Boosting potassium-ion batteries by few-layered composite anodes prepared via solution-triggered one-step shear exfoliation

Earth-abundant potassium is a promising alternative to lithium in rechargeable batteries, but a pivotal limitation of potassium-ion batteries is their relatively low capacity and poor cycling stability. Here, a high-performance potassium-ion battery is achieved by employing few-layered antimony sulfide/carbon sheet composite anode fabricated via one-step high-shear exfoliation in ethanol/water solvent. Antimony sulfide with few-layered structure minimizes the volume expansion during potassiation and shortens the ion transport pathways, thus enhancing the rate capability; while carbon sheets in the composite provide electrical conductivity and maintain the electrode cycling stability by trapping the inevitable by-product, elemental sulfur. Meanwhile, the effect of the exfoliation solvent on the fabrication of two-dimensional antimony sulfide/carbon is also investigated. It is found that water facilitates the exfoliation by lower diffusion barrier along the [010] direction of antimony sulfide, while ethanol in the solvent acts as the carbon source for in situ carbonization.

T wo-dimensional (2D) materials have various unique physical properties, which have prompted widespread successful application in the fields of catalysis, nanoelectronics, energy storage and conversion, etc. In particular, 2D materials present extensive prospects for application in energy storage and conversion due to their highly accessible surface area and fast charge transfer kinetics, so that they have been applied in a unique strategy to significantly enhance the rate performance of electrodes 1,2 . Potassium-ion batteries (KIBs) have attracted enormous attention due to their obvious advantages. Besides the abundance of potassium resources, the relatively lower redox potential of K/K + (−2.93 V vs. standard hydrogen electrode) than that of Na/Na + (−2.71 V), implies that KIBs could have a highvoltage plateau and high-energy density. Due to the large size of the K-ion, however, the insertion of K + into electrode materials is hindered, resulting in their relatively low capacity and poor cycling performance. The research on KIBs is still at an early stage, with the electrochemical reaction mechanism of most electrode materials unclear, and only a few cathode materials (such as Prussian blue 3 , K x MnFe(CN) 6 4 , confined selenium (c-PAN-Se) 5 ), and anode materials (graphite/carbon [6][7][8][9] , Sn 4 P 3 / C 10,11 ) could present reasonable capacity, although the cycling stability for all of them is far away from practical application. Therefore, further exploration of suitable electrode materials with high reversible capacity as well as good rate performance and excellent cycling stability is of great importance, and could possibly be achieved via the design and fabrication of 2D structured materials.
Among all the Sb-based anode materials, antimony trisulfide (Sb 2 S 3 ) has drawn extensive attention 12,13 , owning to its higher reversible theoretical capacity (946 mAh g −1 ) compared to that of Sb anode (660 mAh g −1 ) due to its theoretical accommodation of 12 moles of Li + or Na + per Sb 2 S 3 mole. Better mechanical stability is also expected for Sb 2 S 3 due to its smaller volume changes during charge/discharge than those of Sb anode. Moreover, the reversibility of sulfides is better than those of oxides (Sb 2 O 3 and Sb 2 O 4 ), resulting in relatively better cycling stability 14 . Improvement of the cycling stability and rate performance of Sb 2 S 3 is necessary, however, to meet the requirements for real applications because of the unavoidable volume changes and limited ion/ charge transfer. The bulk Sb 2 S 3 crystal has a layered structure with zigzag sheets parallel to the b-axis, which makes the fabrication of 2D Sb 2 S 3 possible. As far as we know, however, there has been no report on the electrochemical behavior of 2D Sb 2 S 3 so far.
Exfoliation of layered bulk crystals to obtain monolayer or few-layer flakes has been a crucial technique in the fabrication of 2D materials 15 , and it has been the primary technique in the synthesis of high-quality flakes for various applications. Among all the exfoliation techniques, high-shear mixing is a more effective approach than sonication for the large-scale fabrication of graphene 16 and other 2D materials (MoS 2 nanosheets 17 , few atomic-layered LiCoO 2 material 2 ), which could be a feasible and promising approach for industrial scale applications. Although this technique is promising, the challenges are still there, such as the relatively unclear mechanism of shear exfoliation and whether it is applicable to other layered structured materials.
In this study, the few-layered antimony sulfide/carbon sheet (SBS/C) anode is prepared via solution-triggered one-step highshear exfoliation in order to boost the electrochemical performance of potassium-ion batteries (PIBs). The fatal issue of huge volume changes in Sb 2 S 3 during electrochemical cycling could be solved by the design and fabrication of few-layered SBS/C anode, while the poor electrical conductivity of Sb 2 S 3 can be improved by incorporation of carbon via in situ carbonization in an ethanolcontaining solvent. Moreover, the detected by-product S after cycling indicates that the irreversible conversion of SBS to S could be another reason for failure of SBS anode in KIBs, which could be overcome by the trapping effect of carbon sheets. In addition, the solvent effect on the exfoliation is also studied in order to optimize the structure and constitution of SBS/C composite. Based on density functional theory (DFT) calculations, the lower diffusion barrier of water than ethanol along the [010] direction of SBS crystal could give a pathway for facilitating exfoliation. Meanwhile, the ethanol in the solvent provides the carbon source for in situ carbonization. Ultimately, by employing a mixture of water (W) and ethanol (E) in a certain ratio as the exfoliation solvent, the as-prepared SBS/C (E/W = 2:1) composite anode delivers a specific capacity of 404 mA h g −1 after 200 cycles (at a current density of 500 mA g −1 ) and presents outstanding rate capability with 76% capacity retention at current densities from 50 to 500 mA g −1 .

Results
Electrochemical mechanism of SBS anode for K + storage and failure mechanism. To understand the K + ion storage mechanism of Sb 2 S 3 chemistry, the phase evolution of Sb 2 S 3 for KIBs during discharge/charge was studied using in-operando synchrotron X-ray diffraction (XRD, λ = 0.6888 Å), ex situ XRD, and scanning transmission electron microscopy/selected area electron diffraction (SAED) ( 040048897, Pnma (62)) shift toward lower detection angle, 2θ, indicating the insertion of K + into Sb 2 S 3 . Specifically, the 2D colorful contour plot using in situ diffraction patterns in Fig. 1b shows that the peak corresponding to (212) planes is shifted left and then becomes weakened in intensity. When the Sb 2 S 3 anode is discharged from 0.7 to 0.5 V, the main peaks of Sb 2 S 3 gradually become weakened, and two new peaks evolve at 15.17°and 15.68°, which are assigned to Sb (PDF No. 01-071-3736), and suggest the occurrence of a conversion reaction of Sb 2 S 3 . When further discharged to 0.1 V, two new peaks are generated at 13.4°and 13.7°, which can be ascribed to the formation of K 2 S 6 (PDF No. 01-083-9589) as the intermediate state; two new peaks located at 14.17°and 14.46°correspond to the (131) and (221) planes of K 2 S 3 (PDF No. 04-007-0574) as a final discharge product. At the same time, a new diffraction peak at 13.53°has developed, which can be indexed as the (220) peak of K 3 Sb (PDF No. 01-078-6559). Due to the nanocrystallinity of the intermediate products of Sb and K 2 S 6 (Supplementary Fig. 2), the corresponding diffraction peaks are weak and broad. In order to confirm the existence of intermediate products, ex situ SAED was conducted on the electrode after it was discharged to 0.5 V. As shown in the SAED pattern (Fig. 1a), the marked spots in orange are corresponding to the Sb (101) and (002) planes, which is consistent with the weak peaks (15.17°and 15.68°) in the synchrotron XRD pattern, and the diffused green rings with spots belong to the K 2 S 6 (040), which corresponds to the peak at 13.4°in synchrotron XRD. Based on our in situ/ex situ XRD and SAED results, we propose that in KIBs, Sb 2 S 3 may undergo K + intercalation reaction (1) followed by the conversion and alloying reactions (2, 3): K + intercalation reaction: Conversion and alloying reactions: Ex situ transmission electron microscopy (TEM) after discharge and charge were also employed to investigate the morphology and identify the phases of discharge/charge products (Fig. 1c-i and Supplementary Fig. 3). After conversion and alloying reactions of SBS with K ions, the discharge products show a connected globular morphology (Fig. 1c, d) with particle size around 20 nm instead of the initial microsized bulk material ( Supplementary Fig. 4).
According to the electrochemical reactions above (Eqs. (1)-(3)), the volume expansion during discharging is about 300%, estimated based on the density differences between Sb 2 S 3 (4.64 g cm -3 ), K 3 Sb (2.24 g cm −3 ), and K 2 S 3 (2.12 g cm −3 ). Meanwhile, the morphology of the Sb 2 S 3 after charge also shows nanosized particles (Supplementary Fig. 3) with diameters around 25 nm (Supplementary Fig. 3d). The well-defined diffraction rings shown in the SAED pattern ( Fig. 1e) undoubtedly reveal the polycrystalline nature of the charged product. Consistent with the XRD characterization (Fig. 1a), these rings can be satisfactorily indexed to (112) crystal planes of the K 2 S 3 phase (PDF No. 04-007-0574) and (220) crystal planes of K 3 Sb phase (PDF No. 01-078-6559). indicating uniform distributions of the S, K, and Sb in the discharged product, and homogenous distribution of Sb and S in the charged product. The electrochemical performance of bulk SBS is presented in Supplementary Fig. 5 and Fig. 1j. It presents a poor first cycle coulombic efficiency (around 23%) and poor electrochemical stability (from 988 down to 184 mA g −1 after 20 cycles). The failure mechanism of commercial bulk SBS is primarily due to the huge pulverization during cycling, as illustrated in Fig. 1k. Continuous pulverization leads to, rupture of the solid-electrolyte interphase (SEI), decrease in electrical contact, and ultimately, fast deterioration in its electrochemical performance. In order to deal with the problem of serious pulverization, 2D material design was explored and found to be an effective way to overcome the problem, thus enhancing the performance of K-ion batteries.
Few-layered SBS exfoliation and mechanism. The orthorhombic crystal structure of Sb 2 S 3 with Pnma phase (Fig. 2a), is composed of infinite chains of zigzag sheets of (Sb 4 S 6 ) n along the b-axis. The weak bonding between the sheets makes the crystal cleavable along the b-axis direction 18,19 . The highresolution TEM image of exfoliated Sb 2 S 3 also confirms the possible cleavable direction, and Sb 2 S 3 sheets/plates were obtained with (001) Fig. 6), the average particle size of Sb 2 S 3 obtained in ethanol is larger than that obtained in water. Figure 2c shows the thickness distribution of Sb 2 S 3 nanosheets by counting more than 60 sheets for each sample collected from top solution. For the SBS exfoliated in water (W) and the mixed solution (E/W), the typical thickness of the nanosheets is mainly in the range of 2-8 nm and 4-15 nm, respectively, while for the SBS nanosheets obtained from ethanol solvent, the thickness is several times larger than that for the samples exfoliated in water or the mixed solvent, and is in the range of 26-55 nm. The typical AFM images and the cross-sectional height profiles conform the few-layered nature of the SBS nanosheet structures which were exfoliated in the mixed solution (E/W = 2:1) and pure water (W). From the XRD pattern of Sb 2 S 3 after exfoliation in pure ethanol, the identified peaks are all consistent with of the standard PDF pattern, while for Sb 2 S 3 exfoliated in water (W) or ethanol/water solution (E/W = 2:1), as shown in Fig. 2d, some of the characteristic peaks are relatively weaker or even not visible, indicating smaller crystal size, in good agreement with the thickness distribution shown in Fig. 2c. This difference in the peak intensity or peak disappearance in the samples could also be found in separate samples collected from same exfoliation solution ( Supplementary Fig. 8a, Supplementary  Fig. 9a). The AFM results together with XRD results indicated that the bulk Sb 2 S 3 could be more easily exfoliated in water or water/ethanol solution than in pure ethanol. Water may play an important role in shear exfoliation. First-principles calculations were conducted in order to achieve further insight on the water and ethanol dynamics and to test that hypothesis. The calculated adsorption energies ( Fig. 2e and  Carbon sheet production during exfoliation. Carbonization of ethanol could be occurring during high-shear exfoliation because the mechanical shearing and cavitation cause strong collisions between the active material and ethanol. Raman spectroscopy was performed to understand the composition of the exfoliated products (Fig. 3a, Supplementary Fig. 11) 29 are slightly higher for SBS/C exfoliated in ethanol/water solution than for its counterpart exfoliated in pure ethanol, and these functional groups evidenced in the C 1s spectrum are consistent with the FTIR results shown in Fig. 2b. These results (FTIR and C 1s XPS) suggest that there are carboxyl and hydroxyl groups on the surface of the amorphous carbon that can be ascribed to the incomplete carbonization at a relatively low exfoliation temperature. From the TEM and energy dispersive spectroscopy (EDS) results (Fig. 3d, e and Supplementary Fig. 13), it is found that the carbon sheets are relatively thinner compared to the SBS sheets. In addition, from visual inspection of the exfoliated solutions with different solvents, the exfoliated solutions with ethanol are apparently darker than that in water ( Supplementary Fig. 14), due to the presence of carbon.
Electrochemical properties of few-layered SBS and SBS/C electrodes.  and SBS/C anodes were investigated in the range of 0.01-2 V (vs. K + /K). The capacities of SBS/C electrodes were calculated on the basis of the total mass of SBS and carbon. Supplementary Fig. 15 presents typical cyclic voltammograms of the few-layered SBS and bulk SBS for the initial three cycles at 0.1 mV s −1 . In the first cycle of few-layered SBS (NS-3-W), three cathodic peaks at 0.78, 0.45, and 0.31 V are attributed to the intercalation process and the formation of SEI, the conversion reaction with sulfur in SBS, and the alloying of K with Sb, respectively, which are assigned based on in situ XRD analysis and research on SBS anode in lithium-ion batteries (LIBs) and sodium-ion batteries 12,20,30 . For the following cycle, all of the peaks are almost overlap for the few-layered SBS, suggesting good reversibility. In order to understand how the exfoliation solvents influence the electrochemical performance of the SBS samples, the cycling stability and rate capability of the different SBS electrodes are compared (Fig. 4a-d). All samples collected from ethanol in different sections (NS-1-E, NS-2-E, and NS-3-E) present better cycling stability than the samples collected from water (NS-1-W, NS-2-W, and NS-3-W), mainly due to the contribution of carbon from the carbonization of ethanol. The charge/discharge curves of SBS/C (NS-2-E) and bulk SBS at different cycles are compared in Fig. 4e and Supplementary Fig. 16.
The results show that there is a new voltage plateau (around 1.7 V) after a certain number of cycles for SBS/C accompanied by slightly increased capacity, while for bulk SBS, there is no new voltage plateau and the capacity drops quickly. To explore the chemical state of SBS, we analysed the S 2p peaks of the SBS/C (NS-2-E) and bulk SBS electrode surfaces via ex situ XPS (Fig. 4f,  g). The two peaks of SBS/C and bulk SBS located at 161.55 and 162.37 eV show the presence of S 2−13 , while the two peaks at 169 and 170.5 eV are likely to reflect the -SO 2fragments due to the decomposition of the bis(fluorosulfonyl)imide (FSI-) anion in the SEI film 31 . The high ratio/content of -SO 2on the surface of bulk SBS after the 50th charge indicates a thicker SEI film compared with the layered SBS/C electrode due to the continuous pulverization of bulk SBS during cycling. In addition, it should be noted that the element sulfur was detected in both electrodes, with its peak located at around 164 and 165 eV 32 , and the ratio of elemental S 0 to S 2− on the electrode surface after the 50th charge are 16.5% and 9.04% for SBS/C (E) and SBS (bulk), respectively (Supplementary Table 1 and Supplementary Note 2). The high intensity of elemental sulfur in SBS/C could be ascribed to the carbon trapping effect, in which the carbon sheets produced in SBS/C composite may play a crucial role by trapping the polysulfides and preventing them from dissolving in electrolyte and creating the shuttle effect during discharge/charge. Therefore, in the SBS/C composite, the carbon sheets in composite not only increase the electrical conductivity, but also prevent the loss of active material (Sb 2 S 3 or S). On the other hand, without the protection of carbon sheets in bulk SBS and layered SBS (W), the by-product sulfur that is produced could cause the shuttle effect and capacity decay. In terms of the rate capability of SBS/C and layered SBS, although SBS/C shows better rate performance than layered SBS in the low current density range (from 20 to 300 mA g −1 ), layered SBS presents slightly better rate capability at high current densities (from 500 to 1000 mA g −1 ). Carbon sheets in the SBS/C composite play a dominant role in improving the electrical conductivity, which enhances the rate performance at low current densities. At high current densities, however, the much thicker SBS sheets in the composite (around 45 nm in , resulting in improved rate capability, especially at high current densities. (Fig. 4d).
As it was concluded above that water could facilitate the exfoliation by easily absorbing and diffusing into SBS crystal Capacity (mAh g -1 ) Potential (V versus K/K + ) (Fig. 2e, f) and ethanol works as a carbon source, it is reasonable to expect that the mixed solvent of ethanol/water would be the best choice, which not only maintains the few-layered structure of SBS but also introduces carbon sheets at the same time. Therefore, in order to optimize SBS anode for achieving outstanding electrochemical performance, layered SBS/C samples were synthesized via exfoliation in ethanol/water mixtures with various ratios (E/W = 1:1, E/W = 2:1, E/W = 6:1, E/W = 8:1, E/W = 16:1, E/W = 32:1). In Fig. 4h, from the cycling response at different current densities (from 50 mA g −1 to 2 A g −1 ), we find that the SBS/C (E/ W = 2:1) and SBS/C (E/W = 6:1) electrodes present the best rate capability. Moreover, it was found that the initial discharge capacity of SBS/C electrodes at 50 mA g −1 is slightly lower than the following capacity at 100 mA g −1 (Fig. 4h) in the rate test, and a similar unusual phenomena can also be observed in Fig. 4b, d, which can be attributed to the activation process of SBS and SBS/C electrodes in the initial cycles and could be avoided by extending the standing time of fresh cells ( Supplementary Fig. 18). The carbon content in each composite has been investigated in Supplementary Fig. 19 and Supplementary Note 4, (with 3.82% in SBS/C (E:W = 1:1), 3.93% in SBS/C (E:W = 2:1), 3.62% in SBS/C (E:W = 6:1), 3.72% in SBS/C (E:W = 8:1), 4.17% in SBS/C (E:W = 16:1), and 4.94% in SBS/C (E:W = 32:1)), indicating that there is not much difference in carbon content among these samples. In order to distinguish the influence of carbon content and the thickness of SBS sheets on the electrochemical performance of composite electrodes, we designed and fabricated SBS/C composites via a two-step shear exfoliation. It was found that the cycling stability and rate performance improved with decreasing thickness of the SBS sheets, and among them the SBS 6000 electrode shows the best electrochemical performance. We then investigated the influence of the carbon content based on the same thickness of SBS. Supplementary Fig. 20 suggests that although the presence of carbon sheets improved the cycling stability and rate capability significantly, the carbon content in the composites (in the range from 3.5 to 5 wt%) plays a negligible role in influencing the electrochemical performance. Therefore, the differences in electrochemical performance of the few-layered SBS/C electrodes shown in Fig. 4h are mainly caused by the thickness of SBS. Supplementary Figs. 21, 22 show the variations in the size and thickness distributions of the exfoliated sheets in these SBS/C electrodes via SEM and AFM. Less thickness of layered SBS in a composite will lead to the short diffusion time according to the equation t = L 2 /D 33 ( Supplementary Fig. 23) (where t is the diffusion time, L is the diffusion length or the thickness of SBS, and D is the K-ion diffusion constant in SBS), which results in the simultaneous transfer of K ions with improved rate performance. The cycling performances of SBS/C (E/W = 2:1) and SBS/C (E/W = 6:1) electrodes were compared after the rate test, and better cycling stability of SBS/C (E/W = 2:1) was achieved, with a high reversible capacity of 404 mAh g −1 after 200 cycles (Fig. 4h). The long-term cycling performance of SBS/C (E/W = 2:1) electrode was further investigated (Supplementary Fig. 24). It shows excellent cycling stability and highcapacity retention of 79% after 1000 cycles (at a current density of 1 A g −1 ). Meanwhile, we also compared the SBS/C (E/W = 2:1) electrode with previously reported state-of-the-art anodes 11,[34][35][36][37][38][39][40][41][42] for KIBs, excluding carbon-based anodes (Fig. 5). It is shown that our few-layered SBS/C electrode could deliver the highest reversible capacity with unrivaled cycling stability among all the anode materials so far (excluding carbon/graphite anode). The superior cycling stability and rate capability of SBS/C (E/W = 2:1) are mainly due to the synergetic effects between few-layered structured SBS and the carbon sheets in the composite, which not only promote ion/electron transfer, but also maintain the electrode/structure stability and electrode reversibility. Here, we exclude the carbon/graphite anodes from the comparison of electrochemical performance because of their limitations as anode for PIBs. The very low theoretical capacity (279 mA h g −1 , 30% less than that of LIB) and poor capacity retention of the graphite anode mean that it cannot rival nongraphite anode in KIBs. Although amorphous carbon electrode present higher reversible capacity than graphite (250 vs. 200 mAh g −1 ) with relatively better cycling retention, the electrochemical behavior is more like capacitor behavior, with a sloped, inconspicuous, and relatively high voltage plateau.

Discussion
According to the research presented above, high-performance PIBs with a composite of few-layered antimony sulfide/carbon sheets (SBS/C) as anode are introduced. The SBS/C composite was fabricated via one-step high-shear exfoliation in an ethanol/ water solvent (ratio E/W = 2:1). Compared with commercial bulk SBS, the few-layered SBS/C could effectively deal with the issues related to the huge volume changes of Sb 2 S 3 during charge/discharge and its poor electrical conductivity. Few-layer structured SBS anode could minimize the absolute volume changes of SBS and facilitate ultrashort ion transport paths compared with the bulk material. The microsized carbon sheets in the composite not only provide electrical conductivity, but also avoid the loss of active material by trapping the element S that is inevitably produced due to the irreversible reaction between K 2 S 3 and Sb 2 S 3 . After investigating the solvent effect (water and ethanol) on exfoliation, it was found that water could facilitate the exfoliation to produce few-layered SBS based on the experimental results and DFT calculations, while ethanol could promote carbon sheet generation during exfoliation due to the carbonization. The electrochemical performance of layered SBS/C anode was further optimized by investigating the solvent ratio of ethanol/water for exfoliation. With the cooperative action of water and ethanol in a certain ratio as exfoliation solvent, the obtained SBS/C (E/W = 2:1) composite anode could deliver a reversible capacity as high as 404 mA h g −1 after 200 cycles (at a current density of 500 mA g −1 ) and present excellent rate performance with 76% capacity retention from current densities of 50-500 mA g −1 .
At the same time, in order to obtain insight on the electrochemical behavior of Sb 2 S 3 in PIBs, the electrochemical reaction mechanism of the Sb 2 S 3 electrode during charge/discharge was investigated by in situ XRD, ex situ XRD, and TEM. The twostage reactions of SBS in the PIB are proposed. Unlike the reactions of SBS in LIBs/SIBs, it clearly presents an intercalation step before the conversion/alloying reactions. According to the final and their cycling performance after rate testing at a current density of 500 mA g −1 discharge products, the calculated theoretical capacity of SBS in KIBs is as high as 630 mA h g −1 .
Moreover, in terms of fabrication, the solvent effect was discussed for fabricating different thicknesses of layered SBS via shear exfoliation, which gives guidance for exfoliating other materials with control of thickness. The carbonization of organic solvent during high-energy shear exfoliation was proposed and demonstrated for the first time in our work. This strategy for the fabrication of fewlayered material/carbon composite can be extended to other layered crystals, and the shear exfoliation may open a new path towards carbon composite fabrication with selected organic solvents.

Methods
Materials. All the involved chemicals were analytic grade and they were purchased from Sigma Aldrich. They were all directly used without any further purification.
Shear exfoliation. SBS and SBS/C were synthesized via a modified shear exfoliation method using the commercial bulk materials 2 . The L5M high-shear laboratory mixer that was employed is made by Silverson Machines Ltd., UK. Typically, the screw-on slotted and interchangeable disintegrating head equips a rotor with 30 mm in diameter. The gap between the head screen and rotor is about 0.05 mm (illustrated in Supplementary Fig. 25). When it is in operation, the high rotor speed (N) creates a high-shear rate (γ) within the gap. More specifically, in synthesis, the bulk SBS (5 g) was put into a beaker (250 ml) containing ethanol, water, or a mixed solution of ethanol and water (1:1, 2:1, 6:1, 8:1, 16:1, and 32:1) (total volume of 200 ml). The mixtures were kept at room temperature for 2 h before exfoliation. Then, the mixer head was lowered, immersed in the solution, and then rotated at 6000 rpm for a continuous 30 min. During mixing, the beaker was fixed in a water bath with the initial temperature of 20 ℃. After 24 h of standing after mixing, the obtained dispersion was divided into three samples (top dispersion: NS-3; middle dispersion: NS-2; and bottom dispersion: NS-1). For the SBS exfoliated in ethanol, the three separated samples were denoted as NS-1-E, NS-2-E, and NS-3-E, while for the SBS exfoliated in water, the samples were denoted as NS-1-W, NS-2-W, and NS-3-W. These samples were then collected via filtration. In the case of the sample collection from the mixed exfoliation solution, the procedure was similar, but collection was only done from the top of the dispersion for filtration. For the twostep exfoliation, different thicknesses of SBS in water were fabricated with adjusting the rotation rate (4000, 5000, 6000, and 7000 rpm) (SBS 4000, SBS 5000, SBS 6000, and SBS 7000). Then, the solution product of SBS 6000 was chosen as target sample, and different amounts of ethanol (ethanol/water ratio: 1:1, 4:1, and 16:1) were added for step-two exfoliation.
Materials characterization. The microstructure/morphology of the as-prepared Sb 2 S 3 bulks and nanosheets was investigated by XRD (GBC MMA) with Cu Kα radiation; field-emission SEM (FESEM) (JEOL 7500); TEM (JEOL ARM-200F) with high-resolution TEM (HRTEM), and Raman spectroscopy (Jobin Yvon HR800) employing a 10 mW helium/neon laser at 632.8 nm. A commercial AFM (Asylum Research MFP-3D) was used to measure the morphology and thickness of the SBS nanosheets in trapping mode. An Al coated n-silicon probe with resonance frequency of 204-497 kHz and force constant of 10-130 N m −1 was used in the AFM measurements. For synchrotron X-ray powder diffraction, a specially modified CR2032 coin cell was used with holes on both sides. In situ synchrotron XRD measurements were then performed at the Powder Diffraction beamline (Australian Synchrotron), and the XRD patterns were conducted at 0.688273 Å (determined using LaB 6 , NIST SRM 660b).
Electrochemical measurements. The commercial SBS, exfoliated SBS and SBS/C electrodes were assembled in a glove box into coin cells (CR2032). For the anode preparation, slurry containing 60 wt% active material, 20 wt% Super P, and 20 wt% carboxymethyl cellulose was dissolved in an aqueous solution. Then, the working electrodes were prepared by coating the slurry on a copper foil current collector and drying it at 70°C for 12 h. The loading mass of the active materials (SBS or SBS/C) was around 1 mg cm −2 . 1 M potassium fluorosilicate (KSiF 6 ) in ethylene carbonate/ propylene carbonate (1:1 V/V) was applied as the electrolyte. The asfabricated coin cells were charged and discharged in the voltage range of 0.01-2 V for the SBS anode using a Neware instrument.
Theorectical calculations. The calculations were performed based on the DFT approach 43 using the DMol 3 package. The exchange-correlation interaction was tested by using the generalized gradient approximation (GGA) with the Perdew-Wang 91 (PW91) function 44 . The double numerical basis with polarized orbital (DNP) was specified as the atomic orbital basis set 45 . The Brillouin-zone was sampled using the Monkhorst-Pack grid of special k-points 46 . The convergence tolerances of energy, the maximum force, and the maximum displacement were 1.0 × 10 -5 Ha, 0.002 Ha Å −1 , and 0.005 Å for, respectively.

Data availability
The data that support the findings of this study are available from the corresponding author on reasonable request.