In this study, we report a series of free-standing carbon nanofiber films interspersed with diverse coordinated single Sn atoms to explore the metal-heteroatoms and metal-carbon interactions during Na plating/stripping. Beyond the intrinsic activity of single Sn atoms, the surrounding structure of Sn atoms also displays enhanced sodiophilicity due to Sn-induced activation. With the increase of Sn atoms anchored on the carbon texture, the coordination mode of Sn migrates from 3N-Sn-O to N-Sn-3O configuration. By optimizing the Sn coordination environment and activating the surrounding structure, the carbon hosts exhibit robust sodiophilicity, enabling symmetrical batteries to achieve stable cycling for over 1200 h under 100% Na utilization rate, high current density (100 mA cm-2) and deposition capacity (100 mA h cm-2). By integrating Na3V2(PO4)3 (NVP) cathode, anode-free full cells exhibit a stable cycle of 700 cycles under 10 C, indicating a promising prospect for the practical applications of single-Sn-atom-activated carbon matrices with optimized coordination.
The atomic Sn dispersed 3D carbon hosts were obtained by pyrolyzing polyacrylonitrile (PAN) nanofiber precursors containing SnCl (Supplementary Fig. 1). The prepared carbon hosts were named as SnX@CNFs, where X represents the mass fraction of SnCl in the precursor. The SnX@CNFs films display extraordinary flexibility, can even be folded as a boat shape (Fig. 1a and Supplementary Fig. 2) and maintain a 3D network from its precursor (Supplementary Fig. 3), which ensures the continuity of electronic transfer (Fig. 1b and Supplementary Fig. 4). The specific surface area of carbon hosts decreases with the rise of Sn content (Supplementary Fig. 5).
Single Sn atoms are dispersed in the disordered carbon matrix in SnX@CNFs. Wide peaks at around 24° corresponding to the (002) layer of disordered carbon texture appear in the powder X-ray diffraction (XRD) patterns (Fig. 2a) of all SnX@CNFs. The amorphous structure can be confirmed by the high-resolution transmission electron microscopy (HRTEM, Fig. 1c, d and Supplementary Fig. 6) images and Raman spectra (Supplementary Fig. 7). The Sn dispersion state is determined by SnCl mass fraction in the precursors (X). For X ≤ 30, the absence of sharp diffraction peaks (Fig. 2a) indicates non-crystalline Sn, while for X = 40, the diffraction pattern of Sn crystal appears, consistent with Sn nanocrystals observed in the HRTEM image of Sn40@CNFs (Supplementary Fig. 6k), indicating the aggregation of Sn.
The isolated Sn atoms in SnX@CNFs (X ≤ 30) can be directly visualized by (AC-STEM, Fig. 1e-g) with corresponding intensity profiles. The images of the sample edge can better reflect the state of Sn atoms in the carbon layer because of the smaller thickness, which presents as the independent bright spots uniformly dispersed in the carbon matrix. After processing the carbon samples into fiber tips by field ion beam (Fig. 1h and Supplementary Fig. 8), the atomic distribution of Sn, C, N, and O could be manifested by atom probe tomography (APT, Fig. 1i). The results demonstrate that all elements are well-dispersed throughout the fiber matrix, which is consistent with the high-angle annular dark-field scanning transmission electron microscope (HAADF-STEM) with energy-dispersive X-ray (EDX) elemental mapping (Supplementary Fig. 6).
According to X-ray photoelectron spectroscopy (XPS), the Sn content in carbon nanofibers rises with the increasing X (Supplementary Fig. 9a). Wherein, the value for Sn30@CNFs is 22.22 wt%, which is comparable to that obtained from EDX (19.72 wt%, Supplementary Table 1). Because of the residual carbon impurities, thermogravimetric (TG, Supplementary Fig. 10) test results have a positive bias (30.1 wt%). Besides, fluctuations in N and O content with increasing Sn content suggest the change of the Sn-coordination environment (Supplementary Fig. 9b, c). In general, when the Sn content is below approximately 20 wt%, the Sn atoms tend to remain in an atomic state.
Single Sn atoms are incorporated into the carbon skeleton through co-coordination with N and O atoms, and the fine coordination structure is determined by Sn content. By adjusting X from 10 to 40, the coordination environment of Sn atoms transforms from 3N-Sn-O to 2N-Sn-2O and eventually to N-Sn-3O, demonstrating a Sn-concentration-dependent transition. This enables the controllable synthesis of carbon hosts with an optimal Sn coordination mode for guiding sodium deposition. The coordination environment of Sn was preliminarily revealed by XPS (Fig. 2c, d and Supplementary Fig. 11) and soft X-ray absorption spectra (sXAS, Supplementary Fig. 13). With the introduction of Sn, the N 1s XPS spectra of SnX@CNFs shows a new peak at 399.0 eV, in addition to the characteristic peaks at 398.2, 399.8, and 400.8 eV (corresponding to the N-6, N-5, and N-Q, respectively) in CNFs, which indicates the coordination between Sn and N. Similarly, the O 1s XPS spectra of SnX@CNFs also emerge a new peak at 530.8 eV, alongside the peaks at 532.3 and 538.2 eV, which are assigned to C = O and C-O configurations In addition, the C NMR spectra (Supplementary Fig. 14) for both CNFs and Sn10@CNFs show approaching chemical shift at around 112 ppm, assigned to sp hybridized carbon, indicating that the electrons from Sn has tiny shielding effect on C, reflecting the absence of Sn-C coordination. Thus, the primary coordination shell of Sn atoms consists of N and O ligands. Furthermore, as the Sn content in the carbon framework increases, the proportion of N from Sn-N coordination decreases, while the proportion of O from Sn-O coordination increases (Fig. 2b; Supplementary Fig. 12 and Supplementary Table 2), indicating a gradual replacement of N atoms by O in the coordination environment of Sn.
X-ray absorption near-edge structure (XANES) and extended X-ray absorption fine structure (EXAFS) provide the confirmation of the concentration-dependent coordination of Sn atoms. The white line peaks in Sn K-edge XANES spectra of SnX@CNFs (X ≤ 30) are positioned between those of Sn and SnO (Fig. 2e and Supplementary Fig. 15), indicating the +2 valence states, consistent with Sn 3d XPS spectra (Supplementary Fig. 11). The Fourier transformed (FT) k-weighted EXAFS spectra of SnX@CNFs show a predominant R-spacing peak at ≈1.53 Å (Fig. 2f), distinct from both the Sn-Sn peak in metallic Sn and Sn-O characteristic peak of SnO, indicating that Sn exists in an N, O co-coordination state without agglomeration. The peak position of O-Sn-N in the R-space gradually shifts to higher R value as the atomic concentration of Sn increases, indicating an enhanced anchoring effect of O on Sn atoms. Least-squares fitting of Sn K-edge EXAFS curves further confirm the concentration-dependent coordination of Sn, revealing an alternating anchoring mechanism (Fig. 2g and Supplementary Fig. 16). At low Sn content (Sn10@CNFs), the coordination configuration of Sn is 3N-Sn-O. For Sn20@CNFs, this changes to 2N-Sn-2O, and as the concentration of single Sn atoms increases, more O atoms coordinate with Sn to form N-Sn-3O in Sn30@CNFs with the mean bond lengths of 2.07 and 2.10 Å for Sn-O and Sn-N, respectively (Supplementary Table 3). The reason for this coordination configuration transition is that, at high terminal temperature, SAs prefer to coordinate with N, however, when increasing the amount of Sn source, the average amount of N bonding with each Sn decreases, resulting in a lower N/Sn ratio for each defect. Notably, the EXAFS curves of Sn, SnO, and SnO standard samples are also fitted in Supplementary Fig. 17. The peaks in EXAFS spectra of Sn reference are fitted as Sn-Sn scattering path (Supplementary Table 4), revealing the absence of Sn clusters in all SnX@CNFs samples. Both the EXAFS fitting curves of SnO and SnO show different dominant peaks from those of all SnX@CNFs samples, indicating the Sn atoms in SnX@CNFs also do not exist in the form of SnO and SnO. Furthermore, the wavelet transform-EXAFS (WT-EXAFS, Fig. 2h) with high resolution of both k and R spaces shows a shift in the maximum from 5.7 Å for Sn10@CNFs to 7.2 Å for Sn30@CNFs, which are significantly different from those recorded from Sn foil (Sn-Sn at 8.6 Å), SnO and SnO (7.0 and 11.5 Å arose from Sn-O and Sn-Sn, respectively), indicating the offset of the bond position arises from the influence of the coordination configurations.
Encouraged by the controllable coordination mode of Sn, the plating/stripping process of Na on the as-prepared carbon textures is expected to be optimized precisely. Density functional theory (DFT) calculations were initially performed to investigate the adsorption energies of Na atoms on SnX@CNFs. According to the experimental Sn coordination structures, three CNOSn models were optimized on the base of the pristine model, a 5 × 5 supercell of graphene containing 50 C atoms (Supplementary Data 1), as shown in Fig. 3a-d and Supplementary Fig. 19. For comparison, corresponding CNO models without Sn atom were also established (Supplementary Fig. 18). Sodium atoms are respectively placed at different adsorption sites numbered from 1 to 26 for calculation.
Except its own sodium adsorption activity, Sn atoms can activate the surrounding carbon structure, and its active effect is dominated by the coordination environment. The graphite-like sodiophobic carbon network is thermodynamically unfavorable for Na adsorption. In the CNO models (Supplementary Fig. 18 and Fig. 3e-h), apart from several adsorption sites located around the N-O macrocycle showing notably negative adsorption energies (E), the remaining adsorption sites exhibit a weak adsorption activity, indicating the non-spontaneous adsorption process. In contrast, the introduction of Sn atoms greatly enhances the Na adsorption capability of all C44NxOySn models. Sn atoms themselves exhibit high Na affinity. The E values of the sites located on top of Sn (No. 26 in Fig. 3a, d and No. 16 in Fig. 3b, c) is much lower than those at the same sites in the CNO. Similarly, the E of hollow sites in the four heterocycles connected to Sn is reduced. More interestingly, Sn atoms can activate surrounding inert sites. In the CNOSn models, the hollow sites of heterocycles not connected to Sn (N hollow or O hollow), and even the carbon hollow sites far from Sn atoms, exhibit lower E, demonstrating a Na adsorption activity, while these sites in the CNO models exhibit a thermodynamic disadvantage of Na adsorption. The activation of Sn on surrounding structures creates additional active sites for Na adsorption. Furthermore, the average Na adsorption energy on 3N-Sn-O sites surpass those of cis-2N-Sn-2O, trans-2N-Sn-2O and N-Sn-3O, emphasizing the importance of coordination configuration on Na adsorption.
The experimental results regarding Na plating/stripping dynamics on SnX@CNFs provide further confirmation of the theoretical calculations. Nucleation overpotential reflects the energy barrier of initial Na deposition, lower value suggests improved Na affinity of the hosts. Among all samples, at three different current densities, 0.5 mA h cm (Fig. 3i and Supplementary Fig. 20), 1 mA h cm (Supplementary Fig. 21) and 3 mA h cm (Supplementary Fig. 22), Sn10@CNFs (3N-Sn-O structure) shows a lower value than Sn20@CNFs (2N-Sn-2O) and Sn30@CNFs (N-Sn-3O). Although Sn30@CNFs have the highest loading mass of Sn single atoms, the increasing of single atom concentration did not further contribute to lower nucleation overpotential, which indicates there is another more important factor affecting overpotential than single atom content, emphasizing the importance of optimizing the coordination structure. Moreover, the cycling stability of SnX@CNFs hosts in half cells also correlates with the Sn excitation capabilities, which is dominated by its coordination environment. Under a low current density (0.5 or 1 mA cm) with a deposition capacity of 0.5 or 1 mA h cm, all the SnX@CNFs (X = 10, 20, and 30) exhibit a better plating/stripping stability with a cycle life of 2000 h compared with the original CNFs (Supplementary Figs. 20 and 21). At higher current density of 3 mA cm and higher capacity of 3 mA h cm (Fig. 3j), Sn10@CNFs electrode shows the most stable cycling performance with the average coulombic efficiency (CE) stabilizing at 100.0%. While, although the cycle life of Sn30@CNFs is also expanded, the fluctuating CE reflects a relative high nucleation barrier and incomplete stripping process (average CE: 99.5%). Thus, among all structures, Sn atoms can activate the surrounding structures, turning the inert position to activity sites, and Sn coordinated with more N atoms (3N-Sn-O structure) can further improve the nucleation behavior of Na.
The activation effect of Sn on surrounding structures arises from the strong metal-support interaction in the form of the injection of Sn valence electrons into the π orbitals of the carbon matrix. The Bader charge analysis directly reveals the electron transfer of Sn with ON-C (Supplementary Fig. 23). For Sn coordinated with more N atoms (3N-Sn-O), the higher charge suggests that the negatively charged N atoms draw electrons from Sn. While for Sn coordinated with more O atoms (N-Sn-3O), the charge transfer between Sn and O is weaker, resulting in a relatively low Sn charge. Consequently, with the increase of O ligand in the coordination environment, the activating capability of Sn gradually decreases, because the charge transfer between Sn and O is weaker than that of Sn and N. Thus, Sn coordinated with more N atoms shows a stronger interaction with the support carbon matrix.
The Sn10@CNFs with optimized coordination configuration (3N-Sn-O) were selected to study the sodium adsorption mechanism. Multi-stage active sites are used to describe the adsorption positions of CNOSn model (Fig. 4a-c, Stage I: the top site of atomic Sn, Stage II: the surrounding hollow sites of Sn, such as those containing N and O atoms, Stage III: hollow sites composed solely of C atoms). Charge density difference reveals the adsorption of Na on the Stage I site (Fig. 4a). Electron depletion around Na and accumulation around CNOSn reflect a strong interaction between Na⁺ and the carbon matrix via Sn, indicating the effective absorbance of Na on this active site. Experimentally, Sn10@CNFs exhibits an upward edge energy shift after sodiation in ex situ XANES (Supplementary Fig. 24), reflecting the charge transfer happens between Na and Sn. FT-EXAFS depicts detailed information about the interactions between Sn and Na atoms at different plating/stripping states (Fig. 4d). Upon Na nucleation (Fig. 4e), a predominant R-spacing peak emerges at ≈3.4 Å, which is distinguished from single scattering because it is longer than the Sn-Sn scattering path and thus attributed to second shell or multiple scattering. After plating 0.5 mA h cm of Na, a neo-peak appears at ≈2.5 Å in R-space, which is absent in Sn, SnO, SnO and various pristine SnX@CNFs. Fitting analysis attributes this peak to the Sn-Na scattering path, confirming the interaction between Sn active site and Na. The Sn-Na scattering path completely disappears after charging, indicating the complete stripping.
The Na adsorption on the Stage II sites was first explored by charge density difference (Fig. 4b). Electron depletion around N and O was observed after Na adsorption, indicating the interaction between Na and N. The shift of N 1s peaks in XPS towards higher binding energy upon Na plating (Fig. 4f) confirms this interaction, occurring specifically around the initial binding energy of 398.8 eV. Given the proximity of Sn-N binding energy to this region, the involvement of Sn-coordinated nitrogen atoms in Na adsorption is confirmed. Analogously, the obvious change of the dominant peak in O 1s XPS spectra also indicates the activity of O-configurations (Fig. 4g). Noticeably, at the stripping state, the influence of Na on the conjugated system is almost eliminated, manifested as the N 1s and O 1s binding energy returning to the initial position. As a comparison, the binding energies of N 1s and O 1s of CNFs show inconspicuous changes before and after Na plating, reflecting the Na-phobic property of N and O that were not coordinated with Sn. For Sn30@CNFs (N-Sn-3O configuration), although its N and O are active, the irreversible change of N 1s and O 1s binding energy reflect the incomplete sodium stripping (Supplementary Fig. 25), which is the root for the fluctuation of CE in half batteries with Sn30@CNFs electrode (Fig. 3j). In short, the reversible change in binding energy during plating/stripping processes of Sn10@CNFs reveals the Na adsorption ability of N and O which constitutes the Stage II active sites.
Besides, the charge density difference after adsorbing Na around the Stage III site hints the activity of carbon matrix (Fig. 4c). C magic angle spinning solid-state nuclear magnetic resonance (MAS-ssNMR) spectroscopy was conducted to confirm the Na adsorption process around C-contained structure (Fig. 4h). Different from the spectra of CNFs, where the C chemical shift decreases after sodiation due to the shielding effect caused by electron accumulation (Fig. 4i), the C chemical shift of Sn10@CNFs increases significantly after Na plating. This increase is attributed to the de-shielding effect of adsorbed Na on C atoms, corresponding to the blue region adjacent to C atoms in Fig. 4a-c. Notably, both CNFs and Sn10@CNFs show the peak locating at ca. 121 ppm which is attributed to the chemical shift of C in C=O after sodiation, indicating the carbonyl structure are similar in all samples, the movement of the chemical shift of the broad peak is not aroused from the C=O formation.
Projected density of states (PDOS) gives the confirmation of the interactions between Na and Sn, N, O, C elements based on the optimized structures (Supplementary Data 1). Compared with the absence of PDOS overlapping between Na and N, O elements in unmodified CNO models (Fig. 4k and Supplementary Fig. 26b, d), the activation effect of Sn towards the carbon matrix promoted the adsorption capacity of the originally inert sites for Na. For CNOSn models after adsorbing Na (Fig. 4j and Supplementary Fig. 26a, c), the overlapping of electronic distribution regions in s orbitals for Na and p orbitals for C, N, O, and Sn directly reflect the binding formation, indicating all elements contribute to the Na adsorption. Thus, except for the highly anticipated single atom sites themselves, the activation effect of single atoms on the surrounding structure is also crucial. In Sn single atom dispersed carbon texture, N, O, and C contribute abundant active sites and form multi-stage Na adsorption sites.
Benefiting from the multi-stage active sites motivated by coordination-optimized Sn, uniform deposition and stripping performance were achieved under high current density. The ab initio molecular dynamics (AIMD) simulations were first conducted to reveal the homogenous deposition of Na on Sn-modified carbon texture. The simulations begin with a sparse distribution of Na atoms, followed by a sequential deposition of additional atoms on two models (Fig. 5a): one containing a defect with three nitrogen and one oxygen (3N-O) and the other with additional Sn atoms in the 3N-O model (3N-Sn-O). The initial and final configurations for the molecular dynamics trajectories are provided in Supplementary Data 2. AIMD simulation gives the result that, on the 3N-O model, the aggregation of Na was observed at as early as 10 ps, indicating Na nucleation prefers agglomerating as a cluster instead of uniform distribution. However, the absence of Na cluster aggregation in the 3N-Sn-O model during the simulation reflects the strong interaction between the 3N-Sn-O modified carbon texture and Na atoms, manifesting as dispersed and uniform nucleation. Thus, Sn activated carbon substrate is expected to correct the vertical growth into planar deposition, avoiding the dendrites formation.
As indicated by AIMD, the dendrite-free Na plating/stripping process along Sn-separated fibers was recorded by ex situ SEM and in situ optical microscope. On CNFs, the irregularly nucleated Na presents as agglomerated bulk, subsequently accumulating and aggregating on plated Na (Fig. 5b). This uneven deposition promotes dendrite formation, as observed in the in situ optical microscopy (Fig. 5c). Dendrites on CNFs appeared after 45 min of plating under 2 mA h cm, and the formed dendrite detach from the host after 60 min, turning into dead Na. Furthermore, Na residues on stripped CNFs highlight the challenge of achieving complete stripping (Fig. 5b). In contrast, the nucleation performance of Sn-activated carbon skeletons is significantly improved. On both Sn10@CNFs (3N-Sn-O coordination, Fig. 5b) and Sn30@CNFs (N-Sn-3O structure, Supplementary Fig. 27), the Na accumulation process along the fibers reflects guidance for sodium deposition, which effectively utilizes the space of the 3D host. Even when uniformly grown Na completely fills the 3D network, the fiber morphology remains clearly visible without mossy Na dendrites in both top-view (Supplementary Fig. 28) and side-view (Supplementary Fig. 29). Meanwhile, the absence of dendrite formation observed on Sn10@CNFs during the electroplating process (Fig. 5c) further confirms that the sufficient multi-stage active sites break the tendency of sodium aggregation, transforming irregular nucleation into uniform deposition (Supplementary Fig. 30). Even when the deposition capacity is increased to 250 mA h cm at a current density of 5 mA cm, the capacity-voltage curve of the battery remains stable (Supplementary Fig. 31), and the morphology of plated Na is still flat (Supplementary Fig. 32), demonstrating the sodium affinity of the carbon skeleton. In short, dendrite-free fiber-guided nucleation morphology exhibits advantages during the plating/stripping process, allowing Sn10@CNFs to carry a high Na capacity and achieve complete stripping without residue.
Electrolyte consumption caused by unstable interface and side reactions is one of the key reasons for battery failure. The interface morphology of stripped carbon hosts after one cycle was also recorded to reveal the surface reactions. For CNFs, the by-products composed of C, O, and Na with an unconsolidated structure between fibers were observed (Supplementary Fig. 33a-c). Conversely, no distinct surface structure differing from the compact matrix was observed on Sn10@CNFs after 1 plating/stripping cycle, indicating relatively weak surface side reactions (Supplementary Fig. 33d-f). However, with the migration of the Sn coordination environment, by-products reappear and become increasingly pronounced from Sn20@CNFs to Sn40@CNFs (Supplementary Fig. 33g-o). Depth profiling XPS provides detailed surface component information about the hosts after one cycle. For CNFs (Supplementary Fig. 34), the SEI generated by the decomposition of ether-based electrolytes is mainly composed of organic components (e.g., O=C-O, C=O and C-O), which is the root for the fluffy and porous structure (Supplementary Fig. 33b). However, the SEI of Sn10@CNFs is mainly accompanied by beneficial components (Supplementary Fig. 35), such as C-F and Na-F, corresponding to the close-packed surface observed under TEM (Supplementary Fig. 33e). For Sn30@CNFs, the reappearance of the fluffy structure composed of organic components (Supplementary Figs. 33k and 36) is one of the reasons for its relative unstable electrochemical performance.
Symmetric cells were assembled to evaluate the long cycle plating/stripping stability of SnX@CNFs under a high DoD (100%). Supplementary Fig. 37 shows a comparison of the cycling stability of CNFs/Na and SnX@CNFs/Na electrodes at a current density of 10 mA cm with a fixed capacity of 10 mA h cm (pre-deposition: 10 mA h cm). The Sn10@CNFs/Na symmetric cells exhibit a relative constant voltage hysteresis of around 22 mV during its cycle life of 2800 h, which also reflects the diminishing side reactions on the interface. The Sn20@CNFs electrode, featuring a 2N-Sn-2O coordination mode exhibits higher voltage hysteresis with a slight increase. The cycle life of Sn30@CNFs/Na symmetric cell is limited by the significantly rising hysteresis voltage resulting from relatively intense side effects. In sharp contrast, the CNF/Na symmetric cells fail within a few dozen hours, characterized by massive irreversible side reactions and dendrite growth. At a high current density (50 mA h cm) with a capacity of 50 mA h cm (Fig. 5d and Supplementary Fig. 38), Sn10@CNFs/Na symmetric cells also demonstrate a long cycle life exceeding 1200 h under 100% DoD. When further increasing the fixed area capacity to 100 mA h cm, Sn10@CNFs/Na symmetric cells can even operate at 100 mA cm for 1200 h under fully stripped conditions with a tighter voltage hysteresis of around 70 mV, showcasing substantial advantages compared to other recently reported sodium hosts/anodes (Fig. 5g and Supplementary Table 5). Thus, the dendrite-free Sn10@CNFs electrode demonstrates a potential for application in sodium metal batteries (SMBs), benefiting from suppressed side reactions, enhanced sodiophilicity and regulated Na plating/stripping process.
The complete stripping performance of Sn10@CNFs underscores their application properties. Full cells were assembled by incorporating with SnX@CNFs/Na as the anode and NVP as the cathode (Supplementary Fig. 39). Upon overloading the pre-deposited Na on SnX@CNFs hosts, the charge/discharge profiles of Sn10@CNFs/Na||NVP full cells for the initial cycle at 100 mA g show a reversible capacity of 105.8 mA h g with the corresponding initial CE of 98.4% (Supplementary Fig. 40b). The Sn10@CNFs/Na||NVP full cells exhibit a high-rate performance with the specific capacities stable at 102.1, 99.8, 94.5, 89 mA h g at 200 mA g, 500 mA g, 1000 mA g, 2000 mA g and 5000 mA g, respectively, higher than those of the CNFs/Na||NVP full cells (Supplementary Fig. 40a). Furthermore, batteries with Sn10@CNFs/Na anodes can demonstrate a long cycle life in different electrolytes, delivering a high discharge capacity of 84 mA h g (Supplementary Fig. 41) after 2000 cycles at 1 A g. In comparison, the capacity of Na||NVP cells rapidly decay to below 60 mA h g after 570 cycles. Moreover, Sn10@CNFs/Na can withstand higher current density up to 2 A g, with a capacity of 78 mA h g after 2000 cycles (Supplementary Fig. 42b). The charge/discharge profiles show the characteristic of Na plating/stripping process on the anode of full-cell with overloaded Na (Supplementary Fig. 42a).
The above results indicate that Sn10@CNFs, serving as the host material for Na metal, are suitable for anode-free SMBs. When employing Sn10@CNFs without pre-deposition of Na as the host and high-loaded NVP (about 8.5 mg cm) as the cathode, with sufficient electrolyte, the anode-free SMB shows stable cycling performance (1500 cycles) and competitive reversible capacity (73 mA h g) under 5C (Fig. 5e). The "soft short" at the 1588th cycle (Supplementary Fig. 43b, c) indicates battery failure. The capacity loss of each cycle was summarized in Supplementary Fig. 43d, the average Na consumption (within 1500 cycles) could be calculated as 0.003 mA h per cycle, corresponding to 0.05% of active Na in electrolyte. Within 1500 cycles, the average CE of Sn10@CNFs||NVP anode-free SMB is 99.6%, the CE fluctuates with the range from 92 to 106% (Supplementary Fig. 43e, f). The dominant plateau region and limited slopping region (Fig. 5f) reflects the sodium is stored by being plated on the Sn10@CNFs hosts instead of being depleted on interphase. Supplementary Fig. 44b shows the electrochemistry characteristic of sodium-ion batteries, showcasing as wide oxidation-reduction peaks in cyclic voltammetry (CV) curves and slope curves in the voltage profiles. In contrast, the electrochemical features of anode-free batteries are different, with sharp oxidation-reduction peaks in CV curves and sustained charge/discharge plateaus (Supplementary Fig. 45b). Even under the charge/discharge condition of 10 C (Supplementary Fig. 45d), the anode-free cells demonstrate a long cycle stability (700 cycles) with an average CE of 98.1% and a high specific capacity of 83 mA h g in the last cycle. Moreover, even under 20 C (Supplementary Fig. 46), the Sn10@CNFs can support the battery operation for over 700 cycles with an average CE of 97.4%. Sn10@CNFs hosts provide a stable place for Na plating/stripping, enabling the slow Na consumption to compensate Na loss in irreversible plating/stripping (reflected by the charging capacity is slightly higher than the discharging capacity of the previous cycle). Because the charging capacity is slightly higher than the discharging capacity, the batteries can achieve a stable cycle under a CE less than 100% (Supplementary Table 7 and Supplementary Fig. 8). The performance of the anode-free batteries based on Sn10@CNFs host is comparable with most SMBs using NVP as cathode (Supplementary Table 6 and Fig. 5h).
To further emphasize the practicality of Sn10@CNFs hosts, we paired them with a different cathode material, NaNiFeMnO (NFM) (Supplementary Fig. 47). The electrochemical performance of NFM in half cells is shown in Supplementary Fig. 48. In anode-free Sn10@CNFs||NFM cells (Supplementary Fig. 49), the Sn10@CNFs delivered stable cycling for 300 cycles at 1 C (138 mA h g) and maintained meaningful performance even at 2 C (Supplementary Fig. 50). Overall, Sn10@CNFs effectively regulate Na deposition behavior, while the optimized Sn coordination structure suppresses surface side reactions, enabling 100% Na utilization under high deposition capacity and current density.
Anode-free pouch cells were assembled to further evaluate the suppression of dendrite formation and negligible side reactions on the Sn10@CNFs host, confirming its practical applicability (Supplementary Fig. 51). The Sn10@CNFs||NVP anode-free pouch cell, initially tested at 0.5 C (Supplementary Fig. 52), delivered a cumulative capacity (2.55 A h) with a cycle life of 200 cycles and an average CE of 98.1%. Even at a higher rate of 1 C (Fig. 5i), the cell maintained a cycle life of 120 cycles with an average CE of 98.2%. Charge/discharge profiles (Fig. 5j) indicate that Na plating/stripping is the primary contributor to capacity. At 1 C, the cumulative capacity of 1.2 A h is enabled by dendrite-free sodium deposition and minimal side reactions, effectively preventing short-circuiting and capacity degradation.