The waste-activated carbon (sourced from Ningxia Coal Industry Co., Ltd., China Energy and Investment Group) was treated in a tube furnace at 800 °C for 5 h under a continuous argon flow to remove impurities. After cooling to room temperature, the resulting product was dispersed in 0.2 M hydrochloric acid and magnetically stirred for 30 min to unclog/clear pore channels and increase the specific surface area. The mixture was then centrifuged, washed repeatedly with deionized water, and dried in a vacuum oven at 60 °C for 10 h to obtain porous carbon. In the following step, melamine and the prepared porous carbon (mass ratio 5:1) were placed separately into two crucibles and heated to 850 ^oC for 3 h in an Ar flow, then cooled naturally to room temperature, yielding N-AC.

The functional separator was obtained by a conventional filtration process. In specific, a suspension of 9 mg N-AC and 1 mg of graphene oxide in 100 mL of ethanol was sonicated for 30 min at room temperature. The resulting suspension was then poured into a Buchner funnel, using a GF separator (Whatman GF/C, diameter 90 mm, 100 sheets/box) as the filter. After filtration, the glass fiber separator was removed and dried under vacuum at 60 °C for 24 h, yielding the functional activated carbon modified GF separator (GF@N-AC).

The carbon-black sulfur (CB/S) composite was mixed with Super P and polyvinylidene fluoride (PVDF) in a mass ratio of 8:1:1 in N-methyl-2-pyrrolidone (NMP) solvent to make a slurry. The slurry was then uniformly casted onto aluminum foil, and dried at 60 °C for 12 h to punch into electrodes. To evaluate the effectiveness of the functionalized separator, the CR2032 coin cell is assembled in an argon-filled glovebox (O₂ and H₂O lower than 0.1 ppm). CR2032 coin cells were assembled using a CB/S cathode, a sodium metal anode, either a pristine glass fiber separator or a nitrogen-doped activated carbon-coated glass fiber separator, and an electrolyte of 1 mol/L sodium perchlorate in ethylene carbonate/diethyl carbonate (1:1, v/v). The sulfur loading of the cathode is ~0.86 mg cm^-2. All coin cells were tested on CT2001A Battery Tester. (Wuhan Land Electronics Co., Ltd, China).

Thermogravimetric analysis (TGA): TGA was carried out under N₂ flow from room temperature to 800 °C at a rate of 10 ^oC min^-1 using an STA 449C thermogravimetric analyzer, which is located at the Materials Characterization Center of the National Institute of Clean-and-Low-Carbon Energy.

X-ray diffraction (XRD): The pristine sample was measured in powder form. All patterns were recorded on Bruker D8 ADVANCE X-ray powder diffractometer utilizing Cu Kα radiation (λ = 0.15418 nm, 40 kV, 40 mA).

Raman spectroscopy: The samples for Raman spectroscopy were prepared in the same way as the XRD analysis. Raman spectroscopic analysis was performed with a laser confocal Raman spectrometer (RAMAN, Horiba Jobin Yvon, LabRam HR-800), located in the Materials Characterization Center of National Institute of Clean-and-Low-Carbon Energy, utilizing an excitation wavelength of 532 nm.

X-ray photoelectron spectroscopy (XPS): The samples for XPS measurement were prepared in the same way as the Raman spectroscopy analysis. XPS measurements were carried out with ESCALab250Xi located in the Materials Characterization Center of National Institute of Clean-and-Low-Carbon Energy.

Scanning electron microscopy (SEM): The samples for SEM analysis were prepared as follows. The as-prepared functionalized separators were sectioned into squares with a side length of 2 mm using a precision disc cutter. The samples were mounted onto a conductive specimen stub. All observations were performed on FEI Nova NanoSEM 450 scanning electron microscope in National Institute of Clean-and-Low-Carbon Energy.

Brunauer-Emmett-Teller (BET): Nitrogen adsorption/desorption isotherms at 77 K were obtained on ASAP 2020 surface area and porosity analyzer (Micromeritics Instrument Corporation) located in National Institute of Clean-and-Low-Carbon Energy”.

In this work, density functional theory calculations were carried out using the Vienna Ab initio Simulation Package (VASP, version 5.4.4) with the projector augmented wave (PAW) method^[31-34]. The exchange-correlation energy was described using the Perdew-Burke-Ernzerhof (PBE) functional within the generalized gradient approximation^[35], and van der Waals interactions were accounted for using Grimme’s density functional theory (DFT)-D3 method^[36]. The energy cutoff for the plane-wave basis sets was set to 400 eV, and a 3 × 3 × 1 gamma-centered k-point grid was used to sample the Brillouin zone. The geometry optimization and self-consistent field convergence criteria were set to 0.03 eV Å^-1 and 10^-5 eV, respectively. All supercells include a vacuum layer of 15 Å. The microstructure of activated carbon consists of disordered stacked sp²-hybridized carbon layers, exhibiting significant local aromatic character. Its fundamental structural units share identical chemical bonding environments and electronic structural features with graphene, making graphene an ideal simplified model. By introducing defects such as vacancies and heteroatom doping, the graphene model can effectively simulate the chemical properties of the activated carbon surface. Thus, in this study, a graphene-based model is employed to investigate the microscopic mechanisms of polysulfide transformation on activated carbon. The adsorption energy (ΔE_ad) of sodium polysulfides (SPSs) on the substrate was calculated by using ΔE_ad = ΔE_sys – (ΔE_Na2Sx + ΔE_surf), where ΔE_sys, ΔE_Na2Sx, and ΔE_surf are the total energy of the adsorbed system, the energy of the adsorbate in a vacuum, and the energy of the optimized surface, respectively. In this work, the adsorption energy (ΔE_ad) is defined as ΔE_ad = E_sys – (E_Na2Sx + E_surf). Therefore, a negative ΔE_ad value indicates an exothermic and stable adsorption process. For N-doped carbon, three common configurations were considered, namely graphitic N, pyridinic N, and pyrrolic N.

The detailed preparation process is displayed in Supplementary Figure 1, where the waste N-AC was feasibly coated on the glassfiber separators by a consecutive dispersion-filtration operation. The crystal phase of N-AC, GF@N-AC and GF was investigated by XRD. Displayed in Figure 2A, the pristine GF separator exhibits an amorphous character, whereas N-AC and GF@N-AC display peaks at 25.5° and 43.6°, corresponding to the (002) and (100) planes, respectively, indicating that the structure of N-AC is preserved after coating onto the GF fibers. Furthermore, the XRD pattern of the CB/S cathode is presented in Supplementary Figure 2, confirming the successful preparation of the composite material. Figure 2B presents the nitrogen adsorption-desorption isotherms and pore size distribution of the N-AC. It shows a specific area of 1303.9 m² g^-1 with the pore size centered at 2.21 nm, calculated by Barret-Joyner-Halenda (BJH) model. The Raman spectroscopy [Supplementary Figure 3] reveals distinct D and G bands at 1,336 and 1,588 cm^-1, respectively, with an intensity ratio (ID/IG) of 1.25, indicating a relatively high degree of graphitization^[37-40]. Further insights into the surface chemistry of N-AC are provided by XPS analysis. As shown in Supplementary Figure 4, the XPS survey of the N-AC shows the presence of the carbon and nitrogen, corresponding element contents of 91.89% and 8.11%, respectively. The high-resolution C 1s spectrum [Figure 2C] can be deconvoluted into five distinct peaks at 284.2, 285.2, 285.9, 287.1, and 289.5 eV, ascribing to C=C, C-N, C-O, C=O, and COOH, respectively^[41,42]. Whereas the N 1s spectrum can be split into four types of nitrogen species-pyridinic (41.1%), pyrrolic (31.3%), and graphitic (12.7%) and oxidized (14.9%) nitrogen [Figure 2D and Supplementary Table 1]. These oxygen- and nitrogen-containing groups enhance the chemical affinity of the material for polysulfides, which we will present below.

The pristine separator features abundant pores [Figure 2E], enabling efficient sodium-ion transport. However, this configuration allows polysulfide species to migrate freely to the anode, triggering the shuttle effect, which results in active material loss and deteriorates electrochemical performance. In this regard, we proposed to leverage the GF separator with nitrogen-doped activated carbon [Supplementary Figures 5 and 6]. Cross-sectional SEM imaging confirms that the coating thickness of the porous N-AC [Figure 2F] is approximately 149 μm. The mechanical performance of the GF@N-AC separator was assessed by bending tests together with post-bending structural characterization. As shown in Supplementary Figure 7, the separator maintained its structural integrity during both forward and backward bending without visible cracking, indicating excellent flexibility and strong adhesion between the N-AC coating and the GF substrate. SEM images further show that the coating remained uniformly attached to the GF separator without noticeable delamination or peeling after bending [Supplementary Figure 8]. In addition, the corresponding energy-dispersive spectroscopy (EDS) maps exhibit a uniform elemental distribution similar to that before bending [Supplementary Figure 9], confirming that the coating layer remained well preserved after mechanical deformation. These results confirm the superior mechanical durability of the modified separator and the strong bonding between the coating layer and the separator substrate.

For a preliminary analysis of the feasibility of GF@N-AC for room-temperature Na-S batteries, the ionic conductivity was assessed [Supplementary Figure 10]. As a result, the ionic conductivity of the GF@N-AC separator is calculated to be 3.1 mS cm^-1, higher that of the GF separator (1.6 mS cm^-1). Hence the modification of commercial GF separator with the porous N-AC could alleviate the side shuttle reactions and improve Na⁺ ions diffusion.

The schematic illustrations of Na-S batteries employing pristine and functional separators are comparatively shown in Figures 3A and B. A lingering challenge for Na-S batteries is the shuttle effect, which arises when soluble polysulfides generated during cycling diffuse through the electrolyte and continuously migrate between the electrodes. This leads to the irreversible loss of active sulfur species, severely compromising the overall electrochemical performance. The pristine GF separator with abundant micron-sized pores, allows not only the transport of sodium ions but also the free detrimental passage of polysulfides. In contrast, the functionalized separator-featuring a high surface area and nitrogen doping-effectively traps polysulfides through physical adsorption and chemical interactions, as analyzed by the leakage experiment in H-type cell. In this scenario, sodium hexasulfide (Na₂S₆) solution was placed on the left chamber, while pure tetraethylene glycol dimethyl ether (TEGDME) was added to the right side. The two chambers separated by either a pristine or a functional separator were placed at room temperature. Consequently, the pristine separator allows substantial migration of polysulfides after 12 h [Figure 3C]. In contrast, the H-cell equipped with the functionalized separator remained clear and transparent, confirming its superior blocking capability against polysulfide shuttle [Figure 3D]. Furthermore, we also evaluated the overall performance of GF@N-AC and GF separators when used in a full Na-S battery. As shown in Supplementary Figure 11, the GF@N-AC based cell exhibits smaller R_s (1.94-2.48 Ω) and R_ct (42.10-42.91 Ω) during the whole discharge process, demonstrating a faster electron/ion transport in the GF@N-AC based cell, whereas the GF-separator-based cells have higher R_s (5.52-5.89 Ω) and R_ct (216.3-274.5 Ω). This indicates that the functionalized separator prominently improves the internal sluggish kinetics caused by soluble sodium polysulfides, thereby promoting Na⁺ transport rate.

To investigate the inhibitory mechanism of activated carbon-functionalized separators against SPSs, we separately introduced 0.1 g of N-AC into the base solution (0.5 M Na₂S₆ in TEGDME), as illustrated in Figure 4A. consequently, the upper section of the solution exhibited significant clarification after 6 hours, demonstrating the effective adsorption capability of N-AC towards Na₂S₆ polysulfides.

To gain deeper insights into the interaction mechanisms between all sodium polysulfides Na₂S_x (x = 1, 2, 4, 6) and nitrogen-doped activated carbon, we systematically performed DFT calculations. Given the structural complexity of amorphous activated carbon, a simplified graphene-based model system was adopted [Figure 4B], which is a well-accepted method to study activated carbon properties^[43]. With pyridinic nitrogen-doped graphene as the representative, Figure 4C and Supplementary Figure 12 illustrate the optimized adsorption configurations of Na₂S_x species^[44]. Notably, the DFT results reveal that the interaction mechanism involves two key aspects: (1) the positively charged Na⁺ ions form coordination with the electron-rich nitrogen sites, and (2) the electron-rich sulfur atoms (S_x^2-) preferentially orient toward the vacuum region. DFT calculations also reveal a continuous interaction between other sodium polysulfides (Na₂S, Na₂S₂, Na₂S₄) and various nitrogen-doped graphene surfaces [Figure 4D]^[45,46]. In specific, pyridinic- and pyrrolic-N doped graphene demonstrate substantially lower adsorption energies for Na₂S_x, especially for the short-chain polysulfides (Na₂S, Na₂S₂) due to pronounced electron transfer [Figure 4E]^[47]. The detailed adsorption energy values are summarized in Supplementary Table 2. Furthermore, density of states (DOS) analysis [Figure 4F] confirms that nitrogen doping introduces abundant electronic states near the Fermi level, facilitating efficient electron transport during electrochemical processes^[48]. In brief, defective graphene layers within activated carbon reveal strong adsorption capability towards all polysulfides Na₂S_x (x = 1, 2, 4, 6), contributing to inhibite shuttle effect in Na-S batteries.

The above analyses underscore the capability of the functionalized separator in suppressing the shuttle effect of sodium polysulfides, thus we tentatively applied it to Na-S batteries. The thermogravimetric curve confims that the sulfur content in the cathode is approximately 70% [Supplementary Figure 13]. Figure 5A presents galvanostatic charge-discharge profiles of the cells with the functionalized separator. A significantly capacity of 985.5 mAh g^-1 is reached initially with 662.5 mAh g^-1 maintained after 100 cycles at 0.1 C (1 C = 1,672 mA g^-1). The charge-discharge curves are nearly overlapped from the 50th to 100th cycle with operating voltage around 1.3 V, demonstrating the excellent reversibility of the RT Na-S battery with the modified separator. Further insights into redox mechanisms can be inferred from the cyclic voltammetry (CV) curves (inset of Figure 5A): the two peaks at 1.48 and 0.98 V are relevant to the transition from solid sulfur to dissolved liquid long-chain Na₂S_x (4< x ≤ 8), while the peak at 0.69 V corresponds to the long-chain polysulfides to insoluble short-chain polysulfides (Na₂S₂ and Na₂S). Subsequently, two peaks in the recharging process arise from the oxidation of polysulfides to solid sulfur reversibly. Compared to GF separator, CV curves measured from GF@N-AC separator overlapped, implying high reaction reversibility. The CV curves of cells based on GF@N-AC separator displayed a smaller polarization than GF separator (inset of Figure 5A and Supplementary Figure 14). As displayed in the primary comparative cycling measurements [Figure 5B], the cell employing the functionalized separator retains 67.3% capacity after 100 cycles. In contrast, the cell using the pristine separator delivered a low capacity of 590.2 mAh g^-1, which rapidly declined to only 285.6 mAh g^-1 after merely 10 cycles. However, compared to the moderate thickness of N-AC (149 μm), the cell with increased thickness coating (192 μm) shows worse cycling (rapidly drops to 605.3 mAh g^-1 after only 10 cycles). Therefore, a moderate coating thickness is essential for alleviation of the shuttle effect of soluble sodium polysulfides, but over thick coating compromises the mechanical integrity and cycling stability.

Furthermore, Figure 5C depicts the rate performance of batteries. A highest discharge capacity of 660.4, 470.8 and 390.1 mAh g^-1 (at 0.5, 1 and 2 C, comparatively) is achieved in the cell with the modified separator, while for the cell with the pristine separator, an inferior capacity of below 200 mAh g^-1 can only be reached at 0.5 C. It is worthy note that the performance (regarding reversible capacity and rate capability) with separator modification strategy proposed in this work is competitive/superior to other previous strategies^[49-57] [Figure 5D]. Supplementary Figure 15 shows the EIS spectra at open-circuit voltage (OCV) of room-temperature sodium-sulfur batteries employing a GF@N-AC separator. The corresponding charge-transfer resistance (R_ct) is 103.4 Ω, which is comparable to that of the cell with the pristine separator (120.6 Ω), indicating that the separator has a limited effect on interfacial charge transfer. The fitted data of the equivalent circuit for cells with GF@N-AC and GF separator are supplemented in Supplementary Table 3. Finally, the long-term cycling at higher 0.5 C rate was performed [Figure 5E]. The cell with the functional separator delivers an initial reversible capacity of 698.2 mAh g^-1, with 545.4 mAh g^-1 remaining after 260 cycles. This corresponds to a minimal degradation rate of 0.08%/cycle and near 100% Coulombic efficiency, exceeding previous reports^[58,59], by using a separator modified by waste activated carbon. Impressively, the assembled coin cell with GF@N-AC separators is able to light an electric-light emitting diode (LED) with the logo of “nice” (National Institute of Clean-and-Low-Carbon Energy) [inset of Figure 5E]. These results highlight the effectiveness of the functional separator in mitigating the polysulfide shuttle effect and ensuring stable Na-S battery operation.

Additionally, after 50 cycles, the modified separator retains an intact coating with negligible morphological change [Supplementary Figure 16]; Unlike the pure GF separator, which shows a rougher surface from electrolyte residue. EDS mapping analysis [Supplementary Figures 17 and 18] confirms sulfur presence only on the modified separator, indicating effective polysulfide adsorption. This is also consistent with the smoother sodium anode surface observed with the modified separator [Supplementary Figure 19], demonstrating its efficacy in suppressing shuttle and enhancing performance.