Using Ag nanoclusters to achieve mutually reinforcing water desalination and sterilization Authors: Qianhui Ma, Lingyu Zhang, Bo Xiao, Zhong Hua, Xinglong Luan, Jianhui Xia, Jiahui Li, Ting Feng, Yong Liu, Xun Yuan Published: 2026 ABSTRACT Sustainable water purification requires simultaneous desalination and disinfection, yet integrated solutions remain challenging due to mutual interference between the two processes. Here, we report a synchronous capacitive desalination-sterilization (CDIS) system driven by a rationally engineered dual-functional anode of silver nanoclusters embedded in carbon nanofibers (Ag NCs@CNF). This electrode uniquely combines efficient chloride-ion storage with molecular-ike electrocatalytic activity, enabling concurrent high-capacity desalination (84.25 mgCl g−1) and complete sterilization ( > 4.0 log reduction, detection limit) via in situ reactive species (RS) generation. Beyond performance, we provide the unambiguous mechanistic insight into microbe-ion dynamics in capacitive systems. Contrary to the spatial-competition paradigm, we reveal that performance decay primarily stems from electrostatic shielding by live bacteria, while inactivated cells undergo a fragmentation-aggregation process that eliminates shielding. Remarkably, bacterial inactivation further enhances cycling stability by scavenging RS and mitigating anode oxidation. This work establishes an efficient integrated electrochemical purification platform and offers a theoretical framework for designing sustainable water-treatment technologies. Introduction The provision of safe drinking water through sustainable desalination is of paramount importance for human survival1. The ultimate benchmark lies in producing water that is not only free of excessive salts but also microbiologically safe2. However, the practical conversion of contaminated saline water into potable water faces intertwined challenges: the inherent complexity of individual desalination and disinfection steps is compounded by their mutual interference, often compromising each other’s efficacy in sequential processes3,4. Single-technology solutions are frequently inadequate. For instance, pressure-driven membrane processes like reverse osmosis (RO), while effective for removing dissolved ions and microbes, are inherently susceptible to biofouling5, leading to performance degradation and increased energy costs. Conversely, even hybrid approaches face dilemmas: biofilms formed during desalination hinder performance, while high salinity environments can impair subsequent disinfection methods (e.g., by reducing UV transmittance)6. Therefore, the development of an integrated technology that synergistically combines desalination and antibacterial functions without sacrificing either is a critical imperative for future water security. Capacitive deionization (CDI)7–13, particularly in its Faradaic variant (FDI)14–19, has emerged as an energy-efficient platform11,20–23 for brackish water desalination24–27. Notably, its electrochemical nature offers inherent potential for water disinfection28–32, as microbes can be captured and inactivated at the anode33, for instance, via in situ generation of reactive species (RS)3. This positions CDI as a unique platform for building a truly integrated desalination-sterilization system. While the idea of combining desalination and disinfection in CDI has been proposed previously3,33, a systematic understanding of the synergy between these two functions remains elusive, impeded by several fundamental challenges: (i) at the material level, it is difficult to engineer electrodes that simultaneously support efficient, reversible Faradaic chloride storage and possess high catalytic activity for RS generation, as these functions often impose competing demands on material properties. (ii) At the mechanistic level, the nature of the interaction between anions and microbes at the electrode interface is unclear. Given the vast size disparity between sub-nanometer hydrated ions and micrometer-scale bacteria, direct spatial competition seems improbable, yet undeniable interference occurs—the underlying mechanism of which is obscured. (iii) Critically, the physicochemical transformation of bacteria during electrochemical inactivation and its consequent impact on long-term ion storage performance are poorly understood. To address these challenges, we designed ligand-free silver nanoclusters (an emerging class of functional materials34–41, defined by a core size below 2 nm42–47) embedded within a three-dimensional carbon nanofiber matrix (Ag NCs@CNF) via an electrospinning-carbonization approach coupled with a concentration-mediated size-focusing strategy. This material serves as a dual-functional anode in a synchronous capacitive desalination-sterilization (CDIS) system (Fig. 1). The highlight of this work is threefold. First, although dual-function electrodes have been explored before, we integrate within a single ligand-free Ag NCs@CNF electrode the intrinsic reversible chloride-ion storage of Ag NCs with their molecular-like electrocatalytic activity for RS generation, enabling synergistic anion capture and microbial inactivation without external antibacterial agents. Second, this integrated system provides an ideal platform to probe the previously obscured dynamics between ions and microbes in a capacitive field, revealing that the apparent competition is dominated by electrostatic shielding from live bacteria rather than physical spatial blocking—a mechanistic insight absent from prior reports. Third, and most notably, we find that bacterial inactivation not only preserves desalination but unexpectedly enhances cycling stability through a fragmentation-aggregation process: inactivated cell debris assembles into detached microspheres that minimize shielding and, remarkably, scavenge RS to mitigate anode oxidation. Our results demonstrate superior performance in both desalination (84.25 mg Cl g−1) and sterilization (>4.0 log reduction, detection limit). Mechanistic studies reveal that the competition at the electrode is dominated by charge shielding from live bacteria, not physical spatial blocking. Furthermore, inactivated cells undergo a unique fragmentation-aggregation process, forming detached micro-spheres that minimize shielding, thus negligibly affecting ion storage. Remarkably, the scavenging of RS by bacteria mitigates anode oxidation, thereby improving long-term stability. Taken together, this work does not merely present another combined desalination–sterilization system. It reports the anode design that concurrently achieves Faradaic chloride capture and efficient RS-enabled sterilization using ligand-free Ag NCs, and, more importantly, it provides foundational mechanistic insights into the competitive and shielding interactions between microbes and ions within a capacitive field—including a counterintuitive stability enhancement mechanism. These findings transform the traditionally competing desalination vs. disinfection relationship into a synergistic and mutually reinforcing process, paving a promising avenue for the rational design of advanced electro-driven integrated water purification technologies.Fig. 1Schematic illustration of the CDIS system.The overall configuration and working principle of the capacitive desalination and sterilization (CDIS) system, integrating a dual-functional Ag NCs@CNF anode for concurrent desalination and sterilization. Schematic illustration of the CDIS system. The overall configuration and working principle of the capacitive desalination and sterilization (CDIS) system, integrating a dual-functional Ag NCs@CNF anode for concurrent desalination and sterilization. Results and discussion To achieve simultaneous Faradaic anion storage and catalytic RS generation, we engineered a dual-functional anode by confining ultrasmall Ag NCs within a conductive three-dimensional CNF matrix. The Ag NCs@CNF material was synthesized via a combined electrospinning-carbonization approach, integrated with a straightforward concentration-mediated size-focusing strategy. As schematically illustrated in Fig. 2a, simply modulating the Ag⁺ concentration in the polyacrylonitrile (PAN) solution (as the precursor of CNFs) allows precise control over the size of the embedded Ag species, enabling the targeted fabrication of either NCs (Ag NCs@CNF) or nanoparticles (Ag NPs@CNF).Fig. 2Synthesis and morphology of Ag@CNF composites.a Schematic illustration of the fabrication procedure for Ag NCs@CNF and Ag NPs@CNF via a concentration-mediated strategy. Morphological and elemental characterization of Ag NCs@CNF: b FESEM, c TEM, d HAADF-STEM images (the nanoclusters distributed on the material surface are highlighted by circles), and e corresponding STEM-EDS elemental mapping. Characterization of Ag NPs@CNF: f FESEM, g TEM, h HAADF-STEM images, and i corresponding STEM-EDS mapping. The insets in (d, h) show the particle size distribution histograms of Ag NCs and Ag NPs, respectively. Synthesis and morphology of Ag@CNF composites. a Schematic illustration of the fabrication procedure for Ag NCs@CNF and Ag NPs@CNF via a concentration-mediated strategy. Morphological and elemental characterization of Ag NCs@CNF: b FESEM, c TEM, d HAADF-STEM images (the nanoclusters distributed on the material surface are highlighted by circles), and e corresponding STEM-EDS elemental mapping. Characterization of Ag NPs@CNF: f FESEM, g TEM, h HAADF-STEM images, and i corresponding STEM-EDS mapping. The insets in (d, h) show the particle size distribution histograms of Ag NCs and Ag NPs, respectively. The nanostructures of the resulting materials were systematically characterized. Field-emission scanning electron microscopy (FESEM) images confirm that both Ag NCs@CNF and Ag NPs@CNF retain the interconnected 3D nanofiber network of the pristine CNF (Figs. 2b, f, and S1a), ensuring high conductivity and accessible surface area. Crucially, transmission electron microscopy (TEM) and high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) analyses reveal a homogeneous dispersion of ultrasmall Ag NCs (~1.8 nm in diameter) within the CNF matrix for Ag NCs@CNF (Figs. 2c, d, and S1b), underscoring the efficacy of our synthesis in achieving sub-2 nm NCs. The uniform distribution of Ag is further corroborated by STEM-energy-dispersive X-ray spectroscopy (EDS) elemental mapping (Fig. 2e). In contrast, the Ag NPs@CNF sample exhibits larger, less uniformly dispersed NPs with a broader size distribution (9–27 nm, Fig. 2g, h, inset), as also evidenced by its EDS map (Fig. 2i). The pivotal advantage of our concentration-mediated strategy is the tunable size control in the critical NCs regime. Reducing the Ag+ precursor concentration from 0.01 M to 0.005 M yielded smaller-sized NCs with average sizes of 1.49 nm (named as s-Ag NCs@CNF; Fig. S2), whereas increasing it to 0.03 M led to distinct NP formation (Ag NPs@CNF, Fig. 2f–i). This size evolution is governed by a proposed reduction-melting-aggregation mechanism (Fig. 3a). Of note, below 400 °C, carbonization has not yet initiated, and the Ag+ species reduction remains negligible, thus the system maintains its solid PAN fiber structure (Fig. S3a). However, during high-temperature carbonization (>600 °C; Fig. S3b), embedded Ag+ ions are reduced by carbon, entering a mobile state while etching the pore walls. At low concentrations (0.005–0.01 M), mobile Ag atoms anchor at etching-induced defect sites, forming size-confined NCs with a narrow distribution. At high concentrations (above 0.03 M), excessive Ag aggregation intensifies etching, leading to pore collapse, loss of confinement, and the formation of larger, polydisperse NPs. This mechanistic hypothesis is strongly supported by porosity analyses. N2 physisorption (Fig. 3b, c) shows that Ag NCs@CNF maintains a high specific surface area (446.04 m2 g−1) and a pore size distribution similar to pristine CNF, indicating minimal structural disruption. Conversely, Ag NPs@CNF suffers a drastic loss in surface area (215.09 m2 g−1) and a pronounced reduction in smaller pores, consistent with pore-wall over-etching and architectural degradation.Fig. 3Formation mechanism and physicochemical characterization.a Schematic of the proposed reduction-melting-aggregation mechanism governing the size-controlled growth of Ag species within CNFs. b N2 adsorption-desorption isotherms and c corresponding pore-size distributions of CNF, Ag NCs@CNF, and Ag NPs@CNF. d XRD patterns and e TGA profiles (measured in air) of the indicated samples. f Normalized Ag K-edge XANES spectra of Ag NCs@CNF together with reference standards (Ag foil, AgO, Ag2O, AgCl). g Fourier-transformed (FT) k3-weighted EXAFS spectra in R-space and h wavelet-transform EXAFS (WT-EXAFS) contour plot of Ag NCs@CNF. Formation mechanism and physicochemical characterization. a Schematic of the proposed reduction-melting-aggregation mechanism governing the size-controlled growth of Ag species within CNFs. b N2 adsorption-desorption isotherms and c corresponding pore-size distributions of CNF, Ag NCs@CNF, and Ag NPs@CNF. d XRD patterns and e TGA profiles (measured in air) of the indicated samples. f Normalized Ag K-edge XANES spectra of Ag NCs@CNF together with reference standards (Ag foil, AgO, Ag2O, AgCl). g Fourier-transformed (FT) k3-weighted EXAFS spectra in R-space and h wavelet-transform EXAFS (WT-EXAFS) contour plot of Ag NCs@CNF. The crystallographic structures of the synthesized materials were examined by X-ray diffraction (XRD). As shown in Figs. 3d and S4a, the patterns of s-AgNCs@CNF and AgNCs@CNF exhibit only a broad diffraction peak centered near 26°, corresponding to the (002) plane of graphitic carbon, with no discernible peaks attributable to metallic Ag. This absence of sharp Bragg reflections is characteristic of ultrasmall NCs that lack long-range crystalline order48. In contrast, the AgNPs@CNF sample displays a set of well-defined peaks that match the face-centered cubic phase of metallic Ag (PDF04-0783), confirming the formation of larger, crystalline NPs at higher precursor concentrations. Furthermore, thermogravimetric analysis (TGA, Figs. 3e and S4b) determined the Ag contents to be 6.8, 8.4, and 15.4 wt% for s-AgNCs@CNF, AgNCs@CNF, and AgNPs@CNF, respectively, consistent with the precursor concentration trend. To probe the chemical state and local coordination environment of Ag within the NCs, AgL3-edge X-ray absorption near-edge structure (XANES) spectroscopy was performed. The spectra of AgNCs@CNF were compared with those of reference compounds—Ag foil (Ag0), AgO (Ag2+), Ag2O (Ag+), and AgCl (Ag+). The AgL3-edge corresponds to the dipole-allowed 2p3/2 → 4d transition, whose white-line intensity and position are highly sensitive to the oxidation state and local coordination of Ag. As illustrated in Fig. 3f, g, the white-line intensity of AgNCs@CNF is notably lower, and its line shape is broader and smoother than that of Ag2O. This attenuation and broadening indicate a reduced density of unoccupied 4d states, which is characteristic of a significant contribution from metallic Ag0 alongside oxidized Ag+ species. These results confirm that the Ag NCs exist in a mixed valence state, a feature that is crucial for their dual functionality in Faradaic ion storage49,50 and electrocatalytic RS generation3. Superior electrochemical activity is essential for efficient Faradaic deionization. As shown in Figs. 4a and S5, the cyclic voltammetry (CV) profiles of all Ag-containing electrodes exhibit well-defined redox couples near 0 and 0.25 V, corresponding to the reversible Ag⁰/Ag+ transition, superimposed on a quasi-rectangular background that indicates dominant pseudocapacitive behavior. Remarkably, AgNCs@CNF delivers a maximum specific capacitance of 418.38 F g−1 (Fig. 4b), significantly exceeding that of pristine CNF (272.64 F g−1), s-Ag NCs@CNF (395 F g−1), and AgNPs@CNF (349.54 F g−1). This enhancement originates from two synergistic effects: (i) the embedded ultrasmall Ag NCs provide abundant redox-active sites while maintaining a large electrode–electrolyte interface for rapid charge transfer, and (ii) their ultrahigh surface-to-volume ratio promotes surface-dominated ion diffusion, maximizing the utilization efficiency of Ag atoms. Electrochemical impedance spectroscopy (EIS) further confirms the favorable charge-transfer kinetics of the NCs-based electrode. The Nyquist plots (Fig. 4c) show a smaller high-frequency semicircle for AgNCs@CNF, corresponding to a charge-transfer resistance (Rct) of 1.18 Ω, which is markedly lower than that of AgNPs@CNF (1.93 Ω). Quantitative analysis of the capacitive contributions via scan-rate-dependent CV (Figs. 4d–f, S5, S6, and Supplementary Note I) reveals that AgNCs@CNF exhibits a significantly higher surface-controlled contribution than AgNPs@CNF, underscoring that size reduction to the NCs regime effectively shifts the storage mechanism toward surface-dominated processes.Fig. 4Electrochemical and catalytic properties.a CV curves at 5 mV s−1, b specific capacitance, and c Nyquist plots (inset: equivalent circuit) of CNF, Ag NCs@CNF, and Ag NPs@CNF. Deconvolution of surface- and diffusion-controlled contributions for d Ag NCs@CNF and e Ag NPs@CNF at scan rates of 10, 20, and 50 mV s−1. f Quantitative capacitive contribution fractions at increasing scan rates. In situ RS concentrations measured at the AgNCs@CNF surface over 60 min in g ultrapure water and h 1435 mg L−1 NaCl. i Speciation analysis of RS generated at the AgNCs@CNF surface in 1435 mg L−1 NaCl, showing the dominant contribution of H2O2 (~95%) and a minor fraction of RCS (~5%). Electrochemical and catalytic properties. a CV curves at 5 mV s−1, b specific capacitance, and c Nyquist plots (inset: equivalent circuit) of CNF, Ag NCs@CNF, and Ag NPs@CNF. Deconvolution of surface- and diffusion-controlled contributions for d Ag NCs@CNF and e Ag NPs@CNF at scan rates of 10, 20, and 50 mV s−1. f Quantitative capacitive contribution fractions at increasing scan rates. In situ RS concentrations measured at the AgNCs@CNF surface over 60 min in g ultrapure water and h 1435 mg L−1 NaCl. i Speciation analysis of RS generated at the AgNCs@CNF surface in 1435 mg L−1 NaCl, showing the dominant contribution of H2O2 (~95%) and a minor fraction of RCS (~5%). Beyond chloride storage, the anode material must also enable catalytic generation of reactive species (RS) to achieve simultaneous sterilization. Using an in situ sampling method (Figs. S7–9 and Supplementary Note II), we evaluated the RS production capability of AgNCs@CNF in ultrapure water and saline electrolytes (1435 mg L−1 NaCl). As shown in Fig. 4g, h, AgNCs@CNF generates substantial RS under both ultrapure water and saline conditions, with RS levels slightly elevated at higher salinity. Scavenger-based assays were employed to distinguish the contributions of individual reactive species generated at the anode—specifically H2O2, hydroxyl radicals (•OH), and reactive chlorine species (RCS)51,52. The results (Fig. 4i) demonstrate that H2O2 constitutes the dominant species (~95%) in 1435 mg L−1 NaCl, accompanied by a minor fraction of RCS (~5%) and no detectable ·OH, whereas only H2O2 was observed in ultrapure water. This robust and salt-tolerant RS generation, dominated by H2O2, confirms the potential of AgNCs@CNF for reliable disinfection across diverse water matrices, satisfying the dual requirements of high-capacity desalination and effective antimicrobial action in a single integrated electrode. To further clarify the contribution of the CNF scaffold to RS generation, we evaluated bare CNF under identical conditions (Fig. S10). The pure CNF system generated only ~16.71 μM of RS, confirming that the AgNCs serve as the primary catalytic source of RS in the AgNCs@CNF composite electrode. The dual-functional capability of the Ag NCs@CNF anode was evaluated in a continuous-flow synchronous CDIS system (Figs. 5a and S11, Table S1) using a mixed feed of saline water (1435 mg L−1 NaCl) and Escherichia coli (E. coli, 104 CFU mL−1). The desalination performance was shown in Fig. 5b–f. As reflected by the effluent NaCl concentration in Fig. 5b, a rapid decrease upon charging and a sharp regeneration peak during discharge were observed, confirming effective capacitive cycling. Remarkably, the Ag NCs@CNF anode delivers a high desalination capacity (Fig. 5c) of 84.25 mg Cl g−1 at 350 mA g−1, which is over 5 times that of pristine CNF (15.02 mg Cl g−1). This superior performance is attributed to abundant accessible chloride storage sites provided by the embedded Ag NCs. Meanwhile, the system maintains high charge efficiency over a wide current range (Fig. 5c), indicating minimal parasitic losses and efficient ion transport. Moreover, enhanced desalination kinetics are further corroborated by Kim-Yoon-Ragone analysis, which reveals both higher capacity and faster salt removal rates for Ag NCs@CNF across various currents than that of CNF (Fig. 5d). Notably, the Ag NCs@CNF-based CDIS system exhibits desalination performance, both in capacity and rate, compared to state-of-the-art Ag-based systems (Fig. 5e and Table S2), even in the presence of E. coli, underscoring its potential for real-world water desalination applications.Fig. 5Integrated performance of the CDIS system.a Schematic diagram of the CDIS system for simultaneous capacitive desalination and sterilization. b Effluent NaCl concentration profiles for Ag NCs@CNF and pristine CNF at 350 mA g−1. c Desalination capacity and charge efficiency of the Ag NCs@CNF-based system at different constant currents. d Kinetic Ragone plot comparing desalination rate and capacity. e Desalination rate performance vs. electrode mass benchmarking against state-of-the-art FDI Ag-based electrode reported in the literature49,55–59. f Time-dependent inactivation of E. coli for Ag NCs@CNF and CNF. g Sterilization efficiency of Ag NCs@CNF under different NaCl concentrations. h Relative contributions of RS-mediated inactivation, intrinsic antibacterial effects of Ag species, and physical contact-based inactivation to the overall disinfection efficiency. i Inactivation rate of the bacteria adsorbed on the surface of Ag NCs@CNF and pristine CNF during the first 60 min. j Sterilization efficiency of Ag NCs@CNF at different productivity. k Performance trade-off relationships of productivity vs. energy consumption for Ag NCs@CNF-based CDIS, traditional membrane CDI, and flow-through CDI (ΔC = 5 mM, water recovery = 50%. Note: aAll sterilization tests followed a hybrid protocol: the system was first operated in constant current mode until the cut-off voltage was reached, and then switched to constant voltage mode to complete the total 60-min duration; bSamples with no detectable colonies were assigned a value at the detection limit (4 log reduction, 99.99% kill), limited by the detection limit of the plate counting method (1 CFU mL−1 under ideal conditions). (Data are presented as mean ± SD (n = 3), with the mean as the central value; error bars were derived from the standard deviation of three independent measurements). Integrated performance of the CDIS system. a Schematic diagram of the CDIS system for simultaneous capacitive desalination and sterilization. b Effluent NaCl concentration profiles for Ag NCs@CNF and pristine CNF at 350 mA g−1. c Desalination capacity and charge efficiency of the Ag NCs@CNF-based system at different constant currents. d Kinetic Ragone plot comparing desalination rate and capacity. e Desalination rate performance vs. electrode mass benchmarking against state-of-the-art FDI Ag-based electrode reported in the literature49,55–59. f Time-dependent inactivation of E. coli for Ag NCs@CNF and CNF. g Sterilization efficiency of Ag NCs@CNF under different NaCl concentrations. h Relative contributions of RS-mediated inactivation, intrinsic antibacterial effects of Ag species, and physical contact-based inactivation to the overall disinfection efficiency. i Inactivation rate of the bacteria adsorbed on the surface of Ag NCs@CNF and pristine CNF during the first 60 min. j Sterilization efficiency of Ag NCs@CNF at different productivity. k Performance trade-off relationships of productivity vs. energy consumption for Ag NCs@CNF-based CDIS, traditional membrane CDI, and flow-through CDI (ΔC = 5 mM, water recovery = 50%. Note: aAll sterilization tests followed a hybrid protocol: the system was first operated in constant current mode until the cut-off voltage was reached, and then switched to constant voltage mode to complete the total 60-min duration; bSamples with no detectable colonies were assigned a value at the detection limit (4 log reduction, 99.99% kill), limited by the detection limit of the plate counting method (1 CFU mL−1 under ideal conditions). (Data are presented as mean ± SD (n = 3), with the mean as the central value; error bars were derived from the standard deviation of three independent measurements). In addition to superior desalination, the Ag NCs@CNF-based CDIS system exhibits exceptional and rapid antibacterial activity (Figs. 5f–i and S12, S14–16). Under a hybrid operation protocol, a sustained >4.0 log reduction (detection limit) against Gram-negative E. coli was achieved within 60 min in single-pass mode (Figs. 5f and S12), while, in stark contrast, the pristine CNF offers only 1.56 log reduction (Figs. 5f and S13), which is probably due to the vastly different RS generation capability of the two electrode systems (Figs. 4h and S10). The system demonstrates broad-spectrum efficacy, achieving >4.0 log reduction (detection limit) against Bacillus subtilis (B. subtilis, Figs. S16 and S15) and the fungus pathogen Candida albicans (C. albicans, Figs. S15 and S17). This consistent high-level performance across phylogenetically diverse microorganisms—which possess vastly different cell-envelope structures (e.g., thick peptidoglycan, outer membrane, or chitinous wall)—underscores the universal biocidal mechanism of the anodically generated RS. RS inflict non-selective oxidative damage to essential cellular components, effectively bypassing structural resistance conferred by specific cell-wall architectures53. Notably, subtle differences in sterilization kinetics were observed: E. coli exhibited a slightly more stable and less fluctuating disinfection profile compared to B. subtilis and C. Albicans (Figs. 5f and S13). This can be rationally attributed to the conjunction of its thinner cell wall (characteristic of Gram-negative bacteria), which may facilitate faster RS penetration and cellular disruption, and its more negative ζ-potential, which enhances the electrostatic driving force for initial bacterial capture and enrichment at the anode surface under the applied electric field3. Ultimately, these kinetic nuances notwithstanding, the in situ RS generation provides a potent and overarching mechanism that ensures robust and reliable inactivation across the entire spectrum of tested microorganisms. In addition, this robust sterilization performance shows negligible dependence on salinity, with >4.0 log reduction (detection limit) maintained in both ultrapure water and 1435 mg L−1 NaCl solution (Fig. 5g), confirming its suitability for real-world varying water compositions. A fundamental question, however, is how the three putative inactivation mechanisms—RS-mediated oxidation, intrinsic antibacterial effects of Ag species, and physical contact—quantitatively contribute to the overall sterilization efficiency. To deconvolute these contributions (Figs. 5h, i, and S18), we employed an AuNCs@CNF control electrode (synthesized identically, Fig. S19a), since Au lacks intrinsic antibacterial activity. By matching its RS output to that of AgNCs@CNF (~96.0 μM), any residual difference in bactericidal efficiency can be attributed solely to Ag-specific effects. Antibacterial tests (Figs. S19b and S20) show that this intrinsic Ag contribution accounts for ~0.32% of total inactivation. To further isolate physical contact from RS oxidation, we examined a bare CNF electrode, which contributes negligible RS and lacks Ag-based activity (Figs. 5f and S10). The apparent sterilization efficiency of CNF encompasses both physical adsorption and contact-mediated inactivation. To distinguish these processes, fluorescence microscopy was used to quantify the fraction of surface-adsorbed bacteria that were inactivated versus merely adsorbed (Figs. 5i and S21). On CNF, ~89.50% of adsorbed bacteria remained viable, indicating predominant physical adsorption, whereas on AgNCs@CNF all adsorbed cells were inactivated. From these data, the physical contact-mediated bactericidal efficiency was calculated to be 10.14% (i.e., 10.50% of the CNF apparent sterilization efficiency) (Supplementary Note III). Thus, the overall sterilization is semi-quantitatively attributed to RS-mediated inactivation (~89.54%), physical contact (~10.14%), and intrinsic Ag effects (~0.32%), demonstrating that RS are the dominant antimicrobial agents in this system. To underscore the unique merits of our material design, we conducted systematic comparisons with two representative counterparts. First, direct comparison with bismuth nanoclusters embedded in carbon nanofibers (Bi NCs@CNF)48—a classic Faradaic CDI anode—revealed that while it delivers a respectable desalination capacity (61.97 mg Cl g−1), its RS generation is minimal (Fig. S22a), resulting in only 0.55 log reduction (Fig. S22b). This confirms the difficulty for conventional Faradaic materials to achieve concurrent desalination and disinfection. To further probe how surface properties regulate dual functionality, we synthesized glutathione (GSH)-protected Ag28(GSH)20 NCs supported on a typical cobalt-based zeolitic imidazolate framework, ZIF-67 (Ag28(GSH)20 NCs@ZIF-67) as a contrasting anode3. Although this material exhibits excellent antibacterial activity, its desalination capacity in bacteria-containing medium is only 24.02% that of our ligand-free Ag NCs@CNF (Fig. S23). This marked difference stems from the dense coverage of GSH ligands on the NC’s surface, which occupy a substantial fraction of Ag active sites and drastically reduce the proportion of surface Ag atoms available for chloride storage, thereby limiting desalination capability. Together, these comparisons clearly delineate the inherent challenge of achieving high-performance desalination and sterilization simultaneously: Bi NCs@CNF lacks disinfection-active sites, while Ag28(GSH)20 NCs@ZIF-67 sacrifices ion-storage capacity due to surface ligand blocking. In turn, this highlights the rational design of our ligand-free Ag NCs@CNF anode, whose exposed Ag surface atomically integrates abundant chloride-storage sites and highly active RS-generation sites, enabling intrinsic synergy between desalination and disinfection. To evaluate the practical viability of the CDIS system as an integrated purification platform, we scaled up the device threefold and assessed its productivity and total energy consumption (desalination plus disinfection). Following the methodology of Hawks et al.54, we varied either the flow rate at a fixed current or the current at a fixed flow rate to achieve a defined concentration reduction (ΔC = 5 mM) and water recovery (~50%) while maintaining >3-log bacterial inactivation (Figs. 5j, k, and S24). Even under microbial challenge, the productivity at a given energy consumption substantially exceeds that of membrane CDI and flow-through CDI under comparable conditions54, underscoring the practical advantage of the integrated design. To elucidate how microbial presence influences desalination, we systematically compared the performance of Ag NCs@CNF and CNF anodes in both sterile and E. coli-laden (104 CFU mL−1) saline streams (Fig. 6). As shown in Fig. 6a–c, for the bactericidal Ag NCs@CNF anode, the presence of bacteria caused a negligible change in desalination capacity (84.25 vs. 85.6 mg Cl g−1), salt removal rate (0.33 vs. 0.33 mg Cl g−1 s−1), or charge efficiency (78.9 vs. 90.3%). This invariance confirms that inactivated bacteria do not block chloride storage sites or hinder ion transport. In stark contrast, the CNF anode—which only physically adsorbs bacteria—suffered severe performance degradation in the bacterial medium, with its desalination capacity, rate, and efficiency dropping by approximately 46, 45, and 57%, respectively. This decline is attributed to the accumulation of live bacteria on CNF, forming an electrostatically shielding layer that impedes ion access to the electrode surface.Fig. 6Comprehensive evaluation of system stability.Desalination performance under microbial load and concurrent disinfection durability in a saline environment. a Effluent NaCl concentration profiles, b desalination capacity, c charge efficiency, d kinetic Ragone plot, and e cycling stability over 200 cycles for Ag NCs@CNF and CNF in sterile and E. coli-containing feeds. f Corresponding cycling sterilization stability of the two anodes. Note: samples with no detectable colonies were assigned a value at the detection limit (4 log reduction, 99.99% kill), limited by the detection limit of the plate counting method (1 CFU mL−1 under ideal conditions). (Data are presented as mean ± SD (n = 3), with the mean as the central value; error bars were derived from the standard deviation of three independent measurements). Comprehensive evaluation of system stability. Desalination performance under microbial load and concurrent disinfection durability in a saline environment. a Effluent NaCl concentration profiles, b desalination capacity, c charge efficiency, d kinetic Ragone plot, and e cycling stability over 200 cycles for Ag NCs@CNF and CNF in sterile and E. coli-containing feeds. f Corresponding cycling sterilization stability of the two anodes. Note: samples with no detectable colonies were assigned a value at the detection limit (4 log reduction, 99.99% kill), limited by the detection limit of the plate counting method (1 CFU mL−1 under ideal conditions). (Data are presented as mean ± SD (n = 3), with the mean as the central value; error bars were derived from the standard deviation of three independent measurements). Remarkably, long-term cycling revealed an unanticipated synergy in the integrated system (Figs. 6d, e, and S25, S26). In the CNF system, bacterial presence critically degraded performance, leading to a near-complete loss of desalination capacity within 70 cycles. By contrast, operation in a sterile environment allowed CNF to retain 61.28% of its initial capacity after 200 cycles (Fig. 6d). Remarkably—and contrary to conventional expectation—the AgNCs@CNF system not only maintained its desalination function in the bacterial medium but exhibited enhanced cycling stability, retaining 92.95% of its capacity after 200 cycles, outperforming even its bacteria‑free counterpart (88.33% retention). This enhanced durability is closely associated with its sustained antibacterial efficacy (Fig. 6e): the AgNCs@CNF-based system retains >4.0 log reduction (detection limit) even after 200 cycles, while that of the CNF control drops below 70% after only 80 cycles (a consequence of the reversible physical adsorption of bacteria on the CNF electrode surface (Fig. 5i), resulting in the release of viable bacteria at adsorption equilibrium). To further exclude detrimental effects from Ag leaching—especially in the presence of divalent or reactive anions, such as SO42− and NO3−—we performed comparative desalination experiments coupled with ICP-MS analysis of the effluent. Under mixed-anion conditions, the dissolved Ag concentration was only 2.250 ppb at the 1st cycle and 1.750 ppb at the 200th cycle, both far below the WHO drinking-water guideline of 100 μg L−1. This confirms that Ag leaching is negligible and does not compromise cycling stability. Moreover, though the chloride removal capacity decreased by approximately 18% compared to the pure chloride condition due to co-ion competition Fig. S27, (left), the cycling stability barely changed (Fig. S27, right and Supplementary Note IV), indicating coexisting anions have a minimal impact on cycling performance. These distinctive observations prompted us to probe the fundamental interactions between ions and bacteria within the capacitive system, focusing on three pivotal questions: (i) Does electrostatic shielding genuinely occur, and how does it quantitatively govern the electric field and ion transport? (ii) Why does bacterial inactivation lead to such a pronounced divergence in desalination performance? (iii) What is the origin of the enhanced cycling stability of the Ag NCs@CNF system in the presence of microorganisms? To elucidate the fundamental mechanisms governing microbe-ion interactions and transport dynamics, we systematically investigated microbial colonization and the physicochemical evolution of the electrode surfaces. FESEM imaging (Fig. 7a) revealed a striking morphological contrast: the pristine CNF surface was densely covered with viable, rod-shaped E. coli (left panel), while the Ag NCs@CNF surface exhibited only scattered microspheres (6–8 μm in diameter) composed of aggregated cellular debris (right panel). This distinct architecture originates from a dynamic fragmentation-aggregation process (Fig. 7b). Upon contact with the Ag NCs@CNF anode, bacteria are rapidly inactivated and fragmented via locally generated RS (see Fig. 4g–i). The resulting cellular debris subsequently assembles into discrete spherical aggregates under the combined influence of the applied electric field and hydrodynamic flow. Specifically, the electric field within the capacitive region compresses inactivated bacterial debris against the electrode surface, while the hydrodynamic flow simultaneously drives the debris to roll up into spherical aggregates. This behavior emerges only because the fragments produced by inactivation exhibit substantially reduced size and diminished surface charge, which weaken their electrostatic adhesion to the electrode3. Consequently, the detached debris is continuously shaped by the combined electric and flow fields into microspheres that ultimately detach from the electrode surface.Fig. 7Systematic dissection of microbe–ion interactions within the capacitive field.a FESEM images of the bacteria distribution on the surface of the CNF (left image) and Ag NCs@CNF electrode (right image) after 60 min sterilization test (no regeneration applied). b Schematic contrasting bacterial status on electrode surfaces: live bacteria accumulated on CNF versus inactivated (dead) bacteria on Ag NCs@CNF, with the corresponding calculated surface coverage percentages indicated. Finite-element simulation of c electric field distribution and d Cl− concentration distribution on the surface of CNF (left part) and Ag NCs@CNF (right part) in the presence of both NaCl and E. coli. WT-EXAFS contour plots at the Ag K-edge in e discharged and f charged states. g Evolution of Ag–Cl and Ag–Ag coordination numbers during cycling. h Nyquist plots from cycling desalination tests with and without E. coli. i Zeta potential variation after 200 cycles and corresponding local RS concentration at the Ag NCs@CNF surface in the presence and absence of E. coli. Systematic dissection of microbe–ion interactions within the capacitive field. a FESEM images of the bacteria distribution on the surface of the CNF (left image) and Ag NCs@CNF electrode (right image) after 60 min sterilization test (no regeneration applied). b Schematic contrasting bacterial status on electrode surfaces: live bacteria accumulated on CNF versus inactivated (dead) bacteria on Ag NCs@CNF, with the corresponding calculated surface coverage percentages indicated. Finite-element simulation of c electric field distribution and d Cl− concentration distribution on the surface of CNF (left part) and Ag NCs@CNF (right part) in the presence of both NaCl and E. coli. WT-EXAFS contour plots at the Ag K-edge in e discharged and f charged states. g Evolution of Ag–Cl and Ag–Ag coordination numbers during cycling. h Nyquist plots from cycling desalination tests with and without E. coli. i Zeta potential variation after 200 cycles and corresponding local RS concentration at the Ag NCs@CNF surface in the presence and absence of E. coli. Complementary EXAFS analysis (Fig. 7e–g and Table S3) indicated that chloride storage in Ag NCs@CNF proceeds primarily via a reversible surface Faradaic reaction (Ag + Cl− → AgCl + e−), evidenced by a marked increase in Ag–Cl coordination upon charging. Crucially, the RS generation and disinfection efficacy remained stable even when part of the Ag NCs transformed into AgCl (Figs. 4h and 5j), demonstrating that chloride coordination does not compromise catalytic activity. This confirms the intrinsic compatibility of desalination and disinfection functions at the material level. As illustrated in the right panel of Fig. 7b, geometric modeling and mass conservation calculations—treating E. coli as a capsule (length = 2 μm, diameter = 0.5 μm, cell-wall thickness ≈ 0.01 μm)—enabled quantification of surface coverage. The results show that live bacteria occupy ~95% of the CNF surface, whereas the inactivated microspheres on Ag NCs@CNF reduce the effective coverage to only ~0.1% (see Supplementary Note V for detailed calculations). Finite-element simulations based on these coverage values (Fig. 7c, d) provided definitive evidence: the dense layer of viable bacteria on CNF creates a pronounced charge-shielding barrier (left panel of Fig. 7c), markedly attenuating the local electric field and severely hindering Cl− transport to the electrode surface, thereby establishing an ion-depletion zone (left panel of Fig. 7d). In contrast, the sparse aggregates of inactivated debris on Ag NCs@CNF induce negligible shielding (right panels of Fig. 7c, d), preserving both the electric field and the ion-concentration gradient. These findings conclusively demonstrate that electrostatic shielding by accumulated live bacteria is the primary mechanism impairing desalination, whereas the aggregation and detachment of inactivated cells effectively eliminate such interference. The mechanism behind the counterintuitive enhancement of cycling stability by microbial presence, however, required further clarification. In situ electrochemical impedance spectroscopy (EIS) revealed a sharp contrast in cycling stability: the impedance of the Ag NCs@CNF anode surged by 263.85% over 200 cycles under sterile conditions, while it remained largely unchanged in the presence of bacteria (Figs. 7h and S28). We attribute this enhanced stability to the ability of the bacterial inactivation process to scavenge excess RS at the anode surface, thereby mitigating oxidative degradation of the electrode material. To test this, we monitored changes in surface physicochemical states: sterile operation caused a significant increase in surface charge (Δζ = + 1.23 mV) due to continuous anodic oxidation, while bacterial presence resulted in a markedly smaller change (Δζ = + 0.7 mV) coupled with a lower local RS concentration (Fig. 7i). This strongly supports our hypothesis that bacteria, by scavenging RS, alleviate oxidative degradation of the electrode. This transforms the conventional view of microbial interference, turning a typically detrimental factor into a beneficial contributor to long-term durability. To rigorously establish the mechanistic basis of microbe–ion interplay, we designed complementary control experiments that decouple the contributions of electrostatic shielding and RS scavenging. We compared Ag NCs@CNF, which integrates Faradaic chloride storage with strong RS generation and bactericidal activity, with Bi NCs@CNF, a Faradaic analogue that exhibits comparable anion storage yet negligible RS generation and minimal antibacterial action (Fig. S22). For each system, cycling stability was evaluated under three parallel conditions: sterile, live E. coli-containing, and dead E. coli-containing electrolytes (heat-inactivated and ultrasonically dispersed to ensure uniform distribution of non-viable cells). The Ag NCs@CNF system showed similar durability in the presence of dead and live bacteria, both outperforming the sterile condition (Fig. S29a). In stark contrast, Bi NCs@CNF retained comparable stability under sterile and dead-cell conditions but suffered pronounced degradation when live bacteria were present (Fig. S29b). Three conclusions emerge. First, under sterile conditions, Ag NCs@CNF exhibited inferior capacity retention relative to Bi NCs@CNF (88.33% vs. 91.69% after 200 cycles), directly demonstrating that RS themselves compromise long‑term stability. Second, the consistently higher stability of Ag NCs@CNF in bacteria-containing electrolytes relative to sterile operation, regardless of cell viability, supports RS scavenging by bacterial membrane structures as the origin of the improved durability (consistent with the reduced local RS concentration measured in the presence of bacteria; Fig. 7i). Third, the observation that dead bacteria do not degrade Bi NCs@CNF performance, whereas live bacteria cause severe decline, explicitly identifies electrostatic shielding—rather than physical spatial blocking—as the primary mechanism by which live bacteria impair desalination. Collectively, these findings provide direct, internally consistent evidence that electrostatic shielding governs performance deterioration, that RS intrinsically undermine electrode stability, and that bacterial RS consumption paradoxically enhances cycling durability. The inclusion of noble metals often raises concerns over material cost. To assess practical feasibility, we compared the lifecycle cost-effectiveness of bare CNF and AgNCs@CNF electrodes (Table S4 and Supplementary Note VI). Although AgNCs@CNF incurs a higher initial material cost, this is offset by three significant advantages: (i) ~3-fold higher per-cycle desalination capacity (84.25 vs. 27.87 mg Cl g−1), (ii) ~5.7-fold higher RS yield (96.0 vs. 16.7 μM), and (iii) a tenfold longer operational lifetime (200 vs. 20 cycles, defined as >50% capacity loss; Fig. 6e). Normalized over 200 cycles of continuous operation, the AgNCs@CNF electrode achieves desalination and RS costs of 6.48 × 10−3 and 9.10 × 10−2 USDper mgCl and per μM, respectively—representing 88% and 92% reductions relative to bare CNF (46.26 × 10−3 and 12.35 × 10−2). This analysis justifies the incorporation of a small amount of Ag as a cost-effective means to achieve durable, high-performance integrated desalination and disinfection. Furthermore, the concentration-mediated size-focusing strategy is versatile and can be extended to low-cost metals. As a proof of concept, we synthesized Fe NCs-impregnated CNF (FeNCs@CNF) using earth-abundant Fe. Preliminary evaluation confirms that FeNCs@CNF delivers >4.0 log reduction (detection limit) with an RS yield of ~106.7 μM and a desalination capacity of 98.65 mg Cl g−1 (Figs. S30 and S31), demonstrating that the synthetic strategy is readily adaptable to non-noble metals and offers a viable route toward cost-effective, noble-metal-free electrodes for practical applications. Taken all together, this study provides a comprehensive experimental and theoretical elucidation of the microscopic competition between microbes and ions in capacitive deionization. It delivers three key mechanistic advances: (1) it moves beyond the vague notion of spatial blocking, identifying electrostatic shielding as the principal cause of performance decay. (2) It discovers the dynamic fragmentation-aggregation process of inactivated bacteria, explaining their minimal impact. (3) It reveals the counterintuitive synergy wherein bacterial scavenging of RS mitigates electrode oxidation, enhancing stability. These insights not only furnish a fundamental explanation for the superior performance of the Ag NCs@CNF-based CDIS system but also establish a theoretical foundation for designing next-generation, biofouling-resistant, and highly stable electro-driven water purification systems. In summary, we have developed a synchronous CDIS system driven by a dual-functional Ag NCs@CNF anode that enables simultaneous, efficient salt removal (84.25 mg Cl g−1) and complete microbial inactivation (>4.0 log reduction, detection limit). Beyond its superior performance, this work provides the unamibuguous mechanistic elucidation of microbe-ion interactions in capacitive systems, challenging the prevailing spatial-competition paradigm. We reveal that (i) electrostatic shielding by live bacteria is the primary cause of performance decay; (ii) inactivated cells undergo a dynamic fragmentation-aggregation process that eliminates persistent shielding; and (iii) remarkably, bacterial inactivation enhances long-term cycling stability by scavenging RS and mitigating anode oxidation. This study not only establishes a breakthrough electrochemical platform for integrated water purification but also offers a theoretical framework for designing sustainable, fouling-resistant electro-driven systems. Methods Chemicals Silver nitrate (AgNO3, 99.8%), polyacrylonitrile (PAN, average MW 150,000), N, N-dimethylformamide (DMF, >99.5%), and sodium chloride (NaCl, 99.5%), N-methyl-2-pyrrolidone (NMP, 98.0%) and polyvinylidene difluoride (PVDF, 2.5%) were purchased from Sinopharm Chemical Reagent Co., Ltd. Standard strains of Bacillus subtilis (B. subtilis), Escherichia coli ATCC 25922 (E. coli), and Candida albicans (C. albicans) were purchased from the China General Microbiological Culture Collection Center. Luria–Bertani (LB, 1.0%) and Luria–Bertani agar were purchased from Qingdao Hope Biol Co., Calcein Acetoxymethyl Ester (Calcein AM) and Propidium iodide (PI, >94.0%) were purchased from Beyotime Biotechnology, Deionized water (DI, 18.2 MΩ) was used as the universal solvent unless otherwise indicated. All materials were used as received without further treatment. Synthesis of CNF, s-Ag NCs@CNF, Ag NCs@CNF, and Ag NPs@CNF The Ag NCs@CNF material was synthesized via a combined electrospinning–carbonization approach, incorporating a concentration-mediated size focusing strategy. In a typical procedure, silver nitrate (C(Ag+) = 0.005, 0.01, or 0.03 M) and polyacrylonitrile (PAN, 1 g) were dissolved in N, N-dimethylformamide (DMF, 10 mL) under stirring at 60 °C until a homogeneous yellowish solution was obtained (≥8 h). The solution was then loaded into a 10 mL plastic syringe equipped with a blunt-tip needle (diameter: 0.7 mm) for electrospinning. A DC voltage of 22 kV was applied between the needle and an aluminum roll collector, with a tip-to-collector distance of 15 cm and a solution feed rate of 0.7 mL h−1. After electrospinning, the as-spun fiber mat was collected and dried in an oven for at least 12 h. The obtained PAN nanofiber mat containing silver was subsequently stabilized in air at 260 °C for 2 h with a ramp rate of 2 °C min−1. Finally, three distinct silver-containing carbon nanofiber samples were prepared by carbonizing the stabilized fibers at 1000 °C for 2 h under an inert atmosphere, using a heating rate of 5 °C min−1. These samples were denoted as s-Ag NCs@CNF, Ag NCs@CNF, and Ag NPs@CNF, respectively. The pristine CNF was also synthesized as the control sample, following identical steps except for the omission of silver nitrate. Synthesis of Bi NCs@CNF, Au NCs@CNF, and Cu NCs@CNF Following the same method described above, Bi NCs@CNF, Au NCs@CNF, and Cu NCs@CNF were synthesized using bismuth nitrate (C(Bi+) = 0.01 M), chloroauric acid, and copper sulfate (C(Cu2+) = 0.01 M) as the metal sources. The resulting samples were denoted as Bi NCs@CNF, Au NCs@CNF, and Cu NCs@CNF, respectively. Material characterization The surface morphology, elemental distribution, crystal structure, and specific surface area of CNF, s-Ag NCs@CNF, Ag NCs@CNF, and Ag NPs@CNF were systematically characterized using a suite of analytical techniques. Field-emission scanning electron microscopy (FESEM, JEOL JSM-LV5610), transmission electron microscopy (TEM, JEMF200), and high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM, JEOL ARM 200F equipped with a probe-side spherical aberration corrector) were employed to elucidate nanoscale architectures. Elemental mapping was conducted via energy-dispersive X-ray spectroscopy (EDS, Oxford). Crystalline phases were identified by X-ray diffraction (XRD, Panalytical PRO PW3040/60) using Cu-Kα radiation (30 kV, 25 mA) across a 2θ2θ range of 10–90° at a scan rate of 5° min−1. Specific surface areas were determined through nitrogen adsorption–desorption isotherms measured at 77 K using an Autosorb-iQ analyzer, with calculations based on the multipoint Brunauer–Emmett–Teller (BET) method. Extended X-ray absorption fine structure (EXAFS) data were processed in k-space and Fourier-transformed using a k³-weighted EXAFS function to enhance oscillations at high k-values. Detailed fitting parameters are summarized in Table S3. Wavelet transform analysis of EXAFS spectra was carried out using the FORTRAN-based program HAMA, with the Morlet wavelet as the mother function (parameters: \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\sigma=1,\,\kappa=10$$\end{document}σ=1,κ=10). All EXAFS spectra are presented without phase correction. Electrochemical characterization Electrochemical properties were evaluated using a CS380 workstation (CorrTest Instruments) in a three-electrode configuration with 1 M NaCl electrolyte (Ag/AgCl reference). Cyclic voltammetry (CV) and galvanostatic charge-discharge (GCD) measurements were conducted to assess capacitive behavior, while electrochemical impedance spectroscopy (EIS) was performed from 0.1 Hz to 100 kHz with a 5 mV AC amplitude to analyze charge transfer characteristics. The specific capacitances (Csp in F g−1) of the CNF, s-Ag NCs@CNF, Ag NCs@CNF and Ag NPs@CNF were determined from both CV and GCD measurements using the following equations:1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{{\rm{From}}}}\; {{{\rm{CV}}}}\; {{{\rm{curves}}}}:{{{\mbox{C}}}}_{{{\mbox{s}}}_{{\mbox{P}}}}=\frac{\overline{i}}{v\times {{\mbox{m}}}\times {\Delta}{{\mbox{V}}}}$$\end{document}FromCVcurves:CsP=i¯v×m×ΔVWhere,\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\,{C}_{{{\mbox{s}}}_{{\mbox{P}}}}$$\end{document}CsP, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\bar{i}$$\end{document}ī (A), vv (V s−1), m (g), \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\Delta}{V}$$\end{document}ΔV, and represent the specific capacitance, the average current, scan rate, mass of active material of the electrode, and operating potential window, respectively.2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{{\rm{From}}}}\; {{{\rm{GCD}}}}\; {{{\rm{profiles}}}}:{{{\mbox{C}}}}_{{{\mbox{s}}}_{{\mbox{P}}}}=\frac{I\times {\Delta }t}{\,m \times {\Delta }V}$$\end{document}FromGCDprofiles:CsP=I×Δtm×ΔVwhere I, Δt, m, and ΔV, refer to the current (A), discharge time (s), mass of active material (g), and potential window (V), respectively. Desalination experiments The desalination performance was evaluated using a hybrid capacitive deionization system operated in a single-pass mode. A NaCl solution (concentration: 1435 mg L−1) was continuously pumped through the hybrid capacitive deionization cell at a constant flow rate of 10 mL min−1 using a peristaltic pump. The effluent conductivity was monitored in real-time using a precision conductivity meter (INESA Scientific Instrument, DDS308F) connected to the cell outlet, while the conductivity results were further corrected using ion chromatography (Thermo Scientific, ICS-600). All experiments were conducted under constant current conditions at four different current densities (350, 500, 700, 750, and 850 mA g−1) with a voltage cutoff of 1.2 V to prevent side reactions. The Ag concentration at the outlet of the CDIS was analyzed using inductively coupled plasma mass spectrometry (ICP-MS, Nexion 2000) to ensure no significant ion leaching occurs. In our experiment, the desalination capacity, desalination rate, charge efficiency (Λ), productivity (P, L h−1 m−3) and energy consumption (Ev,W h m3) are defined below:3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{{\rm{Desalination}}}}\; {{{\rm{capacity}}}}\left({{{\rm{mgCl}}}}{{{{\rm{g}}}}}^{-1}\right)=\frac{Q{\int }_{0}^{t}({c}_{0}-{c}_{t})\,{dt}}{m}$$\end{document}DesalinationcapacitymgClg−1=Q∫0t(c0−ct)dtm4\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{{\rm{Desalination\; rate}}}}({{{\rm{mgCl}}}}{{{{\rm{g}}}}}^{-1}\,{{{{\rm{s}}}}}^{-1})=\frac{Q{\int }_{0}^{t}({c}_{0}-{c}_{t})\,{dt}}{{mt}}$$\end{document}Desalination rate(mgClg−1s−1)=Q∫0t(c0−ct)dtmt5\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{{\rm{Charge}}}}\; {{{\rm{efficiency}}}}(\Lambda )=\frac{\Gamma \times {{{\rm{F}}}}}{\Sigma }$$\end{document}Chargeefficiency(Λ)=Γ×FΣ6\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{{\rm{Productivity}}}}({{{\rm{L}}}}{{{{\rm{h}}}}}^{-1}{{{{\rm{m}}}}}^{-3})=\frac{{V}_{{single}}}{A\times t}$$\end{document}Productivity(Lh−1m−3)=VsingleA×t7\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{{\rm{Energy\; consumption}}}}({{{\rm{Wh}}}}{{{{\rm{m}}}}}^{3})={\int }_{t}\,\left(I\times {V}_{t}\right){dt}$$\end{document}Energy consumption(Whm3)=∫tI×Vtdtwhere c0 (mg L−1) and ct (mg L−1) are the feed and transient Cl− concentrations. Q (L s−1) is flowrate, m (g) is the total mass of the anode, including both the CNF and the Ag NCs, t (s) is the desalination time, Γ is the desalination capacity (mol g−1), ∑ (charge, C g−1) is obtained by integrating the corresponding current, F is the Faraday constant (96,485 C mol−1), Vsingle (L) is the volume desalinated, Vt (V) is the voltage during the desalination, and I (A) is the desalination current. Antimicrobial experiments (a) Antimicrobial assay of the CDIS system The sterilization model solution was prepared according to a standard protocol: a single colony of each microbial strain (E. coli, B. subtilis, and C. albicans) was transferred from a solid Luria–Bertani (LB) agar plate into 12 mL of liquid LB medium and incubated at 37 °C for 24 h under shaking at 200 rpm. The resulting culture was then centrifuged at 4650 × g for 5 min, resuspended in sterile water or 1435 mg L−1 sterilized saline solution, and adjusted to an OD600 (optical density at 600 nm) of 0.1, corresponding to approximately 10⁷ CFU mL−1. The bacterial suspension was subsequently diluted proportionally to obtain a final concentration of 104 CFU mL−1. Sterilization experiments were performed in single-pass mode using a hybrid electrochemical protocol via a CS310H workstation. The system was initially operated under constant-current conditions until reaching a cut-off voltage of 1.4 V, after which it was switched to constant-voltage mode to complete a total operation time of 60 min. In each experiment, the model bacterial solution (10⁴ CFU mL−1) was continuously pumped through the CDIS unit cell. Effluent samples were collected at the outlet, and the residual bacterial concentration was quantified via the standard plate colony counting method. The antimicrobial activity of the CDIS system was calculated as follows:8\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{{\mathrm{Antimicrobial}}}}\; {{{\rm{efficiency}}}}\left(\%\right)=\frac{{{CFU}}_{{initial}}{-{CFU}}_{{survive}}}{{{CFU}}_{{initial}}}\times 100\%$$\end{document}Antimicrobialefficiency%=CFUinitial−CFUsurviveCFUinitial×100%9\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{{\mathrm{Log}}}}\; {{{\rm{reduction}}}}={{{\mathrm{log}}}}_{10}\left({{CFU}}_{{initial}}\right)-{{{\mathrm{log}}}}_{10}\left({{CFU}}_{{survive}}\right)$$\end{document}Logreduction=log10CFUinitial−log10CFUsurvivewhere CFUinitial and CFUsurvive represent the initial bacterial count in solution and the number of surviving bacteria in the solution at the outlet, respectively. All experiments were performed at least in triplicate to ensure the repeatability and stability of the experimental data, and the average results are presented in this study, with error bars displaying the standard deviation (s.d.). (b) Cell viability assay The viability of E. coli adsorbed on the anode surface was evaluated using a LIVE/DEAD® BacLight™ Bacterial Viability Kit, followed by inverted fluorescence microscopy (Nikon TE2000). At specified disinfection time points (0, 10, 30, and 60 min), the system was short-circuited, after which bacteria were washed off the electrode for staining. The kit utilizes SYTO 9 to stain all bacterial cells green, while propidium iodide (PI) selectively enters membrane-compromised cells, staining them red and simultaneously quenching SYTO 9 fluorescence at those sites. Stained samples were washed with PBS, resuspended, mounted on glass slides, and immediately imaged under bright-field, green fluorescence (FITC, live cells), and red fluorescence (TRITC, dead cells) channels. Spatial distribution and survival ratios were derived from overlaid dual-channel images. For comparative assessment of antibacterial performance, Ag NCs@CNF and pristine CNF were examined in parallel over 60 min. Detection limit: given that the initial concentration used in this experiment was 104 CFU, the detection limit—defined as the maximum log reduction at which a single colony would still be detected—corresponds to 4 log reduction. Samples with no detectable colonies were assigned a value at the detection limit (4 log reduction, 99.99% kill), limited by the detection limit of the plate counting method (1 CFU mL−1 under ideal conditions). RS detection The concentrations of various reactive species (RS) in water were determined using spectrophotometry combined with the scavenger method. The iodometric method for evaluating the oxidative performance of a solution is based on the following principle: under acidic conditions, RS (evaluated based on H2O2 equivalents) oxidize potassium iodide to produce iodine (I2), which subsequently forms a blue complex with starch. The absorbance of this complex is measured at 580 nm, and the RS concentration is indirectly quantified using a standard curve (Fig. S8). (a) Sample measurement Zero point five milliliters of the in situ collected (Fig. S8) water sample, and 0.5 mL each of deionized water, sulfuric acid (3 M), potassium iodide (100 g L−1), and starch indicator (10 g L−1) were sequentially added. After a reaction time of 20 min, the absorbance was measured at 580 nm. The RS concentration, evaluated based on H2O2 equivalents, was calculated from the absorbance difference between the sample and the blank using a standard curve. (b) RS contribution analysis To distinguish the contributions of different reactive species, the scavenger method was employed: tert-butanol (TBA, 1 M) was used to quench ·OH, and glycine (0.5 M) was used to quench reactive chlorine species. The RS concentration of the quenched sample was then determined by following the same sample measurement procedure described above, replacing deionized water with the respective scavenger solution. This allowed indirect calculation of the contribution of ·OH or reactive chlorine to the total RS concentration. Finite element simulations This study utilizes the dilute species transport, chemical reaction, and electrostatics modules in COMSOL for numerical simulations. Diffusion is governed by the Nernst–Planck–Poisson equation.10\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\frac{\partial c}{\partial t}+\nabla \cdot {{{\boldsymbol{J}}}}+{{{\boldsymbol{u}}}}\cdot \nabla c=R$$\end{document}∂c∂t+∇⋅J+u⋅∇c=R11\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{{\boldsymbol{J}}}}=-{D}_{i}\nabla c+{z}_{i}{u}_{m,i}{Fc}\nabla \varphi$$\end{document}J=−Di∇c+zium,iFc∇φ12\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${u}_{m,i}=\frac{{D}_{i}}{{RT}}$$\end{document}um,i=DiRTc is the substance concentration, J is the mass transfer flux, R is the reaction source (in this model the reaction source is 0), Di is the substance diffusion coefficient, u is the fluid velocity, zi is the charge of the substance, um,i is the mobility of substance i, satisfying the Nernst–Einstein relation, ϕ is the electric potential, F is the Faraday constant. R is the ideal gas constant, T is the temperature (293.15 K in this study) The electrostatic module is described by the Poisson equation:13\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\nabla \bullet \left({\varepsilon }_{0}{\varepsilon }_{r}E\right)={\rho }_{v}$$\end{document}∇∙ε0εrE=ρv14\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$E=-\nabla {{\varnothing }}$$\end{document}E=−∇∅where ε0 is the vacuum permittivity, εr is the relative permittivity of the material, ϕ is the electric potential, E is the electric field, and ρV is the volumetric space charge density, which is set to zero in this study. The ambient relative permittivity is set to 78.5. The relationship between ionic concentration and the internal charge density follows:15\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\rho }_{V}=F{\sum}_{i}{z}_{i}{c}_{i}$$\end{document}ρV=F∑izici The diffusion coefficient of ions is set to 1 × 10−9 [m2 s−1] throughout the study. Functional form and fitting parameters All decay curves were fitted using a one-phase exponential decay function (ExpDec1): \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$y={y}_{0}+A\cdot \exp (-x/t)$$\end{document}y=y0+A⋅exp(−x/t), where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${y}_{0}$$\end{document}y0 is the offset (baseline at infinite time), AA is the amplitude, and tt is the time constant. The decay rate kk was calculated as 1/t1/t, and the half-life \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${t}_{1/2}$$\end{document}t1/2 as \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$t\cdot {{\mathrm{ln}}}(2)$$\end{document}t⋅ln(2).” Linear regression was performed using the ordinary least-squares (OLS) method, fitting the data to the equation y = ax + b, where a denotes the slope and b denotes the y-intercept. Supplementary information Supplementary Information Transparent Peer Review File Supplementary Information Transparent Peer Review File Source data Source Data Source Data