is whether another critical issue-the sluggish redox kinetics, is also alleviated by our design. As shown already in a few reports, [ 7,31] TMD materials can enhance the Li-S battery performances, to a certain extent, when compared with the pure S or Li 2 S 6 . These results are still far from satisfactory, especially when considering the practical conditions of high mass loadings and high rates. In the following, we show strong evidence that the nitrogen doping drastically improves the performances of Co 9 S 8 -based electrodes in Li-S batteries.

LiPSs Anchoring-Diffusion-Conversion Processes
Figure 3a-c shows the schematics of LiPSs anchoringdiffusion-conversion processes on the N-Co 9 S 8 , Co 9 S 8 electrodes and on a pure carbon paper. In order to investigate the mechanisms behind these processes, we first performed experiments with the N-Co 9 S 8 /Li 2 S 6 , Co 9 S 8 /Li 2 S 6 , and pure Li 2 S 6 cathodes, respectively. Cyclic voltammetry (CV) tests were used to reveal the electrocatalytic activities of N-Co 9 S 8 nanoparticles for the charge transfer to LiPSs. As shown in Figure 3d, there are two pairs of distinct and reversible redox peaks for the N-Co 9 S 8 /Li 2 S 6 cathode. The reduction peaks at 2.35 V and 2.1 V belong to the transition from sulfur (S 8 ) to high-order LiPSs (Li 2 S χ , 4 ≤ χ ≤ 8), and then high-order LiPSs to Li 2 S 2 / Li 2 S, respectively. In the reverse oxidization process, two anodic peaks at 2.35 V and 2.4 V are related to the transition from solid Li 2 S 2 /Li 2 S to LiPSs, and then to Li 2 S 8 /S 8 . [44,45] These cathodic/ anodic peak positions are consistent with the previously reported results. In the first three cycles, there are almost no peak shifts, indicating that the formation of sulfur species in the N-Co 9 S 8 /Li 2 S 6 chemistry is completely reversible. [46,47] The CV curves of the Co 9 S 8 /Li 2 S 6 cathode are similar to those of the N-Co 9 S 8 /Li 2 S 6 cathode (Figure 3e). On the other hand, the redox peaks for the pure Li 2 S 6 cathode are significantly broader and only one reduction peak can be observed, which suggests a sluggish kinetic process during charging/discharging (Figure 3f).
When comparing the peak potentials of these three electrodes (Figure 3g), it is clear that the N-Co 9 S 8 /Li 2 S 6 cathode has the highest reduction potential and the lowest oxidation potential, and then followed by the Co 9 S 8 /Li 2 S 6 cathode and the pure Li 2 S 6 cathode. This indicates that the N-Co 9 S 8 nanoparticles can reduce the polarization, which is due to their electrocatalytic effect for the S/Li 2 S during the charge/discharge processes. [47,48] Similar result is also observed for the onset potentials ( Figure 3h). For the onset potential values of the cathodic peaks R 1 and R 2 : N-Co 9 S 8 /Li 2 S 6 > Co 9 S 8 /Li 2 S 6 > pure Li 2 S 6 . For the anodic peaks O 1 and O 2 : N-Co 9 S 8 /Li 2 S 6 < Co 9 S 8 / Li 2 S 6 < pure Li 2 S 6 .
More interestingly, when comparing the peak currents (Figure 3i), it is found that the N-Co 9 S 8 /Li 2 S 6 cathode exhibits the largest current densities in both the cathodic and anodic processes. The peak currents of the cathodic peak O 1 (O 2 ) for N-Co 9 S 8 is about 40% (36%) larger than that for Co 9 S 8 , and nearly 240% (146%) larger than that for the pure Li 2 S 6 . This is a strong evidence that N-Co 9 S 8 nanoparticles can significantly improve the sluggish electron transfer to LiPSs, and hence the redox rate of Li-S chemistry during charge/ discharge.

Electrochemical Performances
The nitrogen doping simultaneously improves the immobilization and the catalytic effects of the Co 9 S 8 nanoparticles. As a result, the Li-S batteries based on the N-Co 9 S 8 /Li 2 S 6 cathode show unprecedented electrochemical performances. Figure 4a shows the first three galvanostatic charge/discharge curves of the N-Co 9 S 8 /Li 2 S 6 cathode at a current density of 0.2 A g −1 within a potential window of 1.7-2.8 V (Li 2 S 6 , corresponding to ≈1 mg cm −2 of S), and the curves are constant with the previous reports on the Li-S batteries based on other hosts. [8,32,49] The charge/discharge curves of Co 9 S 8 /Li 2 S 6 and Li 2 S 6 cathodes (with the same S mass loading) are shown in Figure S6, Supporting Information. All these three cathodes display two typical discharge plateaus, which indicate the multistep reduction reactions of sulfur during the discharge process. The high plateau (≈2.35 V) is attributed to the transformation from S 8 to the high order LiPSs (Li 2 S χ , 4 ≤ χ ≤ 8), while the low plateau (≈2.1 V) is related to the reduction of high order LiPSs to Li 2 S 2 / Li 2 S. The plateaus in charge profiles are due to the conversions from Li 2 S 2 /Li 2 S to S 8 . [44,45] Although these three cathodes have similar charge/discharge profiles, the specific capacity of N-Co 9 S 8 /Li 2 S 6 cathode is much larger than that of the Co 9 S 8 /Li 2 S 6 and the pure Li 2 S 6 cathodes (especially during the low potential plateau ≈2.1 V, as shown in Figure 4b). This proves that the LiPSs dissolution and the shuttle effects have been effectively suppressed by the N-Co 9 S 8 nanoparticles, thus leading to a high utilization rate of LiPSs. A similar effect is also observed in the SEM (EDXS elemental maps) of the Li anode after three cycles ( Figure S7, Supporting Information). Again, these results give a strong evidence that the active N-doping in N-Co 9 S 8 nanoparticles can effectively immobilize the LiPSs, therefore suppressing the "shuttling effect" in Li-S batteries. Furthermore, when comparing the polarizations of the charge/discharge profiles in the second cycle (Figure 4c), it is clear that the N-Co 9 S 8 /Li 2 S 6 cathode shows a much lower polarization value (131 mV) than that of the Co 9 S 8 /Li 2 S 6 cathode (175 mV) and the pure Li 2 S 6 cathode (225 mV). This suggests a more kinetically efficient reaction process due to the nitrogen doping. These results agree very well with the CV results in Figure 3.
To further investigate the electrochemical performances, these three electrodes were cycled at varied current densities of 0.2, 1, 2, 5, 10, and 20 A g −1 . The N-Co 9 S 8 /Li 2 S 6 cathode delivered a stable discharge capacity of 1233 mAh g −1 , 1096 mAh g −1 , 1009 mAh g −1 , 864 mAh g −1 , 730 mAh g −1 , and 604 mAh g −1 , respectively (Figure 4d). The two typical plateaus in the charge and discharge curves are clear and stable, even at the largest current density of 20 A g −1 . And after the current density returned to 0.2 A g −1 , the reversible specific capacity was recovered to 1010 mAh g −1 (Figure 4e). The stable charge/discharge plateaus, together with the excellent reversibility, confirm that the N-Co 9 S 8 nanoparticles can effectively immobilize the LiPSs and accelerate the electron transfer to LiPSs even at high current densities. In stark contrast, the Co 9 S 8 /Li 2 S 6 cathode and pure Li 2 S 6 cathode (especially, the latter) exhibited much lower rate capacities under the same conditions (Figure 4e). We obs the rate performance of N-Co 9 S 8 /Li 2 S 6 cathode represents a very high level when compared with that of the previously reported Li-S batteries. Especially under large current densities (≥5 A g −1 ), the capacities are the highest among all available reports (Figure 4f). [50][51][52][53][54][55][56] For many applications (such as in electric vehicles), a high cycling rate with a high mass loading of active material is  Figure 4g shows the long-term performance of these three cathodes with a high mass loading (S: 2 mg cm −2 ) cycled at 1 A g −1 . After 1000 cycles, the N-Co 9 S 8 /Li 2 S 6 cathode delivered a specific capacity of 605 mAh g −1 -with only 0.037% capacity decay per cycle on average, and maintained a high CE (average CE ≈98.5%). In the meanwhile, both Co 9 S 8 /Li 2 S 6 and pure Li 2 S 6 cathodes showed not only significantly lower capacities but also a much faster capacity decay. For example, when compared with Co 9 S 8 /Li 2 S 6 (pure Li 2 S 6 ) at the 200th cycle, the newly developed N-Co 9 S 8 /Li 2 S 6 showed 174% (250%) improvement in the specific capacity.
Considering the theoretical capacity of sulfur, the sulfur utilization for the N-Co 9 S 8 /Li 2 S 6 cathode is greater than that of the Co 9 S 8 /Li 2 S 6 and Li 2 S 6 cathodes ( Figure S8, Supporting Information). Moreover, the cycling performance of the N-Co 9 S 8 / Li 2 S 6 cathode with an even higher S mass loading (≈5 mg cm −2 ) was also tested. It showed an ultrahigh specific capacity of 870 mAh g −1 (cycled at 1 A g −1 ) and >600 mAh g −1 was remained after 300 cycles ( Figure S9, Supporting Information). This unprecedented performance at high rates with high mass loading suggests the strong potential of N-Co 9 S 8 nanoparticles for use in practical Li-S batteries.

Catalytic Oxidation of Li 2 S
Encouraged by the excellent performances of our newly developed N-Co 9 S 8 nanoparticles, we tested the commercial Li 2 S as cathode for Li-S battery. There are several intrinsic advantages of Li 2 S-based Li-S batteries, such as the inhibition of volume expansion during cell operations and the possibility of using non-lithium-metal anodes. However, it is known that Li 2 S suffers from a low Li-ion diffusivity, low electrical conductivity, and high charge transfer resistance, which result in a high overpotential during the initial charging. [7] Here, we mixed the commercial Li 2 S with N-Co 9 S 8 nanoparticles, carbon black, and PVDF to prepare the cathodes (see Experimental Section). The assembled coin cells were first charged to 4.0 V from Figure 4. Electrochemical performances of the Li-S batteries. a) Typical charge/discharge profiles of the N-Co 9 S 8 /Li 2 S 6 cathode for the first three cycles at 0.2 A g −1 , b) High plateau discharge capacity of the N-Co 9 S 8 /Li 2 S 6 , Co 9 S 8 /Li 2 S 6 , and Li 2 S 6 cathodes in the 2nd, 10th, 50th, and 100th cycle, c) charge/ discharge profiles of the N-Co 9 S 8 /Li 2 S 6 , Co 9 S 8 /Li 2 S 6 , and Li 2 S 6 cathodes in the second cycle, d) charge/discharge profiles of the N-Co 9 S 8 /Li 2 S 6 cathode at various current densities of 0.2-20 A g −1 , e) rate capability of the N-Co 9 S 8 /Li 2 S 6 , Co 9 S 8 /Li 2 S 6 , and Li 2 S 6 cathodes at various current densities of 0.2-20 A g −1 , f) comparison of the rate capability of N-Co 9 S 8 /Li 2 S 6 cathode with that of Li-S batteries from recent publications, g) cycling stability of the N-Co 9 S 8 /Li 2 S 6 , Co 9 S 8 /Li 2 S 6 , and Li 2 S 6 cathodes with 2 mg cm −2 S mass loading cycled at 1 A g −1 for 1000 cycles. open-circuit to delithiate Li 2 S, with the aim to convert Li 2 S to the low-order LiPSs, high-order LiPSs and finally S. [40,57] Figure 5a shows the first three cycles of N-Co 9 S 8 /Li 2 S cathode. After the initial delithiation of Li 2 S, the following cycles show curves that are similar to that of a typical Li-S battery. The charge/discharge curves for the Co 9 S 8 /Li 2 S and pure Li 2 S cathodes can be found in Figure S10, Supporting Information. Figure 5b shows the initial delithiation of Li 2 S in these three different cathodes, and the pure Li 2 S electrode exhibits a high potential barrier until 4.0 V, indicating a sluggish activation process and a high charge transfer resistance. Interestingly, the N-Co 9 S 8 /Li 2 S cathode shows a voltage jump after a relatively higher initial potential barrier and maintains a long charge plateau at ≈2.5 V. This is an indication of a strongly reduced charge transfer resistance due to the catalytic effect of nitrogen doping. It is worth mentioning that this potential barrier is much lower than that in the previous reports. [57,58] A similar charging phenomenon is observed for the Co 9 S 8 /Li 2 S cathode, but with a higher potential barrier and a smaller specific capacity. Electrochemical impedance spectra (EIS) measured at the open-circuit voltage also support the reduced charge transfer resistance in the N-Co 9 S 8 /Li 2 S cathode ( Figure S11, Supporting Information). [59,60] And from EIS, the diffusion coefficient of Li + ions in N-Co 9 S 8 /Li 2 S, Co 9 S 8 /Li 2 S, and Li 2 S electrodes at their initial state is calculated to be 6.7 × 10 −11 cm 2 s −1 , 8.4 × 10 −12 cm 2 s −1 , and 7.6 × 10 −15 cm 2 s −1 , respectively. [61] It is clear that the value in the N-Co 9 S 8 /Li 2 S electrode is nearly four orders of magnitude higher than that in the pure Li 2 S electrode. Therefore, the N-Co 9 S 8 nanoparticles can facilitate an efficient migration of Li + thus accelerate the transformation of Li 2 S or S 8 . All these electrocatalytic measurement results are consistent with that of the Li 2 S 6 cathodes shown above.
The cycling performances of the commercial Li 2 S-based cathodes are shown in Figure 5c. The N-Co 9 S 8 /Li 2 S cathode delivers an excellent discharge capacity of 825 mAh g −1 after 100 cycles. On the contrary, the Co 9 S 8 /Li 2 S and Li 2 S cathodes show a rapid capacity decay. The rate capability shows a similar result (Figure 5d). It is worth mentioning that the contributions of N-Co 9 S 8 or Co 9 S 8 to the total capacity can be neglected ( Figure S12, Supporting Information). We envision that, thanks to the excellent results of N-Co 9 S 8 /Li 2 S cathode, a safe Li-S full battery can be built by pairing the new cathode with an Si or C anode (instead of the highly active lithium metal). Also, it can lead to a much higher specific energy when compared with the commercial LiCoO 2 /graphite system. [62,63]

Conclusion
We demonstrate that nitrogen-doped Co 9 S 8 nanoparticles are a highly attractive material for Li-S batteries. Combining the theoretical calculations with the experimental results, we prove that the nitrogen doping can simultaneously and significantly improve the immobilization and the catalyzation of the polysulfides by the Co 9 S 8 nanoparticles. We observe an effective suppression of the shuttling behavior, and as well as a significant acceleration of the redox reaction kinetics. As a result, the N-Co 9 S 8 nanoparticles/Li 2 S 6 composite cathodes exhibit a high reversible capacity (1245 mAh g −1 at 0.2 mA g −1 ), fast reaction kinetics (a record value of 604 mAh g −1 at 20 A g −1 ), as well as a low capacity decay of 0.037%/cycle over 1000 cycles. The excellent performances are retained even at high mass loadings (up to 5 mg cm −2 ) and high current densities. A proof-of-concept experiment is also shown: the N-Co 9 S 8 nanoparticles were
Synthesis of N-Doped Co 9 S 8 Nanoparticles: N-doped Co 9 S 8 nanoparticles were synthesized through a simple hydrothermal reaction. Briefly, 1.746 g Co(NO 3 ) 2 · 6H 2 O, 0.9 g CH 3 CSNH 2 , and 0.42 g PVP were dissolved in 300 mL distilled water and stirred to form a homogeneous solution, then 72 mL 0.5 m NaOH was added into this solution. The mixture was transferred to a three-necked flask and heated to 100 °C with a heating rate of 5 °C min −1 under nitrogen atmosphere and stirred for 2 h. Then, 0.03636 g urea (as nitrogen source) was added into this solution, which was kept at 100 °C for another 30 min. After cooling, the product was washed with ethanol for three times, and then dried using a freeze dryer. For comparison, Co 9 S 8 nanoparticles were synthesized using a similar approach but without adding urea.
Preparation of the Li 2 S 6 Catholyte: Sulfur and Li 2 S at a molar ratio of 5:1 were added to an appropriate amount of 1,3-dioxolane (DOL) and 1,2-dimethoxyethane (DME) by vigorous magnetic stirring at 50 °C overnight in the glove box in order to obtain 1 m lithium polysulfide (Li 2 S 6 ) catholyte.
Polysulfide Adsorption Test: Li 2 S 6 solution was diluted to a concentration of 10 mmol L −1 (calculated based on the S content). Then the N-Co 9 S 8 and Co 9 S 8 nanoparticles powders were added to the Li 2 S 6 solution (with a mass ratio of 2:8), respectively. The mixtures were stirred and then waited for 6 h to ensure a thorough adsorption process. The glass vial only with Li 2 S 6 solution was used for comparison.
Materials Characterization: XRD patterns of the N-Co 9 S 8 /Co 9 S 8 nanoparticles were recorded by using Cu-Kα radiation on a Bruker D8 Advance Discovery X-ray Diffractometer. The morphology, microstructure, and composition of the N-Co 9 S 8 /Co 9 S 8 nanoparticles were investigated by the emission scanning electron microscopy (FE-SEM, JEOL JSM-6700F) and transmission electron microscopy (FEI, ChemiSTEM 80-200, C s -corrected) under the high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) mode, operated at 200 kV. XPS analysis was performed with a hemispherical analyzer of 100 mm radius (Leybold Heraeus). Ultraviolet/visible absorbance spectroscopy was performed in the spectral range of 200-800 nm using a Cary 5000 UV-vis variable wavelength spectrophotometer to evaluate the lithium polysulfide absorption capability of N-Co 9 S 8 and Co 9 S 8 nanoparticles.
Electrochemical Measurements: N-Co 9 S 8 or Co 9 S 8 nanoparticles were mixed with polyvinylidene difluoride (PVDF) in N-methyl-2-pyrrolidinone (9:1 by weight) by vigorous magnetic stirring to form a homogeneous slurry, respectively. It was subsequently blade coated onto the hydrophilic carbon papers and vacuum-dried at 60 °C for 24 h. The mass ratio of N-Co 9 S 8 or Co 9 S 8 nanoparticles over Li 2 S 6 is 2:8 (based on the S content), and the specific capacities were also calculated based on the weight of Sulfur in the cathodes. CR2032 coin cells were assembled in an argonfilled glove box (MBRAUN UNI lab: O 2 < 0.1 ppm, H 2 O < 0.1 ppm). The electrolyte was 1.0 m lithium bis-trifluoromethaesulfonylimide (99.95% trace metals basis, Sigma-Aldrich) dissolved in 1,3-dioxolane (Sigma-Aldrich) and 1,2-dimethoxyethane (Sigma-Aldrich; 1:1 ratio by volume) with 0.1 m lithium nitrate (LiNO 3 , Sigma-Aldrich) as the additive.
Lithium metal was used as counter/reference electrode and Celgard 2400 membrane as the separator. Li 2 S 6 catholyte was used as active material drop on the N-Co 9 S 8 /carbon paper or Co 9 S 8 /carbon paper or pure carbon paper electrodes (the sulfur mass loading is calculated according to the volume of the catholyte ≈1 mg cm −2 ). The electrolyte/ sulfur ratio for low sulfur mass loading (≈1 mg cm −2 ) is ≈6 µL mg −1 , for high sulfur mass loading (≈5 mg cm −2 ) is ≈4 µL mg −1 . The galvanostatic charge/discharge tests were carried out in the LAND CT 2001A charge/ discharge system within a voltage range of 1.7-2.8 V for Li 2 S 6 , and for reduction/oxidation of Li 2 S, the first charge is up to 4.0 V, then charge/ discharge between 1.6 and 2.8 V. The current density for Li 2 S 6 tests was ranged from 0.2 to 20 A g −1 and for Li 2 S was ranged from 0.2 to 5 A g −1 .
The cyclic voltammogram (CV) and EIS measurements were conducted using Metrohm Auto-lab at a scan rate of 0.1 mV s −1 with the voltage range of 1.7-2.8 V and the frequency range was controlled from 100 kHz to 10 mHz with 10 mV fluctuations, respectively.
First-Principles Calculations: The first-principles calculations were performed based on DFT as implemented in VASP code [64,65] to study the equilibrium structures and adsorption energy of LiPSs (Li 2 S 2 and Li 2 S 4 molecules) on the (311) and (440) surfaces of Co 9 S 8 crystal. The projector-augmented wave (PAW) method [66,67] was applied to treat the valence and core electrons interaction. Electron correlation and exchange effects were treated with a generalized gradient approximation (GGA) proposed by Perdew, Burke, and Ernzerhof (PBE). [68] A cutoffenergy of 520 eV was used for the plane wave basis set. The total energy convergence criterion was 10 −6 eV. All systems were fully relaxed until the residual Hellmann-Feynman forces were smaller than 0.01 eVÅ −1 . The dipole correction [69] was considered to treat the impact of the variation of potential distribution due to the adsorption of LiPSs on (311) and (440) surfaces. In addition, the Van der Waals interactions were included during all calculations using the optB86b-vdw functional. [70,71] The calculated lattice constant of cubic Co 9 S 8 crystal is 9.79 Å, which is very close to the previously computed value of 9.80 Å. [56] The slabs of (311) and (440) surfaces were constructed based on fully relaxed Co 9 S 8 crystal, as shown in Figure 5d,e. A vacuum layer of at least 20 Å perpendicular to the plane of the 2D systems was applied to avoid the interaction between neighboring images. During the optimizations of adsorbed models, the outermost layer of slabs and LiPSs were fully relaxed, and other parts were fixed. The Brillouin zone was sampled using a 1 × 2 × 1 and 2 × 3 × 1 Monkhorst-Pack k-point scheme for (311) and (440) surfaces, respectively. The adsorption energy of LiPSs adsorbing on surfaces is defined as follows: where E LiPSs and E surf is the total energy of the LiPSs, the surface slabs, respectively. E LiPSs/surf is the total energy of the attached system of LiPSs and surface slab. According to Equation (1), the E ads with a positive value means that the corresponding adsorption structure is energetically favorable to form. The larger the value, the easier the adsorption. It should be noted that the adsorption energy after surface doping by nitrogen atoms was also calculated.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.