Durable antimicrobial coatings for hospital sink surfaces using hydrophobized TEMPO-oxidised cellulose nanofibre Pickering emulsions Authors: Yoshitaka Ito, Toshiaki Takeuchi, Narumi Komami, Yoshinori Hasegawa, Takehisa Yano Published: 2026 ABSTRACT Effective microbial control in water-associated environments, such as hospital sinks, is critical for preventing the spread of antimicrobial resistance in healthcare settings. Antimicrobial coatings represent a promising means for microbial control; however, their application on water-associated surfaces is challenging. Therefore, durable antimicrobial coatings that retain efficacy under challenging conditions, such as frequent water flow, are needed. To address this challenge, this study focused on water-resistant coatings formed by a silicone oil Pickering emulsion composed of hydrophobized 2,2,6,6-tetramethylpiperidine-1-oxyl (TEMPO)-oxidised cellulose nanofibres (CNF). We evaluated the performance of antimicrobial agent incorporation into the coatings under conditions simulating hospital sink surfaces. Most formulations achieved 99% (≥ 2-log10) reductions in the viable counts of Escherichia coli and Staphylococcus aureus. Among the agents with promising performance, benzisothiazolinone (BIT) localised as particulate inclusions within CNF matrices, and the coating conferred antimicrobial activity against multiple clinically relevant drug-resistant strains. Durability tests, including surfactant exposure and abrasion during cleaning, confirmed sustained antimicrobial performance after treatment. Under conditions simulating frequent water flow, the coating also exhibited superior antimicrobial performance compared with selected commercial coatings. Overall, this environmentally benign and easy-to-apply coating offers a promising strategy for mitigating antimicrobial-resistant bacteria in water-associated environments and demonstrates strong potential for practical infection-control applications.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-026-50866-8. Introduction Antimicrobial resistance (AMR) and AMR-associated microbial infections are major global health threats. These infections not only contribute to increased mortality but also exacerbates clinical management challenges, prolongs infections, and increases the burden on healthcare systems. In 2021, these infections caused an estimated 1.14 million deaths worldwide1, and this number continues to increase. Without effective interventions, annual deaths are projected to reach 10 million by 20502. Accordingly, the appropriate use of antibiotics and environmental hygiene aimed to prevent and control AMR infections is essential3. Such environments include hospitals, where AMR carriers are prevalent, and AMR bacteria are frequently detected4–6. This renders AMR control in hospitals a critical priority. In particular, handwashing sinks and neighbouring patients are recognised as AMR reservoirs because resistant bacteria are often detected within them and can spread to surrounding areas7. Consequently, these sites are the primary targets for intervention. Although various microbial control strategies have been explored, no fundamental solution has been established. Among these strategies, disinfection is one of the most commonly used approaches for sink surfaces. This typically involves chemical methods using antimicrobial agents8–10 or physical methods, such as scrubbing, heat treatment, or sink replacement11,12. However, these methods provide temporary reductions in microbial counts and present major limitations, such as limited efficacy, labour-intensive cleaning procedures, and high costs. In addition, newly introduced bacteria may adhere during cleaning intervals and residual microorganisms can proliferate before the next cleaning13. Antimicrobial coatings are considered a potentially effective solution to these challenges. Such coatings aim to permanently inhibit bacterial adhesion and growth and generally employ one of three strategies: preventing bacterial adhesion, killing adhered bacteria, or a combination of these strategies. Approaches for preventing adhesion include hydrophobic surface formation14–17, hydrophilic surface formation18,19, and control of surface hydration layers20,21. Bactericidal strategies involve the grafting of antimicrobial agents onto surfaces22 or using antimicrobial polymers23. Combined approaches include hydrophobic surfaces incorporating silver nanoparticles24, essential oils integrated into chitosan-based polymers25, and hydrophilic groups introduced into cationic polymers26. However, such coatings present challenges in water-associated environments such as sink surfaces. Specifically, organic solvents or reactive chemicals used for coating may damage the equipment and pose environmental risks. Moreover, chemical vapour deposition, although effective, requires specialised equipment, rendering in situ applications impractical. Additionally, reapplication of pre-installation coatings after delamination or leaching is difficult. Furthermore, all coating materials must be safe for both the patients and healthcare workers. Therefore, antimicrobial coatings in hospital sinks require safe materials, ease of application, water resistance, and application without specialised equipment. In this context, cellulose nanofibres (CNF) have attracted attention for their sustainability and mechanical properties. CNFs are nanomaterials derived from cellulose, the most abundant biomass resource on Earth, and were first reported in the 1980s27. The properties of CNFs have enabled their applications in surfactant-free Pickering emulsions28–30 and the reinforcement of composite materials31–33. Moreover, their chemical characteristics, including abundant hydroxyl groups, have also facilitated research on functional materials, including antimicrobial surfaces. CNF-based antimicrobial coatings using silver nanoparticles34, triclosan35, or grafted functional groups, such as aminoalkyl36, aminosilane37, and quaternary ammonium compounds38, have also been reported. However, these approaches face limitations in water-associated environments because the CNF/antimicrobial composites may be washed away. Therefore, grafting is required for robust attachment; however, this process sometimes involves organic solvents. To address these issues, oil-in-water Pickering emulsions were previously developed39. The emulsions were composed of hydrophobized 2,2,6,6-tetramethylpiperidine-1-oxyl (TEMPO)-oxidised CNFs combined with amino-functional silicones that encapsulated silicone oils. These aqueous dispersions formed organogel films upon coating and drying and exhibited excellent water resistance39. This system satisfies the key requirements for hospital sink coatings, including ease of application, safety, and durability. However, the range of antimicrobial agents that can be incorporated into these emulsions and the level of antimicrobial activity, including that against AMR bacteria, remain unclear. Therefore, this study aimed to evaluate the performance of antimicrobial coatings based on these hydrophobized TEMPO-CNF/silicone oil Pickering emulsions under conditions simulating hospital sink surfaces. Various antimicrobial agents, including surfactants, polymers, metal-based particles, and preservatives, were comprehensively assessed. Moreover, antimicrobial efficacy was tested against AMR bacteria under simulated hospital sink conditions, which included exposure to handwashing surfactants, abrasion during cleaning, and frequent water flow. Results Antimicrobial activity of CNF coatings containing antimicrobial agents CNF coatings are formed by applying and drying a formulation of TEMPO-CNF-stabilized Pickering emulsions, which exhibits submicron droplet sizes (300–600 nm) and highly negative ζ-potentials (approximately − 50 mV) derived from dissociated carboxyl groups39. To identify the antimicrobial agents that confer activity to this CNF coatings, coatings were prepared by incorporating various agents and their performance was evaluated. The overall procedure, from coating formation to antimicrobial testing is illustrated in Fig. 1a. Antibacterial activity was primarily evaluated using colony counting, which was considered more appropriate than fluorescence-based assays such as Live/Dead assays because it directly reflects bacterial proliferation. Fig. 1Antibacterial activity of cellulose nanofibre (CNF) coatings containing antimicrobial agents and characteristics of benzisothiazolinone (BIT)- or chloromethyl isothiazolinone/methyl isothiazolinone (CMI/MI)-containing films. (a) Experimental workflow from coating to antibacterial assay. (b) Minimum inhibitory concentration (MIC, ppm) of each antimicrobial agent against Escherichia coli and Staphylococcus aureus, and antibacterial activity of CNF coatings containing each antimicrobial agent on stainless-steel surface. X-axis shows chemical structure of each antimicrobial agent; Y-axis shows viable counts (log CFU/mL). CNF only: CNF coating without antimicrobial agent, Untreated: uncoated stainless steel. Dashed line represents lower detection limit. n = 4. (c) Water contact angles on glass, stainless steel, and polystyrene (PS) with CNF coatings without antimicrobial agent (CNF), CNF coatings containing BIT (CNF/BIT), CNF coatings containing CMI/MI (CNF/CMI, MI), and uncoated test pieces. n = 6. (d) Representative images used to measure contact angles shown in (c). Coated surfaces shown from top to bottom: glass, stainless steel, and PS. (e) Elution profiles of BIT or CMI/MI from CNF-coated polystyrene plates after water addition; conc.: concentration. n = 3. Antibacterial activity of cellulose nanofibre (CNF) coatings containing antimicrobial agents and characteristics of benzisothiazolinone (BIT)- or chloromethyl isothiazolinone/methyl isothiazolinone (CMI/MI)-containing films. (a) Experimental workflow from coating to antibacterial assay. (b) Minimum inhibitory concentration (MIC, ppm) of each antimicrobial agent against Escherichia coli and Staphylococcus aureus, and antibacterial activity of CNF coatings containing each antimicrobial agent on stainless-steel surface. X-axis shows chemical structure of each antimicrobial agent; Y-axis shows viable counts (log CFU/mL). CNF only: CNF coating without antimicrobial agent, Untreated: uncoated stainless steel. Dashed line represents lower detection limit. n = 4. (c) Water contact angles on glass, stainless steel, and polystyrene (PS) with CNF coatings without antimicrobial agent (CNF), CNF coatings containing BIT (CNF/BIT), CNF coatings containing CMI/MI (CNF/CMI, MI), and uncoated test pieces. n = 6. (d) Representative images used to measure contact angles shown in (c). Coated surfaces shown from top to bottom: glass, stainless steel, and PS. (e) Elution profiles of BIT or CMI/MI from CNF-coated polystyrene plates after water addition; conc.: concentration. n = 3. Before the coating experiments, the intrinsic antimicrobial activity of each agent was assessed by determining the minimum inhibitory concentration (MIC) against Escherichia coli (Gram-negative) and Staphylococcus aureus (Gram-positive). The agents in this study are listed in Table S1. The MIC values varied among the agents; however, all MIC values were ≤ 128 ppm, with the exception of C8 benzalkonium chloride against E. coli and isopropyl methyl phenol (IPMP) against both species (Fig. 1b, upper panel). Based on these results and their practical applicability, the coatings were designed such that complete leaching would yield a maximum antimicrobial concentration of 2,000 ppm. Antimicrobial testing on stainless-steel surfaces revealed that CNF coatings without antimicrobial agents showed bacterial counts comparable with those of the untreated controls, whereas most antimicrobial agent-containing coatings achieved 99% (≥ 2-log10) reductions for both E. coli and S. aureus (Fig. 1b, lower panel). Direct application of antimicrobial agents without CNF incorporation resulted in the partial reductions for certain cationic surfactants, benzisothiazolinone (BIT) and zinc pyrithione (ZPT); in contrast, most CNF-based coatings exhibited greater reductions (Fig. S1a). Notably, despite the incorporation of agents at concentrations well above their MIC values, viable bacteria were detected under certain conditions, specifically on coatings containing didecyl dimethyl ammonium chloride (DDAC), chloromethyl isothiazolinone/methyl isothiazolinone (CMI/MI), and IPMP. Although CMI/MI exhibited lower MICs than BIT, CMI/MI reduced but did not eliminate viable bacteria. By contrast, BIT-containing coatings yielded no detectable viable bacteria even when the BIT concentration was reduced from 0.1 wt% to 0.005 wt% (Fig. S2). Considering the frequent use of hospital sinks, a 3-h waiting period after coating may be impractical in actual settings. Therefore, BIT was incorporated into the CNF coating, the coating was applied to stainless steel for 1 min, rinsed before completely drying, and still achieved no viable bacterial counts (Fig. S3). To clarify these differences and validate BIT, the study focused on CNF coatings incorporating BIT for subsequent investigations. Static water contact angles and antimicrobial agent leaching of CNF coatings on various surfaces To evaluate the influence of substrate type on coating characteristics, the coatings were applied to glass, stainless steel, and polystyrene surfaces, and the water contact angles were measured (Fig. 1c). Untreated surfaces exhibited contact angles of 29.2 ± 2.3° (glass), 61.5 ± 3.0° (stainless steel), and 88.8 ± 0.7° (polystyrene). In contrast, CNF-coated surfaces showed angles between 83.8–99.7°, regardless of the incorporation of antimicrobial agents. Statistical analysis using Dwass–Steel–Critchlow–Fligner test revealed no significant differences among the coated surfaces (p > 0.05). Representative images of the contact angle measurements are shown in Fig. 1d. The BIT- and CMI/MI-containing coatings exhibited comparable contact angles, indicating similar surface wettability. Accordingly, the study hypothesised that differences in antimicrobial performance may reflect variations in leaching behaviour. Consequently, the partition coefficient (logP), a measure of hydrophobicity, showed pronounced differences between BIT (0.4)40 and CMI/MI (0.75)41. To quantify the leached BIT or CMI/MI over time, CNF coatings containing BIT or CMI/MI were applied to polystyrene wells, water was added, and the antimicrobial concentrations were monitored by measuring the absorbance. The initial coating concentration was set to 100 ppm. BIT exhibited release concentrations of 25 ppm after 1 h and 42 ppm after 24 h, whereas CMI/MI showed a markedly lower release (2 ppm at 1 h and 4 ppm at 24 h; Fig. 1e). Distribution of BIT within CNF coatings Microscopic observations were conducted to examine the distribution of BIT within the CNF coatings. As BIT exhibits low water dispersibility, the study hypothesised that it may not completely dissolve in the CNF matrix. Therefore, confocal microscopy was employed to detect undissolved BIT particles. In coatings without BIT, no distinct structures were observed in the XY plane, whereas the XZ plane revealed a layer approximately 40–50 μm thick, corresponding to the coating, with no additional features (Fig. 2a, c). In contrast, coatings containing BIT exhibited numerous particulate structures ranging from several hundred nanometres to approximately 30 μm in the XY plane, which were distributed throughout the coating in the XZ plane (Fig. 2b, d). Fig. 2Distribution of benzisothiazolinone (BIT) in cellulose nanofibre (CNF) coatings. (a–d) Confocal microscope images. (a, c) CNF coatings without BIT; (b, d) CNF coatings with BIT. (a, b) XY-plane images; (c, d) XZ-plane cross-sectional images. (e–p) Scanning electron microscopy (SEM) and SEM-energy dispersive X-ray spectroscopy (EDX) images. (e–j) CNF coatings without BIT; (k–p) CNF coatings with BIT. (e, h, k, n) Backscattered electron images (SEM). (f, i, l, o) Silicon (Si) mapping via EDX. (g, j, m, p) Sulphur (S) mapping via EDX. (e–g, k–m) XY-plane images; (h–j, n–p) XZ-plane cross-sectional images. Yellow arrow indicates presence of particles; blue arrow indicates aluminium foil coated with CNF coatings. Scale bar: 10 μm for C and D; 50 μm for others. Distribution of benzisothiazolinone (BIT) in cellulose nanofibre (CNF) coatings. (a–d) Confocal microscope images. (a, c) CNF coatings without BIT; (b, d) CNF coatings with BIT. (a, b) XY-plane images; (c, d) XZ-plane cross-sectional images. (e–p) Scanning electron microscopy (SEM) and SEM-energy dispersive X-ray spectroscopy (EDX) images. (e–j) CNF coatings without BIT; (k–p) CNF coatings with BIT. (e, h, k, n) Backscattered electron images (SEM). (f, i, l, o) Silicon (Si) mapping via EDX. (g, j, m, p) Sulphur (S) mapping via EDX. (e–g, k–m) XY-plane images; (h–j, n–p) XZ-plane cross-sectional images. Yellow arrow indicates presence of particles; blue arrow indicates aluminium foil coated with CNF coatings. Scale bar: 10 μm for C and D; 50 μm for others. To confirm whether these particles were BIT, scanning electron microscopy (SEM)-energy dispersive X-ray spectroscopy (EDX) analysis was conducted to detect BIT-specific elemental signatures. In coatings without BIT, backscattered electron images showed no distinct features in the XY plane, whereas the XZ plane revealed a coating layer approximately 40–50 μm thick, which was consistent with the laser microscopy observations (Fig. 2e, h). Elemental mapping via EDX demonstrated the uniform detection of silicon (Si) across the entire coating surface, thereby confirming the presence of silicone components (Fig. 2f, i). However, sulphur (S) mapping, used as an elemental marker for BIT, showed no detectable signals in either the XY or XZ planes (Fig. 2g, j). In contrast, coatings containing BIT exhibited particulate structures in the backscattered electron images, which corresponded in size to those observed via confocal microscopy (Fig. 2k, n). EDX elemental mapping confirmed a uniform silicon distribution across the coating (Fig. 2l, o), and sulphur was localised at positions corresponding to the observed particles, thereby indicating the presence of BIT (Fig. 2m, p). Antimicrobial activity of BIT-containing CNF coatings against AMR and susceptible bacteria To assess the applicability of the BIT-containing CNF coatings under conditions simulating hospital sink surfaces, their antimicrobial activity against drug-resistant bacteria was evaluated. If resistant strains exhibit greater tolerance to an antimicrobial agent than susceptible strains, they are more likely to survive following the treatments, resulting in an increased proportion of resistant bacteria in the environment and, consequently, an elevated risk of nosocomial infections. To address this concern, susceptible strains of the same species were tested for comparison. Accordingly, MIC testing of BIT against these strains revealed variability among species; however, no differences were observed between the resistant and susceptible strains (Table S2). Antimicrobial testing using BIT-containing CNF coatings demonstrated that regardless of species or resistance status, all tested strains achieved 99% (≥ 2-log10) reductions in viable counts on BIT-containing coatings, whereas no reduction was observed on CNF coatings without antimicrobial agents (Fig. 3). Fig. 3Antibacterial activity of benzisothiazolinone (BIT)-containing cellulose nanofibre (CNF) coatings against drug-resistant and drug-susceptible strains. X-axis shows BIT-containing CNF coatings (CNF/BIT) and untreated stainless steel (Untreated). Y-axis shows viable counts (log CFU/mL). Susceptible: drug-susceptible strains; Resistant: drug-resistant strains. Dashed line represents lower detection limit. Bacterial strains are listed in Supplementary Table. n = 4. Antibacterial activity of benzisothiazolinone (BIT)-containing cellulose nanofibre (CNF) coatings against drug-resistant and drug-susceptible strains. X-axis shows BIT-containing CNF coatings (CNF/BIT) and untreated stainless steel (Untreated). Y-axis shows viable counts (log CFU/mL). Susceptible: drug-susceptible strains; Resistant: drug-resistant strains. Dashed line represents lower detection limit. Bacterial strains are listed in Supplementary Table. n = 4. Antimicrobial performance under conditions simulating hospital sink surfaces To evaluate the applicability of BIT-containing CNF coatings on simulated hospital sink surfaces, antimicrobial performance was assessed not only under simplified laboratory conditions, but also under conditions characteristic of hospital sink surfaces that may affect coating integrity or antimicrobial retention. Three representative conditions were constructed to simulate frequent sink use: exposure to detergents containing surfactants, abrasion during cleaning, and repeated water contact. Experimental models were designed to reproduce these conditions and to assess antimicrobial activity, as outlined in Fig. 4. Escherichia coli and Staphylococcus aureus were used as representative Gram-negative and Gram-positive bacteria, respectively. Fig. 4Experimental scheme of antimicrobial evaluation under simulated hospital sink surface conditions. Experimental scheme of antimicrobial evaluation under simulated hospital sink surface conditions. For exposure to detergents, CNF coatings without BIT exhibited viable counts of 4–5 log CFU/mL, whereas BIT-containing CNF coatings achieved ≥ 2 log₁₀ reductions in viable cell counts for both bacterial species under all tested conditions, relative to water-treated controls. Notably, treatment with a mouthwash formulation resulted in ≥ 2 log₁₀ reductions in S. aureus counts even in the absence of BIT (Fig. 5a). Fig. 5Antibacterial activities under simulated hospital sink surface conditions. Cellulose nanofibre (CNF) coatings after surfactant (a), abrasion (b), or frequent flow (c) were prepared, and viable bacteria on the surfaces are shown for Escherichia coli and Staphylococcus aureus. Dashed lines in (a) and (b) indicate lower detection limit. n = 4. (a) Surfactant treatment. Top strips indicate products diluted to rinse-use concentrations. X-axis indicates benzisothiazolinone (BIT)-containing CNF coatings (CNF/BIT) and untreated stainless steel (Untreated). (b) Abrasion treatment. Top strips indicate treatment methods applied after formation of CNF coatings; wiping with wet sheet (Wet sheet) or sponges (Sponge), with sponges soaked in dish detergent (Dish detergent), and control without abrasion (No abrasion). (c) X-axis indicates water contact treatment. No washing: no treatment after coating. Washing: after coating, samples were rinsed three times by gentle agitation in ion-exchanged water as described in Fig. 1A. Frequent flow cycles: one cycle consists of supplying Tochigi Prefecture tap water for 10 s followed by a 5-s rest; the numbers indicate the number of cycles. Untreated: untreated stainless steel. The colour scales indicate the mean viable counts (log CFU/mL). Y-axis indicates test samples. RA-200/BIT: product with CNF was mixed with BIT. Product A–F: commercially available products that promote antimicrobial coatings. Untreated: untreated stainless steel. Antibacterial activities under simulated hospital sink surface conditions. Cellulose nanofibre (CNF) coatings after surfactant (a), abrasion (b), or frequent flow (c) were prepared, and viable bacteria on the surfaces are shown for Escherichia coli and Staphylococcus aureus. Dashed lines in (a) and (b) indicate lower detection limit. n = 4. (a) Surfactant treatment. Top strips indicate products diluted to rinse-use concentrations. X-axis indicates benzisothiazolinone (BIT)-containing CNF coatings (CNF/BIT) and untreated stainless steel (Untreated). (b) Abrasion treatment. Top strips indicate treatment methods applied after formation of CNF coatings; wiping with wet sheet (Wet sheet) or sponges (Sponge), with sponges soaked in dish detergent (Dish detergent), and control without abrasion (No abrasion). (c) X-axis indicates water contact treatment. No washing: no treatment after coating. Washing: after coating, samples were rinsed three times by gentle agitation in ion-exchanged water as described in Fig. 1A. Frequent flow cycles: one cycle consists of supplying Tochigi Prefecture tap water for 10 s followed by a 5-s rest; the numbers indicate the number of cycles. Untreated: untreated stainless steel. The colour scales indicate the mean viable counts (log CFU/mL). Y-axis indicates test samples. RA-200/BIT: product with CNF was mixed with BIT. Product A–F: commercially available products that promote antimicrobial coatings. Untreated: untreated stainless steel. Following abrasion treatments, the BIT-containing coatings consistently achieved ≥ 2 log₁₀ reductions in viable counts for both species. In contrast, CNF coatings without BIT remained at 4–5 log CFU/mL under most abrasion conditions, except for wet-sheet abrasion, which resulted in ≥ 2 log₁₀ reductions in S. aureus counts (Fig. 5b). When abrasion was performed using a strongly alkaline detergent effective at removing silicone oil, antimicrobial activity was completely lost (Fig. S4). Further evaluation was conducted under conditions that simulated frequent water use. BIT-containing CNF coatings and RA-200 (CNF-based commercial release agent) maintained ≥ 2 log₁₀ reductions in viable counts even after 50 cycles, whereas Products A–F exhibited no consistent trend (Fig. 5c). The BIT applied directly without a coating lost its antimicrobial activity after 10 cycles (Fig. S1b). BIT leaching tests revealed that BIT was no longer detectable after 20 cycles when applied alone, whereas CNF and RA-200 coatings retained detectable levels of BIT even after 100 cycles, demonstrating leaching profiles comparable to those of untreated control coatings (Fig. S5). Discussion In this study, CNF-based coatings that incorporated various antimicrobial agents were developed, and their performance was evaluated under conditions simulating hospital sink surfaces to inhibit bacterial growth and prevent microbial dissemination. These coatings were easily applied and dried, and most demonstrated strong antimicrobial activity against multiple bacterial species, including AMR strains. To our knowledge, few, if any, prior studies have combined broad-spectrum efficacy with ease of application. Moreover, evaluations under conditions relevant to real-world environments (e.g., exposure to surfactants) have rarely been reported. Collectively, these findings indicate that CNF-based antimicrobial coatings hold considerable promise for microbial control in water-associated environments. The range of antimicrobial agents that can be incorporated into the CNF coatings is an important consideration. Although practical concentrations vary among agents, most of the tested antimicrobial agents reduced viable counts below the detection limits when incorporated into CNF coatings. This indicated that diverse chemical structures could be successfully integrated into the coating. The MIC values for E. coli and S. aureus obtained in this study were generally consistent with those in previous reports42–44 and were lower than the concentrations used in the coatings. This suggests that even lower incorporation levels may provide sufficient antimicrobial activity. The mechanism by which antimicrobial agents in CNF coatings reduce viable counts likely involves leaching into the aqueous phase where the bacteria reside. Therefore, factors such as solubility within the CNF matrix, leaching behaviour, and adsorption onto CNF or bacterial surfaces are critical. These physicochemical properties may have influenced the activity exhibited by DDAC and CMI/MI, which exhibited low MIC values yet allowed bacterial survival on the coated surfaces. Specifically, DDAC formulations containing different counterions (chloride vs. ethyl sulphate) exhibited identical MIC values against E. coli and S. aureus; however, their antimicrobial performances differed. Counterion variations in cationic surfactants can influence the critical micelle concentration and aggregation number45. This implies that DDAC solubility in the coating formulation or its adsorption to bacterial surfaces after leaching may differ between salts. In contrast, CMI/MI was not expected to exhibit such micellar behaviour; however, bacterial survival occurred despite incorporation at concentrations exceeding the MIC. To investigate this discrepancy, CMI/MI was compared with BIT, which is an isothiazolinone-class agent that exhibits a low resistance risk46 and shares an isothiazolinone backbone with CMI/MI. The agents were compared by analysing the coating surface properties and leaching behaviour. Contact angle measurements revealed no pronounced differences among the coatings, regardless of the substrate type or antimicrobial agent, thereby suggesting that surface wettability was unaffected. Leaching studies further showed that BIT reached concentrations above its MIC within 1 h of contact with water, thus supporting its antimicrobial efficacy. Although CMI/MI also leached at concentrations above its MIC, these levels were only two- to four-fold higher and substantially lower than those of BIT. This explains the decrease in bacterial counts without reaching the detection limit. The reported partition coefficients (logP) for water/octanol were 0.4 for BIT and 0.75 for CMI/MI40,41. This indicates that CMI/MI may exhibit lower water-phase mobility and remain within the CNF matrix, thereby reducing its effective antimicrobial action. CNF coatings without antimicrobial agents did not reduce the bacterial counts under any of the tested conditions, thus confirming that CNF and silicone oil lacked intrinsic antimicrobial activity. However, hydrophobic surfaces coated with silicone oil may inhibit bacterial adhesion, as suggested in studies where reduced S. aureus attachment was observed on silicone-coated plates after 72 h47 and decreased E. coli adhesion was observed on slippery liquid-infused surfaces formed by condensation of silane compounds with hydrolysed tetraethyl orthosilicate48. However, these effects vary depending on the surface preparation and experimental conditions. In this study, the coatings were cultured in a static system in which bacterial motility was restricted, thereby possibly limiting adhesion inhibition. By contrast, real-world water-associated environments involve frequent flushing and high shear forces. In particular, biofilm formation is suppressed under high shear stress on silicone-coated surfaces49, which suggests that CNF coatings containing antimicrobial agents may provide dual functionality, adhesion inhibition, and antimicrobial activity, resulting in enhanced bacterial reduction. The strong antimicrobial performance of the BIT-containing CNF coatings may be partly explained by the physical state of BIT within the coating. The particles detected by laser microscopy and confirmed by sulphur mapping in the SEM-EDX analysis were likely BIT because no other sulphur-containing components were present. If BIT exists as discrete particles distributed throughout the coating, these particles may gradually dissolve and diffuse into the aqueous phase, thereby exerting its antimicrobial effects. This mechanism suggests that the antimicrobial activity can persist until the particles are completely dissolved, thus potentially extending the functional lifetime of the coating. Although CNF coatings can incorporate various antimicrobial agents, drug-resistant bacteria frequently detected in clinical settings may also exhibit resistance to these agents. This raises concerns regarding the dominance of resistant strains. To evaluate this risk, MIC tests of BIT were conducted against drug-resistant and drug-susceptible strains and antimicrobial assays were conducted using BIT-containing CNF coatings. The test organisms included species classified within the ESKAPE group (Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, and Enterobacter species), which represents major drug-resistant pathogens50. Overall, no substantial differences were observed between the two groups, which suggests that the use of BIT-containing CNF coatings is unlikely to selectively favour the survival of drug-resistant strains under the conditions tested. However, because drug-resistant bacteria can acquire resistance through diverse mechanisms, long-term assessments under actual hospital conditions are necessary for comprehensive risk evaluation. Notably, although Pseudomonas aeruginosa exhibited an MIC of ≥ 128 ppm for BIT, antimicrobial testing still achieved ≥ 2 log₁₀ reductions. This indicated that the concentration of BIT that leached from the coating likely exceeded the highest MIC observed among the tested species. Finally, the impact of conditions representative of real-world sink environments, including surfactant exposure, abrasion, and frequent water flow, were assessed. Based on the available literature, such evaluations have not been reported for other antimicrobial coatings. However, they are critical for validating the practical applicability of these coatings. Suitably, the BIT-containing CNF coatings maintained their antimicrobial activity after exposure to surfactant concentrations simulating wastewater, which suggesting that their efficacy may persist under harsh conditions. However, further testing is warranted. The observation of the mouthwash alone reducing S. aureus counts likely reflects the adsorption of cationic surfactants present in the formulation onto the surface. The reduction observed on the untreated surfaces following wet sheet abrasion may be attributed to the same mechanism. This interpretation is supported by previous findings that demonstrate that gram-positive bacteria are generally more susceptible than Gram-negative species to cationic surfactants51. To clarify the impact of frequent water flow, the BIT-containing CNF coatings were subjected to ≤ 50 consecutive rinsing cycles. Unlike other commercial products, antimicrobial activity was retained, and quantitative analysis of BIT leaching revealed minimal changes in its concentration (Fig. S5). This suggests that the BIT content within the CNF matrix remains largely unaffected under these conditions. More frequent rinsing could potentially reduce BIT levels; however, previous studies exploring hospital sink usage reported 33 uses within 17 h52. Hence, BIT-containing CNF and RA-200 coatings may offer practical durability compared with that of commercial alternatives. In contrast, the inconsistent antimicrobial performance observed among the marketed products may reflect uneven coating applications or poor water resistance, which consequently leads to delamination and warrants further experimental investigation through detailed surface analyses. The BIT-containing CNF and RA-200 coatings exhibited antimicrobial activity even after incomplete drying (Fig. S4), which suggests the possibility of reducing the required drying time. Therefore, further studies are needed to examine the influence of post-application resting time on coating thickness, water resistance, and antimicrobial performance to optimise protocols for high-frequency water-associated environments. This study did not confirm the efficacy of BIT-containing CNF coatings under actual field conditions; therefore, validation through real-world trials is essential. Concerns regarding the emergence of resistance associated with prolonged use of a single antimicrobial agent must also be addressed. Meanwhile, the coatings developed in this study combine wear resistance with controlled removability, allowing routine surface cleaning to be coupled with rotation of incorporated antimicrobial agents. Evaluation of other antimicrobial agents under water-associated conditions, as performed for BIT, may further help mitigate the risk of persistent or emerging resistant strains. Importantly, this study demonstrates that environmentally benign materials can be used to form water-resistant antimicrobial coatings through simple application and drying. To our knowledge, no previous studies have comprehensively evaluated antimicrobial performance against diverse drug-resistant bacteria under conditions simulating water-associated environments. Accordingly, these findings provide valuable insights with strong potential for practical applications in infection control. Future studies assessing these coatings across various water-associated settings and timescales will help clarify their role in reducing patient exposure to drug-resistant pathogens. Materials and methods Materials Wood cellulose fibres were isolated from bleached softwood kraft pulp (West Fraser, Hinton, Canada). TEMPO, sodium bromide, sodium hypochlorite, 1 M hydrochloric acid, sodium chloride, and sodium borohydride without further purification were purchased from FUJIFILM Wako Pure Chemical Corporation (Osaka, Japan). Silicone oil (KF-96-100cs) with an average molecular weight of 1,000, a density of 0.935 g/cm³, and viscosity of 100 cSt was obtained from Shin-Etsu Chemical Co., Ltd. (Tokyo, Japan). Liquid amino-functional silicone (SS-3551, Dow Corning Toray Co., Ltd., Tokyo, Japan) with a linear structure, primary amine content of 0.3 mmol/g, secondary amine content of 0.3 mmol/g, average molecular weight of 30,000, and density of 0.97 g/cm³ was used as a chemical-modified silicone. Stainless steel test pieces (JIS: SUS304, ISO: X5CrNi18-10) with dimensions of 0.5 × 30 × 30 mm were obtained from the Engineering Test Service (Osaka, Japan). Glass slides (76 × 26 × 0.9–1.2 mm, frosted area: 15 × 26 mm) were purchased from Matsunami Glass Ind., Ltd (S2215, Osaka, Japan). Polystyrene multi-well plates (24-well format) were obtained from AGC Inc. (1620-024, Tokyo, Japan). Aluminium foil with a standard thickness of 12 μm for generally household use was used. The following antimicrobial agents were obtained from Kao Corporation (Tokyo, Japan): benzalkonium chloride (C8, C8–C18 mixture, C12–C16 mixture), dodecyltrimethylammonium chloride, tetradecyltrimethylammonium ethyl sulphate, didecyldimethylammonium chloride, and didecyldimethylammonium ethyl sulphate. ProClin™ 200 (CMI/MI = 3:1, 1.5 wt%) was purchased from Merck (Darmstadt, Germany), whereas Proxel™ BDN (BIT, 33 wt%) and Proxel™ IB (PHMB, 20 wt%) were obtained from Arxada (Tokyo, Japan). Silver oxide, isopropyl methyl phenol, and ZPT were obtained from FUJIFILM Wako Pure Chemical, whereas Tinosan™ HP 100 (dichlosan, 30 wt%) was purchased from BASF (Tokyo, Japan). Preparation of CNF coating dispersion An aqueous dispersion of TEMPO-oxidised CNF (TEMPO-CNF-COOH) was prepared following a previously reported method39. Bleached pulp fibres (10 g, dry weight basis) and water (990 g) were placed in a 2-L polypropylene basket, and the resulting pulp slurry was stirred at 25 °C and 100 rpm for 30 min. TEMPO (0.13 g), sodium bromide (1.3 g), and a 10.5 wt% sodium hypochlorite solution (35.5 g) were sequentially added to the slurry. The pH was maintained at 10.5 using 0.5 M sodium hydroxide and a pH-stat system (AUT-701, DKK-TOA Corporation, Tokyo, Japan). After stirring at 25 °C for 120 min, the water-insoluble oxidised fibres were repeatedly washed with water via filtration to obtain the purified TEMPO-oxidised cellulose (TOC) fibres. The TOC fibres were suspended in water (1,000 g) to achieve 2 wt%, and the pH was adjusted to 8 using 0.5 M sodium hydroxide. Subsequently, the TOC slurry was homogenised thrice at 150 MPa using a high-pressure homogeniser (NanoVader L-ES, Yoshida Kikai Co., Ltd., Tokyo, Japan), thereby yielding a transparent TEMPO-CNF aqueous dispersion with 1 wt% solids. The TEMPO-CNF dispersion (1,800 g) was diluted with water (2,200 g) in a 5-L stainless steel beaker to prepare a 0.45 wt% dispersion. Subsequently, 0.1 M NaOH (12 mL) and sodium borohydride (1.2 g) were added, and the mixture was stirred at 25 °C for 3 h to completely reduce the residual C6-aldehyde groups to C6-hydroxyl groups53. Thereafter, 1 M hydrochloric acid (90 g) was added, and the mixture was stirred for 1 h to form a TEMPO-CNF-COOH gel. The gel flakes were washed five times with water via filtration to remove excess acid. The washed gel was resuspended in water and homogenized three times at 150 MPa, thus yielding a transparent TEMPO-CNF-COOH aqueous dispersion with 0.95 wt% solids54. The carboxyl content of the resulting TEMPO-CNF-COOH, measured via conductivity titration, was 1.5 mmol/g55. Subsequently, a silicone oil-in-water emulsion stabilised with TEMPO-CNF-COOH and amino-functionalised silicone was prepared. Briefly, water (11.16 g), TEMPO-CNF-COOH dispersion (78.95 g; containing 0.75 g CNF), silicone oil (7.50 g), and amino-functional silicone (2.39 g; amine-to-carboxyl molar ratio = 1.25) were combined in a 300-mL beaker. Thereafter, the mixture was homogenized ten times at 150 MPa to obtain a stable emulsion. The final emulsion contained 0.75 wt% CNF (total mass: 100.00 g, density: 1.00 g/cm3), with a CNF-to-silicone oil mass ratio of 1:10. Bacterial strains and suspension preparation Eighteen strains representing nine bacterial species were used in this study (Table S1). Among these, nine drug-susceptible strains were obtained from the Biological Resource Center, NITE (NBRC, Chiba, Japan), eight drug-resistant strains from the American Type Culture Collection (ATCC, Manassas, VA, USA), and one strain from the Gifu University Microbial Genetic Resource Center (Gifu, Japan). The frozen stock was streaked onto Soybean-Casein Digest (SCD) agar plates (Shiotani MS, Hyogo, Japan) and incubated at 37 °C for 20–24 h. The resulting colonies were collected and suspended in 5 mL of SCD broth, and the suspension was incubated at 37 °C with shaking at 200 rpm for 20–24 h. MIC assay The MIC assay was performed according to the standard method established by the Japanese Society of Chemotherapy56, with minor modifications. Briefly, the cultured bacterial suspension was collected and resuspended in SCD medium and serially diluted to and adjusted optical density OD₆₀₀ to 0.008. Thereafter, polystyrene round-bottom 96-well plates (3788, Corning Inc., Corning, NY, USA) were prepared by dispensing 50 µL of SCD medium into the wells. Test solution was diluted with SCD medium to 256 ppm, and 100 µL of the solution was added and serially diluted by two-fold to 0.25 ppm. SCD medium without the test solution was used as controls. Subsequently, 50 µL of OD-adjusted bacterial suspension was added to each well, and the plate was incubated at 37 °C for 20–24 h. After incubation, the MIC was determined using precipitants or turbid suspension; the assays were performed in triplicate (n = 3). Antimicrobial assay An antibacterial assay was performed according to JIS Z 2801 (ISO 22196) with minor modifications. An overview of this procedure is shown in Fig. 1a. Stainless steel test pieces were coated with 55 µL of CNF coating solution containing each antibacterial agent at final concentration of 0.1 wt%. The coated pieces were then air-dried at 25 °C for 3 h. After drying, each piece was immersed in 200 mL of ion-exchanged water, washed three times with gentle shaking, and removed. This washing procedure was repeated three times, followed by air-drying at 25 °C. Untreated stainless-steel pieces served as controls. Thereafter, bacterial suspension adjusted to OD₆₀₀ = 0.05 in SCD broth was dropped (55 µL) onto the coated surface, and another treated surface was placed on top to sandwich the suspension. Samples were placed in a humidified container and incubated at 37 °C for 20–24 h without agitation. After incubation, the pieces were separated and the contact surfaces were rinsed three times with 2 mL of ion-exchanged water. The surface was then swabbed using a Fuki-Fuki Check II (Eiken Chemical, Tokyo, Japan) by drawing ten circles, and the collected cells were suspended in 10 mL of phosphate-buffered saline. The suspensions were serially diluted (10-, 100-, and 1,000-fold) with sterile saline, and 50 µL of each dilution was spread and incubated on SCD agar at 37 °C for 20–24 h, followed by colony counting. Static contact angle measurement The contact angles were measured using an automated contact angle meter (DM-300, Kyowa Interface Science Co., Ltd., Saitama, Japan). For CNF-coated lubricating films on stainless steel test pieces, a water droplet (1 µL) was placed on the surface, and the contact angle was recorded 5 s after deposition. Measurements were performed in six replicates for each sample. Untreated stainless-steel pieces served as controls. Quantitative evaluation of antimicrobial agent elution from coating Polystyrene 24-well plates were coated by dropping 100 µL of coating solution containing antibacterial agent at 0.1 wt% and air-dried at 25 °C. The CNF coating without antimicrobial agents served as controls. Each well was washed with 2 mL of ion-exchanged water and air-dried. Thereafter, 1 mL of ion-exchanged water was added to each well, and the plate was shaken using a shaker. At 1, 8, and 24 h after water addition, 500 µL of samples were collected. Three samples were collected at each time point, with only one sample per well. The collected solutions were placed in quartz cuvettes and analysed using a spectrophotometer (U-5100, HITACHI, Japan) at 319 nm for BIT57 and 273 nm for CMI/MI58. Confocal microscopy analysis The internal structures of the coatings on the glass substrates were observed using a confocal microscope (OPTELICS HYBRID+, Lasertec Corporation, Kanagawa, Japan) equipped with a xenon white light source. Imaging was performed in water using a water-immersion objective lens (Plan Apo NIR, NA 1.0, 60×; Nikon Corporation, Tokyo, Japan). The cross-sectional samples of the coatings were prepared using a cooling-type cross-sectional polisher (IB-19520-CPP; JEOL, Japan). Films formed on aluminium foil were cooled to − 100 °C and milled at 4 kV for 4 h to obtain flat cross-sections. Cross-sections were observed using the same microscope equipped with a xenon light source and a CFI Plan EPI objective lens (NA 0.95, 150×, Nikon). SEM-EDX analysis The surface and cross-sections of the coatings on the aluminium foil were observed using a field-emission scanning electron microscope (S-4800, Hitachi High-Tech Corporation, Tokyo, Japan). The samples were coated with a thin osmium conductive layer and observed at 15 kV. Elemental analysis was performed using an energy-dispersive X-ray spectrometer (Ultim Max 65; Oxford Instruments, Oxfordshire, UK). Elemental maps were acquired at 500× (surface) and 1200× (cross-section). Quantification was conducted in the standardless mode with ZAF correction to obtain the concentration distributions of silicon (Si) and sulphur (S). Antimicrobial performance under simulated hospital sink surface conditions An overview of this procedure is shown in Fig. 4. The following commercially available products manufactured by Kao Corporation were used for surfactant exposure: dishwashing detergent (CuCute™), hand soap (Biore™ Foaming Hand Soap), mouthwash (PureOra™), and toothpaste (Clear Clean™ Regular). The products were diluted to concentrations representative of those found in wastewater. Alternatively, each product was diluted with ion-exchanged water to obtain surfactant concentrations of 50, 250, 500, and 470 ppm. Each diluted solution (200 mL) was prepared in a container, and stainless-steel test pieces coated and dried with the CNF coating solution were immersed and washed with gentle shaking three times. Untreated stainless-steel pieces served as controls and were exposed to the same diluted solutions using an identical procedure. After exposure, all test pieces were rinsed with ion-exchanged water and used for subsequent antibacterial tests. For the abrasion treatment, stainless-steel test pieces coated and dried with CNF coating solution were wiped using either a medical wet sheet (SafeKeep™, Kao Professional Services, Tokyo, Japan), a urethane sponge impregnated with ion-exchanged water, or a urethane sponge impregnated with dishwashing detergent. Each surface was wiped by drawing 20 circles. After the first 10 wipes, the sheet or sponge was rotated to ensure that a fresh surface contacted the test piece, and an additional 10 wipes were applied. Untreated stainless-steel pieces served as controls and were subjected to the same abrasion procedures. After treatment, all pieces were then rinsed with ion-exchanged water and used for subsequent antibacterial tests. For the frequent water-flow treatments, a flow path was constructed, as shown in Fig. 4. Tap water from Tochigi (average hardness: ~50 mg/L) was connected to an automatic irrigation timer linked to a four-way joint and garden hoses. The timer was set to cycles of 10-s flow and 5-s rest (1, 3, 10, or 50 cycles). Stainless-steel test pieces coated and dried with the CNF coating solution or six commercial antimicrobial coating products (products A–F), were mounted on a metal plate with waterproof double-sided tape. Tap water was directed to flow over the plate and across the test pieces. After completing the programmed cycles, the pieces were air-dried and used for subsequent antibacterial tests. Untreated stainless-steel pieces served as controls. Statistical analysis and figure preparation Statistical analysis of contact angle differences was performed using the non-parametric Steel–Dwass–Critchlow–Fligner multiple comparison test with tie correction, as the Shapiro–Wilk test showed that part of the dataset did not satisfy the assumption of normality. All statistical analyses were conducted using R software (version 4. 4. 3)59. Figures were generated using the ggplot2 package (version 3. 5. 1)60 in R. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1