GAI et al.

Abstract

Electrochemical properties of lithium–sulfur (Li–S) batteries are mainly hindered by both the insulating nature of elemental sulfur (i.e., molecular S₈) and the shuttling effect or sluggish redox kinetics of lithium polysulfide intermediates (Li₂S_n, 3 ≤ n ≤ 8). In this paper, a three‐dimensional mesoporous reduced graphene oxide‐based nanocomposite, with the embedding of metallic Co nanoparticles and the doping of elemental N (Co/NrGO), and its simply ground mixture with powdered S at a mass ratio of 1:6 (Co/NrGO/S) are prepared and used as cathode‐/separator‐coated interlayers and working electrodes in assembled Li–S cells, respectively. One of the effective cell configurations is to paste composite Co/NrGO onto both the S‐loading cathode and separator, showing good cycling stability (1070 mAh g⁻¹ in the 100th cycle at 0.2 C), high‐rate capability (835 mAh g⁻¹, 2.0 C), and excellent durability (905 mAh g⁻¹ in the 250th cycle at 0.5 or 0.2 C). Compared with the experimental results of Co‐absent NrGO, electrochemical properties of various Co/NrGO‐based cell configurations clearly show multiple functions of Co/NrGO, indicating that the absence of Co/NrGO coatings and/or Co nanoparticles may be inadequate to achieve superior S availability of assembled Li–S batteries.

INTRODUCTION

Commercial companies represented by Tesla Motors have made huge breakthroughs in the design of high-efficiency lithium-ion batteries (LIBs) for electric vehicles; however, there is still a long way to develop high-capacity electrode materials in improving the high-energy and high-power densities of LIB devices.^1–3 In recent decades, lithium–sulfur (Li–S) batteries have been recognized as one of the most promising energy storage systems due to their high specific capacity (1675 mAh g⁻¹) and high-energy density (2600 Wh kg⁻¹) in theory.^4–6 It is also well known that the electrical insulating nature of elemental S (i.e., molecular S₈) and its lithiated product of insoluble lithium sulfides (i.e., Li₂S₂ and Li₂S), accompanied by the intermediate formation of electrolyte-soluble lithium polysulfide (Li₂S_n, 3 ≤ n ≤ 8) and its back-and-forth migration between the cathode and the anode, induces low utilization of molecular S₈ and rapid capacity decay of Li–S batteries.^7,8 During galvanostatic cycling processes, the sluggish redox kinetics of Li₂S_n may cause its accumulation in an electrolyte solution, which can further accelerate its shuttling effect between the cathode and the anode.^9,10

One of the well-developed carbonaceous materials used in Li–S batteries is graphene, which possesses high electrical conductivity, large surface area, and excellent mechanical strength to load or to immobilize S-based species.^11–14 However, the weak host–guest interactions between nonpolar graphene and polar guests can hardly be associated with the entrapment of intermediate Li₂S_n.^15–17 Therefore, on the one hand, heteroatom-doped graphene-based materials are tentatively used to enhance the chemical entrapment of Li₂S_n,^18–20 and on the other hand, introduction of metal-based active centers into the graphene-based frameworks of the sulfur cathode may yield an unexpected catalytic ability to improve the kinetics of polysulfide conversion. In terms of metal-based active centers, transition-metal oxides, sulfides, carbides, nitrides, or phosphides have a strong affinity toward Li₂S_n because of its catalytic conversion, which eliminates the shuttling effect in Li–S batteries.^21–30

To the best of our knowledge, parts of the above-mentioned catalysts still suffer from an inferior electronic conductivity, and thus their replacement by metallic nanoparticles or their doping with a polar carbonaceous material may more or less overcome this issue.^31–37 When porous carbon nanosheets modified with metallic Co and heteroatoms N and B (Co-NBC) are selected as sulfur-loading frameworks, the corresponding S/Co-NBC cathode yields an initial discharge capacity of 823 mAh g⁻¹ at 0.5 C and then reaches a reversible value of 440 mAh g⁻¹ after 500 galvanostatic cycles.³⁸ Another example is the use of Co-embedded, N-doping carbon nanotubes (Co@NCNTs) as a host of sulfur, and the corresponding S/Co@NCNT yields a reversible capacity of 658 mAh g⁻¹ at 0.5 A g⁻¹ in the 200th cycle.³⁹ Aside from the catalytic ability of embedded Co nanoparticles therein, powdered S and nanostructured Co@NCNT/Co-NBC are ground at a mass ratio of 3:1 (i.e., 75 wt% of S), and then a 12-h melting–diffusion of elemental sulfur at 155°C under an inert atmosphere is essential to demonstrating a structure-integral synergistic effect on the enhanced performances of the S-loading composites. To further understand the multi-functional mechanism of each Co embedded, N doped carbonaceous composite, the following still need to be explored for potential application purposes: simplified preparation of a S-containing composite with a mass loading higher than 75.0 wt%, use of reduced graphene oxides (rGOs) as a high-conductivity matrix, porous-carbon permutation of varying electrode/cell configuration for high sulfur availability and so on.

In this paper, a three-dimensional (3D) mesoporous rGO-based nanocomposite with both embedding of metallic Co nanoparticles and doping of elemental N (Co/NrGO) is prepared, and is then simply ground with powdered sulfur to obtain a high S-containing counterpart of Co/NrGO/S (85.7 wt%) as Li–S battery cathodes. Then, the pristine composite of 3D mesoporous Co/NrGO is treated as a coating material of the cathode, separator, and/or both of assembled cells to evaluate its multiple functions in different ways. Aside from the structural characterization of Co/NrGO and its Co-absent counterpart (NrGO), the multiple functions of pristine Co/NrGO and a plausible mechanism of high-performance Li–S batteries are mainly examined and discussed in the context.

EXPERIMENTAL SECTION Chemical and material preparation

All chemicals, such as cobalt acetate tetrahydrate Co(CH₃COO)₂·4H₂O, sucrose C₁₂H₂₂O₁₁, urea CO(NH₂)₂, lithium sulfide (Li₂S), and powdered sulfur (or molecular S₈) were of analytical grade and were used as received. Ultrapure water (18.25 MΩ cm) was used throughout the aqueous solution preparation. using a modified Hummers method,⁴⁰ an aqueous suspension of graphene oxide (GO) was freshly prepared before each experiment. Li₂S₈ or Li₂S₄ polysulfide stock solution, in the equal-volume mixed solvent of 1,2-dimethoxyethane DME and 1,3-dioxolane DOL, was prepared by dissolving Li₂S and S₈ at a molar ratio of 1:8 or 1:4 under vigorous stirring for 48 h, and the operational procedure and subsequent storage were performed in an argon-filled glovebox at room temperature.

A typical synthesis procedure of the pristine composite Co/NrGO (or NrGO) is as follows. First of all, in the presence or absence of Co(CH₃COO)₂·4H₂O (105 mg), sucrose (210 mg) was dissolved in a GO suspension (2 mg mL⁻¹, 70 mL) under magnetic stirring for 1 h, and then the aqueous mixture was transferred to a 100-mL Teflon-lined stainless-steel autoclave for a 12-h hydrothermal reaction at 160°C. Subsequently, the resulting black-colored rGO-based hydrogel was rinsed in ultrapure water three times, soaked in a CO(NH₂)₂ solution (25 mg mL⁻¹, 100 mL) for 24 h, and then taken out for drying at 80°C. Finally, the dried solid block was transferred to a N₂-atmosphere furnace for a programmed (5°C min⁻¹) heat treatment at 350°C for 2 h and then at 900°C for 5 h in sequence, resulting in a pristine composite of Co/NrGO (or NrGO).

The cathode material of the S-loading composite Co/NrGO/S (NrGO/S) was prepared by manually grinding pristine composite Co/NrGO (NrGO) and powdered S at a mass ratio of 1:6 for 1 h.

Structural characterization

X-ray diffraction (XRD) data were collected on a Rigaku D/max-2400 powder X-ray diffractometer with Cu Kα radiation (40 kV, 120 mA) in the 2θ range of 8° and 80°. The morphology and chemical composition of the samples were determined using a scanning electron microscope (SEM; ZEISS Sigma 300), operated at 3 kV, equipped with an energy-dispersive spectroscope (EDS), operated at 10 kV. Transmission electron microscopy (TEM, JEOL JEM 1011) was used to characterize the structure of samples, and high-resolution TEM images and corresponding selected area electron diffraction (SAED) patterns were determined on an FEI Tecnai G2 F20 TFM. Thermogravimetric analysis and differential scanning calorimetry (TGA-DSC) tests were conducted on a Mettler Toledo TGA/SDTA851 machine with a step of 10°C min⁻¹. Raman spectra were obtained using a LabRAM-HR confocal laser micro-Ramma spectrometer (632.8 nm). X-ray photoelectron spectroscopy (XPS) data were collected on an ESCALAB 250 XI electron spectrometer using Al Kα radiation. Nitrogen adsorption–desorption isotherms (i.e., Brunauer–Emmett–Teller or BET isotherms) were constructed using a Micromeritics ASAP 2460 sorptometer, and the corresponding pore size distributions were analyzed using the Barrett–Joyner–Halenda (BJH) method.

Electrochemical characterization

First, 70 mg of Co/NrGO/S or NrGO/S, 10 mg of acetylene black, and 8 mg of polyvinylidene fluoride (PVDF) were dispersed in N-methylprolinodone (NMP) to form a homogeneous slurry. Then, the slurry was uniformly cast onto an Al foil, dried at 60°C for 12 h, and then cut into 12-mm disks, resulting in a Co/NrGO/S or NrGO/S working electrode with a S loading of 1.1 - 2.3 mg cm⁻².

The surface coating slurry of the Co/NrGO/S or NrGO/S working electrode was prepared by dispersing Co/NrGO or NrGO (10 mg) and a sodium alginate (SA, 2 mg) binder in water, which was pasted onto the working electrode and then dried at 60°C for 12 h. The surface coating slurry of the separator (Celgard 2300 polypropylene membrane) was prepared by dispersing Co/NrGO (8 mg) and hydrophobic PVDF (2 mg) in NMP, and then this slurry was homogeneously pasted onto each separator (diameter ~19 mm) with a mass density of ca. 0.7 mg cm⁻².

CR2032-type coin cells were assembled in an Ar-filled glovebox using the electrolyte of lithium bis(trifluoromethanesulfonyl)imide LiTFSI (1.0 M) and LiNO₃ (2.0 wt%) in a DME:DOL (1:1, volume/volume) mixed solvent. Metallic lithium and the pristine or modified membrane Celgard 2300 were used as the counter/reference electrode and the separator, respectively.

All of the electrochemical measurements were performed at room temperature. Galvanostatic charge–discharge tests were conducted on a Land CT2001A battery system within a narrow potential range of 1.8–2.7 V (vs. Li⁺/Li, ibid). Cyclic voltammetry (CV) performances were carried out on an LK 2005A Electrochemical Workstation at 0.1 mV s⁻¹ in the range of 1.5–3.0 V. Electrochemical impedance spectroscopy (EIS) measurements were conducted in the frequency range from 100 kHz to 100 mHz with an AC voltage amplitude of 5 mV.

Electrochemical kinetics measurement

The as-obtained composite Co/NrGO or NrGO and hydrophobic binder PVDF were dispersed in the NMP solvent at a weight ratio of 4:1, and then the uniform slurry was coated on the Al foil and dried at 60°C overnight, yielding the working electrodes of Co/NrGO or NrGO to assay the conversion kinetics of as-synthesized Li₂S₈ and Li₂S₄. Either the Li₂S₈-based symmetrical or Li₂S₄-based asymmetrical cell model is also the CR2032-type coin, that is, the only difference is the recipe of electrolyte solution (1.0 M LiTFSI + 0.1 M Li₂S₈ or 1.0 M LiTFSI + 0.05 M Li₂S₄).

CV tests of a symmetrical cell (e.g., Co/NrGO || 1.0 M LiTFSI + 0.1 M Li₂S₈ || Co/NrGO) were scanned at 5 mV s⁻¹ between −0.8 and 0.8 V, while linear sweep voltammetry (LSV) tests of an asymmetrical cell (e.g., Co/NrGO || 1.0 M LiTFSI + 0.05 M Li₂S₄ || Li) were performed at 0.2 mV s⁻¹ within the potential window from −50 to 50 mV.

RESULTS AND DISCUSSION Preparation and structural properties of Co/NrGO

Four steps are artificially distinguished to describe the experimental procedure of preparing the pristine composite Co/NrGO and its S-loading counterpart Co/NrGO/S for high-performance Li–S batteries (Figure 1). Before the hydrothermal formation of an rGO-based hydrogel in the first step (Step 1), dissolving chemicals Co(CH₃COO)₂·4H₂O and C₁₂H₂₂O₁₁ into a GO aqueous suspension may ensure a homogeneous distribution of rGO ultrathin scales in the resulting hydrogel as far as possible.

Figure 1. Schematic experimental procedure of preparing a pristine composite Co/NrGO and its S-loading counterpart Co/NrGO/S for a high-performance Li–S battery configuration

After removing uncross-linked components (or residual ions) from the 3D network of the rGO-based hydrogel, thorough soaking of the “purified” hydrogel in a CO(NH₂)₂ aqueous solution and then further carbonization under a N₂ atmosphere induced the formation of metallic Co nanoparticles within the 3D interconnected rGO-based skeleton doped by elemental N.^41,42 Especially in the second step, the soaked CO(NH₂)₂ may simultaneously serve as a self-removed pore former to achieve porous structure optimization, as a nitrogen source to ensure elemental N doping and as a reducing agent to promote the complete transformation of Co²⁺ ions into metallic Co nanoparticles.

The third step (Step 3) involves manual grinding of powdered elemental S and pristine composite Co/NrGO to form the final composite Co/NrGO/S, which markedly differs from the common encapsulation of liquid-/solution-state sulfur into a porous matrix.^6,9,15 This process is relatively simple and facile because it facilitates the mass production of a S-loading porous carbon composite at an arbitrary mass ratio, and simultaneously, the limited pore volume and polydisperse pore sizes of the carbonaceous framework may be ignored in view of the S-loading. The final step (Step 4) involves using the pristine composite Co/NrGO to modify the electrode/cell configuration of as-assembled Li–S cells according to the formula “Co/NrGO-coated Co/NrGO/S || electrolyte | Co/NrGO-coated Separator | electrolyte || metallic Li,” and in a sense, coating Co/NrGO onto Co/NrGO/S may compensate for the packaging defect of S powder in the previous step.

As for Step 4, three aspects about the modification of the electrode/cell configuration should also be mentioned: (i) the cathode coating slurry of Co/NrGO and the hydrophilic binder sodium alginate were used for the possible infiltration of elemental sulfur into the coating interlayer during the drying process; (ii) the slurry of Co/NrGO and hydrophobic binder PVDF, coated onto the cathode side of the separator, were tentatively used to block the diffusion of electrolyte-soluble Li₂S_n into the anode part of each cell; and (iii) using the control sample of Co-absent NrGO can highlight the catalysis of metallic Co nanoparticles to accelerate redox kinetics of the electrolyte-soluble intermediate Li₂S_n.

In the absence of cobalt acetate, the hydrothermal reaction of GO and sucrose successfully induced the formation of an rGO-based loose hydrogel, which could be further transformed into the pristine composite of the NrGO aerogel (Figure S1). By contrast, the structure of the Co²⁺-containing rGO-based hydrogel was well knit and could hardly be broken by supernatant agitation, and the metallic Co nanoparticles formed in the Co/NrGO framework had an average particle size of 82.4 ± 1.2 nm (Figure S2). Low-resolution SEM observation reveals pristine Co/NrGO is a 3D nanostructured architecture made up of loosely stacked wrinkled scales of rGO-based nanosheets (Figure 2A). These carbonaceous nanosheets were randomly decorated with spherical nanoparticles (Figure 2B), and then by SEM-attached EDS line-scan analyses of elemental C, O, Co, and N, the selected nanoparticle was found to be metallic cobalt, rather than cobalt oxide, cobalt nitride, or others (inset in Figure 2B).

Figure 2. (A, B) SEM and (C, D) TEM images of pristine Co/NrGO. The insets in panels (B) and (D) show the EDX line-scan elemental analyses and the corresponding SAED pattern, respectively. (E) SEM image and (F–I) corresponding elemental C, N, Co, and S mappings of the Co/NrGO/S composite. SAED, selected area electron diffraction; SEM, scanning electron microscope; TEM, transmission electron microscopy

After the 3D nanostructured architecture of pristine Co/NrGO was sonically dispersed in ethanol, TEM images further confirms the presence of both the ultrathin nanosheet-like feature of rGO-based scales and the random doping of metallic Co nanoparticles at a nanoscale (Figure 2C,D). The inset of Figure 2D shows the corresponding SAED pattern, which not only presents a diffraction ring of the (002) crystal face of graphitized carbons but also yields three diffraction signals of the (111), (220), and (311) crystal planes of metallic cobalt in sequence.⁴³

As mentioned above, grinding of powdered sulfur with pristine Co/NrGO is too simple to encapsulate elemental S into the inner gaps among carbonaceous scales. However, the homogeneous covering of elemental S onto Co/NrGO nanosheets may be similar to the possibly homogeneous distribution of elemental Co at a macroscopic scale, but may differ from the microscopically heterogeneous distribution of the metallic Co nanoparticles shown in Figure 2C and 2D. An overall SEM image and the corresponding elemental mappings of the final composite Co/NrGO/S actually reveal a homogeneous distribution of elemental C, N, Co, or S at a mesoscale, indicating an effective approach to the S-loading composite formation (Figure 2E–I). Moreover, cross-sectional SEM images of a Co/NrGO-coated Co/NrGO/S working electrode and a Co/NrGO-coated separator were visually used to estimate the thicknesses of the Co/NrGO/S cathode, the cathode coating, and the separator coating (28, 21, and 35 μm, respectively) (Figure S3).

In Figure 3A, the XRD pattern of as-obtained NrGO clearly shows two diffraction peaks, at around 26.1° and 43.0°, attributable to the (002) and (100) crystal planes of graphitic carbon, respectively, while that of pristine Co/NrGO also shows metallic Co reflections (JCPDS No. 15-0806) of (111), (200), and (220) crystal faces at the 2 theta positions of 44.2°, 51.5°, and 75.9°, respectively.^15,38,44 Under an air atmosphere, the TGA-DSC profiles of Co/NrGO showed two exothermic peaks at 461°C and 498°C, and the corresponding weight losses can be assigned to the calcination of the N-doped rGO-based carbonaceous framework and the high-temperature oxidation of nanocrystalline Co, respectively (Figure 3B). If the air-atmosphere transformation of elemental Co into Co₃O₄ is complete (Figure S4), herein, the Co content in as-obtained Co/NrGO can be calculated to be ~15.8 wt% according to TGA-DSC results.

Figure 3. (A) XRD pattern, (B) TGA-DSC curves, (C) N2 adsorption–desorption isotherms and pore size distribution, and (D–F) C 1 s, N 1 s, and Co 2p XPS spectra of Co/NrGO. The XRD pattern of the control sample NrGO is also presented in panel (A). (G) XRD pattern, (H) Raman spectra, and (I) TGA curves of Co/NrGO/S and NrGO/S. TGA-DSC, thermogravimetric analysis and differential scanning calorimetry; XPS, X-ray photoelectron spectroscopy; XRD, X-ray diffraction

Nitrogen adsorption–desorption isotherms of pristine Co/NrGO (Figure 3C) are classified as Type IV isotherms, and a distinct hysteresis loop within the relative pressure P/P₀ range of 0.4–1.0 indicates a mesoporous feature of the nanostructured composite.⁴⁵ On analysis, the specific surface area and the pore volume of Co/NrGO were found to be 526 m² g⁻¹ and 1.31 cm³ g⁻¹, and the pore size distribution shows an average pore size of 8.8 nm (inset in Figure 3C).

An overall XPS spectrum of Co/NrGO proves the coexistence of C, N, O, and Co elements (Figure S5). The peak-fitted four signals in the high-resolution C 1s XPS spectrum of Co/NrGO at 278.4, 285.0, 285.5, and 291.2 eV, can be assigned to the presence of C─C/C═C, C–N, C–N, C═N bonds, and O═C─O⁻ functional groups, respectively (Figure 3D).⁴⁶ As reported in various studies,^14,47–50 nitrogen doping of a nanostructured carbon can significantly improve its surface interactions with the electrolyte-soluble intermediate Li₂S_n of cycled Li–S batteries, facilitating catalysis of the embedded Co nanoparticles for the efficient redox reactions of polysulfide Li₂S_n. The spectra of pyridinic─N (398.7 eV), Co─N (399.5 eV), pyrrolic─N (400.5 eV), and graphitic─N (401.6 eV) of Co/NrGO are fitted out and marked in Figure 3E, and the Co─N species may functionalize in confining the catalytic centers (i.e., metallic nanoparticles) within the conductivity-enhanced carbonaceous network.

As labeled in the Co 2p XPS spectrum of Co/NrGO (Figure 3F), fitting of the Co 2p_3/2 (774–790 eV) and Co 2p_1/2 bands (791–800 eV) reveals the co-existence of the Co 2p_3/2 (778.8 eV) and Co 2p_1/2 (794.4 eV) peaks of metallic Co and the Co 2p_3/2 peak (780.4 eV) of the Co–N bond.^50–52 Importantly, due to the doping of metallic Co nanoparticles, pristine Co/NrGO acquires a much higher electronic conductivity (32.36 S cm⁻¹) than its Co-absent counterpart NrGO (4.48 S cm⁻¹).

XRD patterns of the final S-loading composites Co/NrGO/S and NrGO/S are almost the same as that of elemental S (Figure 3G), but the Raman spectrum of each S-loading composite clearly shows the typical characteristics of a partially graphitized carbon: a D band at 1340 cm⁻¹ and a G band at 1590 cm⁻¹ (Figure 3H). As for the N₂-atmosphere TGA result of each S-loading composite (Figure 3I), only the weight loss of sulfur evaporation can be detected from room temperature to 500°C, and the S content of 85.7 wt% is exactly the same as that of the Co/NrGO:S (or NrGO:S) with mass ratio at 1:6. By considering the measured pore volume of Co/NrGO (i.e., 1.31 cm³ g⁻¹) and the mass density of S powder (i.e., 2.36 g cm⁻³) simultaneously, even if all of the inner pores are open and can be completely filled with sulfur, the maximum S content of Co/NrGO/S is not higher than 75.6 wt%. To our knowledge, herein, the actual S content of Co/NrGO/S or NrGO/S is much higher than that (i.e., <60 wt%) of a porous carbonaceous framework obtained by filling liquid-/solution-state sulfur into the inner pores.

Enhanced electrochemical properties of Co/NrGO/S

According to recent advances in the research on high-performance Li–S batteries,^{3,5,22,24–30,34–39,53,54} porous carbon generally serves as a sulfur host and a second collector in the cathode configuration and as a separator coating to block polysulfide migration, and then due to the decoration of a transition metal or its compound, the modified carbonaceous nanostructure acts as an active site for the catalytic conversion of intermediate Li₂S_n. To assay the multiple functions of the pristine composite Co/NrGO for high-performance Li–S batteries shown in Figure 1, another pristine composite of NrGO obtained in the absence of elemental Co was used for comparison.

As shown in Figure S3, the thickness of the cathode and separator coating is about 21 and 35 μm, respectively irrespective of which pristine composite (i.e., Co/NrGO or NrGO) is used to assemble Li–S cells. Within the potential range of 1.5–3.0 V at 0.1 mV s⁻¹, CV profiles of Co/NrGO/S and NrGO/S in the 3rd full cycle show similar lithiation–delithiation characteristics to those of elemental S: two distinctive cathodic peaks for the reduction of molecular S₈ to electrolyte-soluble intermediate Li₂S_n and then to electrolyte-insoluble Li₂S₂/Li₂S, and one anodic peak for the gradual oxidation of Li₂S₂/Li₂S to molecular S₈ (Figure 4A). By comparison with the cathodic/anodic peak position of NrGO/S in voltage, those of Co/NrGO/S localize at the relatively high-/low-voltage positions, indicative of the lower electrode polarization owing to the presence of Co nanoparticles. As far as the peak intensity (or peak area) is concerned, each value of Co/NrGO/S is visually higher than that of the NrGO/S counterpart, corresponding to a higher electrochemical activity of Co/NrGO/S.

Figure 4. (A) CV curves at 0.1 mV s−1, (B) Voltage profiles at 0.2 C, (C) EIS results at open-circuit voltage, and (D) Zre ~ω−0.5 linear relationships of Co/NrGO/S and NrGO/S. EIS, electrochemical impedance spectroscopy

At 0.2 C (1 C = 1675 mA g⁻¹), the 3rd charge–discharge cycling curves in Figure 4B show a discharge/charge capacity of 1505.8/1482.2 and 1144.8/1101.9 mAh g⁻¹ for Co/NrGO/S and NrGO/S, respectively, comparatively indicating higher sulfur utilization of the former than that of the latter. Aside from this, the relatively long-and-flat discharging/charging plateau of Co/NrGO/S with a smaller value of plateau voltage difference (e.g., Co/NrGO/S ~ 0.15 V, NrGO/S ~ 0.23 V) confirms the relatively high electrode polarization of NrGO/S, as shown in Figure 4A.

At open-circuit voltage, the Nyquist plot of either Co/NrGO/S or NrGO/S is composed of a depressed semicircle within the high- and middle-high-frequency region and a straight line in the low-frequency region (Figure 4C). In general, the intercept of the horizontal axis is indicative of electrolyte solution resistance R_e, the diameter of the depressed semicircle represents the sum of surface-film resistance R_f and charge-transfer resistance R_ct, and the slope of the straight line relates to Warburg impedance Z_w for the diffusion of Li⁺ ions within the bulk electrode. Fitting the Nyquist plot of Co/NrGO/S or NrGO/S yields the total resistance (i.e., R_e + R_f + R_ct) of 74.6 (66.1 + 6.2 + 2.3) or 136.6 (104.2 + 30.2 + 2.2) Ω. This means that the reversible conversion kinetics of molecular S₈ to Li₂S₂/Li₂S is in inverse proportion to the electronic conductivity of a S-hosting Co/NrGO or NrGO (i.e., 32.36 or 4.48 S cm⁻¹).

According to the linear relationship of Z_re with angular frequency ω in the low-frequency region (i.e., Z_re = R_e + R_ct + σω^−0.5),^55,56 the Li⁺-ion diffusion coefficient D_Li within the bulk electrode can be estimated when the originally fictitious Li⁺-ion concentration in bulk Co/NrGO/S or NrGO/S is replaced by that of an electrolyte solution (1.0 mol L⁻¹). The Z_re - ω^−0.5 linear plots of Co/NrGO/S and NrGO/S are shown in Figure 4D, and the corresponding D_Li values are calculated to be 4.25 × 10⁻¹¹ and 2.82 × 10⁻¹¹ cm² s⁻¹, respectively. The Li⁺-ion diffusion of Co/NrGO/S is faster than that of NrGO/S, coinciding well with the former's superior electrochemical properties shown in Figure 4A–C.

At a current rate (C-rate) of 0.2 C, the Co/NrGO/S cathode with a S-loading of 1.2 mg cm⁻² presents an initial Coulombic efficiency (CE) of 95.3% and delivers ultrahigh discharge/charge capacities of 2154.8/1635.3, 1557.9/1508.8, and 1505.8/1483.2 mAh g⁻¹ in the 1st, 2nd, and 3rd cycles, respectively (Figure 5A). These, as well as the high values of residual capacity (1124.4 mAh g⁻¹) in the 100th cycle and of constant CE (99.8%) since the 7th cycle, indicate a seldom seen ultrahigh availability of elemental sulfur using the high-performance cell configuration (termed as Config. I) shown in Figure 1. Moreover, the initial discharge capacity (2154.8 mAh g⁻¹) is much higher than the theoretical value of the S cathode (1675 mAh g⁻¹), and to the best of our knowledge, this phenomenon has been attributed to the effect of an efficient porous carbon on the initially complete transformation of molecular S₈ into its fully lithiated Li₂S.^57,58

Figure 5. Cycling and/or rate performances of Co/NrGO/S cathodes: (A) with the different S-loading amounts at 0.2 C; (B) compared with NrGO/S at 0.5 C; (C, D) at various C-rates; and (E) long-term cycling stability at 0.5 C and then at 0.2 C

From Figure 5A, it can also be seen that even at a high mass loading of elemental S (i.e., 2.3 mg cm⁻²) and at 0.2 C, the discharge capacity of Co/NrGO/S is 1453.7, 1212.9, and 1168.9 mAh g⁻¹ in the 1st, 2nd, and 3rd cycle, respectively, and in the 100th cycle, the residual capacity is still as high as 903.0 mAh g⁻¹. Correspondingly, the CE values change from 89.4% in the 1st cycle to 96.9% in the 2nd cycle and then to a constant of 99.7% since the 8th cycle. Although the specific capacity of the high S-loading cathode in each cycle is smaller than that of the 1.2 mg cm⁻² S-loading electrode, the former's sulfur availability is also not poor. However, it should be emphasized that the high S-loading of 2.3 mg cm⁻² was conducted only once, and the S-loading in other working electrodes was fixed at 1.2 mg cm⁻².

In Figure 5B, at a relatively high C-rate of 0.5 C, the Co/NrGO/S cathode always shows higher sulfur availability and better cycling stability than NrGO/S cathode, and in the 100th discharge–charge cycle, their residual capacities are 950.0 and 582.0 mAh g⁻¹, respectively. Rate performances of Co/NrGO/S show high discharge capacities of 1290, 1180, 1089, 1029, and 835 mAh g⁻¹ at the C-rates of 0.2, 0.5, 1.0, 1.5, and 2 C, respectively (Figure 5C,D). As soon as the applied C-rate returns back from 2 to 0.2 C, the reversible capacity of this Co/NrGO/S cathode recovers to a high value of 1158 mAh g⁻¹ (Figure 5C). By contrast, the inferior rate performances of the control NrGO/S (Figure S6) may be attributed to a sluggish redox mechanism of intermediate Li₂S_n (3 ≤ n ≤8) because of the absence of metallic Co.

A long-term cycling performance of Co/NrGO/S (Figure 5E), recorded at 0.5 C for initial 200 cycles and then at 0.2 C for subsequent 50 cycles, shows an initial charge capacity of 1382.0 mAh g⁻¹, a 2nd-cycle discharge capacity of 1396.9 mAh g⁻¹, and an initial CE of 101.1%. Upon continuous cycling, the reversible capacity gradually decreased and then reached a value of 793.4 mAh g⁻¹ at 0.5 C in the 200th cycle. When the C-rates changed from 0.5 to 0.2 C, the discharge capacity recovered to 896.0 mAh g⁻¹ in the 201th cycle and then to 902.3 mAh g⁻¹ in the 250th cycle. By comparison, the corresponding CE value slightly fluctuated for the initial 13 cycles and then reached a constant of 99.7%. In a word, all these listed in Figures 4 and 5 imply a potential application of pristine Co/NrGO in the high-performance Li–S batteries by comparison with that of NrGO.

Multifunctional mechanism of Co/NrGO

By considering the reversible transformation of molecular S₈ into electrolyte-soluble Li₂S_n and then into electrolyte-insoluble Li₂S₂/Li₂S, using the multilayered Co/NrGO-coated Co/NrGO/S as a working electrode may simultaneously lead to the encapsulation of solid-state sulfur species and the entrapping of intermediate Li₂S_n within the inner part of the 3D mesoporous carbon matrix. Besides this, the following experimental designs of electrochemical kinetics were used to assay the catalytic function of the as-obtained composite Co/NrGO: absence of elemental S in working electrodes, presence of intermediate Li₂S_n in the electrolyte solution, and substitution of the Co-absent composite NrGO for Co/NrGO. Herein, intermediates Li₂S₈ and Li₂S₄ were chosen for the symmetrical cells (e.g., Co/NrGO || 1.0 M LiTFSI + 0.1 M Li₂S₈ || Co/NrGO) and the asymmetrical cells (e.g., Co/NrGO || 1.0 M LiTFSI + 0.05 M Li₂S₄ || Li), respectively.

At 5 mV s⁻¹, within the voltage range of −0.8 to 0.8 V, CV tests of NrGO- and Co/NrGO-based symmetrical cells showed no current signals, when the electrolyte-soluble intermediate Li₂S₈ was absent (Figure 6A). Under the same operation conditions, the presence of Li₂S₈ in the electrolyte of NrGO-based symmetrical cells induces the full-cycle CV curve to be an extremely narrow rectangle without redox peaks, indicating an electrochemically inert nature of NrGO. Also in Figure 6A, addition of Li₂S₈ to Co/NrGO-based symmetrical cells comparatively reveals two pairs of distinct redox peaks in a full-cycle CV curve, which agrees well with the results reported in the literature^7,38,59,60: the reduction of Li₂S₈ to Li₂S₆ (0.09 V) and then to Li₂S (−0.28 V) in the cathodic scan, and the oxidation of Li₂S to Li₂S₆ (0.28 V) and then to Li₂S₈ (−0.09 V) in the anodic scan. Importantly, this assures a catalytic activity of pristine Co/NrGO for the reversible conversion of Li₂S₈.

Figure 6. (A) CV profiles and (B) Tafel plots of symmetric Li2S8- and asymmetric Li2S4-solution cells. (C) Cycling performance of the general Co/NrGO electrode at 0.2 C. In each panel, Sy./Asy. is the abbreviation of symmetrical/asymmetric. (D) Schematic illustration of Co/NrGO to catalyzing the reversible transformation of S8 into Li2S

As for the LSV tests of NrGO-Li and Co/NrGO-Li asymmetrical cells operated at 0.2 mV s⁻¹ within the potential window from −50 to 50 mV, comparative Tafel plots are obtained in the presence of intermediate Li₂S₄. As can be seen from Figure 6B, the exchange/corrosion current densities of Co/NrGO and NrGO are estimated to be 3.03 and 1.08 mA cm⁻², respectively. Taking an electrochemically cycled cathode of Co/NrGO/S or NrGO/S into consideration, if the disappearance of either solid-state molecular S₈ or the fully lithiated product Li₂S is attributed to the corrosion process of a S-loading composite,^10,61,62 the much higher corrosion current density of S-absent Co/NrGO (3.03 mA cm⁻²) than that of S-absent NrGO (1.08 mA cm⁻²) can be associated with the former's superior catalytic ability for the enhanced reversible conversion kinetics of intermediate Li₂S₄.

When pristine composite Co/NrGO is directly used as a general cathode in the Li₂S₈/Li₂S₄ intermediate-absent Co/NrGO-Li asymmetrical cells, within the voltage range of 1.7–2.8 V, the electrochemical cycling (Figures 6C and S7) shows an initial discharge capacity of 14.4 mAh g⁻¹ at 0.2 C and a reversible constant of 4.6 mAh g⁻¹ from the 8th to the 200th cycle. Therefore, the slight lithiation of the S-free 3D mesoporous carbonaceous matrix (e.g., 14.4 mAh g⁻¹) may be omitted by comparison with the ultrahigh initial discharge/charge capacity of Co/NrGO/S (i.e., 2154.8/1635.3 mAh g⁻¹) shown in Figure 5A. As schematically illustrated in Figure 6D, the high-performance feature of the S-loading composite Co/NrGO/S mainly originates from the catalytic ability of the embedded metallic Co nanoparticles for the reversible transformation of S₈ into Li₂S.

When the separator coating of pristine Co/NrGO in the cell Config. I (Figure 1) is removed and then pasted onto the Co/NrGO-coated Co/NrGO/S cathode, and the resulting another cell configuration is termed as the Config. II. Also, the removal or rearrangement of Co/NrGO coatings results in three extra Co/NrGO/S-based Li–S cell configurations of Config. III, Config. IV, and Config. V. Furthermore, schematic drawings of these cell configurations and their galvanostatic cycling results are visually presented in Figures 7 and S8.

Figure 7. (A) Cycling performance of Co/NrGO/S in the Li-S cells of Configs. I and II at 0.2 C, and (B) their rate capabilities at various C-rates. (C, D) Schematic drawings of Configs. I and II for showing the different arrangements of Co/NrGO coatings

In the 2nd, 40th, and 80th cycles at 0.2 C, the discharge capacities of Co/NrGO/S in Config. I/II are obtained as 1557.9/1345.1, 1278.5/1054.0, and 1193.7/947.7 mAh g⁻¹, respectively (Figure 7A). For comparison the corresponding three discharge capacities of Config. III, IV and V are marked in Figure S8. That is, there is a decreasing trend of the specific capacity in each cycle (e.g., in the 40th cycle): 1278.5 mAh g⁻¹ (Config. I) > 1054.0 mAh g⁻¹ (Config. II) > 976.0 mAh g⁻¹ (Config. III) > 827.3 mAh g⁻¹ (Config. IV) > 676.6 mAh g⁻¹ (Config. V). In terms of exactly the same electrochemical reactions of elemental S toward metallic Li, the differences in the performance among these cell configurations highlight that aside from the specific nanostructured characteristics of pristine Co/NrGO as a S-hosting matrix, the use and permutation of Co/NrGO coatings are also of great importance in achieving high sulfur availability for Li–S batteries.

Rate performance of Co/NrGO/S in Configs. I and II comparatively show that at each C-rate, the former's reversible capacity in each cycle is always higher than the latter's, for example, the specific capacity of Config. I ~ 1089.1 mAh g⁻¹ and Config. II ~ 737.0 mAh g⁻¹ at 1.0 C (Figure 7B). If there is a spacing between the SA-based Co/NrGO coating of the cathode and the PVDF-based Co/NrGO coating of the separator (Figure 7B–D), the superior rate performance of Co/NrGO/S in the Li–S cell Config. I can be primarily attributed to the higher electron conductivity of the SA-based Co/NrGO coating than that of the close-packed bilayers of both PVDF- and SA-based Co/NrGO coatings. However, this is not in agreement with the results in the literature, showing that a porous carbon-based coating of the separator acts as a secondary current collector of the S-loading cathode.^27,34,52,54

Comparison of the cycling performances of Config. I and II cells (Figure 7A,C,D) with those of others (Figure S8) may imply that it is too difficult to distinguish the soluble Li₂S_n-entrapping ability of an SA- or PVDF-based Co/NrGO coating at different locations. Furthermore, considering the same Co/NrGO as the S-loading host and the coating material, it may be impossible to recognize or distinguish the catalytic effectiveness of Co nanoparticles embedded in each of these coatings.²⁷ Therefore, the high-performance mechanism of Co/NrGO/S-based Li–S batteries in Figure 5 is too comprehensive to be clarified by the nanostructured characteristics and/or multiple functions of pristine Co/NrGO. That is, even if the production cost of Co/NrGO is not included, it is still a long way to realize the commercialization of Co/NrGO/S-based Li–S batteries with a high sulfur availability.

CONCLUSION

Both the 3D rGO-based mesoporous composite (Co/NrGO), with embedded metallic Co nanoparticles and elemental N doping, and its grinding mixture with powdered S (Co/NrGO/S) were prepared and used as cathode-/separator-coated interlayers and working electrodes in assembled Li–S batteries, respectively. In comparison with the pristine composite of Co-absent NrGO, a high-performance mechanism of the Co/NrGO/S-based cell Config. I (i.e., Config. I) includes the following aspects of as-obtained Co/NrGO: (i) to host/encapsulate the re-cycling elemental S; (ii) to catalyze/promote the reversible conversion kinetics of intermediate Li₂S_n (3 ≤ n ≤ 8); (iii) to partly trap/immobilize the electrolyte-soluble Li₂S_n and then to restrain its shuttling between cathode and anode; (iv) to serve/act as the secondary current collector in working electrode. The S-availabilities of the Co/NrGO/S electrode in these cell configurations (Configs. I–V) comparatively show that both the synthetic strategy of the 3D mesoporous carbonaceous matrix Co/NrGO and its appropriate permutation in a cell configuration play an equally important role in the development of high-performance Li–S batteries.

ACKNOWLEDGMENTS

The authors are grateful for the financial support of the National Natural Science Foundation of China (21673131) and the Natural Science Foundation of Fujian Province (2019J01800).

CONFLICT OF INTEREST

The authors declare no conflict of interest.