Switchable encapsulation of polysulfides in the transition between sulfur and lithium sulfide

Encapsulation strategies are widely used for alleviating dissolution and diffusion of polysulfides, but they experience nonrecoverable structural failure arising from the repetitive severe volume change during lithium−sulfur battery cycling. Here we report a methodology to construct an electrochemically recoverable protective layer of polysulfides using an electrolyte additive. The additive nitrogen-doped carbon dots maintain their “dissolved” status in the electrolyte at the full charge state, and some of them function as active sites for lithium sulfide growth at the full discharge state. When polysulfides are present amid the transition between sulfur and lithium sulfide, nitrogen-doped carbon dots become highly reactive with polysulfides to form a solid and recoverable polysulfide-encapsulating layer. This design skilfully avoids structural failure and efficiently suppresses polysulfide shuttling. The sulfur cathode delivers a high reversible capacity of 891 mAh g−1 at 0.5 C with 99.5% coulombic efficiency and cycling stability up to 1000 cycles at 2 C.

W ith the rapid development of electronic devices and electric vehicles, the demand for high-energy-density batteries has increased significantly over the past few decades [1][2][3] . The lithium−sulfur (Li-S) battery is the most promising candidate to replace traditional lithium-ion batteries due to its high theoretical cathode capacity (1675 mAh g −1 ) and energy density (2500 Wh kg −1 ) 4,5 . Although significant improvements have been made in recent decades, serious challenges remain. The most notorious challenge is the rapid capacity decay and low coulombic efficiency that results when lithium polysulfides (LiPS) dissolve and diffuse in the electrolyte during the discharge/charge processes 6 .
Technology is often inspired by nature. It is well known that hemostasis is a host's defence mechanism to preserve the integrity of the closed high-pressure circulatory system when bleeding occurs. This complex process involves several steps: vasoconstriction, platelet activation, thrombus formation, and thrombolysis. Among them, thrombus formation and thrombolysis are self-healing and recoverable processes in the blood vessel [28][29][30] . As shown in Fig. 1a, when an injury occurs, the platelets in the blood are activated and start to form a platelet plug or thrombus, which can sufficiently seal the damaged blood vessel wall to restrict bleeding. When the injured blood vessel is completely healed, the signal of dissolution of the clot is automatically received in the blood vessel, and the body recovers homeostasis.
The recoverable processes of thrombus formation and thrombolysis in the blood vessel inspired us to explore a unique LiPS encapsulation strategy where a skilfully designed electrolyte additive (analogous to the platelet) can interact only with LiPS to form a recoverable protective layer in situ. Benefiting from the additive's existence within the electrolyte, the protective layer is skilfully formed at the electrode-electrolyte interface rather than tightly surrounding each single sulfur particle, as previously reported in most studies. In addition, since the additive otherwise maintains its dissolved state when the electrode presents sulfur, the in situ formed protective layer thus senses no structural damage caused by the volume change of the sulfur electrode during S-Li 2 S transitions. The LiPS "clotting factors" (LiPSCFs) additive should meet at least three criteria: good dispersibility in the electrolyte, strong interaction with LiPS, and excellent restraining ability of the recoverable protective layer for the subsequent dissolution of LiPS.
Here in the present study, nitrogen-doped carbon dots (N-CDs) are chosen as LiPSCFs due to their low cost, facile fabrication and excellent dispersibility in electrolyte. The recoverable protective clotting layer of LiPS/N-CDs is clearly visualized in Li-S batteries both ex situ and in situ and blocks the subsequent dissolution and diffusion of LiPS. More interestingly, both discharge and charge involve the formation and dissolution of a protective LiPS/N-CDs clotting layer, similar to the recoverable processes of thrombus formation and thrombolysis in blood vessels but with N-CDs functioning as activated platelets. As a result, this design efficiently avoids structural failure caused by the large volume change of the sulfur electrode and suppresses LiPS shuttling. Battery testing shows that the sulfur cathode can deliver a high reversible capacity of 891 mAh g −1 at 0.5 C with 99.5% coulombic efficiency and cycling stability up to 1000 cycles.

Results
The strong interaction between the N-CDs and electrolyte and the formation of the LiPS/N-CDs clotting layer. To study the optical properties of the as-prepared N-CDs, photoluminescence measurements for the N-CDs in aqueous solution were carried out using different excitation wavelengths ( Supplementary Fig. 1). Photoluminescence spectra of N-CDs show the strongest intensity at 380 nm excitation. As shown in Fig. 2a, the N-CDs have excellent dispersibility in the electrolyte of LiTFSI and LiNO 3 dissolved in the cosolvents of DOL/DME. Even at a high N-CDs content of 5 wt%, no solid precipitation can be found, demonstrating that the N-CDs satisfy the condition of good dispersibility as an ideal electrolyte additive. In contrast, the N-CDs are almost nondispersible/insoluble in the cosolvents of DOL/ DME ( Supplementary Fig. 2). This might be a result of the strong interaction between the N-CDs and lithium salt (LiTFSI and LiNO 3 ) due to the polar oxygen-and nitrogen-containing groups on the surface of the N-CDs (Supplementary Figs. 3, 4 and  Supplementary Tables 1, 2), which improves the dispersibility of the N-CDs in the cosolvents of DOL/DME. In this situation, it is desirable to obtain a protective clotting layer via the strong interaction between the N-CDs and LiPS to restrain the subsequent dissolution and diffusion of LiPS. To achieve this goal, Li 2 S 6 was dissolved in the cosolvents of DOL/DME and added isometrically into the electrolyte containing 0 and 0.3 wt% N-CDs. As shown in Fig. 2b, a LiPS/N-CDs clotting layer is generated immediately upon adding the Li 2 S 6 solution into the electrolyte with 0.3 wt% N-CDs. More interestingly, there was almost no color or volume change for the electrolyte containing 0.3 wt% N-CDs after 30 min, suggesting that the as-generated clotting layer can effectively prevent Li 2 S 6 diffusion into the electrolyte. Then, the mixture was violently shaken for 5 min. As shown in Supplementary Fig. 5, it can be seen that the LiPS/ N-CDs clotting layer was smashed, and the mixture retained the color of the Li 2 S 6 solution, demonstrating that not all Li 2 S 6 participated in the reaction with the N-CDs. In contrast, for the electrolyte without N-CDs, diffusion of the Li 2 S 6 into the electrolyte is obvious.
To further demonstrate the effect of N-CDs on LiPS clotting and the formation of the LiPS/N-CDs clotting layer, the electrodes (carbon paper) with 2 mg cm −2 sulfur loading were discharged at 0.5 mA in the electrolyte containing 0 and 0.3 wt% N-CDs, respectively ( Supplementary Fig. 6). During discharge, the electrolyte without N-CDs changed from colorless to bright yellow-green as a result of the dissolved LiPS. By comparison, the 0.3 wt% N-CDs electrolyte remained light brown throughout, suggesting that the generated sulfur species were effectively confined in the cathode.
Therefore, the clotting layer formation could be seen more clearly, and nickel foam with a sulfur loading of 10 mg cm −2 was employed as the sulfur cathode and assembled onto the optically transparent Li-S cell. Figure 2c shows visual changes of the electrolyte and sulfur cathode in the optically transparent Li-S cell with 0 and 0.3 wt% N-CDs electrolyte during discharge. For the cell without N-CDs, the color of the electrolyte appears reddish brown after 2 h of discharge, owing to the high concentration of long-chain LiPS. However, for the cell with 0.3 wt% N-CDs electrolyte, the color of the electrolyte after 2 h of discharge is nearly the same as that at the beginning. More importantly, the sulfur cathode is covered with a thin LiPS/N-CDs clotting layer after 1 h of discharge that becomes thicker as the discharge time increases, further confirming that N-CDs possess a significant clotting function for LiPS and that the    as-generated LiPS/N-CDs clotting layer suppresses LiPS dissolution and diffusion. Studies have shown that the LiPS concentration is highest at a potential of 2.38 V during the discharge/charge process, resulting in a maximum shuttle current 31,32 . Here, 2032 coin Li-S cells with sulfur loading of~2 mg cm −2 were assembled without adding LiNO 3 electrolyte, and the shuttle current was measured via 12 h potentiostatic testing at 2.38 V. As shown in Fig. 3a, the shuttle current of the cell with 0.5 wt% N-CDs is almost negligible (7 μA), whereas the cell without N-CDs has a 65 μA shuttle current. It is clear that N-CDs provide considerable clotting function for LiPS, significantly preventing the subsequent dissolution and diffusion of LiPS and leading to enhanced electrochemical performance. As shown in Fig. 3b, the cell without N-CDs delivered low discharge capacity (736 mAh g −1 ) and coulomb efficiency (52%) arising from the fatal shuttling effect of LiPS, whereas the cell with 0.5 wt% N-CDs exhibits superior discharge capacity (1042 mAh g −1 ) and high coulombic efficiency of 93%, which can be ascribed to the clotting effect of N-CDs, significantly blocking the shuttle effect. The cell with 0.5 wt% N-CDs was taken apart when the discharge voltage tapered to 2.1 V, and the surface structure of the sulfur cathode was analysed by scanning electron microscopy (SEM) and energy dispersive spectroscopy (EDS). The LiPS/N-CDs clotting layer covering the super P surface can be clearly observed (Fig. 3c).
The formation mechanism of the LiPS/N-CDs clotting layer and superior electrochemical performance of Li-S cells with 0.5 wt% N-CDs. As shown in Supplementary Fig. 7a-d, N-CDs with an average diameter of~5 nm are uniformly dispersed in the electrolyte. The high-resolution transmission electron microscopy image shows that the N-CDs exhibit the characteristic lattice fringes of 0.208 nm, corresponding to the (100) plane of graphite 33,34 . After being mixed with Li 2 S 6 , the N-CDs in the electrolyte agglomerated to form larger clusters of less than 20 nm in size, suggesting that the N-CDs, similar to platelets, can serve as LiPSCFs to form the LiPS/N-CDs clotting layer.
The chemical interaction between the N-CDs and Li 2 S 6 (as a model compound of LiPS) was studied using X-ray photoelectron spectroscopy. As observed in Supplementary Fig. 7e, the highresolution S 2p spectrum of Li 2 S 6 shows two sulfur 2p 3/2 peaks, at 161.4 and 163.0 eV, with an intensity ratio of 1:2, pointing to the terminal (S T −1 ) and bridging sulfur (S B 0 ) atoms, respectively 23 . After interacting with the N-CDs, the bonding energy of S T −1 shifts to 161.1 eV, suggesting an increased valence electron density on the sulfur atoms due to the interaction between the N-CDs and Li 2 S 6 35 . The new peak at 167.2 eV represents the binding energy of thiosulfate, which can be attributed to the surface redox reaction between chemisorbed oxygen on the N-CDs and Li 2 S 6 . Thiosulfate can serve as an internal mediator to anchor and convert long-chain LiPS 36 . Supplementary Fig. 8 presents the FTIR spectra of Li 2 S 6 , N-CDs, and the interaction system of N-CDs and Li 2 S 6 . For pristine Li 2 S 6 , the weak absorption peak at 482 cm −1 can be assigned to the typical stretching vibration of the S-S band. The peak position of the S-S band shifted to 503 cm −1 in the presence of N-CDs, indicating that the strong interaction between N-CDs and LiPS induces rotation and distortion of the S-S band of absorbed LiPS to form a stable configuration 37 . To evaluate the electrochemical performance of Li-S cells using N-CDs as the electrolyte additive, a series of electrochemical measurements was performed on the assembled 2032 coin cells with sulfur loading of~2 mg cm −2 . As shown in Fig. 4a, b and Supplementary Fig. 9, the cell with 0.5 wt% N-CDs electrolyte exhibits higher rate performances than the cells with 0, 0.3, and 1.0 wt% N-CDs electrolyte, especially when the current density is high. The delivered discharge capacity of the cell with 0.5 wt% N-CDs is retained at 1068, 814, and 701 mAh g −1 at a current density of 0.2, 0.5, and 1.0 C, respectively. Even at a high current density of 2.0 C, there is still a high reversible capacity of 547 mAh g −1 . When the current density returns to 0.2 C after 40 cycles, the average discharge capacity is as high as 920 mAh g −1 , demonstrating the beneficial clotting property of N-CDs for LiPS at high current rates. Supplementary Fig. 10 compares the galvanostatic discharge-charge profiles of Li-S cells with 0 and 0.5 wt% N-CDs electrolyte at 0.2 C. It can be seen that there are no obvious differences in the voltage profile, except for the higher specific capacity for the Li-S cell with 0.5 wt% N-CDs electrolyte. This may be ascribed to the fact that the LiPS/N-CDs clotting layer formed by the interaction between LiPS and N-CDs is just for addressing polysulfide dissolution and diffusion without changing the charge and discharge principles of Li-S batteries. As shown in Supplementary Fig. 11, two typical cathodic peaks at 2.33 and 1.98 V are ascribed to the reduction of sulfur to Li 2 S x (4 ≤ x ≤ 8) and Li 2 S 2 /Li 2 S, respectively. In the charge process, the overlapping anodic peaks are ascribed to the oxidation of Li 2 S 2 / Li 2 S to Li 2 S x and sulfur. Additionally, the cyclic voltammetry curves almost overlap with each other for the first five cycles, implying superior capacity retention capability and excellent cycling stability.
The long-term cycling performances of the cells with 0 and 0.5 wt% N-CDs were tested at rates of 0.5, 1.0, and 2.0 C, respectively. As shown in Fig. 4c, the cell with 0.5 wt% N-CDs delivers an initial discharge capacity of 891 mAh g −1 at 0.5 C. After 500 cycles, a capacity of 589 mAh g −1 is retained, corresponding to a low capacity decay rate of 0.067% per cycle with a coulombic efficiency of 99.5%. It is worth noting that at much higher rates of 1.0 and 2.0 C, discharge capacities of 447 mAh g −1 over 600 cycles and 348 mAh g −1 over 1000 cycles, respectively, are still retained (Fig. 4d and Supplementary Fig. 12). The capacity fluctuations are probably due to the volume fluctuations and redistribution of sulfur during the discharge/ charge processes 23 . In contrast, for the cell without N-CDs, the sulfur cathode suffers from much faster capacity decay (0.149% per cycle) even at a low C rate. Moreover, the cell with 6.6 mg cm −2 sulfur loading and 0.5 wt% N-CDs still delivers a high areal capacity of 5.8 mAh cm −2 and exhibits no obvious capacity decay after 200 cycles ( Supplementary Fig. 13).
To further verify the practicability of this strategy, a Li-S pouch cell with 0.5 wt% N-CDs electrolyte was assembled using carbonized cotton fiber foam (CFF)/Li 2 S 6 with a low specific surface area of 3.1 m 2 g −1 (Supplementary Fig. 14) as the cathode, which can reduce the weight ratio of electrolyte/active sulfur to 8:1 38 . As shown in Fig. 4e, a high area capacity of 6.0 mAh cm −2 was achieved. After 120 cycles, the cell can still deliver a quite stable and high areal capacity of 4.7 mAh cm −2 due to the N-CDs clotting function for LiPS. Based on the total weight of their components and the average working voltages (Supplementary  Table 3), the practical energy density of the Li−S pouch cell was calculated to be 177 Wh kg −1 .
Morphological changes in the sulfur cathode with 0 and 0.5 wt% N-CDs in the electrolyte during the first discharge/charge process were observed using scanning electron microscopy SEM. At 0.5 wt% N-CDs, the pristine sulfur cathode consists of Super P and sulfur agglomerations before discharge ( Supplementary  Fig. 15). When the discharge voltage is tapered to 2.1 V, the surface of the sulfur cathode becomes smooth and shiny compared to the pristine sulfur cathode (Fig. 5a), suggesting the formation of a clotting layer of LiPS/N-CDs. Proceeded to discharge to 2.0 V, the surface of the sulfur cathode is mostly covered by a clotting layer of LiPS/N-CDs (Fig. 5b). The elemental mapping images indicate that the clotting layer contains C, N (2.6 wt%), O, F, and S ( Supplementary Fig. 16a), implying the coexistence of N-CDs and LiPS in the clotting layer. For the fully discharged sulfur cathode (1.7 V), the LiPS/N-CDs clotting layer is nearly fully replaced by hierarchical porous flower-like Li 2 S crystals (Fig. 5c), which may facilitate rapid electron transfer and ion transport 39 .
A formation mechanism for hierarchical porous flower-like Li 2 S crystals is proposed as follows: the LiPS in the LiPS/N-CDs clotting layer is in situ reduced to Li 2 S, and the N-CDs function as the active sites for Li 2 S crystal growth, giving rise to the temporary disappearance of the clotting layers. This formation mechanism is confirmed by elemental mapping ( Supplementary  Fig. 16b, c): C and N elements of N-CDs are uniformly dispersed in the flower-like Li 2 S crystals, and the content of N element is as high as 2.2 wt%, indicating that a considerable amount of N-CDs was used as nucleation sites.
During charging, as shown in Fig. 5d, e, the hierarchical porous flower-like Li 2 S crystals are first oxidized to LiPS accompanied by the in situ formation of the LiPS/N-CDs clotting layer via the interaction between LiPS and N-CDs (2.4-2.3 V). For the fully charged sulfur cathode (2.8 V), almost all the LiPS is oxidized to sulfur, causing the disappearance of the LiPS/N-CDs clotting layers and the redispersion of N-CDs into the electrolyte (Fig. 5f). Moreover, there is no noticeable change in the surface morphology of the sulfur cathode even after 50 cycles ( Supplementary Fig. 17), suggesting that the introduction of N-CDs can not only restrain the dissolution and diffusion of LiPS but also facilitate the homogeneous electrochemical deposition of insulating sulfur, thus enhancing both the capacity and cycling performance. In contrast, for the sulfur cathode without N-CDs, there is serious dissolution and diffusion of LiPS throughout discharge and charge (Fig. 5g-l), resulting in the loss of active materials from the surface of the sulfur cathode and poor electrochemical performance.
To further clarify the formation mechanism of the LiPS/N-CDs clotting layer clotting and the existing state of the N-CDs during the different discharge/charge processes for the assembled 2032 coin cells, XPS spectra were obtained for the sulfur cathode of the Li-S cell with 0.5 wt% N-CDs electrolyte at different charge/ discharge states. Figure 6a 7 V), the LiPS in the LiPS/N-CDs clotting layer is in situ reduced to Li 2 S, and the N-CDs function as nucleation sites for Li 2 S crystal growth, giving rise to the temporary disappearance of the clotting layers. For the fully charged sulfur cathode (2.8 V), the Li 1s XPS spectra are nearly the same as that of the initial state of the sulfur cathode, implying the renewed dispersion of N-CDs into the electrolyte, which can be further confirmed by the XPS spectra of N 1s and O 1s. As shown in Fig. 6b, c, the peaks of pyridinic N (398.9 eV), pyrrolic N (399.7 eV) 40  Previous research has found that pyridinic N exhibits the strongest binding to LiPS among all kinds of doped N species 16,21 . However, the research does not provide a precise model for the simultaneous interaction of LiPS with pyridinic N and oxygen-containing groups simultaneously. It is therefore essential to theoretically understand the interaction of LiPS with both pyridinic N and oxygen-containing groups on the N-CDs using density functional theory.
Our research found that the configuration of Li 2 S 4 simultaneously interacting with C=O and pyridinic N is the most favorable structure for the formation of a strongly coupled clotting layer that preserves active materials and restrains the shuttle effect. Figure 7a-c and Supplementary Fig. 18 show the optimized structures and binding energies of Li 2 S 4 binding to oxygen-containing groups (-C=O, -COOH, C-OH, C-O-C) on the N-CDs and pyridinic N (doped at the center/edge of the N-CDs). Li 2 S 4 bridging simultaneously to C=O and pyridinic N sites shows the highest binding energy (3.694 eV) of the proposed structures.
This preferred configuration of Li 2 S 4 simultaneously interacting with C=O and pyridinic N is further confirmed by the differential charge density and the corresponding partial density of states (PDOS). As shown in Fig. 7d, a clear charge transfer from a Li atom to the edge of the N-CDs can be observed, indicating a strong interaction between Li 2 S 4 and the edge of the N-CDs, which could be mainly caused by the strong electronegativity of oxygen and nitrogen. It can be seen from the PDOS   On the basis of our experimental results and theoretical analysis, a possible LiPS clotting mechanism is proposed in Fig. 8. First, the LiPS generated in the process of discharge interact with the N-CDs in the electrolyte to form a LiPS/N-CD clotting layer at the surface of the sulfur cathode, which prevents the subsequent dissolution and diffusion of LiPS and thus solidifies the sulfur cathode. Even though the expansion of the sulfur cathode may crack the initial clotting layer during discharge, causing the diffusion of LiPS, the highly dispersed N-CDs can capture the escaped LiPS and heal the gap. Then, during the discharge process, the LiPS in the LiPS/N-CDs clotting layer are reduced in situ to Li 2 S. With the N-CDs serving as nucleation sites, the as-generated Li 2 S grows into hierarchical porous flower-like Li 2 S crystals, leading to the temporary disappearance of the clotting layers. During the charge process, the evolution of the protective layer is totally reversible. The discharge/charge process is analogous to thrombus formation and thrombolysis in the blood vessel, with N-CDs acting as activated platelets. The LiPS/N-CDs clotting layer can repeatedly generate and dissolve at the surface of the sulfur cathode during discharge/ charge, efficiently solidifying the intermediate LiPS and leading to enhanced electrochemical performance.

Discussion
Inspired by the recoverable processes of thrombus formation and thrombolysis in the blood vessel, we designed N-doped carbon dots as analogous "blood platelets" in Li-S battery electrolytes. Upon the formation of dissolvable LiPS, N-CDs become activated and stimulate the formation of a protective layer at the sulfur   electrode-electrolyte interface to suppress sulfur loss. This layer disappears when dissolvable LiPS are absent, as when the electrode is fully charged or discharged. The polysulfide-solidification capability of N-CDs is due to the abundance of N atoms and surface oxygen-containing groups, which strengthen the interaction of N-CDs with LiPS. As a result, high sulfur utilization and long-term cycling stability in Li-S batteries are achieved. We expect the encapsulation concept inspired by blood clotting to also be applied to battery electrochemistry beyond Li-S to prevent the loss of active materials.

Methods
Synthesis of N-CDs. The N-CDs were synthesized through rapid one-step pyrolysis of ethanolamine according to the method described in Dong et al. 33 . A mixture of 4.5 mL hydrogen peroxide aqueous solution (30%) and 3.0 mL ethanolamine in a 50 mL beaker was placed into an oven and calcined at 250°C for 10 min until dark colloidal N-CDs formed.
Preparation of sulfur cathode. The sulfur cathodes were fabricated using a slurry coating procedure. Sublimed sulfur, Super P, and polyvinylidene fluoride with a mass ratio of 70:20:10 were mixed into N-methyl-2-pyrrolidinone to form homogeneous slurry. The slurry was uniformly coated onto the carbon fibers and dried at 60°C under vacuum overnight, resulting in an areal sulfur loading of 2 mg cm −2 . Therefore, the formation of LiPS entrapments could be observed in situ, and a nickel foam electrode with a high areal sulfur loading of~10 mg cm −2 was prepared using the same procedure. In addition, electrodes with a high areal sulfur loading of 6.6 mg cm −2 for the coin cell and pouch cell were prepared by adding 0.5 M Li 2 S 6 in DOL/DME into the carbonized CFF 43 , followed by drying under vacuum overnight. The pouch cell was assembled by using 67.5 mg of S active material and 540 mg of 0.5 wt% N-CDs electrolyte.
Characterization. Transmission electron microscopy and SEM images of the samples were acquired using an FEI Tecnai G2 F30 STWIN operating at 200 kV and a JEOL 7800F, respectively. X-ray photoelectron spectroscopy analyses were performed using a Thermo Escalab 250 system with a monochromatic Al Kα source (hν = 1486.6 eV). Photoluminescent spectra were recorded on a fluorescent spectrophotometer (F-4600, Hitachi). Fourier transform infrared spectra were obtained on a Bruker VERTEC 80v vacuum FT-IR spectrometer. The Bruker-Emmett-Teller (BET) surface area was determined by ASAP 2020 micromeritics.
Electrochemical measurements. CR2032 coin cells were assembled in an argonfilled glove box (O 2 and H 2 O levels < 0.1 ppm) for the electrochemical measurements, where the as-prepared sulfur electrode, Celgard 2500 membrane, and Li foil were used as the cathode, separator, and anode, respectively. The electrolyte was 1.0 mol L −1 lithium bis (trifluoromethanesulfonyl) imide (LiTFSI) and 1.0% lithium nitrate (LiNO 3 ) dissolved in the cosolvents of 1,3-dioxolane (DOL) and 1,2-dimethoxyethane (DME) (v/v = 1:1) with N-CD additives of 0, 0.3, 0.5, 1.0, and 5.0 wt%. The ratio of added electrolyte volume/active sulfur weight was set at 20 μL mg −1 . Galvanostatic charge-discharge tests were carried out within the cutoff voltages of 1.7−2.8 V using a LAND CT2001A cell test instrument. The current density and the specific capacity were calculated according to the mass of elemental sulfur (1C = 1675 mA g −1 ). For shuttle current measurement, cells with a sulfur loading of~2 mg cm −2 were assembled without adding LiNO 3 . In the cycling process at 0.2 C, the shuttle current of cells was detected via potentiostatic hold on charge at 2.38 V. Cyclic voltammetry measurements were conducted on a Chenhua CHI-760 electrochemical workstation with a scan rate of 0.1 mV s −1 in the potential range of 1.7-2.8 V.
Computational method. Our first-principles calculations were based on density functional theory with generalized gradient approximation for exchangecorrelation potential given by Perdew et al. 44 as implemented in the Vienna ab initio simulation package 45 . The projector-augmented wave method 46 was used to treat the core electrons, and the valence electrons were represented using the planewave basis set. The planewave cut-off energy was set at 500 eV. The first Brillouin zone was sampled using a Γ-centred Monkhorst-Pack grid 47 . The convergence criteria for energy and force were set to 10 −5 eV and 0.02 eV/Å, respectively. To simulate the surface adsorption of large molecules, a slab system with a vacuum space of 20 Å along the z direction was adopted. To simulate edge adsorption, a nanoribbon system with a vacuum space of 20 Å along the y and z directions was adopted.

Data availability
All relevant data supporting the key findings of this study are available within the Article,