Manufacturing N,O-carboxymethyl chitosan-reduced graphene oxide under freeze-dying for performance improvement of Li-S battery

Lithium-sulfur (Li-S) batteries can provide far higher energy density than currently commercialized lithium ion batteries, but challenges remain before it they are used in practice. One of the challenges is the shuttle effect that originates from soluble intermediates, like lithium polysulfides. To address this issue, we report a novel laminar composite, N,O-carboxymethyl chitosan-reduced graphene oxide (CC-rGO), which is manufactured via the self-assembly of CC onto GO and subsequent reduction of GO under an extreme condition of 1 Pa and −50 °C. The synthesized laminar CC-rGO composite is mixed with acetylene black (AB) and coated on a commercial polypropylene (PP) membrane, resulting in a separator (CC-rGO/AB/PP) that can not only completely suppress the polysulfides penetration, but also can accelerate the lithium ion transportation, providing a Li-S battery with excellent cyclic stability and rate capability. As confirmed by theoretic simulations, this unique feature of CC-rGO is attributed to its strong repulsive interaction to polysulfide anions and its benefit for fast lithium ion transportation through the paths paved by the heteroatoms in CC.

These authors contributed equally to this work. * Authors to whom any correspondence should be addressed.
Original content from this work may be used under the terms of the Creative Commons Attribution 4.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.

Introduction
The demands for rechargeable batteries with high energy density are increasing due to rapidly upgrading electrical devices [1][2][3][4][5]. Currently commercialized lithium ion batteries that provide an energy density of less than 200 Wh·kg −1 can not meet these demands [6][7][8][9][10]. Lithium-sulfur (Li-S) batteries have a theoretical energy density of 2600 Wh·kg −1 , together with the obvious advantage of abundant and nontoxic sulfur, providing an ideal solution to the energy density limitation of lithium ion batteries, and thus have been attracting much attention [11][12][13][14][15]. However, challenges remain before Li-S batteries can be applied in practice [16][17][18]. The main issue is the soluble lithium polysulfides (LiPSs) that cause the shuttle effect due to their penetration through the separator [1,19,20].
Correspondingly, modifying a separator with coating materials that can block the penetration of LiPSs has been adopted to suppress the shuttle effect of LiPSs [42][43][44][45][46]. These efforts seem effective in blocking the penetration of LiPSs [47][48][49], but the introduction of other materials in the separator might also block the lithium ion transportation [50]. Therefore, it is necessary to search for the coated materials that can not only suppress LiPS penetration, but also can favor the lithium ion transportation for the Li-S battery performance improvement.
rGO has been widely used to improve the electronic conductivity of the sulfur cathode and the utilization of sulfur, and is also considered as the ideal coated material for modifying the separator of Li-S batteries [51][52][53], but the folded and compact multi-layer rGO sheets usually a exhibit small surface area and are not beneficial for lithium ion transportation. It has been known that inserting some large molecules with heteroatoms into the interval between the two-dimension rGO sheets can unfold rGO [54,55]. The introduced heteroatoms provide active sites not only for repelling polysulfides [56] but also for transferring lithium ions [50]. However, the syntheses of rGO from graphene oxide (GO) usually involve a hydrothermal reaction [45], calcining [57] and chemical reduction [58]. These methods are complicated and therefore simpler methods need to be developed for the application of rGO in practice.
The formation mechanism of rGO from GO reduction involves the removal of oxygen-containing groups in GO. This process can also be achieved by the reaction of oxygencontaining groups with -NH 2 groups in other compounds, which is accompanied by removing H 2 O molecules [59][60][61]. Under extreme conditions, for example, under 646.5 Pa and 0 • C, H 2 O molecules can be easily removed due to its formation in a gaseous state [62,63]. Based on this knowledge, we propose a novel formation technology of rGO from GO, and manufacture a laminar N,O-carboxymethyl chitosan-rGO composite (CC-rGO) to modify the separator of the Li-S battery. CC-rGO is manufactured via the self-assembly of CC onto GO and the subsequent reduction of GO under freezedrying, which has been confirmed by physical characterization. The as-synthesized CC-rGO is coated on a commercial polypropylene (PP) membrane, resulting in a modified separator that can completely suppress the penetration of polysulfides and also accelerate the lithium ion transportation. Consequently, the Li-S battery using this modified separator exhibits a high initial specific capacity of 948.7 mAh·g −1 and maintains a specific capacity of 677.9 mAh·g −1 after 500 cycles at 0.5 C (0.75 mg·cm −2 ). These excellent performances are attributed to the unique combination of CC and rGO. The introduction of CC unfolds the stacked graphene through the combination of -NH 2 groups in CC with -COOH groups in GO, increasing the surface area of rGO and yielding a shield to block the penetration of polysulfides when the composite is coated on PP. The CC in CC-rGO is rich in electronegative N and O heteroatoms that are able to repel the electronegative polysulfide anions and favor lithium ion transportation [64][65][66][67]. Additionally, the intervals among rGO nanosheets formed by inserted CC molecules provide channels for lithium ion transportation. The formation of rGO does not need any other chemicals, while CC is low-cost, easily accessible and environmentally friendly [68]. Therefore, our composite provides an effective and eco solution to the shuttle effect of LiPSs and will help apply Li-S batteries in practice [69].

Composite manufacturing and cell assemblies
The GO and CC aqueous solutions were prepared as described in our previous work [70]. The concentrations used in this work were 5.0 g·l −1 and 10.0 g·l −1 for GO and CC solutions, respectively. The CC-rGO composites were synthesized under freeze-drying in a lyophilizer (SCIENTZ-10N, China). Firstly, CC and GO solutions were mixed with the volume ratios of 8:1, 4:1, 1:1, 1:4, and 1:8 under stirring. Then, the mixed solutions were freeze-dried at 1 Pa and −50 • C until all the water was sublimated to obtain products labeled as CC-rGO 81 , CC-rGO 41 , CC-rGO 11 , CC-rGO 14 , and CC-rGO 18 , respectively.
About 60 wt% CC-rGO was mixed with 20 wt% AB and 20 wt% PVDF in N-methyl-2-pyrrolidone (NMP) under stirring in a blender (THINKY MIXER ARE-310, Japan) for 0.5 h to produce a uniform slurry. AB was used because it is an electronic conductor and is beneficial for the transformation of the deposited LiPSs on the separator. The resultant slurries were coated on one side of a commercial PP membrane and dried at 60 • C in vacuum oven for 12 h to obtain modified separators labeled as CC-rGO 81 /AB/PP, CC-rGO 41 /AB/PP, CC-rGO 11 /AB/PP, CC-rGO 14 /AB/PP, and CC-rGO 18 /AB/PP, respectively. The loaded mass of CC-rGO and AB were controlled to ∼0. 22 and ∼0.073 mg·cm −2. For comparison, the AB/PP separator without CC-rGO was also prepared in the same process as CC-rGO/AB/PP separators. About 10 wt% PVDF was dissolved in NMP to obtain a clear and transparent solution, then mixing with a content of 60 wt% S 8 and 30 wt% AB under grinding for 0.5 h to prepare a homogeneous slurry that was coated on charcoal-coated Al foil and dried at 60 • C in vacuum oven for over 12 h, to prepare sulfur cathodes with sulfur and AB loading mass of ∼0.75 and ∼0.225 mg·cm −2 , respectively. The obtained cathode was tailored to wafers that were assembled into CR2025 Li-S coin batteries together with a Li anode, an electrolyte and the above obtained separator in an Ar-filled glove box. The electrolyte was 1 M bis(trifluoromethane) sulfonimide lithium salt (LiTFSI) and 2 wt% LiNO 3 dissolved in a mixed solvent of dimethoxy ethane (DME) and 1,3-dioxolane (DOL) (v/v, 1:1), and the ratio of electrolyte to sulfur was controlled at ∼25 µl·mg −1 .

Characterization, measurement and calculation
The blocking capacity of the samples to polysulfides were characterized in a 10 mM Li 2 S 6 solution that was prepared by dissolving S 8 and Li 2 S at a molar ratio of 5:1 into a mixture of DME and DOL (v/v, 1:1) [71]. To evaluate the blocking capacity of the separators to LiPSs, the mixture of DME and DOL with and without Li 2 S 6 was added into two sides of an H-type glass cell, respectively, and the color change of the side without Li 2 S 6 was observed.
Field emission scanning electron microscopy (FEISEM, Quanta 250 FEG, America) and field emission transmission electron microscopy (FEITEM, Talos F200X, America) were used to observe the morphology of samples. X-ray diffraction (XRD, Ultima IV x-ray diffractometer, Japan) with Cu Kα radiation scanning from 10 • to 80 • (2θ) at 10 • ·min −1 . A confocal Raman microscope (Raman, WITec alpha 300 R, Germany) at an excitation laser of 532 nm was used to confirm the structure of samples. X-ray photoelectron spectroscopy (XPS, Kratos Axis Ultra Dld, Japan) and Fourier transform infrared spectroscopy (FTIR, Tensor 27, Germany) were used to analyze the composition of samples. Brunauer-Emmett-Teller (BET) specific surface area and pore size distribution of the samples were analyzed by nitrogen adsorption and desorption in a Gold APP Instruments (V-Sorb 2800P, China). All the samples were degassed at 120 • C for 4 h before measuring. The contact angles were measured with a contact angle tester (JC2000C, China) by dropping the electrolyte onto separators. The electrokinetic characteristics of samples were determined by a nanoparticle analyzer (SZ-100-HORIBA, Japan), and the surface zeta potentials of separators were measured by an electrokinetic analyzer (SurPASS™ 3, Anton Paar GmbH, Austria). The surface charge properties of separators were analyzed by measuring their zeta potentials in 0.001 mol·l −1 KCl aqueous solutions with pHs ranging from 1 to 6. And their surface zeta potentials were obtained according to the Helmholtz-Smoluchowski equation with the Fairbrother and Mastin substitution [72].
A Land cell test system (Land CT2001A, China), a Solartron analytical-1470E CellTest system (AMETEK, America), and a Metrohm Autolab PGSTAT302N (AUT84217, Netherlands) were used to determine the electrochemical properties of the samples. The charge-discharge performances of the batteries were performed in a voltage range from 1.7 to 2.8 V (vs. Li/Li + ). The cyclic voltammetry (CV) curves were obtained at various scan rates between 1.7 V and 2.8 V. The electrochemical impedance spectra (EIS) was obtained at a frequency range from 0.01 to 100 000 Hz and fitted by using Zview electrochemical impedance software.
Theoretical calculations were carried out using a Vienna ab initio simulation package (VASP) based on density functional theory (DFT) [73,74]. The projector augmented wave method was used to describe the influence of the core electrons, and the Perdew-Burke-Ernzerhof exchange correlation function with the DFT-D3 correction was introduced to the generalized gradient approximation [75,76]. An amidated rGO model was chosen to simplify CC-rGO, with only one formamide group in a 4 × 4 supercell. The thickness of the vacuum layer was set to 30 Å for all the systems. In each calculation, a Γ-centered K-point mesh with a resolved value of 0.06 π Å −1 was adopted to sample the Brillouin zone, and the cut-off energy for the plane wave expansion was set to 500 eV. A climbing-image nudged elastic band (CI-NEB) method was used to obtain the path with optimal energy for the lithium ion transportation [77]. In this process, three and six points were interpolated between the start and end points of the lithium ion transportation path on the surface of rGO and CC-rGO or across CC-rGO, and a fast inertial relaxation engine algorithm was used to optimize these points to obtain the energy profiles [78]. The peak value of the energy profiles was defined as the energy barrier for a lithium ion to overcome.
The interaction of CC-rGO with Li 2 S 8 was calculated based on the 4 √ 3 × 4 √ 3 superbatteries, while that of Li 2 S x (x = 1, 2, 4, and 6), and the transportation of a lithium ion on CC-rGO was obtained with the 4 × 4 superbatteries. All the relaxation calculations were finished until the maximum force was less than 0.02 Å −1 and the total energy was converged within 2 × 10 −7 eV, while the energy convergence standard of CI-NEB calculations was 1 × 10 −7 eV. In this calculation process, the interaction energy (E a ) is defined as the energy difference between the system (E CC-rGO/Li2Sx (x = 1, 2, 4, 6, 8) ) and the summation of pure Li 2 2,4,6,8) ) and the substrate (E CC-rGO ) and calculated by the equation (1): (1)

Physical and chemical characteristics of CC-rGO composite
The syntheses process of laminar CC-rGO composite under freeze-drying is illustrated in figure 1(a). When the GO and CC aqueous solution were mixed, the disordered GO sheets were self-assembled with CC molecules due to the electrostatic interaction between -COOH groups in GO and -NH 2 groups in CC. Moreover, the acidic -COOH and alkaline -NH 2 groups can be partly polymerized to form amide bonds and free H 2 O molecules. This polycondensation is too slow to proceed under room temperature, which can be accelerated via the removal of H 2 O molecules in gaseous state under extreme conditions. Under our freeze-drying conditions (1 Pa, −50 • C), GO can be reduced to rGO due to the quick removal of H 2 O molecules in a gaseous state and thus CC-rGO composite is successfully manufactured. The stacked GO was unfolded due to the entering of CC into the interval of the stacked GO, resulting in thin CC-rGO nanosheets, which can be also connected together horizontally with CC via self-assembling and enlarging the scale of the nanosheets (figure 2). This selfassembling is accompanied by the reduction of GO to rGO, resulting in CC-rGO. Compared with the currently available syntheses of rGO from GO, which need additional chemicals, our freeze-drying is simple and environmentally-friendly, and therefore provides an eco-way to prepare a thin and large rGO. To our knowledge, the formation of rGO from GO reduction under freeze-drying has never been reported in literature. As presented in figures 1(f) and 2(b), the CC-rGO 11 presents laminar structure with a much larger unfolded surface area than a GO sheet (figures 1(b) and 2(a)), which is beneficial for blocking polysulfides. In the high-magnification FEISEM images, the self-assembled laminar CC-rGO 11 can be observed, with a hairy layer of CC adhered to rGO surface ( figure 1(g)). The FEITEM images confirm the morphology of the GO sheet (figures 1(d) and (e)) and CC-rGO 11 (figures 1(h) and (i)), with CC-rGO being more thicker that the GO sheet (about 96.7 nm). The FEITEM elemental mappings show the uniform dispersion of C (87.1%), O (12.4%), and N (0.5%) elements on CC-rGO 11 (figures 1(j)-(l)), indicating that CC molecules have been uniformly assembled onto the rGO network. Since the N element is provided by CC, the presence of N in CC-rGO 11 confirms the self-assembling between CC and rGO.
In figure 3(a), the diffraction peak of GO sample locates at 11 • is characteristic of the crystal structure of GO [79]. As the CC is assembled onto GO, the resultant CC-rGO 11 shows the characteristic peaks of rGO located at 22.8 • [80], confirming the formation of rGO from GO due to the reaction of the -COOH groups in GO with -NH 2 groups in CC, which forms amide bonds that is accompanied with water removal under freeze-drying conditions (1 Pa, −50 • C).
The Raman spectra of GO and CC-rGO in figure 3(b) shows the two peaks located at 1350 cm −1 and 1581 cm −1 , which are typical D and G bands of carbon, respectively [81]. The peak intensity ratio of (I D /I G ) reflects the relative content of active sites in materials [82]. The peak intensity ratio of CC-rGO 11 (1.0) is much higher than GO (0.81), suggesting that more active sites have been produced in GO after introducing CC. These active sites correspond to the N and O heteroatoms, which are helpful for repelling polysulfide anions and accelerating the lithium ion transportation [83]. Figure 3(c) shows the FTIR spectra of CC, GO, and CC-rGO 11 . The absorption peak at 1602 cm −1 corresponding to N-H stretching can be detected in both the CC-rGO 11 and CC, confirming that CC has been successfully assembled onto rGO. The peak at 1733 cm −1 in GO can be attributed to the C=O stretching from -COOH group, while that in CC-rGO 11 can be ascribed to the C=O stretching from amide bonds formed from the reaction of the -COOH groups in GO with -NH 2 groups in CC. In addition, the C-H stretching peak at 2916 cm −1 and N-H stretching peak at 3413 cm −1 in CC can be still detected in CC-rGO 11 . These results confirm that CC has been successfully assembled onto rGO via the formation of amide bonds.
XPS was used to characterize the composition of the CC-rGO 11 . The obtained results are presented in figures 3(d)-(f) and 4. Three obvious peaks of C1s at 284.9, 286.4, 287.8, and 290.7 eV can be observed, corresponding to C=C, N-C, -C=O/N-C=O, and O-C, respectively ( figure 3(d)). These species are from rGO, amido bonds, and the small amount of -COOH in rGO [52]. Meanwhile, two strong peaks of N1s at 399.3 and 401.3 eV, corresponding to N-C and N-H, respectively, can be identified (figure 3(e)), which are attributed to amide bonds in CC-rGO 11 [84]. Besides, the O1s spectrum was fitted with three peaks at 531.1, 532.6, and 532.7 eV, which can be ascribed to C=O from N-C=O, C-O and C=O from -COOH, respectively [85]. These results demonstrate that rGO and CC have been combined together via the formation of amide bonds from the interaction of -COOH groups in GO with -NH 2 groups in CC, yielding stable laminar CC-rGO with abundant N-containing and O-containing active sites for suppressing polysulfide penetration and accelerating lithium ion transportation.
Coating CC-rGO with AB on a PP membrane results in a CC-rGO/AB/PP separator. The performances of the resultant separators are compared with those of the PP and AB/PP. The cross-section thickness of the CC-rGO 11 /AB/PP separator is about 12 µm (figure 5). The PP separator has a porous structure and presents a large BET surface area ( figure 6). These pores are blocked and the surface area of PP is reduced when AB or CC-rGO 11 is coated on PP. Since AB has a larger surface area than CC-rGO 11 (figure 7), AB/PP presents a larger BET surface area than CC-rGO 11 /AB/PP ( figure 6). Although the CC-rGO 11 /AB/PP separator has a smaller surface area, it exhibits a much better wettability than PP and AB/PP separators ( figure 8). Obviously, CC-rGO 11 /AB/PP is more compatible to the electrolyte, which should be related to its surface with abundant N-containing and O-containing        active sites which are important for battery performance [86]. However, when the CC content in CC-rGO is higher, the electronic conductivity of CC-rGO will be deteriorated. As shown in figure 9, the CC-rGO 11 /AB/PP separator presents the best charge/discharge performances. Although the increase of CC contents in CC-rGO is beneficial for the cyclic stability improvement, it reduces the specific capacity of the Li-S batteries, which might be ascribed to the reduced ionic conductivity of the separators. Therefore, the separator with the CC-rGO 11 composite manufactured with a volume ratio 1:1 of CC to GO solutions was considered for further investigations. Figure 10(a) compares the cyclic stability of Li-S batteries based on a sulfur loading mass of 0.75 mg·cm −2 and using different separators at 0.5 C, with their selected charge/discharge curves presented in figure 11. As shown in figures 11(c)-(e), the voltage plateaus at 2.3 V and 2.1 V in the discharge curves are ascribed to the transformation of S 8 into LiPSs (Li 2 S n , 4 ⩽ n ⩽ 8) and then to Li 2 S 2 /Li 2 S, respectively. The Li-S battery using the CC-rGO 11 /AB/PP separator exhibits an initial specific capacity of 948.7 mAh·g −1 , retaining a capacity of 677.9 mAh·g −1 with an average coulombic efficiency of 99.9% after 500 cycles. Comparatively, the Li-S batteries using PP and AB/PP separators deliver an initial capacity of only 749.2 and 799.3 mAh·g −1 , and retain a capacity of only 142.4 and 378.6 mAh·g −1 with an average coulombic efficiency of 89.6% and 99.3%, respectively. These comparisons suggest that the PP separator is poor in the suppression of the shuttle effect of LiPSs, which can be improved to some extent by coating AB and significantly improved by coating CC-rGO 11 /AB. Particularly, when the sulfur loading mass is increased to 3.41 mg·cm −2 (figures 12(a) and (b)) and 10 mg·cm −2 (figures 12(c) and (d)), the Li-S battery using CC-rGO 11 /AB/PP separator still run well (figure 12), which manifests the significance for the application of the CC-rGO 11 /AB/PP separator in practice.

Improved performances of CC-rGO modified separator
The Li-S battery using a CC-rGO 11 /AB/PP separator also exhibits excellent rate capacity (figures 10(b) and 11(b), (d), (f)), delivering stable specific capacities of 1158.5, 921.1, 804.9, 708.6, and 648.1 mAh·g −1 at 0.2, 0.5, 1, 2, and 3 C, respectively. On the contrary, both the Li-S batteries using PP and AB/PP separators present a sharp reduction in specific capacity with the increase of current rates. When the current rate is increased to 3 C, the specific capacity of the Li-S batteries using PP and AB/PP separators are only 190.9 and 158.4 mAh·g −1 , respectively. Especially, the discharge voltage plateaus disappear for the Li-S batteries using PP and AB/PP separators at the current rates higher than 1 C, but remain for that using a CC-rGO 11 /AB/PP separator ( figure 11). Obviously, the separator modified CC-rGO 11 significantly improved not only cycling stability but also the rate capability of Li-S battery, indicating that the CC-rGO plays an important role for the utilization of sulfur and the suppression of the shuttle effect of LiPSs [87,88]. Compared with the polymer-coated separators reported in literatures (table 1), our CC-rGO 11 /AB/PP separator provides an Li-S battery with comparable electrochemical performances, featuring a facile syntheses process without using toxic organic solvents.
The improved cycling stability and rate capability of Li-S battery CC-rGO 11 /AB/PP separator using can be confirmed by the CV curves in figures 10(c)-(e) and 13. It can be found from these CV curves that there appear two couples of oxidation and reduction peaks. The reduction peaks located at around 2.30 V and 2.02 V are ascribed to the transformation of S 8 into LiPSs (Li 2 S n , 4 ⩽ n ⩽ 8), and then into Li 2 S 2 /Li 2 S, corresponding to two discharge voltage plateaus in figure 11, respectively [98]. Compared with the Li-S batteries using PP and AB/PP separators, the Li-S battery using a CC-rGO 11 /AB/PP separator exhibits the largest peak currents and less current change during initial five cycles, indicating the excellent cyclic stability and rate capability of the battery using CC-rGO 11 /AB/PP separator. On the other hand, the oxidation peak current for the transformation of LiPSs (Li 2 S n , 4 ⩽ n ⩽ 8) is significantly higher than that for the transformation of Li 2 S 2 /Li 2 S, suggesting that more soluble LiPSs (Li 2 S n , 4 ⩽ n ⩽ 8) has been transformed into S 8 and that polysulfide anions are blocked by the separator, rather than migrating to Li anode.
CV is often used to understand electrochemical processes of sulfur cathode, which involves the lithium ion diffusion in solution and follows the Randles-Sevcik equation that describes the current-scan rate relationship for a freelydiffusing species in solution [19]. Figure 13 presents the CV curves of Li-S batteries using different separators at various scan rates (figures 13(a), (c) and (e)) and the corresponding relations of peak currents with the square root of scan rates (figures 13(b), (d) and (f)). The peak currents, A1/C2 for the transformations between S 8 and LiPSs and A2/C1 for those between LiPSs and Li 2 S 2 /Li 2 S, were extracted by subtracting the background current, which follows the Randles-Sevcik equation. The larger absolute slope in the plot of the peak current (I p ) to the square root of the scan rate (v 1/2 ) represents fast lithium ion diffusion speed [99]. As shown in figures 13(b), (d) and (f), the cathode using a   CC-rGO 11 /AB/PP separator exhibits much faster lithium ion diffusion speed than these using PP and AB/PP separators, confirming that the CC-rGO 11 /AB/PP separator is beneficial for lithium ion transportation for an Li-S battery. This benefit can be ascribed to the improved Li + transference number (t Li + ) of the CC-rGO 11 /AB/PP separator. The t Li + of the CC-rGO 11 /AB/PP separator is 0.51, which is much higher than that of PP (0.26) and AB/PP (0.36) separators (figure 14  and table 2). Consequently, the CC-rGO 11 /AB/PP separator shows much higher ionic conductivity (0.33 mS·cm −1 ) than the PP (0.18 mS·cm −1 ) and AB/PP (0.26 mS·cm −1 ) separators ( figure 15 and table 3). In addition, the Li-S battery using CC-rGO 11 /AB/PP separator presents a A2 peak current obviously higher than its A1 peak current, illustrating the remarkable oxidation conversion from Li 2 S 4 into longchain Li 2 S 6 /Li 2 S 8 and the effective suppression of LiPSs penetration.
To confirm the contribution of CC-rGO to the suppression of polysulfide anions and the acceleration of lithium ion transportation, more physical and electrochemical characterizations combining with theoretical calculations were performed. The surface zeta potentials of the CC-rGO 11 and CC-rGO 11 /AB/PP separators were measured and the obtained results are presented in figures 16 and 17. The CC-rGO 11 carries negative charges, which has not been changed after it is coated on PP (figure 17) in the solutions with pH less 3. Compared with the AB/PP separator (IEP AB/PP = 3.53 mV), the CC-rGO 11 /AB/PP separator exhibits stronger electronegativity (IEP CC-rGO11/AB/PP = 2.95 mV), suggesting that the CC-rGO 11 /AB/PP separator tends to combine species with a positive charge and to repel species with a negative charge, which are responsible for the suppression of polysulfide anions penetration and the acceleration of lithium ion transportation.
To further confirm the suppression of polysulfide anions by CC-rGO 11 /AB/PP separator, the Li anodes and separators taken from the cycled Li-S batteries using different separators were performed with FEISEM, FEITEM and XPS characterizations. As revealed in figure 18 where, I initial and I steady represent initial and steady current recorded during potentiostatic polarization, R initial and R steady represent the cell resistances before and after polarization, respectively. The cell resistance is the sum of Rct, R f and R b representing charge transfer, film and Omic resistances, respectively.
identified in the CC-rGO 11 at the charged state ( figure 21(d)), confirming that the polysulfides blocked by CC-rGO 11 can be transformed into Li 2 S reversibly, which ensures the high capacity retention of the sulfur cathode.
The contribution of CC-rGO 11 coating can be further verified by XPS spectra and Raman spectra of the CC-rGO 11 /AB/PP separator after 25 cycles at 0.5 C. Figure 22 presents the XPS results. The relative magnitude and shape of the C1s, N1s and O1s peaks for the cycled CC-rGO 11 /AB/PP separator are different from those for the fresh CC-rGO 11 , confirming that the chemical interaction existing between CC-rGO 11 composite and LiPSs happens after cycling. On the other hand, the peak locations of C1s (C=C, N-C), N1s and O1s keep unchanged after cycling, except for the C1s (-C=O/N-C=O, O-C) that are related to deposition of LiPSs, confirming that the structure of CC-rGO 11 keeps stable. The obtained Raman results are presented in figure 23(a). The I D /I G value (0.90) of CC-rGO 11 at the fully discharged state   is smaller than that (0.99) at the fully charged state, suggesting that partial active sites of CC-rGO 11 have been occupied by LiPSs during discharging and can be recovered during charging. Similarly, the S2p XPS spectra in figure 23(b) shows that the CC-rGO 11 /AB/PP separator at the fully discharged state presents much stronger peaks than that at the fully charged state, especially the Li-S peak, signifying the reversible transformation between LiPSs and Li 2 S. The interactions between CC-rGO and LiPSs can be further confirmed by DFT calculation. The calculated interaction energy between Li 2 S x (x = 1, 2, 4, 6, 8) and CC-rGO     one side to the other side of the battery. In contrast, the battery using CC-rGO 11 /AB/PP (figures 25(g)-(i)) separator keeps its electrolyte color unchanged, demonstrating that CC-rGO 11 /AB/PP separator can completely suppress the polysulfide penetration.
To understand the mechanism for the CC-rGO/AB/PP separator to accelerate lithium ion transportation, the energy profiles for lithium ion transporting on and across the CC-rGO were calculated with VASP. The obtained results are presented in figures 26 and 27, where simplified models for calculation are included. It can be found that the energy barrier for lithium ion to cross CC-rGO (7.04 eV) is much higher than those for rGO (0.30 eV) and CC-rGO (0.29 eV) surface, suggesting that it is easier for lithium ion to transport through the surface of the laminar CC-rGO, which should be ascribed to the active sites of the heteroatoms on the CC-rGO surface [101]. The CC not only functions as a binder to connect rGO sheets in two dimensions, forming a shield to block the polysulfide penetration, but also provides abundant active sites with its negative heteroatoms ′ lone pair electrons for repelling negative polysulfides and accelerating lithium ion transportation, leading to the excellent cycling stability and rate capability of a Li-S battery.
It should be mentioned that the application of our CC-rGO 11 /AB/PP separator is accomplished with a high sulfur loading and a thin layer of polysulfides barrier (12 µm). The former is higher and the latter is thinner than the commercial viability of polysulfide-retaining barriers, in which sulfur       loading and polysulfides-barrier thickness are recommended to be 2-3 mg·cm −2 and 40-60 µm, respectively [69].

Conclusions
In summary, we have successfully manufactured a laminar polymer-rGO composite that can efficiently suppress the shuttle effect of LiPSs resulting in a performance improvement of a Li-S battery. CC is self-assembled on GO, which is accompanied by the formation of rGO from GO reduction under an extreme condition of 1 Pa and −50 • C, resulting in the laminar CC-rGO composite. When a conventional PP membrane is coated with CC-rGO and used as a separator, the Li-S battery presents excellent cycling stability and rate capability. This feature is attributed to different contributions from CC and rGO. Firstly, the CC functions as a binder to connect rGO nanosheets in two dimensions, forming a shield to block the polysulfide penetration. Secondly, the CC provides abundant active sites by its heteroatoms with lone pair electrons for repelling polysulfides and accelerating lithium ion transportation. The formation process of rGO is environmentally friendly and therefore the manufactured laminar CC-rGO composite is promising to be applied at large scale for highperformance Li-S batteries.