Enhancing low-temperature durability and sodium-ion transport of anode-free sodium metal batteries through utilization of a solvent adsorption separator

Abstract

A battery with high-energy density at low-temperature has been actively pursued in energy storage systems for decades. Anode-free sodium metal batteries (AFSMBs) have emerged as a promising battery configuration for enhanced energy densities by eliminating conventional anode materials. Nevertheless, their practical implementation in low-temperature environments remains constrained by two critical challenges: insufficient dynamics for sodium plating/stripping processes during cycling and instability of the solid electrolyte interphase. Herein, a strategy of multifunctional separator design by employing a solvent adsorption separator with Na supplementation (SAS-N) is proposed to enhance the low-temperature performance of AFSMBs. SAS-N acts as a supplemental sodium reservoir to mitigate irreversible sodium depletion and enhance interfacial compatibility through improved electrolyte wettability. Furthermore, SAS-N modulates more contact ion pair solvation structures, facilitating the formation of an inorganic-rich solid electrolyte interphase (SEI). This reconstructed interface simultaneously stabilizes electrochemical reactions at the electrode/electrolyte interface and accelerates sodium-ion transport kinetics at low temperature. SAS-based AFSMBs demonstrate ultralong-term cyclability, retaining 95.06% capacity over 600 cycles at 25 °C while sustaining 92.53% capacity retention through 1000 cycles under harsh −20 °C operation. This work provides a new approach of separator engineering to improve the low-temperature performance of AFSMBs.

---

Broader context

With the accelerating global transition towards carbon neutrality, the demand for low-cost, high-energy, and safe electrochemical energy storage technologies is rapidly increasing. Sodium-based batteries are emerging as promising candidates to replace lithium-ion systems owing to the abundance and broad distribution of sodium resources. Among them, anode-free sodium metal batteries (AFSMBs) eliminate excess metallic sodium and conventional anode materials, delivering higher energy density and simplified manufacturing. However, their viability remains severely limited, especially at low temperatures, where sluggish sodium plating/stripping kinetics and unstable solid electrolyte interphases (SEI) cause fast capacity decay and safety concerns. Previous strategies have focused on electrolyte formulation or current collector engineering, but often require complex molecular designs or costly materials, with limited effectiveness under sub-zero conditions. Herein, we report a multifunctional separator strategy based on a solvent adsorption separator with sodium supplementation (SAS-N) that provides a fundamentally new route to address the dual challenges of interfacial instability and sodium depletion at low temperatures. The SAS-N not only improves electrolyte wettability and sodium-ion transport, but also modulates local solvation structures to promote robust, inorganic-rich SEI. This simple and scalable design enables remarkable cycling performance of AFSMBs even at −20 °C.

Introduction

Rechargeable batteries with high energy density, low cost, and safety are urgently needed for smart grid-scale energy storage and electric vehicles. Sodium-ion batteries (SIBs) have the advantage of low-cost compared with lithium-ion batteries (LIBs) due to the abundant availability of sodium resources.^1–4 The energy density of SIBs, however, is not ideal, lower than that of commercial LIBs (160 Wh kg⁻¹ for LiFePO₄). To improve the energy density of SIBs, using sodium metal as an anode is one of the best methods, due to the high theoretical specific capacity (1165 mAh g⁻¹) and low redox potential (−2.714 V vs. standard hydrogen electrode).^5–7 Although these properties hold promise for significantly higher energy densities, sodium metal batteries face commercialization challenges including safety concerns, manufacturing limitations in producing ultrathin metallic sodium foils, and elevated production costs. Anode-free sodium metal batteries (AFSMBs), an emerging design, eliminate the use of metallic sodium anodes, and have significant advantages in terms of safety, energy density and production cost.^8–10 Therefore, AFSMBs are considered promising candidates for next-generation energy storage systems. The cycle life of AFSMBs with limited sodium, however, is not satisfactory, due to the low cycle reversibility derived from the uneven sodium plating and side reactions with the electrolyte.¹¹

To improve the performance of AFSMBs, various reliable solutions have been proposed to regulate sodium deposition by modifying the current collectors and regulating the electrolyte. For current collector modifications, sodiophilic metal coatings (e.g. In,¹² Sb,¹³ Sn,¹⁴ Zn,¹⁵) have been employed to regulate sodium deposition kinetics by guiding uniform metal nucleation, effectively suppressing dendrite growth and significantly extending cycling durability. With respect to electrolyte optimization, many attempts have been made to adjust the SEI components by regulating the solvation structure of sodium ions, including the use of diluent solvents,¹⁶ electrolyte additives,¹⁷ and high-concentration or locally high-concentration electrolytes (HCE or LHCE).¹⁸ These reported strategies enhanced the cycling performance of AFSMBs to some degree at room temperature, but their performance at low temperatures is still poor. Notably, dendrite growth is aggravated at low temperature owing to insufficient dynamics, resulting in lifespan degradation of the battery.¹⁹ To date, only a few works have focused on improving the performance of AFSMBs at low temperature, by modulating the electrolyte to enhance the ionic conductivity and Na deposition reversibility. These works on electrolyte optimization, however, require specially designed molecules (such as 2-methyl tetrahydrofuran (MeTHF) and 1,2-diethoxyethane (DEE)) that involve complex synthesis processes and a high price.^20,21 Therefore, it is crucial to develop new approaches for improving the performance of AFSMBs at low temperatures.

The separator, one crucial internal component of rechargeable batteries, is electronically insulated, and its pores act as electrolyte channels for sodium ion transport between the cathode and anode. Commercially used polypropylene (PP) separators exhibit the shortcomings of uneven porosity and insufficient wettability in sodium-based batteries, resulting in nonuniform distribution of Na ions and subsequent uneven Na deposition (Scheme 1(a)). Shen et al. demonstrated that a well-designed separator modified with 2D diamond and sodiophilic zinc nanoparticles accelerates sodium ion migration and promotes uniform flux, thereby effectively preventing dendrite growth.²² Additionally, a separator modified with Na₃P could provide an additional sodium source to improve the coulombic efficiency (CE) of sodium metal batteries.²³ Therefore, it is crucial to modify commercial separators using simple and efficient methods to enhance battery performance. So far, no work has investigated whether the separator design could influence the low-temperature performance of AFSMBs. Additionally, the impact mechanism of the modified separator on the performance of AFSMBs is unclear.

Scheme 1 Schematic diagram of the Na ion deposition behaviors using a conventional (a) PP separator and (b) SAS-N.

Herein, we propose a solvent adsorption separator with Na supplementation (SAS-N) strategy to enhance the low-temperature performance of anode-free sodium metal batteries (AFSMBs). This strategy introduces functional surface properties to the separator, enabling it to act as a supplementary sodium reservoir that compensates for irreversible sodium loss during cycling. Additionally, the SAS-N improves electrolyte wettability, which ensures a uniform sodium-ion flux and reduces interfacial resistance, effectively suppressing dendrite formation. Furthermore, SAS-N regulates the local solvation environment by promoting the formation of contact ion pair (CIP) configurations, facilitating the development of an inorganic-rich solid electrolyte interphase (SEI) that stabilizes the electrode/electrolyte interface and enhances sodium-ion transport kinetics through the interphase, thereby improving battery performance at low temperatures (Scheme 1(b)). The synergistic effects of ion flux regulation, interface stabilization, and solvation structure modulation enable AFSMBs with SAS-N to achieve outstanding long-term cycling stability—maintaining 95.06% capacity retention over 600 cycles at 25 °C and 92.53% after 1000 cycles at −20 °C. These findings highlight the potential of separator engineering through solvent interaction control as a robust and scalable approach to improve the low-temperature electrochemical performance of next-generation AFSMBs.

Results and discussion

Influence of the solvent adsorption separator on the electrolyte solvation structure

The solvation structure at the interface plays a vital role in boosting the performance of AFSMBs, due to its impact on the kinetics of the sodium ions.^7,24 To regulate this solvation environment and enhance interfacial ion dynamics, we meticulously designed a high-performance solvent adsorption separator, by ultrasonically spraying functional Na₂Ti₃O₇ (NTO) onto a polypropylene (PP) separator (denoted SAS-N) (Fig. S1). To verify the sodium-compensation function of the NTO, a PP separator coated with H₂Ti₃O₇ (denoted SAS) was prepared as a control sample (Fig. S2–S13, and related discussion). To investigate the impact of these separators on the electrolyte solvation structure, PP, SAS-N, and SAS were immersed in the electrolyte, followed by FTIR and Raman spectroscopic analysis. Deconvoluted Raman peaks at approximately 740, 744, and 747 cm⁻¹ were attributed to solvent-separated ion pairs (SSIPs), contact ion pairs (CIPs), and aggregates (AGGs), respectively, as identified by spectral fitting.²⁵ The electrolyte infiltrated with different separators shows an SSIP-dominated solvation structure, whereas when using SAS-N and SAS, the peak deconvolution revealed a marked decline in SSIP proportion, accompanied by an increase in CIPs and AGG (Fig. 1(a)). Investigating temperature-dependent solvation structures is critical for the rational design of low-temperature electrolytes. Remarkably, Raman spectroscopic analysis conducted at −20 °C revealed that although the proportion of CIP had decreased compared with that at 25 °C, it was still higher than that of electrolytes with a PP separator (Fig. 1(b) and (c)). This unique solvation structure can promote an anion-derived inorganic-rich SEI during cycling.^26,27 In the FTIR spectrum, a free solvent peak at 847.9 cm⁻¹ is observed for the electrolyte infiltrated with PP separators (Fig. S14). Upon using SAS-N and SAS, a red shift occurs, attributed to stronger interactions between solvents and SAS-N and SAS, which leads to a solvation structure dominated by coordinated solvents. Consequently, the free solvent is reduced in the modified electrolytes with SAS-N. To further explore the interaction between separators and electrolytes, SAS-N and SAS were immersed in 1 M NaPF₆ in G2 for 24 hours to ensure adequate solution infiltration, and were subsequently subjected to XPS analysis to characterize the chemical states of the separator surfaces. XPS analysis of the Ti 2p spectra revealed a 0.2 eV positive shift in the binding energy of Ti⁴⁺, suggesting preferential adsorption of G2 solvent molecules on SAS-N and SAS (Fig. S15).

Fig. 1 Experimental and theoretical analysis of the electrolyte solvent structure. Fitted Raman spectra for electrolyte infiltrated with various separators at (a) 25 °C and (b) −20 °C. (c) Percentages of solvation structures with different separators at 25 °C and −20 °C. (d) Adsorption energies of PP, SAS-N and SAS with G2 molecules. (e) A snapshot obtained from classical molecular dynamics simulations with SAS-N. (f) Structures of SSIP, CIP, and AGG. (g) Percentages of solvation structures with different separators. (h) Front-view of the snapshot. (i) G2 concentration distribution across the separator interface and electrolyte bulk phase.

To gain a detailed understanding of the solvation structures and Na⁺ coordination environments, density functional theory (DFT) and molecular dynamics (MD) simulations were systematically conducted. As illustrated in Fig. 1(d), the adsorption energies of PP, SAS-N and SAS with G2 molecules were analyzed and compared, revealing that SAS-N exhibits the strongest adsorption energy towards G2 molecules. This indicates that SAS-N possesses superior capability to adsorb solvent molecules from the electrolyte, consistent with the experimental observations mentioned above. Furthermore, MD simulations were performed on both the blank electrolyte (BE) and electrolytes with SAS-N and SAS to quantitatively assess the relative proportions of SSIPs, CIPs, and AGG (Fig. 1(e)–(g) and Fig. S16, S17). The results demonstrate a notable decrease in SSIP content (from 93.21% to 84.64%) coupled with increased proportions of CIPs (from 6.72% to 14.55%) and AGGs (from 0.07% to 0.81%) when SAS-N is implemented, which aligns well with Raman spectroscopic characterization.

To visually elucidate the interaction between SAS-N and solvent molecules, Fig. 1(h) presents a front-view representation of the electrolyte solvation structure. This visualization clearly demonstrates substantial adsorption of solvent molecules at the SAS-N interfaces. Statistical analysis of the G2 concentration distribution across the separator interface to the electrolyte bulk phase further corroborates SAS-N's exceptional adsorption capacity for the G2 solvent (Fig. 1(i)). The reduction of free solvent molecules in the electrolyte facilitates the enhanced formation of CIPs and AGGs. Radial distribution function (RDF) analysis (Fig. S18) and coordination number calculations support a greater number of anions residing within the first solvation shell of Na⁺ in electrolytes with SAS-N/SAS compared to the BE, a finding that is fully consistent with the statistical trends observed for SSIP, CIP, and AGG populations.

Na plating and stripping reversibility

To investigate the effect of NTO loading on performance (detailed in Fig. S19 and Table S1), we selected a loading of 0.35 mg cm⁻² for all subsequent studies. To study the effect of a modified separator on sodium deposition, ex situ SEM and laser confocal scanning microscopy (LCSM) were carried out to observe the evolution of sodium deposition on Cu. Fig. 2(a)–(c) show SEM images of deposited sodium metal on Cu with three different separators at the fixed capacity of 0.16 mAh cm⁻². It can be seen that there are fewer nucleation sites on the Cu foil with a PP separator and the initially deposited sodium clusters are larger. In contrast, more nucleation sites appeared on Cu foils with the SAS-N and SAS. Meanwhile, the highest number of nucleation sites appeared on Cu foil with SAS-N. After plating 0.5 mAh cm⁻² of sodium, compared to the PP separator, the deposited sodium shows more uniformity and higher density when using the SAS-N and SAS (Fig. 2(d)–(f)). Even at a high deposition capacity of 5 mAh cm⁻², Cu substrates still exhibit smooth and uniform sodium deposition with SAS-N and SAS, while the sodium deposition on the Cu substrate is uneven and presents some noticeable cracks with the PP separator (Fig. S20). Moreover, at the deposition capacity of 5 mAh cm⁻², the sodium plating thickness with the PP separator is 63.7 μm, significantly thicker than 57.7 μm with the SAS and 53.4 μm with the SAS-N (Fig. 2(g)–(i)). These results indicate that SAS-N facilities uniform and dense sodium deposition, suppressing dendrite growth. Furthermore, LCSM was used to analyze the surface morphology after sodium deposition. When using PP as a separator, deposited sodium showed a rougher surface morphology, indicating the growth of numerous dendrites. In comparison, the surface of the deposited sodium was more uniform with SAS-N and SAS. The surface roughness of the deposited sodium with the SAS-N (16.7) was significantly lower than that with the SAS (27.88) and PP (61.69) separators (Fig. 2(j)–(l) and Fig. S21). These findings suggest that SAS-N can regulate sodium ion transport, achieving uniform sodium metal deposition. To further investigate the impact mechanism of different separators on the sodium plating process, COMSOL was used to simulate the electric field distribution and sodium ion concentration distribution between the separator and the anode current collector. When using the PP separator, more sodium ions preferentially deposit around the protruding tips. As the Na⁺ concentration field and the electric field intensity at the tips increase, gradual formation of sodium dendrites occurs (Fig. 2(m) and (o)). When using SAS, uneven distributions of sodium ion concentration and electric field at the tip are also observed (Fig. S22). SAS-N exhibits a more uniform sodium ion concentration and electric field distribution (Fig. 2(n) and (p)).

Fig. 2 Investigation of sodium deposition morphology. SEM images of Cu substrate deposition for 0.16 mAh cm⁻² sodium with (a) PP, (b) SAS, and (c) SAS-N. SEM images of Cu substrate deposition for 0.5 mAh cm⁻² sodium with (d) PP, (e) SAS, and (f) SAS-N. Cross-section of Cu substrate deposition for 5 mAh cm⁻² sodium with (g) PP, (h) SAS, and (i) SAS-N. CLSM images of plating 5 mAh cm⁻² sodium with (j) PP, (k) SAS, and (l) SAS-N. COMSOL simulation results of the Na⁺ concentration field based on (m) PP and (n) SAS-N. COMSOL simulation results of the electric field based on (o) PP and (p) SAS-N.

To further validate the cyclic reversibility of asymmetric cells with different separators, the Aurbach test was conducted to accurately assess the efficiency of sodium deposition and stripping. Compared to asymmetric cells using PP and SAS separators, the CE was significantly improved with the use of SAS-N (99.5%), demonstrating higher reversibility of sodium metal deposition and stripping (Fig. 3(a)). Additionally, it was observed that the nucleation overpotential of the Na||Cu cell with the SAS-N at a current density of 1 mA cm⁻² was 18.0 mV, which is significantly lower than that of the SAS (27.4 mV) and PP separator (30.1 mV) (Fig. 3(b)). Moreover, as the current density gradually increased, the SAS-N still exhibited a low nucleation overpotential (Fig. 3(c)). Long-cycling stabilities of sodium deposition and stripping in asymmetric cells were measured. The asymmetric Na||Cu cell with the SAS-N achieved stable cycling for 1200 cycles while maintaining a high CE of 99.93% at 1 mA cm⁻². Additionally, the SAS-N exhibited a stable voltage–time curve (Fig. 3(d) and Fig. S23). We also investigated the sodium plating/stripping behavior of the three types of separators at a higher current density (5 mA cm⁻²) of 1 mAh cm⁻². When using the PP separator, the CE fluctuated significantly, indicating poor reversibility and the formation of numerous Na dendrites and dead sodium. In contrast, the CE remained stable with the use of SAS-N, with stable cycling observed up to approximately 4500 cycles (Fig. S24). Furthermore, to evaluate the effect of the upper cutoff voltage on cycling reversibility, asymmetric cell CE tests were also conducted at a higher cutoff voltage of 1.0 V. As shown in Fig. S25, the cells maintained a similarly stable CE trend compared to those tested at 0.5 V, while still exhibiting excellent performance over extended cycling. This result confirms that the Na plating/stripping reversibility is robust across a wide voltage window, and that even at 1.0 V the cell can deliver superior long-term stability without significant degradation. To evaluate the performance of the SAS-N functional separator under practical sodium loading, half-cell cycling tests were conducted at an increased sodium deposition capacity of 4 mAh cm⁻². As shown in Fig. S26, the SAS-N separator maintains stable cycling with high coulombic efficiency even at this elevated capacity. These results demonstrate the separator's effectiveness in regulating sodium plating/stripping behavior and its promising potential for practical sodium metal battery applications. This suggests that the SAS-N can regulate sodium metal plating/stripping at higher current densities while suppressing the growth of sodium dendrites. The rate performance of asymmetric cells was measured ranging from 1 mA cm⁻² to 5 mA cm⁻² (Fig. S27), the overpotential of asymmetric cells with SAS-N is significantly lower than that of cells with PP and SAS, and the voltage profiles are more stable. To further explore the cyclic reversibility of asymmetric cells with different separators, impedance during the deposition and stripping processes was tested in asymmetric cells assembled with different separators. With PP and SAS separators, the charge transfer impedance (τ₂) exhibits larger fluctuations, reflecting sodium dendrite formation and poorer reversibility during sodium metal deposition and stripping (Fig. 3(e), (f) and Fig. S28). This is because during plating, the gradual growth of metallic sodium improves electronic contact and reduces impedance, whereas during stripping, void formation and SEI disruption increase interfacial resistance. In contrast, SAS-N shows smaller and more stable impedance changes with overall lower impedance values (Fig. 3(f)), indicating enhanced interfacial stability and superior kinetic performance. This improvement arises from the ability of SAS-N to regulate the interfacial environment and enhance Na⁺ ion flux, effectively suppressing dendrite formation and interfacial degradation, thereby significantly reducing charge transfer impedance throughout the plating and stripping process.

Fig. 3 Sodium plating/stripping reversibility. (a) CE profiles and (b) nucleation overpotential of asymmetric cells with different separators. (c) Nucleation overpotential of asymmetric cells with various separators. (d) Galvanostatic cycling profile of asymmetric cells with different separators at 1 mA cm⁻² and 1 mAh cm⁻². DRT results from Nyquist curves obtained from in situ EIS measurements based on (e) PP, and (f) SAS-N. SEM images of Cu substrates with (g) PP and (h) SAS-N after stripping to 0.5 V. (i) Cycling and (j) rate performance of the Na||Na symmetric cells with different separators.

To further evaluate the influence of the separators on the reversibility of sodium plating/stripping, scanning electron microscopy (SEM) was conducted on the stripped Cu substrates. When using the PP and SAS, a significant amount of large sodium metal residues was observed on the Cu substrates (Fig. 3(g)), and energy-dispersive spectroscopy (EDS) further confirmed the presence of residual sodium (Fig. S29 and S31). In contrast, the Cu substrate surface was smooth and free of residues when using the SAS-N (Fig. 3(h) and Fig. S31). This further confirms that the SAS-N can suppress the formation of sodium dendrites and dead sodium, enhancing the reversibility of sodium plating and stripping. To examine how modified separators suppress dendrites, side-view observations of sodium metal deposition patterns were recorded during electrochemical processes using in situ optical microscopy. Fig. S32a shows that the Na electrode surface with the PP separator has noticeable Na protrusions, possibly leading to potential safety hazards from Na dendrites in batteries in practice. In contrast, no significant Na dendrites were observed on the Na electrode surface with the SAS-N. Time-series optical images reveal that Na deposits continue to grow laterally on the base plane rather than vertical propagation toward the separator (Fig. S32c). Fig. 3(i) displays the voltage–time profiles of the symmetric cells employing the three different separators. Notably, symmetric cells equipped with the SAS-N demonstrated exceptional cycling stability, maintaining operation for 3000 hours under a current density of 1 mA cm⁻² and an areal capacity of 1 mAh cm⁻². In contrast, cells incorporating the SAS separator displayed pronounced voltage hysteresis after 2500 hours, while those using the PP separator exhibited a significantly higher overpotential, and the overpotential kept increasing as the cycling proceeded. When tested under harsher conditions (2 mA cm⁻² and 2 mAh cm⁻²), the SAS-N-based cells retained stable performance for 2000 hours, whereas the PP separator-based counterparts failed after 200 hours with severe voltage polarization (Fig. S33). Furthermore, the symmetrical battery based on SAS-N exhibited the lowest polarization and stable curves (Fig. 3(j)). These results underscore the superior electrochemical stability enabled by the SAS-N.

Kinetics of ion transportation at low temperature

The ion transport behaviors of different separators were investigated through wettability and various electrochemical measurements. A separator with superb electrolyte wettability can reduce ion-transfer resistance.^28,29 The contact angles of the electrolyte on SAS-N (8.93°) and SAS (12.98°) were significantly smaller than that on PP (43.37°) (Fig. S34), indicating that SAS-N has low ionic transfer resistance. Chronoamperometry and electrochemical impedance spectroscopy (EIS) were conducted to calculate the sodium ion conductivity (σ) and transference number (t_Na⁺) through the separators. The σ of SAS-N was 0.81 mS m⁻¹, outperforming the PP and SAS values of 0.66 mS m⁻¹ and 0.79 mS m⁻¹, respectively (Fig. S35). Furthermore, t_Na⁺ increased from 0.67 for PP to 0.84 for SAS-N (Fig. S36). The polarization curves of PP, SAS-N, and SAS (insets: EIS before and after constant potential polarization) are shown in Fig. S36. It has been reported that enhanced ionic conductivity and electrolyte wettability can effectively mitigate Na⁺ flux concentration gradients. Higher t_Na⁺ values indicate that more sodium ions are involved in the charge transfer process.

Compared to ambient temperature environments, battery performance is more challenging under low-temperature conditions.³⁰ To evaluate the effect of modified separators on the low-temperature performance, Na ion diffusion kinetic behavior is explored by investigating the electrochemical impedance spectroscopy and exchange current density. Tafel curves showed that the exchange current density provided by symmetric cells assembled with SAS-N was 0.32 mA cm⁻², greater than that of PP (0.09 mA cm⁻²) and SAS (0.17 mA cm⁻²), respectively (Fig. 4(a)), confirming that the SAS-N facilitates the interfacial charge transfer kinetics at low-temperature. Moreover, EIS measurements of Na||Na cells at −20 °C were also performed to assess interfacial resistances. Na||Na symmetric cells assembled with SAS-N demonstrated the lowest resistance (Fig. S37). We evaluated the activation energy (E_a) of different separator systems using Na||Na symmetric cells to characterize sodium ion transport kinetics. The results show that SAS-N exhibits a lower activation energy across both low-temperature (−20 °C to 0 °C) and moderate-to-high temperature (30 °C to 60 °C) ranges, significantly enhancing sodium ion mobility compared to SAS and PP separators, demonstrating its superior ion transport performance over a wide temperature range (Fig. 4(b) and Fig. S38, S39). Density functional theory (DFT) calculations of desolvation energies for SSIPs, CIPs, and AGG revealed that CIPs exhibit the lowest desolvation energy (1.99 eV) among the three configurations (Fig. 4(c)). This can be attributed to the direct Na⁺–anion coordination in CIPs, which replaces part of the solvent shell and reduces the number of tightly bound solvent molecules to be removed during desolvation.^31,32 Previous experimental and computational studies have demonstrated that the SAS-N increases the proportion of CIPs in the electrolyte. The lower desolvation energy facilitates sodium ion transfer kinetics, which aligns consistently with the observed activation energy trends. Thanks to the superior diffusion kinetics of sodium ions, asymmetric cells with SAS and SAS-N were able to stably cycle for 500 cycles at a current density of 0.5 mA cm⁻² and an areal capacity of 0.5 mAh cm⁻² at −20 °C, with average coulombic efficiencies of 99.80% and 99.82%, respectively. In contrast, the asymmetric cells with PP could only cycle 60 times (Fig. 4(d)). The lower initial coulombic efficiency at −20 °C is primarily due to slower SEI formation kinetics and increased sodium consumption to form a stable SEI. These results suggest that even in low-temperature environments, the reaction kinetics are notably enhanced with the use of SAS-N, leading to significantly improved reversibility of sodium deposition and stripping. Similarly, symmetric cells assembled with SAS-N demonstrate outstanding stability, sustaining cycling for over 1500 hours. A low overpotential of just 8 mV is also maintained at a current density of 0.5 mA cm⁻² and an areal capacity of 0.5 mAh cm⁻², even at −20 °C. In contrast, it can be observed that the symmetric cell with the PP separator exhibits large voltage fluctuations and essentially could not operate at 0.5 mA cm⁻² and 0.5 mAh cm⁻² (Fig. 4(e)). Even under a higher current density of 1 mA cm⁻² and 1 mAh cm⁻², the symmetric cell with the SAS-N can exhibit stable cycling for 400 hours (Fig. S40). Finally, SEM images were used to compare the morphology of the Na metal electrodes after low-temperature cycling. The Na electrode surface with the PP separator exhibited rod-like dendrites and dead Na (Fig. 4(f)). The Na surface with the SAS was smoother than that with the PP separator but still had some protrusions (Fig. 4(g)). In contrast, the Na electrode with the SAS-N exhibited a smooth surface without any uneven Na protrusions (Fig. 4(h)). The above analysis indicates that symmetric and asymmetric cells based on the SAS-N separator exhibit superior cycling stability across a range of current densities and temperatures, including low temperatures. The excellent long-term stability at −20 °C is primarily attributed to the unique solvation structure regulation by SAS-N. By adsorbing excess solvent molecules, SAS-N modulates the local solvation environment to favor the formation of contact ion pairs (CIPs), which have lower desolvation energy barriers and thus facilitate faster Na⁺ transport even under cold conditions. This enhanced ion transport promotes uniform sodium deposition and effectively suppresses dendrite growth, a key factor in capacity fading and safety issues. Moreover, SAS-N stabilizes the electrode/electrolyte interface and reduces side reactions, maintaining electrode integrity over extended cycling. Collectively, these effects contribute significantly to the outstanding cycling performance observed at low temperatures.

Fig. 4 Kinetic analysis and Na plating/stripping at low temperature. (a) Tafel plots and (b) Arrhenius plots for the calculation of the activation energy with different separators. (c) Desolvation energies for SSIPs, CIPs, and AGG. (d) Cycling performance of the Na||Cu asymmetric cells with various separators at 0.5 mA cm⁻² and 0.5 mAh cm⁻² at −20 °C. (e) Cycling performance of the Na||Na symmetric cells with different separators at 0.5 mA cm⁻² and 0.5 mAh cm⁻² at −20 °C. SEM images of the Na electrodes with (f) PP, (g) SAS, and (h) SAS-N after 100 cycles at 0.5 mA cm⁻² and 0.5 mAh cm⁻² at −20 °C.

The effect of the solvent adsorption separator on the SEI

The composition and microstructure of the SEI significantly influence the interfacial stability and the reversibility of Na plating/stripping.³³ To investigate this, we analyzed the chemical composition of the SEI layer using in-depth XPS and time-of-flight secondary ion mass spectrometry (ToF-SIMS). The Na–Cu substrates were obtained from disassembled Na||Cu asymmetric batteries cycled 10 times (ending with plating) using different separators. In the C 1s spectrum, signals corresponding to CO₃²⁻, C–O, C–H, C–C, and C=O indicate the presence of organic species within the SEI layer. Compared to the SEI formed on Na–Cu substrates in the Na||SAS-N||Cu battery, the SEI layers in the Na||PP||Cu and Na||SAS||Cu batteries exhibit a higher C 1s phase content, suggesting a greater proportion of organic components (Fig. 5(a), (b) and Fig. S41). In contrast, the SEI on Na–Cu substrates from the Na||SAS-N||Cu asymmetric battery shows a consistently higher intensity of inorganic species, such as NaF (Fig. 5(c) and (d)). This indicates that the introduction of SAS-N promotes the formation of a NaF-rich inorganic SEI layer. The inorganic-rich SEI structure provides robust mechanical strength and high stability.^34,35 The significant increase in inorganic content may be attributed to more CIP in the electrolyte with SAS-N.³⁶

Fig. 5 The evolution of the SEI structure. C 1s spectra of the SEI formed on Na–Cu substrates with (a) PP and (b) SAS-N. F 1s spectra of the SEI formed on Na-Cu substrates with (c) PP and (d) SAS-N. ToF-SIMS depth profiles of (e) CO₃²⁻, (f) CHO²⁻, (g) F⁻ and (h) NaF₂⁻ with PP and SAS-N. Spatial distribution of the various ions with (i) PP and (j) SAS-N (for 0.2 nm s⁻¹). (k) Mechanism of SAS-N Na deposition.

To achieve three-dimensional characterization and structural reconstruction of the SEI formed at low temperatures, depth profiling analysis of ToF-SIMS was conducted. The Na–Cu substrates were obtained from disassembled Na||Cu asymmetric batteries cycled 10 times (ending with plating) using different separators at −20 °C. Characteristic ion fragments, including CO₃²⁻ and C₂HO⁻ for organic components and F⁻/NaF₂⁻ for inorganic components, were used to identify the SEI composition. The SEI formed on Na–Cu substrates with the PP separator exhibited a higher abundance of carbonate (CO₃²⁻) and C₂HO⁻ ion fragments compared to that formed with the SAS-N. Moreover, the intensities remained consistently higher with increasing sputtering time, indicating a greater organic content in the SEI formed with the PP separator (Fig. 5(e) and (f)). In contrast, the inorganic phase content in the SEI formed with the PP separator continuously decreased and remained low, whereas the SEI formed with the SAS-N exhibited a sustained increase in the intensity of inorganic fragments over time (Fig. 5(g) and (h)). This confirms that the SEI formed with the SAS-N is predominantly composed of inorganic species, indicating that anion decomposition dominates its SEI formation process. The SEI formed with the SAS separator is also predominantly composed of inorganic species, but its content is lower than that of the SEI formed with the SAS-N (Fig. S42). The 3D diagram more intuitively shows the distribution of the four fragments. It can be seen that the inorganic phase in the SEI using SAS-N is distributed over the whole SEI with a high content (Fig. 5(i) and (j)). Moreover, a closer examination of the in situ EIS evolution reveals that the R_SEI in Na||Cu asymmetric cells with SAS-N not only starts at a lower initial value but also exhibits smaller fluctuations during cycling compared to PP (Fig. S43). This indicates that the SEI formed in the presence of SAS-N maintains its ionic transport pathways more effectively over prolonged operation. The suppressed growth of R_SEI can be attributed to the mechanically robust and chemically stable NaF domains, which mitigate continuous electrolyte decomposition and structural collapse of the SEI. In our case, NaF serves a dual role: mechanically reinforcing the interphase to suppress dendrite penetration, and providing sufficient Na⁺ transport channels through grain boundaries and heterogeneous interfaces.^34,35 This balanced combination of structural stability and ionic accessibility explains why the SAS-N-derived SEI enables both low interfacial resistance and high reversibility during Na plating/stripping. The above analysis shows that SAS-N adsorbs solvent molecules from the electrolyte, thereby increasing the proportion of contact ion pairs (CIPs). CIPs exhibit lower desolvation energy, which facilitates faster sodium-ion transport and promotes the formation of an inorganic-rich solid electrolyte interphase (SEI). Among the inorganic components, sodium fluoride (NaF) is particularly advantageous due to its high ionic conductivity and excellent chemical stability. Compared with organic SEI components, inorganic NaF domains are more ionically conductive, so an SEI rich in NaF helps enhance ion transport.^37,38 The crystalline structure of NaF provides efficient pathways for sodium-ion migration, significantly lowering the energy barriers for ion transport through the SEI. Moreover, the presence of NaF in the SEI reduces the interfacial resistance at the electrode/electrolyte interface, thereby enhancing interfacial stability and further accelerating ion transport kinetics. In addition, the SAS-N serves as a supplementary sodium source to compensate for irreversible sodium loss during cycling and ensures a uniform sodium-ion flux, enabling homogeneous sodium deposition and effectively suppressing dendrite growth (Fig. 5(k)). These combined effects contribute to significantly improved battery performance under low-temperature conditions.

Evaluation of the performance of the anode-free sodium metal battery

To verify the practical performance of the SAS-N, an anode-free sodium metal battery (AFSMB) was assembled using a Na₃V₂(PO₄)₃ (NVP) cathode. For detailed explanations regarding the different choices of anode current collectors in asymmetric and full cell configurations, please refer to (Fig. S44). To provide direct evidence for the sodium supplementation effect of SAS-N, anode-free cells assembled with PP, SAS, and SAS-N separators were disassembled after the first charge, and the morphology of the deposited sodium was characterized by SEM. As shown in Fig. S45, cells using the PP separator exhibited fewer sodium nucleation sites and less deposited sodium, whereas cells with SAS and SAS-N separators showed more abundant and uniformly distributed sodium deposits. Furthermore, EDS analysis of the deposited sodium revealed that the sodium content was highest in the cells employing the SAS-N separator (Fig. S46–S48 and Table S2). These morphological and compositional results strongly support that Na present in the SAS-N separator contributes additional sodium to the anode during cycling. Fig. 6(a) shows the charge–discharge curves of AFSMBs with the three different separators. The AFSMB with the SAS-N exhibited the highest discharge specific capacity and a smaller overpotential, indicating AFSMBs with SAS-N deliver superior performance and sodium diffusion kinetics. Additionally, the AFSMB with the SAS-N demonstrates the highest CE of 97.14%, indicating that SAS-N can provide additional sodium sources to compensate for sodium loss. Cyclic voltammetry (CV) curves of the AFSMBs with the three types of separator were recorded, showing that the AFSMB with the SAS-N showed the highest current response and the smallest potential difference (Fig. 6(b)), indicating faster sodium ion transport and better kinetics, which is consistent with the results of Fig. 6(a). Fig. 6(c) displays the rate performance of the AFSMBs with the three separators at different current densities. The AFSMB with the SAS-N delivered discharge capacities of 105.5, 105.1, 104.2, and 102.4 mAh g⁻¹ at 0.5, 1, 2, and 5C (1C = 117 mA g⁻¹), respectively. Under the same measurement conditions, these values were significantly higher than those of the AFSMBs with the SAS and PP separators. When the current density returned to 0.5C, the AFSMB with the SAS-N maintained a capacity of 105.3 mAh g⁻¹, demonstrating its excellent rate performance. Meanwhile, the AFSMB with the SAS-N demonstrated the lowest overpotential at all current densities (Fig. S49). In situ impedance tests generated impedance curves of AFSMBs at different charge–discharge states (Fig. S50 and S51). Through the distribution of relaxation times (DRT) analysis, the variations in SEI impedance, charge transfer impedance, and diffusion impedance between the electrolyte and the electrode were semi-quantitatively analyzed. The AFSMB with the SAS-N exhibited lower impedance and smaller impedance changes in any state (Fig. 6(f)). In contrast, the AFSMBs with the SAS and PP separators exhibited larger fluctuations in SEI impedance, charge transfer impedance, and diffusion impedance, indicating uneven sodium metal deposition and stripping, resulting in dendrites and dead sodium during the charge–discharge process (Fig. 6(d) and (e)). The long-term cycling of the AFSMBs with different separators was measured under different current densities and NVP mass loading (Fig. 6(g) and Fig. S52–S54). The AFSMB with SAS-N exhibited an initial discharge capacity of 103.54 mAh g⁻¹ at 2C. After 600 cycles, it retained a reversible capacity of 98.43 mAh g⁻¹, with a high capacity retention rate of 95.06% and a very low capacity decay rate of about 0.0082% per cycle (Fig. 6(g)). In contrast, the battery with the PP separator showed a capacity retention rate of only 53.57%. It was observed that the CE of the AFSMB with the SAS separator began to fluctuate after 500 cycles, possibly because the SAS failed to provide additional sodium ions to compensate for the loss of the sodium source during the cycles. XRD measurements after extended cycling (Fig. S55) show no noticeable differences from the pristine separator, indicating that the SAS and SAS-N maintain the crystallographic structure and integrity, thereby supporting sustained ion regulation and deposition behavior during long-term cycling. Whether increasing the mass-loading of the NVP (20 mg cm⁻²) or enhancing the current density (5C), the AFSMB with SAS-N also delivers a higher capacity retention and a steady CE (Fig. 6(g)). The cycling performance of the AFSMBs with the three separators was also evaluated at −20 °C. The AFSMB with the PP separator rapidly dropped to about 15 mAh g⁻¹ after 100 cycles (Fig. S56). In contrast, the AFSMB with the SAS-N could stably cycle for 1000 cycles with a higher capacity retention of 92.53% (Fig. 6(h) and (i)), confirming the excellent long-term cycling stability at low temperatures (Fig. 6(j) and Table S3).^{16,18,20,21,39,40}

Fig. 6 Electrochemical performance of anode-free sodium metal batteries (NVP||Al-C). (a) Voltage curves of AFSMBs with different separators at 0.5C. (b) CV curves of AFSMBs with different separators. (c) Rate performance of AFSMBs with different separators. DRT results from Nyquist curves collected from in situ EIS tests of AFSMBs with (d) PP, (e) SAS, and (f) SAS-N. (g) Long-term cycling performance of AFSMBs with different separators at 25 °C at 2C under different NVP mass loadings. (h) Galvanostatic charge/discharge curves of AFSMBs with SAS-N at different cycles. (i) Long-term cycling performance of AFSMBs with different separators at −20 °C at 1C. (j) Comparison of the cycling number and capacity retention of AFSMBs with SAS-N and reported AFSMBs at different temperatures.

Conclusions

To conclude, we successfully developed a solvent adsorption strategy. The proposed SAS-N effectively regulates the solvation environment by adsorbing solvent molecules, promoting the formation of contact ion pairs (CIPs), which facilitate the formation of an inorganic-rich solid electrolyte interphase (SEI). Meanwhile, SAS-N serves as an extra sodium source to compensate for irreversible sodium loss and enhances electrolyte wettability. These synergistic effects contribute to improved interfacial ion dynamics and stabilized electrode/electrolyte interfaces, especially under low-temperature conditions. As a result, the AFSMBs equipped with SAS-N deliver excellent cycling stability, achieving 95.06% capacity retention over 600 cycles at 25 °C, and maintaining 92.53% capacity retention after 1000 cycles at −20 °C. This solvent adsorption separator strategy offers a facile and effective avenue for optimizing interfacial chemistry and improving the low-temperature performance of anode-free sodium metal batteries.

Author contributions

Conceptualization: W. J. L. Methodology: Z. W. H., L.Y. L., X. W. Investigation: Q. Q. Z., H. Y. L., Z. W. T. Visualization: Z. W. H., Z. W. T. Supervision: C. H. Writing – original draft: Z. W. H., W. J. L. Writing – review & editing: Z. W. H., W. J. L.

Conflicts of interest

There are no conflicts to declare.

Data availability

All data needed to support the conclusions in the paper are present in the paper and the SI. Materials and methods; Computer details; SEM, XRD and XPS of SAS-N; XPS and Raman of the solvation structure of electrolytes with SAS-N; SEM of Na plating/stripping on anode current collectors; Electrochemical data of Na||Cu, Na||Na, and NVP||Al-C with SAS-N. See DOI: https://doi.org/10.1039/d5ee03213j.

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

Acknowledgements

The authors acknowledge financial support from the National Natural Science Foundation of China (Youth program no. 22309209). We are grateful for technical support from the High Performance Computing Center of Central South University.

Notes and references

1. Y. Li, F. Wu, Y. Li, M. Liu, X. Feng, Y. Bai and C. Wu, Chem. Soc. Rev., 2022, 51, 4484–4536.
2. X. Wang, Y. Fan, J. Li, X. Li, W. Li, J. Wang and W. K. Pang, Energy Environ. Sci., 2025, 18, 1096–1129.
3. Z. Li, M. Han, J. Wang, L. Zhang, P. Yu, Q. Li, X. Bai and J. Yu, Adv. Funct. Mater., 2024, 34, 2404263.
4. Z. Zou, Y. Mu, M. Han, Y. Chu, J. Liu, K. Zheng, Q. Zhang, M. Song, Q. Jian, Y. Wang, H. Hu, F. Yu, W. Li, L. Wei, L. Zeng and T. Zhao, Energy Environ. Sci., 2025, 18, 2216–2230.
5. Y. Sun, J.-C. Li, H. Zhou and S. Guo, Energy Environ. Sci., 2023, 16, 4759–4811.
6. C. Chu, R. Li, F. Cai, Z. Bai, Y. Wang, X. Xu, N. Wang, J. Yang and S. Dou, Energy Environ. Sci., 2021, 14, 4318–4340.
7. F. Huang, P. Xu, G. Fang and S. Liang, Adv. Mater., 2024, 36, 2405310.
8. Z. Hu, L. Liu, X. Wang, Q. Zheng, C. Han and W. Li, Adv. Funct. Mater., 2024, 34, 2313823.
9. Y. An, Z. Pei, D. Luan and X. W. Lou, Sci. Adv., 2025, 11, eadv2007.
10. Z. Hu, L. Liu, X. Wang, H. Lu, Q. Zheng, Y. Gao, J. Wang, Y. Qi, C. Han and W. Li, ACS Nano, 2025, 19, 23193–23208.
11. H. Li, F. Wu, J. Wang, J. Wang, H. Qu, M. Chen, H. Zhang and S. Passerini, Energy Environ. Sci., 2025, 18, 3887–3916.
12. C. Xie, K. Liang, H. Wu, Z. Xie, Y. Ren, J. Dai, J. Lu, Y. Tang and H. Wang, Adv. Energy Mater., 2025, 2500351.
13. C. Xie, H. Wu, J. Dai, Z. Fu, R. Zhang, H. Ji, Q. Zhang, Y. Tang, T. Qiu and H. Wang, Adv. Energy Mater., 2024, 14, 2400367.
14. S. Li, H. Zhu, Y. Liu, Q. Wu, S. Cheng and J. Xie, Adv. Mater., 2023, 35, 2301967.
15. J. Ge, C. Ma, Y. Zhang, P. Ma, J. Zhang, Z. Xie, L. Wen, G. Tang, Q. Wang, W. Li, X. Guo, Y. Guo, E. Zhang, Y. Zhang, L. Zhao and W. Chen, Adv. Mater., 2024, 2413253.
16. L. Hu, J. Deng, Y. Lin, Q. Liang, B. Ge, Q. Weng, Y. Bai, Y. Li, Y. Deng, G. Chen and X. Yu, Adv. Mater., 2024, 36, 2312161.
17. Y. Li, Q. Zhou, S. Weng, F. Ding, X. Qi, J. Lu, Y. Li, X. Zhang, X. Rong, Y. Lu, X. Wang, R. Xiao, H. Li, X. Huang, L. Chen and Y.-S. Hu, Nat. Energy, 2022, 7, 511–519.
18. Z. Lu, H. Yang, Q.-H. Yang, P. He and H. Zhou, Angew. Chem., Int. Ed., 2022, 61, e202200410.
19. Z. Bai, Q. Yao, M. Wang, W. Meng, S. Dou, H. K. Liu and N. Wang, Adv. Energy Mater., 2024, 14, 2303788.
20. B. Ge, J. Deng, Z. Wang, Q. Liang, L. Hu, X. Ren, R. Li, Y. Lin, Y. Li, Q. Wang, B. Han, Y. Deng, X. Fan, B. Li, G. Chen and X. Yu, Adv. Mater., 2024, 36, 2408161.
21. Y. Zou, B. Zhang, H. Luo, X. Yu, M. Yang, Q. Zheng, J. Wang, C. Jiao, Y. Chen, H. Zhang, J. Xue, X. Kuai, H.-G. Liao, C. Ouyang, Z. Ning, Y. Qiao and S.-G. Sun, Adv. Mater., 2024, 36, 2410261.
22. N. Shen, S. Dai, G. Zhou, J. Miao, Z. Hu, G. Zhi, X. Li, H. Wang, D. Kong, T. Xu, Z. Zhang, X. Li, H. Y. Yang and Y. Wang, Adv. Funct. Mater., 2025, 35, 2417809.
23. Y. Mao, C. Zhou, H. Gong, S. Zhang, X. Wang, X. Liu, Q. Xiang and J. Sun, Small, 2023, 19, 2303259.
24. Y.-S. Hu and H. Pan, ACS Energy Lett., 2022, 7, 4501–4503.
25. C. Zheng, Q. Yao, Y. Sun, H. Lv, Z. Bai, G. Zhang, N. Wang and J. Yang, Adv. Funct. Mater., 2025, 2501303.
26. Q. Zhu, J. Wang, L. Wu, H. Lan, J. Chen, L. Cheng, B. Zhou, D. Yang, J. Yang, M. Kurbanov, S. Dong and H. Wang, Natl. Sci. Rev., 2025, 12, nwaf124.
27. C. Zheng, Q. Yao, Y. Sun, H. Lv, Z. Bai, G. Zhang, N. Wang and J. Yang, Adv. Funct. Mater., 2025, 2501303.
28. L. Zhou, S. Dong, Z. Qiang, C. Zhang, A. Shi, Y. Ning, Z. Liu, C. Chen, Y. Zhang, D. Li and S. Lou, J. Energy Chem., 2025, 102, 516–523.
29. Y. Zhang, H. Zheng, X. Tong, H. Zhuo, W. Yang, Y. Chen, G. Shi, Z. Chen, L. Zhong and X. Peng, Energy Environ. Mater., 2024, 7, e12735.
30. C. Che, F. Wu, Y. Li, Y. Li, S. Li, C. Wu and Y. Bai, Adv. Mater., 2024, 36, 2402291.
31. K. Tasaki, A. Goldberg and M. Winter, Electrochim. Acta, 2011, 56, 10424–10435.
32. C. Tan, Z. Shen, S. Zhang, Z. Wu, S. He, H. Pan, H. Cheng and Y. Lu, Adv. Funct. Mater., 2025, 2509658.
33. T. Liu, X. Yang, J. Nai, Y. Wang, Y. Liu, C. Liu and X. Tao, Chem. Eng. J., 2021, 409, 127943.
34. P. Liu, L. Miao, Z. Sun, X. Chen and L. Jiao, Adv. Mater., 2024, 36, 2406058.
35. T. Deng, C. Li, G. Lu, Z. Li, C. Xu and R. Wang, Energy Environ. Sci., 2025, 18, 3278–3287.
36. T. Shi, R. Hou, L. Zheng, H. Lu, C. Xu, X. Sun, P. He, S. Li, H. Zhou and S. Guo, Adv. Energy Mater., 2025, 15, 2405803.
37. J. Lv, Z. Tang, Q. Zhang, H. Sun, M. OuYang and Y. Cao, ACS Nano, 2025, 19, 16133–16146.
38. Z. Xue, T. Zhang, X. Li, F. Wang, G. Xu and M. Zhu, Angew. Chem., Int. Ed., 2025, 64, e202415283.
39. Y. Li, J. Wang, Y. Wang, S. Wang, L. Wu, B. Zhou, D. Yang, L. Jiang, L. Kan, Q. Zhu, M. Kurbanov and H. Wang, Adv. Mater., 2025, 37, 2419764.
40. Y.-Y. Zhang, C.-H. Zhang, Y.-J. Guo, M. Fan, Y. Zhao, H. Guo, W.-P. Wang, S.-J. Tan, Y.-X. Yin, F. Wang, S. Xin, Y.-G. Guo and L.-J. Wan, J. Am. Chem. Soc., 2023, 145, 25643–25652.

---