Introduction

Lithium-ion batteries (LIBs) are the most well-known rechargeable electrochemical energy storage devices, and they are a key component of electric mobility and portable electronics1,2,3,4. Sodium-ion batteries (SIBs) are conceptually similar, and they have attracted enormous attention in recent years because of the higher natural abundance of sodium and more favorable distribution of sodium reserves compared with lithium5,6,7,8,9,10,11,12,13. Although graphite is currently the main commercialized anode material for LIBs, its low theoretical charge storage capacity (372 mAh g−1) limits its application in new generation batteries, requiring exploration of new electrode materials with higher capacity and stable cycling performance14. With respect to SIBs, the search for efficient Na-storing anodes is a high priority, because graphite shows low Na-ion capacities of 30–35 mAh g−115, while other carbonaceous materials have low tap densities and exhibit capacities of less than 300 mAh g−116. Additionally, the relatively low potential of carbon sodiation (~0 V vs. Na+/Na) leads to deposition of sodium metal on the carbon electrode surfaces, which may eventually result in compromised safety17.

Over the past decade, much attention has been focused on development of alternative anode materials for both LIBs and SIBs12. In particular, low-cost and environmentally benign Sb2S3 anodes have attracted great interest because of their high capacities and relatively low redox lithiation/sodiation potentials18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45. Theoretically, Sb2S3 can generate a specific capacity as high as 946 mAh g−1 through conversion and alloying reactions (corresponding to 12 mol of lithium/sodium and electrons per formula unit). However, harnessing this storage potential of Sb2S3 is hindered by its poor capacity retention owing to the structural (conversion) and volume (alloying) changes during discharging/charging, which lead to mechanical disintegration of the electrodes and thus loss of electrical connectivity. These difficulties can be mitigated by nanostructuring, particularly when the active material is embedded in an elastic and conductive network that helps to enhance electronic transport and reduce the cycling instability caused by volumetric changes in the conversion–alloying-type anode material36,46,47. Specifically, in the last few years, extensive effort has focused on various forms of nanostructured Sb2S3, such as Sb2S3 nanowires24,48,49,50,51, nanorods45,52,53, nanoparticles (NPs)37,54,55,56,57, nanocables58, and Sb2S3/C nanocomposites23, to maximize the anodic charge-storage capacity and improve the cycling performance. Notably, the electrochemical performance of highly uniform colloidal Sb2S3 NPs has not been reported. Such NPs are an ideal platform for studying the effects of the size and electrode morphology on the charge storage capacity and cycling stability of Sb2S3 anodes.

In this study, we synthesized uniformly sized colloidal Sb2S3 NPs whose size is tunable in 10–200 nm size range, which allowed us to comprehensively investigate the effect of the primary particle size on the electrochemical behaviour of Sb2S3 as the anode material for LIBs and SIBs. We assessed the pros and cons of nano-Sb2S3 anodes in comparison with commercial microcrystalline Sb2S3 (hereafter denoted bulk Sb2S3, Figure S1). We note that although synthesis of Sb2S3 NPs might be prohibitively expensive for practical application in commercial batteries, the insight gained from using such precisely tunable model NPs can guide development of Sb2S3 anodes for both LIBs and SIBs. We found that at current rates of 0.3–12 A g−1, the Li-ion storage capacities for anodes composed of both ca. 20–25 nm (1055–608 mAh g−1) and ca. 180–200 nm Sb2S3 (970–574 mAh g−1) were significantly higher than for their bulk counterpart (683–418 mAh g−1). For Na-ion storage, the capacities of nano-Sb2S3 and bulk Sb2S3 anodes were similar. Regarding the cycling stability, the major finding was that nano-Sb2S3 exhibited significantly higher capacity retention for both Li-ion and Na-ion storage than bulk Sb2S3. Notably, unprecedented Li-ion capacity retention of 55% was achieved for ca. 20–25 nm Sb2S3 NPs at a current density of 2.4 A g−1 after 1200 cycles.

Results and Discussion

The general synthetic route for preparation of ca. 20–25 nm amorphous antimony sulfide NPs using octadecene (ODE) as a solvent in the presence of oleylamine (OAm) as a surface capping ligand is shown in Fig. 1a. In a typical synthesis, the Sb2S3 NPs were synthesized by the hot-injection technique using antimony(III) chloride and bis(trimethylsilyl)sulfide ((TMS)2S) as the antimony and sulfur precursors, respectively. After injection of (TMS)2S into the SbCl3/ODE mixture, the color of the reaction solution rapidly changed to red-orange. The reaction temperature was maintained at 120 °C for 15 min. Transmission electron microscopy (TEM) and X-ray diffraction (XRD) analysis confirmed formation of amorphous spherical ca. 20–25 nm antimony sulfide NPs with a narrow size distribution (Figs. 1c and S2a). A longer reaction time of 30 min resulted in formation of ca. 1.5–2.0 μm long crystalline Sb2S3 nanorods with diameters of ca. 150–200 nm (Figure S3). When the (TMS)2S solution was injected at 100 °C and maintained at this temperature for 15 min, ca. 8–10 nm amorphous Sb2S3 NPs were obtained (Figure S4). Scanning transmission electron microscopy with energy-dispersive X-ray spectroscopy (STEM-EDXS) measurements of the as-synthesized ca. 20–25 nm Sb2S3 NPs revealed that Sb and S were homogeneously distributed throughout each NP (Fig. 1e,f, g, h). From scanning electron microscopy with energy-dispersive X-ray spectroscopy (SEM-EDXS) analysis, the atomic ratio of Sb, S and O was about 1:1.4:0.1 (Figure S5). The presence of detectable oxygen in the EDS spectrum could be because of oxidation of the NPs during synthesis, cleaning, or preparation of the specimen. Notably, similar synthesis of Sb2S3 NPs was reported by Bakr et al.59. using SbCl3 and (TMS)2S in ODE with oleic acid (OAc) as a ligand. The synthesis yielded relatively polydisperse ca. 30–50 nm Sb2S3 NPs with a chain-like structure. In our synthesis, the use of the OAm ligand resulted in slower reaction kinetics, causing more homogenous nucleation and growth of Sb2S3 NPs in comparison with the OAc ligand.

Figure 1
figure 1

Schematics of one-pot synthesis of the (a) ca. 20–25 nm and (b) 180–200 nm Sb2S3 NPs. TEM images of the (c) ca. 20–25 nm and (d) ca. 180–200 nm Sb2S3 NPs. (e) Sb-Lα and (f) S-Kα elemental STEM-EDXS maps of the ca. 20–25 nm Sb2S3 NPs. (g) Reconstructed overlay image of the elemental maps shown in (e) and (f). (h) EDXS spectrum of the small Sb2S3 NPs. The insert shows a high-angle annular dark-field scanning transmission electron microscopy image of the ca. 20–25 nm Sb2S3 NPs.

Larger Sb2S3 NPs of approximately ca. 180–200 nm were synthesized in a similar way to the ca. 20–25 nm Sb2S3 NPs by replacing the antimony(III) chloride precursor and OAm ligand with antimony acetate and OAc, respectively (Figs. 1b,d and S2b, S6; for details see the experimental section). By changing of the (TMS)2S sulfur source to S/OAm (elemental sulfur dissolved in OAm), crystalline Sb2S3 nanoplates were obtained (Figure S7).

The galvanostatic cycling measurements of the Sb2S3 NPs are summarized in Figs. 2 and 3. For electrochemical testing, the Sb2S3 NPs were treated with a 1 M solution of hydrazine in acetonitrile for 2 h60,61. The untreated NPs gave no operational electrodes because of the isolating long-chain capping molecules surrounding the as-synthesized Sb2S3 NPs. In addition to the effect of the active material, the charge storage capacity of the electrode strongly depends on the electrode formulation (the origin and amounts of the binder and conductive additive), electrode thickness, porosity, temperature, electrolyte, and so forth. Therefore, with the aim of distinguishing the size effect from the other factors, the following experimental parameters were fixed for all of the electrodes: (i) the choice and mass fractions of the binder and carbon black and (ii) the electrolyte composition. All of the electrodes contained 64 wt% of the active material, 15 wt% carboxymethylcellulose as a binder, and 21 wt% carbon black as a conductive additive. The electrochemical tests were performed in Li-ion or Na-ion half-cells with elemental lithium or sodium acting as both the counter and reference electrodes, respectively. Further details of electrode preparation and assembly of the batteries are given in the Supporting Information.

Figure 2
figure 2

Electrochemical results of the Sb2S3 electrodes cycled with lithium electrolyte (1 M LiPF6 in ethylene carbonate/dimethyl carbonates (EC/DMC)) in a half-cell configuration using metallic lithium as the counter and reference electrode. (a) CV curves (the first cycle is shown in orange or blue and the second cycle is shown in grey) of the small and large Sb2S3 NPs measured at a scan rate of 1 mV s−1 (see Figure S8 for details). (b) Galvanostatic charge–discharge curves of the small Sb2S3 NPs, large Sb2S3 NPs, and bulk Sb2S3 during the first cycle. (c) Rate capacity and (d) cycling stability measurements of Li-ion half-cells using Sb2S3 anodes made from small Sb2S3 NPs, large Sb2S3 NPs, and bulk Sb2S3. The corresponding galvanostatic charge–discharge curves and Coulombic efficiency measured at current densities of 0.3–12 A g−1 and different cycle number are shown in Figures S9, S10 and S11 respectively.

Figure 3
figure 3

Electrochemical results of the Sb2S3 electrodes cycled with sodium electrolyte (1 M NaClO4 in PC) in a half-cell configuration using metallic sodium as the counter and reference electrode. (a) CV curves (the first cycle is shown in orange or blue and the second cycle is shown in grey) of the electrodes composed of small and large Sb2S3 NPs measured at a scan rate of 1 mV s−1 (see Figure S12 for details). (b) Galvanostatic charge–discharge curves of the electrodes composed of small Sb2S3 NPs, large Sb2S3 NPs and bulk Sb2S3 during the first cycle. (c) Rate capacity and (d) cycling stability of Na-ion half-cells using Sb2S3 anodes composed of small Sb2S3 NPs, large Sb2S3 NPs, and bulk Sb2S3. The corresponding galvanostatic charge–discharge curves and Coulombic efficiency measured at current densities of 0.3–12 A g−1 and different cycle numbers are shown in Figures S13, S14 and S15, respectively.

The cyclic voltammetry (CV) curves of electrodes composed of ca. 20–25 nm and ca. 180–200 nm Sb2S3 NPs (hereafter denoted small and large Sb2S3 NPs, respectively) measured in Li-ion electrolyte at a scan rate of 1 mV s−1 are shown in Fig. 2a. In the first cathodic cycle, the broad peak at about 1.2–1.4 V vs. Li+/Li can be attributed to formation of a solid electrolyte interphase (SEI) layer and the conversion reaction of Sb2S3 NPs (Sb2S3 + 6Li+ + 6e → 2Sb + 3Li2S). Upon further lithiation, two reduction peaks at 0.7 and 0.5 V vs. Li+/Li appeared, which are ascribed to formation of Li2Sb and Li3Sb alloys, respectively. In the reverse scan, the Sb2S3 electrode showed two peaks at 1 and 1.9 V, which are associated with delithiation of the Li3Sb alloy phase following formation of Sb2S3. The discharge voltage profiles of the Sb2S3 NPs are shown in Fig. 2b. The profiles of the Sb2S3 NPs are similar to the CV curves, showing two-step reduction of Sb2S3 eventually resulting in formation of metallic Sb (conversion reaction, voltage range 1.7–1.2 V vs. Li+/Li) and the Li3Sb alloy (alloying reaction, voltage range 0.4–1.0 V vs. Li+/Li). As follows from CV measurements, alloying of Sb in bulk Sb2S3, large and small Sb2S3 NPs takes place differently. In the bulk system, it appears that the lithiation proceeds through the direct formation of Li3Sb alloy. On the contrary, in the case of Sb2S3 NPs, the lithiation takes place through sequential formation of Li2Sb and Li3Sb alloys, respectively.

The Li-ion discharge capacities of Sb2S3 anodes composed of small Sb2S3 NPs, large Sb2S3 NPs, and microcrystalline Sb2S3 at charge/discharge current densities of 0.3–12 A g−1 are shown in Fig. 2c. At a low current density of 0.3 A g−1, the anodes composed of small and large Sb2S3 NPs exhibited theoretical capacities of about 1000 mAh g−1 with Coulombic efficiency of 97%–98% (Figure S10). The capacity retention values of the Sb2S3 anodes composed of small and large Sb2S3 NPs were 60% and 61% at 12 A g−1, respectively. The slightly higher discharge capacity of the anode composed of small Sb2S3 NPs during the first few cycles at a low current density of 0.3 A g−1 can be attributed to formation and stabilization of a SEI layer. For the bulk Sb2S3 system, the anode composed of microparticles of Sb2S3 exhibited only 60% of the theoretical capacity at 0.3 A g−1, but it retained 57% of its initial charge-storage capacity at high current density, similar to the Sb2S3 NP anodes. Regarding the cycling performance, the Sb2S3 NP and bulk Sb2S3 anodes showed stable capacities for the first 200 cycles (Fig. 2d). However, upon prolonged cycling, the capacity of the bulk Sb2S3 anode gradually decreased.

The capacities of the anodes composed of small and large Sb2S3 NPs were stable for 1200 cycles. The anode composed of small Sb2S3 NPs systematically showed at least 5% higher capacity than the anode composed of large NPs. In all cases, the Coulombic efficiency was relatively low for the initial 10–20 cycles (95%–97%), but it then increased to more than 99% upon cycling. As mentioned above, the higher cycling stability of the anode composed of Sb2S3 NPs compared with that composed of bulk Sb2S3 probably originates from the lower kinetic constraints of nanomaterials for conversion and alloying reactions. For instance, for alloying anode materials (e.g., Sn, Si, and Ge), several studies have demonstrated the existence of a critical size of the particles below which they do not fracture62,63. Furthermore, we speculate that the amorphicity of the Sb2S3 NPs aids in isotropic expansion/contraction upon their lithiation/delithiation, eventually resulting in reduction of the amount of anisotropic mechanical stress within the electrode.

In Na-ion cells with Sb2S3 NP electrodes, CV measurements showed three peaks at ca. 1, 0.7, and 0.27 V associated with formation of a SEI layer/intercalation of sodium ions into Sb2S3, conversion, and alloying reactions, respectively (Fig. 3a, see Figure S12 for details). Upon desodiation (reverse scan), the Sb2S3 electrode showed two peaks at 0.8 and 1.3 V vs. Na+/Na, which are associated with dealloying of Sb and reconversion of the Sb2S3 phase. The third peak at a higher potential of 1.6 V can be assigned to deinsertion of Na+ ions from Sb2S3. In general, the CV curves (Fig. 3a) and shape of the voltage profiles (Fig. 3b) suggest conversion and the alloying mechanism of sodiation of the Sb2S3 NPs in the voltage ranges 0.6–1 V and 0.1–0.5 V vs. Na+/Na, respectively.

In Na-ion cells, the nano-Sb2S3 and bulk Sb2S3 electrodes showed similar charge storage capacities of ~580–620 mAh g−1 at current densities of 0.3–1.2 A g−1 (Fig. 3c). The similar capacities of the nano-Sb2S3 and bulk Sb2S3 anodes in Na-ion cells can be explained by the presence of an amorphous surface oxide shell on the Sb2S3 NPs (see Figures S5 and S6 for EDS spectra). This leads to formation of Na2O, eventually resulting in irreversible capacity loss in the first discharge cycle. The much smaller differences among the capacities of the electrodes composed of small Sb2S3 NPs, large Sb2S3 NPs, and bulk Sb2S3 for Na-ion cells than Li-ion cells can be explained by the different properties of Li2O and Na2O. We suspect that Li2O acts as a relatively benign impurity covering the Sb2S3 NPs because of its high Li-ion conductivity. In contrast, Na2O is a much poorer Na+ conductor, leading to exclusion of some Sb2S3 NPs from the reversible charge/discharge storage capacity. The results of stability tests for 500 cycles at a high current density of 2.4 A g−1 are shown in Fig. 3d. In general, the charge storage capacities were consistently higher for nano-Sb2S3 than bulk Sb2S3, although the capacities remained stable for only about 50 and 100 cycles for bulk Sb2S3 and nano-Sb2S3, respectively.

Conclusions

In summary, we have reported facile colloidal synthesis of highly uniform colloidal Sb2S3 NPs with mean particle sizes in the ranges ca. 20–25 nm and ca. 180–200 nm. The underlying chemistry is based on the reaction of antimony(III) chloride/acetate and (TMS)2S in ODE using OAm/OAc as a coordinating ligand at high temperature of 120/130 °C for small/large Sb2S3 NPs. Both the small and large Sb2S3 NPs showed electrochemical cyclic stability superior to that of bulk Sb2S3 in both LIBs and NIBs. In particular, the small NPs exhibited high retention of the capacity upon extended cycling, losing only 55% of their initial capacity over 1200 cycles at a high density of 2.4 A g−1.

Methods

Chemicals

Oleic acid (OAc, Sigma-Aldrich), oleylamine (OAm, Acros, 80–90%), octadecene (Sigma-Aldrich), octadecene (ODE, Sigma-Aldrich), antimony (III) chloride (ABCR), antimony (III) acetate (Sigma-Aldrich), bis[trimethylsilyl]sulfide (Sigma-Aldrich), chloroform and acetone were used as received.

Synthesis of 20–25 nm spherical amorphous NPs

In a typical synthesis 0.5 mL oleylamine, OAm, (Acros, 80–90%) and 4 mL octadecene (ODE) were loaded into 25-mL flask and dried at 100 °C for 30 min. Then, 114 mg (0.5 mmol) SbCl3 were added to the flask under argon. The reaction mixture was heated up to 120 °C and 0.5 mmol bis[trimethylsilyl]sulfide (100 μL, (TMS)2S) in 2 mL dried ODE was then injected into the reaction flask. The color of the solution has changed to red-orange. In 15 min reaction mixture was cool down to room temperature and washed 2 times by chloroform/acetone and separated by centrifugation. After second washing step, Sb2S3 NPs were re-dispersed in oleic acid (OAc)/chloroform mixture (50 μL OAc in 2–3 mL chloroform) and stored under ambient condition. Injection of (TMS)2S solution at 100 °C and maintaining this temperature through the reaction for 15 min leads to formation of 8–10 nm amorphous Sb2S3 NPs. Injection of (TMS)2S solution at 170–180 °C and maintaining this temperature through the reaction for 3–5 min leads to formation of micrometer-sized crystalline rods (Figure S3a). Powder XRD of as-prepared NRs shows that they are highly crystalline and their XRD pattern corresponds to stibnite phase of antimony sulfide (Figure S3b). Crystalline rods could be also obtained at 120 °C in case of longer growth time. In 30 min after injection of (TMS)2S the orange color of the reaction mixture started to change into a gray-black.

Synthesis of 180–200 nm spherical amorphous NPs

In a typical synthesis, 2.5 mL OAc, 2.5 mL ODE and 0.5 mmol antimony (III) acetate were loaded into 25-mL flask and dried at 100 °C for 30 min. The reaction mixture was heated up to 130 °C under argon. At 130 °C, 0.375 mmol (TMS)2S (78 μL) in 2.5 mL dried ODE was then injected into the reaction flask. The color of the solution has changed to orange. In 3–5 min, reaction mixture was cool down to room temperature and final product was washed 2 times by chloroform/acetone and separated by centrifugation. After washing Sb2S3 NPs were re-dispersed in OAc/chloroform mixture (50 μL OAc in 2–3 mL chloroform) and stored under ambient condition.

Synthesis of thin crystalline Sb2S3 nanoplatelets

We have found that another sulfur source such as elemental sulfur in OAm effects on the morphology of Sb2S3 NPs yielding the formation of thin crystalline Sb2S3 nanoplatelets (Figure S7a). Their average size is approximately several hundred nanometers and their XRD pattern suggest that they are highly crystalline (Figure S7b). In a typical synthesis, 5 mL OAm (Acros) and 0.25 mmol (57 mg) antimony (III) chloride were loaded into 25-mL flask and dried at 80 °C for 30 min. The reaction mixture was heated up to 110 °C under argon. At 110 °C, 1 mmol (32 mg) sulfur dissolved in 2 mL OAm (Acros) was then injected into the reaction flask. Then temperature of reaction mixture was increased to 180 °C and kept for 15 min. The final product was washed 2 times by chloroform/acetone and separated by centrifugation. Sb2S3 nanoplatelets were re-dispersed in OAc/chloroform mixture (50 μL OAc in 2–3 mL chloroform) and stored under ambient condition.

Battery components

Carbon black (Super C65, TIMCAL), carboxymethyl cellulose (CMC, Grade: 2200, Lot No. B1118282, Daicel Fine Chem Ltd.), NaClO4 (98%, Alfa Aesar, additionally dried), propylene carbonate (BASF, battery grade), 4-fluoro-1,3-dioxolan-2-one (FEC, Hisunny Chemical, battery grade), 1 M solution of LiPF6 in ethylene carbonate/dimethyl carbonate (EC/DMC, Novolyte, Celgard separator (Celgard 2400, 25 µm microporous monolayer polypropylene membrane, Celgard Inc. USA), glass microfiber separator (GF/D, Cat No. 1823–257, Whatman), Al foil (MTI Corporation), Na foil (Sigma-Aldrich), Li foil (MTI Corp.), Sb2S3 (99.995%, Sigma Aldrich), Coin-type cells (Hohsen Corp., Japan),

Electrochemical characterization of antimony sulfide

Coin-type cells were assembled in an argon-filled glove box (O2 < 1 ppm, H2O < 1 ppm) using one layer separator (glass fiber) for NIBs and two layers of separators (Celgard and glass fiber) for LIBs. Elemental sodium or lithium served as both reference and counter electrodes. As electrolyte 1 M NaClO4 in PC was used for Na-ion batteries and 1 M LiPF6 in EC:DMC (1:1 by wt.) for Li-ion batteries. To improve cycling stability 3% of FEC were added to both electrolytes. Electrochemical measurements were performed using constant current mode for both, charge and discharge steps between 0.01–2.5 V for both Na and Li-ion batteries on a MPG2 multi-channel workstation (Bio-Logic).

Materials characterization

TEM samples were prepared by dropping a solution of Sb2S3 NPs onto standard amorphous carbon-coated TEM grids. TEM images were recorded using JEOL JEM-2200FS microscope operated at 200 kV, STEM images and EDXS spectrum were collected on FEI Talos F200X operated at 200 kV and equipped with Super-X EDS system (4 detector configuration). Scanning electron microscopy (SEM) measurements were done on a Quanta 200 F microscope (Thermo Fisher Scientific) operated at an acceleration voltage Vacc = 20 kV. Energy-dispersive X-ray spectroscopy (EDXS) was performed with an Octane SDD detector (EDAX (Ametec)) attached to the microscope column. Powder X-ray diffraction pattern was collected with STOE STADIP powder diffractometer.