Introduction

Air pollutants like particulate matter (PM), bioaerosols, and volatile organic compounds (VOCs) have emerged as a pressing global challenge1,2, threatening both public health and ecosystems3,4,5,6. In response, various air filtration technologies have been developed to reduce exposure to polluted air7,8,9,10,11,12. However, commercial filters made from synthetic microfibers, such as melt-blown polypropylene (PP) fibers, face significant limitations. These filters require complex fiber and network formation processes and are ineffective against certain high-risk pollutants, particularly nanoscale PM and small-molecule VOCs, due to their micron-sized fibers and limited surface functionality13,14,15,16,17. Moreover, they are non-degradable and lack biocidal properties, contributing to environmental contamination18,19 and increasing the risk of secondary disease transmission when exposed to pathogenic microorganisms20,21,22.

Recent advancements have positioned biomass-derived materials as promising sustainable alternatives to synthetic fibers, thanks to their renewability and chemical versatility23,24,25,26. Notably, fibrous filters made from natural proteins like zein and silk have been developed through electrospinning for effective capture of PM and VOCs27. Nonetheless, their practical application is limited by insufficient mechanical strength and reliance on organic solvents during electrospinning, impacting both production and environmental sustainability28. Collagen fiber networks (CFNs) are protein-based materials ubiquitously found in animal hides, known for their hierarchical structure and strong mechanical properties29,30,31,32. These characteristics enable the production of robust filters without complex manufacturing processes, making CFNs suitable for separating diverse substances, including oily pollutants, natural products, metal ions, and airborne PM33,34,35. However, collagen fibers generally aggregate into thick bundles with diameters larger than 5 μm, primarily due to strong hydrogen bonding between adjacent nanofibrils36,37. This aggregation greatly reduces separation efficiency, especially for smaller airborne PM and indoor VOCs34,38,39. In addition, collagen fibers, as a protein-based material, offer a favorable environment for airborne microorganisms to attach and grow, leading to filter deterioration and potential pathogen spread21,40,41. Therefore, developing effective strategies to enhance fibril dispersion while imparting microbial resistance is essential for achieving efficient and sustainable air purification with collagen-based filters.

In this work, we present a nanofibrillation and functionalization strategy to engineer natural CFNs for high-performance air purification. By combining mechanical action with the chemical introduction of zwitterionic copolymers and Zr4+ ions, we disrupt hydrogen-bond-induced fibril aggregation, transforming collagen fiber bundles into nanoscale fibrils. This enhances air flow through the generated hierarchical channels, boosting purification efficiency. The nanofibrillation process also exposes more active amino groups on the nanofibril surface, improving the capture of indoor VOCs like formaldehyde (HCHO). Additionally, the coordination of sulfonate groups from the zwitterionic polymers with Zr4+ increases charge density around quaternary ammonium groups, imparting antimicrobial properties to the collagen-based filters. A life-cycle assessment highlights the environmental sustainability and economic viability of these filters, demonstrating their potential for commercial production. As such, this nanofibrillation approach enables the development of high-performance air filters from natural collagen fibers, capable of efficiently removing a wide range of pollutants, including PMs, pathogenic bioaerosols, and VOCs.

Results

Preparation and characterization of nanofibrillated CFNs

CFNs derived from animal skins are a renewable biomass resource mainly composed of type I collagen, featuring a natural hierarchical fiber structure42,43. As illustrated in Fig. 1A, the basic unit of CFNs is a right-handed helix structure (~1.5 nm) formed by three intertwined polypeptide chains. These triple helices assemble into nanofibrils (20−200 nm), which further aggregate into fiber bundles ranging from 5 to 100 μm in diameter. These bundles interweave into a three-dimensional network suitable for separation. In pristine CFNs, the nanofibril aggregation is driven by non-covalent interactions, particularly hydrogen bonds between active groups on adjacent nanofibrils (Fig. 1B). However, this aggregation reduces the separation efficiency of collagen-based filters.

Fig. 1: Nanofibrillation of CFNs.
figure 1

A Schematic depicting the 3D hierarchical structure of CFNs and the two-step nanofibrillation process. B Illustration of collagen fibril aggregation driven by hydrogen bonding. C Schematic of PSG copolymer introduction to enhance water adsorption, promoting hydration-assisted fibril dispersion. D Schematic of Zr-based coordination to passivate active sites and stabilize the nanofibrillated CFNs. E Photograph of a large-area Nano-CFN membrane. F Cross-sectional and G top-surface SEM images comparing CFN and Nano-CFN samples. Experiments were repeated three times with similar results. H Statistical analysis of fiber diameters before and after nanofibrillation.

Full size image

To address this limitation, we propose a two-step nanofibrillation strategy involving (1): mechanical stirring and surface hydrophilic modification to disperse collagen fiber bundles, and (2) coordination-driven passivation and stabilization of the nanofibrillated CFNs (Fig. 1A and Supplementary Table S1). Drawing inspiration from the role of water in nanocellulose processing44,45,46, the first step introduces a zwitterionic copolymer, poly(sulfobetaine methacrylate-co-glycidyl methacrylate) (PSG), onto the collagen fiber surface. The zwitterionic groups in PSG promote water adsorption, creating a hydration layer that enhances fiber swelling and dispersion (Fig. 1C)47, while the epoxy groups facilitate polymer grafting through a ring-opening reaction with the carboxyl groups on the CFN surface. The successful synthesis of PSG via free radical polymerization was confirmed by 1H nuclear magnetic resonance (1H NMR) spectroscopy and gel permeation chromatography (Supplementary Fig. S1)48. PSG was then used to modify CFNs derived from goat hide under continuous agitation (Supplementary Fig. S2). This mechanical action triggers fiber dispersion and PSG infiltration, while PSG grafting further accelerates fiber swelling. Together, these actions transform fiber bundles into nanofibrils, resulting in PSG-modified CFNs (PSG-CFN).

However, the initial nanofibrillation effect was temporary. As the sample dried, it reverted to a tightly packed state (Supplementary Fig. S3A) due to reformation of hydrogen bonds49. To address this, Zr(SO4)2 was employed as a stabilizer to treated PSG-CFN under stirring. This zirconium treatment serves two critical purposes (1): Zr4+ ions passivate active groups on the fibril surface, preventing hydrogen bond reformation; and (2) they form multinuclear coordination complexes at the nanofibril junctions, enhancing network stability (Fig. 1D). Zr4+ infiltration was controlled by adjusting the pH of the reaction solution. At pH 2.0, uniform infiltration occurred, and increasing the pH to 4.0 facilitated coordination, leading to more stable nanofibrillated CFNs (denoted as Nano-CFN) (Supplementary Fig. S3B). To verify the effect of zirconium treatment, we compared the thickness and packing density of CFN, PSG-CFN, and Nano-CFN in both hydrated and dried states. As shown in Supplementary Fig. S4, the thickness and packing density of PSG-CFN are comparable to those of Nano-CFN when hydrated. After drying, however, the thickness of PSG-CFN decreased significantly, and its packing density nearly doubled. In contrast, the thickness and packing density of Nano-CFN remained stable, confirming that zirconium treatment preserved the nanofibrillation effect. To further validate this, we treated Nano-CFN with oxalic acid and ethylenediaminetetraacetic acid disodium salt solution, which dissolve Zr-based crystals. As exhibited in Supplementary Fig. S5, the thickness of treated Nano-CFN decreased markedly with reduced water content, affirming the importance of zirconium treatment in stabilizing the nanofibrillated structure.

Significantly, we successfully prepared a scalable Nano-CFN membrane measuring 85 cm × 50 cm (Fig. 1E), showcasing the practicality of our nanofibrillation approach. Scanning electron microscopy (SEM) images reveal the clear transformation from thick bundles to well-dispersed, fluffy nanofibrils (Fig. 1F, G, and Supplementary Fig. S6). The sample thickness increased from 460 μm to 1.1 mm (Fig. 1F), while the average fiber diameter decreased drastically from 5.4 μm to 120 nm (Fig. 1H). Our nanofibrillation strategy has thus produced Nano-CFNs with high stability and well-dispersed nanofibrils, indicating great promise for purifying various air pollutants (Supplementary Fig. S7).

To confirm the roles of PSG and Zr4+ in nanofibrillation, we systematically characterized the samples before and after each modification step. The FTIR spectrum of PSG-CFN (Fig. 2A and Supplementary Fig. S8A) shows characteristic peaks at 1106 cm−1 (C−O−C), 1038 cm−1 (−SO3−), confirming the successful incorporation of hydrophilic PSG into CFNs via the epoxy ring-opening reaction. After zirconium treatment, the −SO3− and −OH peaks shifted to lower wavenumbers, indicating coordination between Zr4+ and active groups from PSG-CFN50. X-ray photoelectron spectroscopy (XPS) survey spectra (Fig. 2B) show a new S peak (167.7 eV) in PSG-CFN, and two new Zr signals (185.0 eV and 330.0 eV) in Nano-CFN, verifying the incorporation of PSG and Zr4+ into CFNs51. Moreover, comparison of the high-resolution N 1s spectra of CFN and Nano-CFN (Fig. 2C and Supplementary Fig. S8B) reveals a new signal at 402.3 eV for C−N+, indicating the presence of quaternary ammonium groups in Nano-CFN52. The deconvoluted S 2p1/2 and S 2p3/2 peaks in PSG-CFN shifted to higher binding energies after zirconium treatment (Supplementary Fig. S8C, D), suggesting coordination between Zr4+ and sulfonate groups50. In addition, new signals at 287.8 eV in C 1s and 531.6 eV in O 1 s spectra for Nano-CFN (Fig. 2D, E, and Supplementary Fig. S8E, F) are attributed to C−O−Zr bonds, implying coordination between collagen fibrils and Zr species53. X-ray diffraction (XRD) patterns reveal enhanced peaks at 8° and 19° after zirconium treatment (Fig. 2F), reflecting disruption of hydrogen bonds between collagen fibrils54. These findings confirm the successful introduction of hydrophilic PSG and effective coordination of Zr4+ with surface functional groups.

Fig. 2: Characterization of nanofibrillated CFNs.
figure 2

A FTIR spectra of collagen fibers before and after each treatment step. B Survey XPS spectra of different collagen fiber samples. C N 1s, D C 1s, and E O 1s high-resolution XPS spectra of Nano-CFN. F XRD patterns of CFN, PSG-CFN, and Nano-CFN. G The computed surface charge density of PSG. H Energy profiles for interactions of Zr species and the dipeptide compound. I DSC water melting and vaporization curves for different samples. J Tensile stress–strain curves and K thermal denaturation temperatures of collagen fibers before and after each treatment step.

Full size image

To elucidate the nanofibrillation mechanism, we performed computational simulations to explore the interactions between collagen fibers and water or Zr(SO4)2. Given the composition of collagen-based samples, seryl-glycine and seryl-taurine dipeptides were used as model compounds for CFN and PSG-CFN, respectively. According to the literature55, Zr(OH)2(aq)]2+ was selected as the model of active Zr species in the coordination system. First, we computed the surface charge density of PSG to identify potential coordination sites. Analysis in Fig. 2G reveals higher electronic density around the −SO3− group, facilitating hydrogen bond formation with water and coordination with Zr species that possess lower electronic density (Supplementary Fig. S9). Next, simulations of dipeptide interactions with water and Zr species (Supplementary Fig. S10) indicate that water primarily interacts with the dipeptide through hydrogen bonding assisted by electrostatic forces56. Notably, the Gibbs free energy (ΔG) between the sulfonate group and water is the lowest, suggesting that PSG modification enhances water binding to form a hydration layer (Fig. 2H). In addition, [Zr(OH)2(aq)]2+ can replace bound water molecules on the −NH2, −OH, −COOH, and −SO3− groups via coordinate bonding at ambient temperature, as indicated by negative ΔG values.

We further examined the water-binding capacity using differential scanning calorimetry (DSC). As shown in Fig. 2I, the vaporization enthalpy of water in PSG-CFN is the highest, indicating increased hydrophilicity and water adsorption due to PSG modification. After zirconium treatment, the enthalpy decreased, suggesting partial replacement of collagen-bound water by Zr4+ coordination. Notably, the higher vaporization temperature of water in PSG-CFN compared to CFN is likely due to the confinement by zwitterionic groups57. DSC analysis (Supplementary Fig. S11) further revealed that PSG-CFN retained more water (1.83 g g−1) than CFN (1.28 g g−1) and Nano-CFN (1.61 g g−1), confirming that PSG modification significantly enhances water binding, aiding fiber swelling and fibrillation, while zirconium treatment helps stabilize the nanofibrillated structure by replacing some bound water.

We also assessed the mechanical performance and thermal stability of the collagen samples. Stress–strain curves and softness measurements revealed that nanofibrillation improved the flexibility, toughness, and comfort (Fig. 2J and Supplementary Fig. S12). Moreover, DSC showed a significant increase in denaturation temperature (Fig. 2K), while thermogravimetric analysis (TGA) revealed higher decomposition and maximum decomposition temperatures (Supplementary Fig. S13). These findings demonstrate that nanofibrillation effectively improves the mechanical properties and thermal stability of collagen networks, enhancing their potential for practical use. It is worth mentioning that the nanofibrillation strategy is adaptable to other metal ions including Fe3+ and Cr3+. However, Zr4+ emerges as the optimal choice due to its relatively environmentally friendly nature, efficient and stable cross-linking enhancement performance (Supplementary Fig. S14), as well as its minimal risk of material failure due to ion leaching (Supplementary Fig. S15).

Air purification performance

Due to its nanofibrillated structure and abundant surface groups, Nano-CFN shows great promise for removing air pollutants like VOCs and PMs (Fig. 3A). To demonstrate this, we first tested its ability to remove HCHO, a common indoor VOC found in building materials and furniture. Using a custom-built filtration device (Supplementary Fig. S16), we compared the HCHO removal efficiency of pristine CFN, Nano-CFN, and a commercial high-efficiency particulate arrestance filter (H13). For a fair comparison, three H13 core layers (3 × H13) were stacked to match the 1 mm thickness of Nano-CFN. As shown in Fig. 3B, both CFN-derived filters outperformed the H13 filter. This results from chemical adsorption of HCHO by amino groups on the nanofibril surface (Fig. 3A)27,58, as supported by increased FTIR peak intensity (C=N) in Nano-CFN after filtration (Supplementary Fig. S17). Notably, Nano-CFN’s removal efficiency nearly doubled from 48% to 93% compared to pristine CFN (Fig. 3B and Supplementary Fig. S18), which can be attributed to the greater exposure of amino groups after nanofibrillation.

Fig. 3: Air purification performance of nanofibrillated CFNs.
figure 3

A Schematic illustration of the experimental setup for air purification, and the filtration mechanisms of Nano-CFN. B Comparison of HCHO removal efficiencies among pristine CFN, Nano-CFN and three layers of standard H13 filters (3 × H13). Data are presented as mean values ± SD (n = 3). C Filtration efficiencies of CFN, Nano-CFN, and a three-layer H13 filter for PM of varying sizes (PM10: 10 μm, PM2.5: 2.5 μm, PM0.3: 0.3 μm). Data are presented as mean values ± SD (n = 3). D SEM image of Nano-CFN after PM removal. Experiments were repeated three times with similar results. E Pore size distribution profiles for CFN and Nano-CFN. F Pressure drops versus air face velocity for CFN, Nano-CFN, and 3 × H13 filters. G Comparison of quality factor (QF) and dust holding capacity (DHC) among CFN, Nano-CFN, and H13 filters. Data are presented as mean values ± SD (n = 3); the error bars indicate measurement variations. H Dynamic HCHO removal efficiency curves over a 5-cycle test using a commercial air purifier equipped with a Nano-CFN filter. I Dynamic PM2.5 removal efficiency under identical conditions. J Comparative analysis of Nano-CFN’s performance against reported protein-based air filters and other common filter materials.

Full size image

Next, we evaluated the filters’ efficiency in removing particles of different sizes using a dynamic filtration system (Supplementary Fig. S19). As shown in Fig. 3C, Nano-CFN significantly enhanced removal across all PM sizes (10 μm, 2.5 μm, 0.3 μm) compared to pristine CFN, underscoring the impact of nanofibrillation. The improvement was particularly profound for smaller particles, with over 97% removal efficiency for PM0.3. Significantly, Nano-CFN also outperformed three layers of the H13 filter, a recognized standard for high-efficiency PM purification. Post-filtration SEM and FTIR analysis further confirmed the effective PM capture by Nano-CFN (Fig. 3D and Supplementary Fig. S20)59.

To investigate the improvement in PM removal, we systematically characterized the porous structure of the collagen-based samples. Mercury intrusion porosimetry (MIP) analysis showed that porosity increased from 40.88% to 75.54% after nanofibrillation (Fig. 3E). Further examination using an automated surface area and porosity analyzer revealed Type IV isotherms (as classified by the International Union of Pure and Applied Chemistry) for all collagen-based filters (Supplementary Fig. S21A), indicating a hierarchical structure with meso- and macro-pores13. Barrett-Joyner-Halenda (BJH) analysis (Supplementary Fig. S21B) showed a significant increase in pore volume for pores under 5 nm, with a dominant diameter of ~2.8 nm post-nanofibrillation. These findings confirm the creation of fluffy and porous collagen networks using our nanofibrillation strategy. The enhanced PM removal efficiency is attributed to the increased air channel curvature within the CFN, which improves PM trapping by enhancing fiber dispersion.

Additionally, nanofibrillation reduced the pressure drop of CFN (Fig. 3F and Supplementary Fig. S21C) and significantly increased its quality factor and dust-holding capacity (Fig. 3G and Supplementary Fig. S21D). This enhances the long-term performance of the filter, maintaining >95% removal efficiency (fine air filters, EN 779: 2012) over 50 repeated PM filtration tests (Supplementary Fig. S22). The long-term PM removal performance of Nano-CFN was further evaluated by examining its filtration resistance and efficiency across repeated use cycles under varying pollution conditions. Even under severe air pollution (PM2.5 > 300 μg m−3), both filtration resistance and efficiency remained stable after 4500 reuse cycles (Supplementary Fig. S23), demonstrating exceptional long-term stability. Additionally, we estimated the service lifespan of Nano-CFN based on established standards in the air filtration industry. Under identical operating conditions, Nano-CFN exhibited superior long-term filtration performance compared to commercial H13 filters (three core layers, Supplementary Fig. S24 and Supplementary Table S2), further highlighting its durability and efficiency.

For practical application, we installed Nano-CFN in a commercial air purifier (Supplementary Fig. S25) to evaluate HCHO removal in a confined space (Supplementary Fig. S26). As shown in Fig. 3H, the HCHO concentration decreased from a severe pollution level (~10 ppm) to a safe level (<0.6 ppm) in 200 min, with stable performance over five uses. We also assessed particle removal using cigarette smoke to simulate PM (Supplementary Fig. S27). Figure 3I shows that a single cycle reduced PM concentration from a hazardous level (>3000 μg m−3) to a safe level (<35 μg m−3) in 15 min, maintaining efficiency over five tests. Compared to other protein-based filters and commercial materials, Nano-CFN demonstrated significant advantages in both VOC and PM removal (Fig. 3J and Supplementary Table S3), establishing it as an effective and versatile air purification platform.

Microbial resistance and bioaerosol elimination

The COVID-19 pandemic highlighted the importance of air in transmitting microorganisms like bacteria and viruses, while discarded masks pose a risk of secondary disease transmission. This calls for filters that can both capture and kill infectious microorganisms. However, collagen-based filters, mainly composed of proteins, can foster microbial growth, limiting their long-term use. To address this problem, we introduced a zwitterionic copolymer containing sulfonate and quaternary ammonium groups during the nanofibrillation process. Zr4+ ions were then used to coordinate with the sulfonate groups to increase charge density around the quaternary ammonium groups, aiming to endow the filter with antimicrobial properties (Fig. 4A)60,61.

Fig. 4: Microbial resistance and bioaerosol elimination of nanofibrillated CFNs.
figure 4

A Schematic showing the microbial resistance properties of Nano-CFN. B Zeta potential of different collagen-based samples, with Zr-CFN representing CFN cross-linked by Zr(SO4)2. C Anti-mildew performance of different filters after storage at 25 °C and 98% relative humidity. D Bacterial inactivation efficiency of different filters against S. aureus and E. coli. E S. aureus and F E. coli aerosol capture efficiency of CFN, H13, and Nano-CFN in 50 replicate tests. G Inactivation efficiency of S. aureus and E. coli aerosols over 50 cycles, demonstrating the antibacterial durability of Nano-CFN. Data are presented as mean values ± SD (n = 3); the error bars indicate measurement variations.

Full size image

To test this hypothesis, we systematically evaluated the antibacterial properties and microbial aerosol filtration efficiency of collagen-based filters. As shown in Fig. 4B, the surface charge of collagen fibers increased following zwitterionic copolymer introduction and was further enhanced by Zr4+ coordination with sulfonate groups. We first assessed the anti-mildew properties of CFN, Nano-CFN, and H13 filters using the inhibition zone method (QBT 4199-2011) with Aspergillus Niger as the model strain. Figure 4C shows that both pristine CFN and H13 filters developed mildew within two to three days, whereas Nano-CFN remained mold-free throughout the entire one-month test, demonstrating its remarkable anti-mildew capability. Next, we evaluated the inactivation efficiency of the filters against S. aureus and E. coli after 24 h of contact. While pristine CFN and H13 filters exhibited substantial bacterial growth, Nano-CFN achieved over 99.99% antimicrobial activity against both strains (Fig. 4D). Bactericidal kinetics further validated Nano-CFN’s effectiveness, deactivating S. aureus and E. coli within 30 min (Supplementary Figs. S28, 29).

Subsequently, we evaluated the filters’ ability to capture microbial aerosols using a custom-built filtration device (Supplementary Fig. S30). The device generates microbial aerosols with a median diameter of 3.9 μm, comparable to those expelled during human coughing24,62. As depicted in Fig. 4E, F and Supplementary Fig. S31, Nano-CFN demonstrated exceptional capture performance, retaining over 99% of S. aureus and E. coli aerosols even after 50 repeated uses. In contrast, three-layer H13 filters suffered a significant decline in efficiency under prolonged testing, likely due to charge dissipation from aerosol-induced humidity. Nano-CFN outperformed both pristine CFN and three-layer H13 filters, a result attributed to its highly porous structure, abundant active groups, and hydrophilicity, all of which collectively contribute to its superior filtration efficiency63,64. Furthermore, cyclic antimicrobial testing revealed that Nano-CFN maintained >98% inactivation efficiency against both S. aureus and E. coli after 50 cycles (Fig. 4G), confirming its long-term antimicrobial durability. These results highlight Nano-CFN’s dual functionality − effectively intercepting and inactivating airborne microbes − making it a robust solution for sustained air filtration.

To elucidate the antimicrobial mechanism of Nano-CFN, we employed confocal laser scanning microscopy (CLSM) with live/dead bacterial staining to assess membrane integrity. CLSM analysis (Supplementary Fig. S32) revealed a dramatic fluorescence shift in Nano-CFN-treated specimens: while pristine CFN controls showed exclusive green Syto 9 fluorescence (indicating viable cells with intact membranes), Nano-CFN-treated bacteria exhibited dominant red propidium iodide (PI) fluorescence with concomitant Syto 9 quenching. This distinct pattern provides direct evidence of membrane disruption, as PI penetration requires membrane compromise. Complementary SEM characterization (Supplementary Fig. S33) provided structural confirmation of this membrane damage. Nano-CFN-treated bacteria displayed severe morphological alterations, including membrane wrinkling, perforation, and cytoplasmic leakage \(-\) all absent in control specimens. The correlation between CLSM-observed membrane permeability and SEM-documented physical breaches establishes a consistent mechanism of action. These results collectively demonstrate that Nano-CFN’s antimicrobial action primarily stems from electrostatic disruption of microbial membranes by surface-bound quaternary ammonium groups.

Environmental impact and economic evaluation

We have demonstrated that nanofibrillated collagen filters are highly efficient at removing air pollutants and possess excellent antimicrobial properties. To explore their commercial potential, we evaluated the environmental impact and economic feasibility of this filter material throughout its lifecycle, from production to use and disposal (Fig. 5A).

Fig. 5: Environmental and economic analysis of nanofibrillated CFNs.
figure 5

A Schematic depicting the life cycle of Nano-CFN from production through usage to disposal. B Enzymatic degradation of Nano-CFN following de-zirconium treatment. C Simulated landfill degradation of H13 and Nano-CFN filters. D Environmental impacts of Nano-CFN compared to PP fiber, PET fiber, and glass fiber, with each category normalized to the highest-impact material. E Cost breakdown associated with materials production and environmental impacts. F Comparative analysis of Nano-CFN against three common filter materials regarding biodegradability, renewability, microbial resistance, VOC adsorption, sustainability, and cost, highlighting its overall advantages.

Full size image

Commercial filters made from petrochemical products are difficult to degrade, their disposal, especially when contaminated with pathogens, can lead to severe environmental pollution and the risk of secondary disease transmission22,65. Therefore, a sustainable filter must remain stable during use but degrade easily after disposal. To assess this, we examined the enzymatic degradability of collagen-based filters using acid protease. As shown in Supplementary Fig. S34A, Nano-CFN showed good enzyme resistance, with no visible changes after 15 days of soaking. However, following a simple de-zirconium treatment using oxalic acid and ethylenediaminetetraacetic acid disodium salt solution, Nano-CFN gradually disintegrated and achieved a 99.0% degradation rate within 15 days (Fig. 5B and Supplementary Fig. S34B). Simulated landfill degradation experiments further revealed that Nano-CFN underwent about 50% degradation after 63 days (Supplementary Fig. S34C, D), likely due to microbial breakdown of peptide chains40,66. After de-zirconium treatment, Nano-CFN fully degraded within 63 days of soil burial (Fig. 5C), while the H13 filter remained intact. These results underscore the desirable biodegradability of collagen-based filters.

To comprehensively evaluate the environmental impact, we conducted a life-cycle assessment (LCA) of Nano-CFN following ISO 14040 standards, with three widely used commercial filter materials, PP, polyethylene terephthalate (PET), and glass fiber, as controls. As shown in Fig. 5D and Supplementary Table S4, Nano-CFN had the lowest impact in 16 of 18 environmental indicators, with significantly lower marine and freshwater ecotoxicity, and reduced human carcinogenic and non-carcinogenic toxicity compared to commercial filters. It should be noted that all components used in Nano-CFN fabrication, including the collagen fiber matrix, PSG and Zr4+, were selected based on their established biocompatibility and environmental safety. Collagen degrades naturally into non-toxic peptides and amino acids, while PSG’s potential degradation intermediates (Supplementary Fig. S35A) were evaluated using the Toxicity Estimation Software Tool (T.E.S.T.). Acute toxicity assessments, including Oral Rat LD50 for terrestrial organisms and T. pyriformis IGC50 (48 h) for aquatic systems, confirmed the low toxicity profile of all predicted breakdown products (Supplementary Fig. S35B, C). These results demonstrate minimal risk to both ecological and biological systems. Furthermore, bioconcentration factor (BCF) analysis revealed low accumulation potential (Supplementary Fig. S35D), indicating that the compounds would not persist significantly in soil or aquatic environments. Together, these computational assessments provide strong evidence that PSG degradation intermediates pose negligible long-term environmental hazards. Additionally, given that animal skin collagen is the primary raw material in the leather industry, we also conducted a comparative LCA contribution analysis between Nano-CFN and conventional leather production. Supplementary Fig. S36 reveals that energy consumption is the largest contributor to Nano-CFN’s environmental impact, but its overall impact was much lower than that of leather production due to its simpler manufacturing process (Supplementary Fig. S37).

To assess the economic feasibility of large-scale Nano-CFN production, we analyzed both production and environmental costs across various filter materials. As shown in Fig. 5E and Supplementary Table S5, the estimated minimum production cost of Nano-CFN ($1140 per ton) is lower than that of conventional leather ($2306 per ton) and several commercial filters, including PP fiber ($1437 per ton) and glass fiber ($2925 per ton). The main production cost for Nano-CFN is the raw material (Supplementary Fig. S38), suggesting that expenses could be further reduced by using lower-cost leather waste. In addition, we evaluated the environmental cost, which quantifies the economic decline of adding one extra kilogram of pollutants to the environment67. Nano-CFN’s environmental cost is lower than that of common fiber materials and markedly lower than conventional leather, emphasizing its reduced environmental impact. Collectively, the cost-effectiveness of Nano-CFN, along with its superior PM and VOC removal, microbiocidal properties, biodegradability, and lower environmental footprint, positions it as a sustainable and highly competitive alternative to synthetic filter materials (Fig. 5F).

Discussion

In summary, we have developed a facile nanofibrillation strategy to engineer natural CFNs for the efficient purification of various air pollutants, including PM, bioaerosols, and VOCs. This approach synergistically combines physical processing with chemical functionalization using zwitterionic polymers that feature water-binding groups, followed by zirconium treatment to stabilize the nanofibril structure. The resulting nanofibrillated CFNs demonstrate increased porosity and functional group exposure, leading to boosted removal efficiencies for formaldehyde and PMs, particularly smaller particles.

The performance of Nano-CFN results from the coordinated action of three features (1): the hierarchical and nanofibrillated structure, (2) abundant surface functional groups, and (3) controlled hydrophilicity. Unlike conventional synthetic fibers with inert surfaces, collagen fibers possess a combination of amino, carboxyl, and hydroxyl groups that actively participate in pollutant capture through multiple interaction mechanisms. The nanofibrillation process increases the exposure of these functional groups and generates interconnected tortuous channels that maximize air-pollutant contact time. For volatile organic compounds like formaldehyde, the amino groups facilitate Schiff base formation, enabling chemical immobilization rather than simple physical adsorption. For particulate matter, nanoscale fibers with tuned hydrophilicity enable mechanical interception and hygroscopic capture. Crucially, these factors are interdependent: nanofibrillation without sufficient functional groups would limit chemical interactions, while abundant functional groups without proper nanofibrillation would reduce accessibility. Similarly, hydrophilicity must be carefully balanced − excessive water absorption could collapse the porous structure, while insufficient hydrophilicity would impair PM capture. By integrating all three attributes, Nano-CFN demonstrates superior performance compared to conventional filters like the commercial H13 HEPA filter − the current industry benchmark for medical and semiconductor cleanroom applications − which operate through fewer purification mechanisms.

In addition, the coordination of Zr4+ ions with sulfonate groups in the zwitterionic polymers enhances the surface charge of quaternary ammonium groups, imparting antimicrobial properties to the filter. Nano-CFN exhibits broad-spectrum antimicrobial activity against drug-resistant strains including standard Gram-positive (S. aureus) and Gram-negative (E. coli) models. Live/dead staining (CLSM) revealed dominant red fluorescence (propidium iodide) in Nano-CFN-treated bacteria, confirming membrane disruption, while SEM showed morphological damage, including wrinkling, perforation, and cytoplasmic leakage. These results collectively demonstrate that Nano-CFN’s antimicrobial action primarily stems from electrostatic disruption of microbial membranes by surface-bound quaternary ammonium groups.

Further analyses demonstrate that this nature-derived material offers not only biodegradability but also economic advantages for large-scale production. This study is important as it introduces a sustainable method for creating multifunctional, high-performance, and cost-effective air filters, with potential applications in diverse settings such as hospitals, laboratories, and factories. While the lab-scale manufacturing process is robust, we acknowledge that industrial-scale implementation will require further optimization of the fibrillation process to ensure product uniformity and maximize energy efficiency. In addition, although our stability tests in controlled environments show promising results, we recognize the need for additional evaluation under real-world conditions, including other complex chemical stressors, to fully assess long-term durability. Beyond air purification, these nanofibrillated CFNs hold promises for broader applications, including nanomaterial separation and water purification. Furthermore, this sustainable approach to utilizing animal biomass resources provides a compelling opportunity to transform traditional, environmentally harmful, and energy-intensive industries like leather manufacturing.

Methods

Nanofibrillation of CFNs

CFN was obtained from goat hide, a byproduct of the meat industry, and first modified with PSG under continuous stirring at 45 °C for 12 h. Subsequently, the pH of the reaction solution was adjusted to 2.0, and Zr(SO4)2 was added. After continuous agitation at room temperature for 4 h, the pH was raised to 4.0 and the mixture was heated to 45 °C for another 2 h. The sample was then left to stand overnight and then thoroughly washed with water to remove any residual chemicals. The resulting product treated with both PSG and Zr(SO4)2 is designated as Nano-CFN. For comparison, control samples modified solely with PSG or Zr(SO4)2 were prepared using identical conditions. More experimental details can be found in Supplementary Table S1.

Characterizations

FTIR spectra were recorded using a Fourier transform infrared spectrometer (Nicolet 6700, Thermo Scientific, USA) equipped with an attenuated total reflectance (ATR) accessory, over a spectral range of 800−4000 cm−1. The XPS spectrum was measured using an X-ray photoelectron spectrometer (AXIS Supra, Kratos, UK), with binding energies calibrated using the adventitious carbon signal (C 1s = 284.8 eV). Crystalline and amorphous regions were analyzed by X-ray diffraction (XRD, Smart lab, Rigaku, Japan) using Cu Kα (λ = 1.542 Å) radiation. Electrostatic potential (ESP) and Gibbs free energy calculations were conducted using Gaussian 09 computational chemistry software. The thickness of CFN and Nano-CFN was measured using a vernier caliper (DL91150, Deli, China). Softness was assessed with a leather softness tester (GT-303, Gotech, China). The top surface and cross-sectional morphologies of Nano-CFN, both before and after cross-linking, were examined using a scanning electron microscope (SEM, JSM-7500F, JEOL, Japan). Porosity, pore size distribution, and total pore volume of CFN and Nano-CFN were determined using a mercury intrusion porosimetry analyzer (MIP, Auto Pore IV 9600, Micromeritics, USA). Mesoporous pore size distribution was analyzed using the Barrett-Joyner-Halenda (BJH) method with an automated surface area and porosity analyzer (BET, ASAP 2460, Micromeritics, USA). Water melting and vaporization curves, binding and free water content, and thermal denaturation temperature (Td) of the samples were determined by differential scanning calorimetry (DSC, 204F1, Netzsch, Germany). Thermal decomposition stability was characterized by thermogravimetric analysis (TGA, 209F1, Netzsch, Germany). Tensile strength was measured using a universal testing machine (AI-7000 SN, Gotech, China).

Filtration performance test

The HCHO adsorption efficiency and PM filtration efficiency were assessed using a custom-built experimental setup at ambient temperature. A 37 wt.% HCHO solution served as the formaldehyde gas source, while cigarette smoke was used to generate PM pollutants. The face velocity was maintained at 0.44 m s−1, corresponding to an airflow rate of 32 L min−1. Key performance metrics, including filtration efficiency, dust holding capacity (DHC), and quality factor (QF), were measured and calculated. To evaluate its practical application, the Nano-CFN filter was tested in a closed space by replacing the composite filter in a desktop air purifier (ZKJ-F15A1, Haier) with the Nano-CFN filter (Supplementary Fig. S25, Supplementary Information). For comparison, the core filter layer from a high-efficiency particulate arrestance (HEPA, KXJFA300-A0, GREE, H13) filter was also characterized. To minimize any potential electrostatic effects on filtration performance, all samples were pretreated with isopropyl alcohol vapor. Additional experimental details can be found in the Supplementary Information.

Microbial aerosol capture

A bacterial suspension of Staphylococcus aureus (S. aureus, Gram-positive bacteria) or Escherichia coli (E. coli, Gram-negative bacteria) was prepared and diluted to a concentration of 106 CFU mL−1 using phosphate-buffered saline (PBS). This suspension was then aerosolized using a compressed air nebulizer (Yuwell, 403D), which produced bacterial aerosols with a median particle size of 3.9 μm and an airflow rate of 10 L min−1. The capture efficiency of the Nano-CFN filter for microbial aerosols was evaluated by counting the colony-forming units (CFU) before and after filtration.

Antimicrobial properties

The antibacterial activity of CFN, Nano-CFN, and H13 filters was evaluated against S. aureus and E. coli as test bacteria following the ASTM E 2149-2001 protocol. At specified time intervals, 20 µL aliquots of the bacterial suspension were sampled and inoculated onto NB agar plates, followed by incubation at 37 °C for 18 h. The number of viable colonies was then counted, and the inactivation efficiency for each bacterial species was calculated.

To assess the anti-mildew properties of collagen-based filters and the H13 filter against Aspergillus Niger (ATCC 6275), the inhibition zone method was employed, in accordance with the QBT 4199-2011 standard (Leather-Test method for antimould). Further experimental details are provided in the Supplementary Information.

Biodegradability

The biodegradability of Nano-CFN and commercial HEPA filters was assessed using enzymatic and landfill degradation tests. The degradation progress was monitored by photographing the samples and recording their weight at predetermined time intervals. Detailed protocols are provided in the Supplementary Information.

Life-cycle assessment

A life-cycle assessment (LCA) was conducted in accordance with ISO 14040 standards using Simapro 9.1.1.168 to compare the environmental impacts of producing various filter materials. The functional unit for this comparison was set as the production of 1 kg of filter material. The system boundaries included raw material procurement and material manufacturing processes. Environmental impacts were evaluated using the ReCiPe2016 approach, which covers 18 environmental impact indicators. In addition, the LCA incorporated an analysis of manufacturing and environmental costs, offering insights into the economic aspects of different filter materials.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.