The energy crisis and environmental pollution have spurred the global pursuit of renewable energy sources (ex., solar and wind), but their intermittent and fluctuating natures calls for the development of low-cost large-scale energy storage technologies to ensure reliable energy supply^[1-3]. Lithium-ion batteries (LIBs) have been widely applied in portable electronics and electric vehicles owing to their exceptional energy/power density and long cycling life^[4-6]. Nevertheless, the scarcity of lithium resources severely limit their large-scale application, creating an urgent demand for alternative battery systems based on earth-abundant elements^[7]. Alternatively, sodium metal batteries (SMBs) have emerged as a promising candidate, thanks to the ubiquitous availability, low cost, and wide distribution of sodium resources^[8-10]. Moreover, sodium metal anodes (SMAs) exhibit analogous physical and chemical properties to lithium metal, including a high specific capacity (1,166 mAh g^-1 in theory) and a low redox potential [-2.71 V vs. Standard Hydrogen Electrode (SHE)], making them ideal for fabricating high-energy-density SMBs^[11-13]. Despite these inherent advantages, the practical application of SMBs is limited by intractable challenges associated with SMAs, such as uncontrolled dendritic Na growth and the continuous destruction-construction of the solid-electrolyte interphase (SEI) film during repeated plating/stripping cycles^[14-17]. These issues lead to low Coulombic efficiency (CE), rapid capacity fading, severe electrolyte consumption, and even safety hazards such as short circuits, which have become the primary bottlenecks hindering the commercialization of SMBs.

To address the above drawbacks of SMAs, extensive strategies have been explored, including the construction of artificial SEI layers^[18], optimization of electrolyte compositions and additives^[19], modification of current collectors^[20,21], and the development of all-solid-state electrolytes^[22-24]. Among these approaches, modifying copper current collectors with sodiophilic materials to in-situ form stable artificial SEI films has garnered tremendous attention, as it directly regulates Na nucleation and deposition behavior at the electrode/electrolyte interface, effectively suppressing dendrite growth and stabilizing SEI formation^[25-27]. Various sodiophilic metallic materials and their compounds have been explored for current collector modification, as they can form reversible Na-based alloys via alloying-dealloying reactions, which effectively lower the Na nucleation energy barrier and induce uniform Na deposition^[25,27,28]. For instance, flexible 3D carbon nanofiber frameworks anchored with Sb nanoparticles or encapsulated with ultrafine Sb₂S₃ nanoparticles have been demonstrated to enable stable Na plating/stripping for over 1,000 h and 2,800 h, respectively, by in-situ forming sodiophilic Na₃Sb sites and high Na⁺-conductive Na₂S matrices^[25,27]. Additionally, CeF₃@N-doped carbon and Na-Sn alloy-based porous carbon composites have been developed to construct NaF-rich SEI layers^[10,26], achieving ultra-long cycling stability for SMAs under various test conditions. These studies confirm that the rational design of sodiophilic interfaces with both preferential nucleation sites and high ion-conductive matrices is the key to realizing dendrite-free Na deposition.

Bismuth (Bi) and its derivatives have attracted increasing interest as sodiophilic modifiers due to their strong Na adsorption affinity and reversible Na-Bi alloying behavior^[29-31]. However, most reported Bi-based modifiers rely on metallic Bi nanoparticles or Bi/carbon composites, while Bi chalcogenides (e.g., Bi₂S₃) have rarely been explored as precursors for in-situ constructing composite sodiophilic interfaces. Bi₂S₃ possesses a unique layered crystal structure and inherent sodiophilicity, and its electrochemical reaction with Na⁺ can theoretically generate both Na-Bi alloy nucleation sites and Na₂S ion-conductive matrices- two key components for stabilizing SMAs- via a one-step conversion-alloying process. Furthermore, the nanostructured design of Bi₂S₃ (e.g., nanowires) could provide a porous framework to accommodate the volume change during Na plating/stripping.

Herein, we report the solvothermal synthesis of one-dimensional Bi₂S₃ nanowires and their application as a sodiophilic artificial interface layer on Cu current collectors for high-stability SMAs. The Bi₂S₃ nanowires undergo in-situ electrochemical conversion-alloying with Na⁺ to form a composite interface consisting of sodiophilic Na₃Bi alloy sites and highly Na⁺-diffusive Na₂S matrices, which is further covered by a NaF-rich artificial SEI film. Density functional theory (DFT) calculations reveal that Bi₂S₃ and its electrochemical derivatives exhibit much higher Na⁺ adsorption energies than bare Cu, while Na₂S possesses a significantly higher Na⁺ diffusion coefficient than conventional SEI components (Na₂O, NaF). Consequently, the construction of Bi₂S₃ nanowire layer efficiently reduced the Na nucleation energy barrier, accelerated Na⁺ diffusion kinetics and provided sufficient space to buffer volume expansion during Na plating/stripping. Thus, the Na/Bi₂S₃@Cu symmetric cell achieves a long lifespan of over 2,500 h with an ultralow overpotential of 15 mV at 2 mA cm^-2 and 2 mAh cm^-2. When pairing with commercial Na₃V₂(PO₄)₃/C (NVP/C) cathode, the SMB full-cells also exhibit excellent cycling performance and high-rate capability. This work not only develops a simple strategy for constructing stable sodiophilic interfaces for SMAs, but also provides new fundamental insights for the design of chalcogenide-based precursors for in-situ forming composite artificial SEI layers, paving the way for the development of high-performance SMBs.

Figure 1A shows the SEM image of the as-synthesized Bi₂S₃ product via solvothermal method, which exhibits well-defined one-dimensional nanowire-like morphology with lengths up to tens of micrometers. Moreover, the nanowires tend to hierarchically bundle at one end. Figure 1B shows the TEM image of a single Bi₂S₃ nanowire with inset showing the Selected Area Electron Diffraction (SAED) pattern, revealing its single-crystalline structure. As shown in the High-Resolution Transmission Electron Microscopy (HRTEM) image [Figure 1C], two sets of lattice fringes with interplanar distances of 0.8 nm and 0.4 nm can be clearly observed, which can be correspondingly indexed to the (101) and (20-2) crystallographic planes of Bi₂S₃^[32]. In addition, the angle between (101) and (20-2) is measured as about 90°, which is consistent with the SAED pattern analysis. Figure 1D displays the HAADF-STEM image of a bundle of Bi₂S₃ nanowires which are apparently rooted together at one end. Furthermore, the corresponding EDS analysis shows that the elemental maps for Bi and S are well overlapped within the nanowires [Figure 1E and F], confirming the homogeneous elemental distribution, uniform stoichiometry, and high phase purity of the as-synthesized Bi₂S₃ nanowires.

Figure 1G presents the XRD pattern of the Bi₂S₃ product, where all the diffraction peaks can be well indexed to the orthorhombic phase of Bi₂S₃ (JCPDS No.97-017-1864) with high purity^[32]. The inset of Figure 1G schematically illustrates the supercell structure of Bi₂S₃, which consists of tightly bound ribbons extending along the (011) direction^[33], and this is consistent with the HRTEM analysis [Figure 1C]. Figure 1H presents the survey XPS spectrum of the Bi₂S₃ product, where characteristic peaks of Bi and S elements are apparently observed. Additionally, O 1s and C 1s signals are present, which are likely attributed to surface adsorption of atmospheric species. Figure 1I displays the XPS spectrum of Bi 4f along with S 2p, in which two pairs of strong doublet peaks locate at 164.6/159.2 eV and 163.5/158.1 eV, corresponding to the Bi 4f_5/2/Bi 4f_7/2 with the chemical state of Bi³⁺ in the Bi-S and Bi-O, respectively. In addition, two minor peaks located at 161.9 eV and 160.7 eV are assigned to S 2p_1/2 and S 2p_3/2, respectively, suggesting the presence of S^2- species^[34]. Moreover, the high-resolution S 2s XPS spectrum exhibits a peak at 225.6 eV [Supplementary Figure 1], further confirming the sulfide state in the Bi₂S₃ nanowires. The coexistence of bismuth and sulfur spectral features confirms the successful formation of Bi-S bonds within the synthesized product.

Figure 2A shows the CV profile of the Bi₂S₃ electrode in the first cycle, highlighting its characteristic sodiation and desodiation behavior. In the cathodic scan, three distinct reduction peaks locate at approximately 1.00, 0.63 and 0.44 V, which are consistent with the voltage plateaus in the galvanostatic discharge profile [Figure 2B]. The minor peak at 1.00 V is related to the Na⁺ intercalation into the Bi₂S₃ lattice (Bi₂S₃ + Na⁺ + e^- → NaBiS₂), while the strong peak at 0.63 V corresponds the reduction of NaBiS₂ into metallic Bi (NaBiS₂ + Na⁺ + e^- → Bi + Na₂S). Furthermore, the sharp peak at 0.44 V is assigned to the alloying reaction (Bi + Na⁺ + e^- → Na₃Bi)^[34]. In the anodic scan, two distinct peaks locate at 0.61 V and 0.76 V, which correspond to the dealloying of Na₃Bi and the oxidation of metallic Bi, respectively.

To further elucidate the structural evolution upon sodiation, ex-situ XRD analyses were conducted on the cycled Bi₂S₃ electrodes at different states. When initially discharging to 0.01 V, Bi₂S₃ was electrochemically reduced into Bi and then Na₃Bi alloy, accompanied by the final formation of Na₂S, which was confirmed by the XRD pattern (Figure 2C, pattern I). At this state of 0.01 V, the diffraction peaks for Bi₂S₃ almost disappeared, while new peaks emerge that match with cubic Na₂S (JCPDS No. 97-064-4959), metallic Bi (JCPDS No. 97-006-4704), Na₃Bi alloy (JCPDS No. 97-067-1311) and cubic NaBiS₂ (JCPDS No. 97-002-8698). Further after plating a capacity of 1 mAh cm^-2, only XRD pattern for Na metal appears, suggesting the Na deposition on top of the Bi₂S₃-derived substrate. When initially stripped and charged to 0.5 V, the initial Coulombic efficiency is as low as 32%, suggesting large amount of Na was implanted into Bi₂S₃ serving as preferential nucleation sites. In addition, XRD pattern reveals the formation of Bi and NaBiS₂ after stripping to 0.5 V in the 5th cycle, and both display high Na⁺ adsorption energies, which facilitate the uniform Na plating in the following cycles. Figure 2D depicts the TEM image of the discharged Bi₂S_¬3 nanowire at 0.5 V, which shows the well retained 1D morphology after sodiation. Figure 2E shows the HRTEM image of the discharged electrode, where metallic Bi nanograins are embedded within the Na₂S matrices, and the enlarged lattice fringes with d-spacing of 0.331 nm can be ascribed to the (200) plane of metallic Bi. Figure 2F displays the HAADF-STEM image of the discharged Bi₂S_¬3 nanowire with corresponding EDS mapping of Bi, S and Na elements, which are uniformly distributed within the nanowire.

Figure 2G reveals the sodium adsorption energies on Cu (111), Bi (001), Na₃Bi (001), Bi₂O₃ (001), Bi₂S₃ (110) and NaBiS₂ (110) surfaces. The calculated adsorption energies are negative, indicating that these adsorptions are exothermic processes. The equivalent values are -1.753, -1.242, -1.375, -1.065 and -0.801 eV for Bi₂S₃ (110), NaBiS₂ (110), Bi₂O₃ (001), Na₃Bi (001) and Bi (001), respectively, which are more negative than of Cu (111) (-0.696 eV), suggesting the stronger adsorption and higher interfacial binding energies on the Bi-based substrates. Thus, the Bi₂S₃ surfaces are highly sodiophilic with greater affinity ability to anchor Na⁺ ions, as compared to metallic Bi and its alloy or oxide counterparts, which are critical for the uniform Na nucleation and subsequent plating. Additionally, Bi₂S₃ upon sodiation would produce NaBiS₂, metallic Bi and finally Na-Bi alloy sequentially, which also show more sodiophilic ability than that for bare Cu substrate, leading to the reduction of the energy barrier for Na nucleation. In order to study the Na⁺ diffusion behavior with different sodium-based compounds, the diffusion coefficients of Na₂O, Na₂S, and NaF were calculated via DFT calculation (see details in supporting information) and compared in Supplementary Figure 2 and Supplementary Table 1. The findings indicate that the diffusion coefficient of Na₂S (7.120 × 10^-9 cm^-2/s) is much higher than those of Na₂O (5.551 × 10^-10 cm^-2/s) and NaF (1.548 × 10^-10 cm^-2/s), suggesting the type of anions has a pronounced affection on the diffusion behaviors of sodium-based compounds, as the anionic structural properties greatly affect the lattice density and ion mobility. Importantly, due to the much higher diffusion coefficient of Na₂S^[25], it could effectively alleviate the volume expansion of Na-Bi alloying and facilitate the homogeneous distribution of sodium ion flux^[35].

To evaluate the reversibility of sodium plating/stripping during cycling, asymmetric cells were fabricated using Bi₂S₃@Cu (or Cu) foils as sodium plating substrates and bare Na foil as counter electrode. As illustrated in Figure 3A, the sodium plating on Bi₂S₃@Cu exhibits a lower nucleation overpotential of only 7 mV than that for the Cu electrode (43 mV) at 1 mA cm^-2. The small overpotential can be attributed to the highly sodiophilic nature of Bi₂S₃ and its derivates (ex., Bi and Na-Bi alloy) with higher adsorption energies, which facilitate the uniform sodium nucleation and enhance homogeneous sodium deposition, and this is supported by the DFT calculations [Figure 2G]. Figure 3B compares the Coulombic efficiencies (CEs) of the Na plating/stripping on Bi₂S₃@Cu and Cu substrates upon cycling at 5 mA cm^-2 with 1 mAh cm^-2, and CE as a critical indicator reveals the electrochemical stability, which is defined as a capacity ratio of sodium stripping to plating in each cycle. Apparently, the sodium plating/stripping on the Bi₂S₃@Cu electrode demonstrates highly stable CEs with values of up to 99.9% during 1,000 cycles, indicating the extremely high reversibility of the sodium plating/stripping process. In contrast, the bare Cu electrode exhibits significant CE variations and loses functionality after tens of cycles, which have been attributed to unstable SEI formation and mossy Na deposition^[36].

Figure 3C displays the voltage-capacity profiles of the sodium plating/stripping on the Bi₂S₃@Cu electrode, which are highly overlapped with voltage hysteresis increasing only slightly from 21 mV in the first cycle to 42 mV after 1,000 cycles at 5 mA cm^-2 and 1 mAh cm^-2. In comparison, the Na plating/stripping on bare Cu substrates displays larger voltage hysteresis, escalating from 68 mV (1st cycle) to 130 mV (500th cycle) under the same condition [Figure 3D], along with inferior cycling stability. These results confirm that the construction of sodiophilic Bi₂S₃ buffering layer on Cu substrate significantly enhances the reversibility of sodium plating/stripping, which efficiently suppress the interfacial side reactions and uncontrolled dendrite growth.

To evaluate the efficacy of Bi₂S₃ in suppressing sodium dendrite growth, symmetric cells were assembled using two identical electrodes. As depicted in Figure 3E, the Na/Bi₂S₃@Cu-based symmetric cell demonstrates exceptional long-term sodium plating/stripping stability with a lifespan of 2,500 h and a low overpotential of 15 mV at 2 mA cm^-2 and 2 mAh cm^-2. In contrast, the bare Na@Cu-based symmetric cell exhibits unstable sodium plating/stripping behavior, with the overpotential abruptly exceeding 0.5 V after 260 h, indicating severe polarization under identical current conditions. Even at 3 mA cm^-2 and 1 mAh cm^-2 [Figure 3F], the Na/Bi₂S₃@Cu-based symmetric cell still achieved a long stable lifespan of over 1,200 h with a small overpotential of 21 mV. Comparatively, the bare Na@Cu-based symmetrical cell displays apparent voltage variations with gradually increased overpotentials, and a steep increase after around 600 h. It’s noteworthy that the voltage-time profiles of Na/Bi₂S₃@Cu-based symmetric cells at various cycles exhibit well-approximated square waves [Figure 3F], indicating the suppression of sodium dendrite growth and interfacial side reactions^[37]. By comparison, the bare Na@Cu-based cell shows highly irregular voltage-time curves, suggesting the formation of disordered sodium dendrites and the continuous SEI breaking/repairing, which would lead to unnecessary electrolyte consumption and finally result in battery failure.

Figure 3G depicts the voltage-time profiles of the Na/Bi₂S₃@Cu-based symmetric cells at 5 mA cm^-2 and 5 mAh cm^-2, which still exhibits highly reversible sodium plating/stripping behavior for over 1,000 h, exhibiting a low overpotential of 33 mV. Moreover, Figure 3H shows the rate performance rate of the Na/Bi₂S₃@Cu-based symmetrical cell, which was test at a capacity of 1 mAh cm^-2 with varied current densities ranging from 2 to 10 mA cm^-2. Remarkably, the cell displayed small overpotentials of 6.7, 10.4, 17.5 and 39.4 mV at current densities of 2, 3, 5 and 10 mA cm^-2, respectively. When recovering the current density to 2 mA cm^-2, an overpotential of 6.6 mV can be achieved. These results highlight the excellent rate performance of the Na/Bi₂S₃@Cu electrode and the robustness of Bi₂S₃ buffering layer on suppressing the sodium dendrite growth and regulating the uniform sodium deposition. To better understand the excellent sodium plating/stripping behavior, EIS was carried out on the symmetric cells. Figure 3I displays the Nyquist plots of Na/Bi₂S₃@Cu symmetric cells after different cycles, which show low interfacial resistances of 1.1, 5.8 and 7.8 Ω after 1st, 200th and even 1350th cycles, corroborating the significantly enhanced electrochemical stability. The excellent sodium plating/stripping performance of Na/Bi₂S₃@Cu electrode could be primarily attributed to the construction of sodiophilic Bi₂S₃ buffering layer with higher sodium adsorption energy, which in situ converted into sodiophilic Bi and Na-Bi grains, as well as the Na₂S matrices with higher sodium diffusion coefficients. These characteristics are highly beneficial for uniform sodium plating. Moreover, Supplementary Figure 3 and Supplementary Table 2 summarized the electrochemical performances of various sodium metal anodes, and the Na/Bi₂S₃@Cu symmetric cell also displayed outstanding overall performance.

Figure 4A schematically illustrates the different Na plating behaviors on the bare Cu and Bi₂S₃@Cu substrates. It’s well recognized that the coarse surface of bare Cu substrate usually leads to the random sodium nucleation and non-uniform deposition with un-avoided dendrite formation^[38,39]. In contrast, owing to the high adsorption energy of Bi₂S₃, the sodium ions are anchored and readily reacted with Bi₂S₃ in the initial plating stage, forming sodiophilic Bi and Na-Bi grains which were embedded into the Na₂S matrices with high sodium ion diffusivity. These sodiophilic grains thus facilitate the homogeneous sodium nucleation and subsequent uniform dendrite-free sodium deposition, which were verified by the ex-situ SEM analysis. Figure 4B shows the SEM image of Na-plated Cu substrate with areal capacity of 1 mAh cm^-2 at 1 mA cm^-2 after 70 cycles, apparently displaying mossy and dendritic Na morphologies, while the silver-like Na metal on Cu substrate displays apparently rough surface and uneven distribution (inset of Figure 4B). After subsequent stripping, dead Na can be observed, further confirming the dendrite formation [Figure 4C].

Supplementary Figure 4 reveals that the pristine Bi₂S₃ nanowires are coated onto the Cu substrate, forming porous nanowire networks, within which the volume can efficiently accommodate the expansion after Na deposition. Consequently, with Bi₂S₃ nanowire networks as an artificial buffering layer, the Bi₂S₃@Cu effectively achieve the uniform Na plating/stripping [Figure 4D-G]. Figure 4D shows the SEM image of Na-plated Bi₂S₃@Cu after 600 cycles at 1 mA cm^-2 and 1 mAh cm^-2, where no Na-dendrite can be observed, and photo of the electrode displays uniform silver-like Na metal layer with smooth surface. Even at a higher areal capacity of 5 mAh cm^-2 after 120 cycles [Figure 4F], both SEM image and electrode photo reveal the uniform Na depositon without dendrite formation. Oppositely, after complete Na-stripping, the photos of the desodiated Bi₂S₃@Cu electrodes display dark in color, and no silver-like “dead Na” appears. Interestingly, the cycled Bi₂S₃ nanowires still retain the 1D morphology [Figure 4E and G], but their volumes are expanded owing to the volume changes after repeated sodiation/desodiation cycling. Remarkably, no mossy or dead Na are observed in the Bi₂S₃@Cu electrode even upon long-term plating/stripping cycling, suggesting Bi₂S₃ effectively inhibits the sodium dendrite formation.

Moreover, in-situ optical microscopy was further applied to dynamically observe the sodium plating/stripping process on the bare Cu and Bi₂S₃@Cu substrates in asymmetric batteries. As shown in Supplementary Figure 5A-D, upon plating at 1 mA cm^-2 for 60 min, the cross-section of Bi₂S₃@Cu maintains flat topography throughout the cycling process, indicating no apparent dendrite generation and the superior dendrite inhibition effect of Bi₂S₃ layer. But for bare Cu substrate, localized bright spots can be apparently observed after about 15 min of plating [Supplementary Figure 5E-H], suggesting the formation of sodium bulks and dendrites, and this can be explained by the tip discharge effect due to the inhomogeneous electric field on the surface^[40]. Such sharp contrast further highlight the critical role of Bi₂S₃ in enabling the homogenous deposition of sodium and good inhibition of dendrite growth^[41].

To investigate the composition and chemical situations of the Bi₂S₃ induced artificial interface layer, ex-situ TOF-SIMS analysis was performed on the Na-stripped Bi₂S₃@Cu after 10 cycles [Figure 5A], revealing the presence of broken inorganic fragments (ex., NaF^-, NaO^-, NaS^-, NaBiS₂^-, etc.) and organic derivatives (C₂HO^-, arising from the decomposition of solvent molecules). As revealed by the 3D-mapping images with in-depth profiling along z-axis, NaF^- and NaO^- species are well distributed within the SEI layer, while C₂HO^- mainly presents in the surface region. In contrast, Bi/S-related species (ex., NaS^-, NaBiS₂^-) predominantly exist in the inner region, indicating the SEI layer is rich in inorganic species, which are covered on the sodiophilic Bi₂S₃-related derivates. Consequently, the Bi₂S₃-induced artificial SEI layer with uniform distribution of different inorganic phases facilitates the well-distributed Na⁺ infiltration at the same rate, thus leading to the uniform Na deposition^[26,42,43].

To investigate the interactions of PF₆ anion with different substrates, DFT simulations were performed to calculate the binding energies (E_b) of PF₆ anion on various substrates. It’s worth noting that the high binding energies between the substrate and anion would facilitate the dissociation of sodium salt, thus releasing more free Na⁺ ions for active charge transfer^[10]. Moreover, the confined anion migration would increase the Na⁺ ion transport number and enhance ionic conductivity, thereby leading to the enhanced migration kinetics. The computational results reveal that Bi₂S₃ and NaBiS₂ demonstrate higher binding energies of -1.78 eV and -1.83 eV than metallic substrates, suggesting a more pronounced affinity with PF₆ [Figure 5B]. On one hand, the strong interaction leads to the surface accumulation of anions and affects the solvation structure, thus avoiding the direct contact between solvent molecules and Na metal. On the other hand, the confined anion migration prolongs the duration of Na⁺ nucleation stage, thereby allowing the refined grain nucleation and homogeneous Na deposition. Moreover, the strong binding with PF₆ anion would contribute to the construction of stable NaF-rich SEI, which thus suppresses the hazardous side reactions. Remarkably as revealed by the TOF-SIMS results, the significant content of NaF on the surface corroborate the calculation results. In addition, to investigate the stability of the artificial SEI layer, XPS analyses were performed on the Na-stripped Bi₂S₃@Cu electrode after 1 and 10 cycles [Supplementary Figure 6]. Note that the electrolyte of 1M NaPF₆ in DIGLYNE contains C, Na, O, F and P, which are the primary components of SEI layer, as confirmed by the XPS results. It’s noteworthy that there is no significant variation of Na-F, Na-O, C-O and P-F bonds for the electrodes after 1 and 10 cycles, implying the high structure stability of the artificial SEI layer. However, in the S 2p XPS spectra, the S 2p peak is quite weak after 1 cycle and disappears after 10 cycles, suggesting Na₂S exists in the inner part, which is in agreement with the TOF-SIMS analysis.

To further demonstrate the efficacy of Na/Bi₂S₃ anode in practical application, full cells were constructed by paring with commercial NVP/C cathode (inset of Figure 6A). Figure 6A compares the cycling performances of the NVP/C‖Na/Bi₂S₃ and NVP/C‖Na full cells at 1 C (1 C = 117 mAh g^-1), and the NVP/C‖Na/Bi₂S₃ exhibits a higher specific capacity of 80 mAh g^-1 with a higher retention of 84.5% after 300 cycles at 1 C, as compared with NVP/C‖Na cell using Na metal as anode (72 mAh g^-1 with 80.0% retention). Figure 6B depicts the rate performances of both cells at different rates ranging from 2 C to 10 C, and NVP/C‖Na/Bi₂S₃ shows higher capacities under different rates than those for NVP/C‖Na, suggesting the better rate capability after intruding Bi₂S₃. Remarkably, NVP/C‖Na/Bi₂S₃ delivered a high specific capacity of 75.7 mAh g^-1 even at 10 C (while 60.8 mAh g^-1 for NVP/C‖Na), which can return to original value (88.3 mAh g^-1) after shifting back to 2 C (while 83.4 mAh g^-1 for NVP/C‖Na). Moreover, Figure 6C and Supplementary Figure 7 show the corresponding charge-discharge profiles at various rates, and the full cells with Na/Bi₂S₃ anode showed well-defined plateau at around 3.4 V but with reduced polarization and low voltage hysteresis, suggesting the more stable and efficient electrochemical process of Na/Bi₂S₃ than that for the bare Na anode.

Figure 6D displays the long-term cycling stability of both full cells at a high rate of 10 C, the NVP/C‖Na/Bi₂S₃ full cell can survive for 2,500 cycles, delivering a capacity retention of 70% with a capacity decay rate of 0.012% per cycle. In contrast, the NVP/C‖Na full cell shows a rapid capacity fading with 37.2 mAh g^-1 and 51% retention after 250 cycles under the same conditions. In addition, EIS was utilized to investigate the reaction kinetics of both kinds of full cells. As revealed by the Nyquist plots [Figure 6E], NVP/C‖Na/Bi₂S₃ displays a lower charge-transfer resistance (R_ct) of about 5 Ω than that of NVP/C‖Na (18 Ω, Supplementary Figure 8). Even after cycling at a high rate of 10 C, the R_ct remains at low values of 13 Ω after 100 cycles and 24 Ω after 1,000 cycles. Contrarily, the NVP/C‖Na cell showed a fast increased R_ct of 89 Ω after only 250 cycles. Such difference reveals that the bare Na anode undergoes great interfacial degradation upon cycling, probably because of thick SEI formation arising from the dendrite growth and inactive Na formation. These observations further reveal that the construction of Bi₂S₃ protective layer greatly contributed to the homogeneous and stable Na plating behavior, which finally enhanced the reaction kinetics and electrochemical performance of SMBs.

In summary, we have demonstrated the fabrication of Bi₂S₃ nanowires via a facile solvothermal route and their application as sodiophilic protective layers, which efficiently induced the uniform dendrite-free Na deposition. The Bi₂S₃ nanowires with higher Na adsorption energy underwent conversion-alloying reactions forming sodiophilic Na-Bi alloy sites and highly Na⁺-diffusive Na₂S matrice, which synergistically lowered the Na nucleation barrier and enhanced Na⁺ transfer kinetics, thus inhibiting the Na dendrite growth. Consequently, the Na/Bi₂S₃-based symmetric cell displayed exceptional long-term cycling stability upon repeated plating/stripping cycling (2,500 h with small voltage hysteresis of 15 mV at 2 mA cm^-2 and 2 mAh cm^-2). Furthermore, the full cells consisting of commercial Na₃V₂(PO₄)₃/C cathode and Na/Bi₂S₃ anode demonstrated better cycling stability and higher rate performance than those for bare Na anode, implying the great potential in practical application. More importantly, this study validated Bi₂S₃ nanowires as an efficient sodiophilic modifier for SMAs, and proposed a scalable strategy for artificial SEI design- using metal chalcogenides as precursors to in-situ construct composite interfaces with alloy nucleation sites and ion-conductive matrices. This design provides valuable insights for developing dendrite-free alkali metal anodes and advancing the practical application of high-performance sodium metal batteries.