Air filters, designed to capture airborne particulate matter (PM), are present in every aspect of our life from air cleaners for residential housing to industrial filters at cleanrooms10-13. The adhesion force between PM and conventional filter media is primarily governed by van der Waals interactions, and hence are typically on the order of nanonewtons. This low adhesion can be insufficient to capture and retain PMs, causing poor filtration. Issues arising from weak adhesion have remained unsolved for decades and continue to be a limitation in filtration performance.
A nasal cavity, which includes nasal hairs, is an effective filtration organism through strong adhesion (Fig. 1a). The key element facilitating PM adhesion is a thin mucus layer that coats the nasal hairs. When PM encounters the mucus layer, a meniscus of mucus is formed on the PM, generating strong adhesion via capillarity. Filtration capability of mucus-coated nasal hair is demonstrated by exposing pollen-laden air (Extended Data Fig. 1); an optical microscope image of hair confirmed that pollen is adhered onto the surface by liquid mucus (Fig. 1b). Inspired by this filtration mechanism, we introduce a bioinspired filter having a thin liquid layer. This liquid is an oil, which increases adhesion for enhanced particulate removal efficiency from air. We term this the Particle Removing Oil-coated Filter (PRO Filter). To realize this concept, it is necessary to create a thin, uniform coating of a non-volatile liquid on the surface of a filter. It was previously demonstrated that tri-methylsiloxy-terminated polydimethylsiloxane (PDMS, e.g. silicone oil), generates a long-term stable thin liquid layer on a PDMS brush grafted surface14,15. In fact, this method has been employed to fabricate lubricant infused surfaces having thin lubricant layers. Because of the mobile lubricant, liquid drops and/or solid particles easily slide off the surface with low friction16-21, finding applications in areas like anti-icing22,23, anti-biofouling24-27, and liquid-drop manipulation28,29. In these cases, low friction implies low adhesion in the lateral direction. However, in the vertical direction (i.e. normal) to the surface, lubricant can induce strong adhesion to drops and/or particles, mainly facilitated by capillary forces. This unique characteristic is exploited in the present study for effective capture of PM.
General approach and comparative performance of PRO filter
PDMS brushes with a thickness of ~10 nm are covalently grafted on the surface of different filter media by polymerization of dimethyldimethoxysilane from surface hydroxyl groups30,31 (Extended Data Fig. 2). Silicone oil is then applied onto the PDMS brush grafted filter by spraying a silicone oil solution in hexane. After the hexane evaporates, a thin, uniform silicone oil layer is formed (Fig. 1c). The scanning electron microscope (SEM) image depicted in Fig. 1d reveals a homogeneous, thin oil layer on polyester fibers without issues from Plateau-Rayleigh instabilities32,33. We confirm that this thin liquid layer does not significantly influence the pore structure or the air permeability. The air permeability of the oil-coated filter (326 ± 3.9 cm3/s/cm2) is nearly identical to the uncoated bare filter (323 ± 6.1 cm3/s/cm2).
Despite the similarities in air permeability, the PRO filter demonstrates superior performance in capturing PM. As a comparative analysis, ~50 µm pollens were blown into a non-woven bare filter with a pore size range of 100~500 µm (Supplementary Video 1). Most of the pollen particles are not filtered and pass through the conventional filter (Fig. 1e). An SEM image of the filter after filtration confirms that pollens were rarely captured by the filter fibers. In contrast, most pollen is filtered by the PRO filter (Fig. 1f). The silicone oil layer acts similarly to the mucus on nasal hair, resulting in strong adhesion between the pollen and the fibers. This clear contrast exhibits the positive effect of the thin liquid layer on enhancing PM capture.
Filtration efficiency
The filtration efficiency (FE) is quantitatively investigated by using a custom-built air filter testing chamber (Supplementary Fig. S3). The chamber is designed to quantify two parameters: i) PM concentrations and ii) the pressure drop across the filter. The FEs for filters of varying materials and media types with different pore sizes are shown in Fig. 2a. These efficiencies are categorized into three distinct size groups that are relevant to inhalable PM causing adverse health effects: PM1.0 (0.3~1.0 μm), PM2.5 (1.0~2.5 μm), and PM10 (2.5~10 μm). The orange bars represent the FEs of bare filters without any treatment, while the green bars represent the FEs of PRO filters. The application of a thin oil layer leads to FEs enhancements of 10~30% compared to their bare filter counterparts. Remarkably, this improvement is achieved with almost no change in air permeability (Extended Data Table 1).
The applied liquid layers are submicron in thickness, resulting in negligible changes to the fibers diameters (~5-100 µm). Consequently, both air permeability and the pressure drop are nearly unchanged from before to after the application of the liquid layer (smaller than 3% difference). It should be noted that increasing the liquid layer thickness could clog the pores. However, our findings indicate that such a thick layer is not required for the PRO filter because FE enhancements are already maximized at liquid layer thicknesses of just a few hundred nanometers (Extended Data Fig. 3). The optimum quantity of liquid required to achieve maximum FE enhancement, while minimally affecting pressure drop, varies according to the specific filter medium.
Particle adhesion
As evidenced by the FE results, a thin oil layer on the filter surface yields remarkable advances in PM capture. To better understand the effect of a thin oil layer on filter efficiency, we measured single particle adhesion on thin oil layers using colloidal probe atomic force microscopy (AFM). This is a useful tool for measuring adhesion in the normal direction by approaching and retracting a particle from planar substrates34 (Fig. 2b). Given that the diameter of typical airborne PMs ranges from 0.3 to 20 µm, we selected microparticles with diameters between 7 and 15 μm as probes and glued them to tipless cantilevers. Force-distance curves generated by the colloidal microprobe involve the adhesion force (FA): the most negative force recorded during retraction (Extended Data Fig. 4b). To detach a particle, this adhesion force has to be overcome. As a control, we first measured particle adhesion on flat bare glass; the adhesion force of a 7 μm silica particle was 15 ± 0.9 nN (Fig. 2c). Hard particles predominantly interact with solid substrates by van der Waals force, with a range of 2~50 nN for a 7 μm particle, see Supplementary Information Section 3. For a model system of the PRO filter, we use a glass substrate uniformly coated with silicone oil (viscosity: 100 cSt) having a thickness of 203 ± 2 nm, referred to here as PRO glass. On PRO glass, the adhesion force increased to 400 ± 8 nN, which is ~25 times greater than that on the bare glass (Fig. 2d).
This enhanced adhesion is attributed to the formation of a liquid meniscus35,36. Laser scanning confocal microscopy shows that a silicone oil meniscus forms between the silica particle and the PRO glass (Fig. 2e). When a solid particle contacts a liquid layer, the liquid wets the particle surface, promptly creating meniscus that generates a capillary force. The capillary force induced by the meniscus on a sphere is governed by the equation: where γ is the surface tension, d is the particle diameter, Θ is the contact angle, and β is the filling angle that describes the location of the three-phase contact line on the particle (details in Fig. 2d). Capillary force between a particle and a thin liquid layer have been considered in previous studies and is generally 1~2 orders of magnitude greater than van der Waals force; such findings agree with our measured adhesion forces37.
The work of adhesion was also characterized from force-distance curves, specifically from the area above the retraction curve (the bright green area in Fig. 2c and d). The work of adhesion on bare glass was 0.2 ± 0.03 fJ, whereas it reached 730 ± 34 fJ on PRO glass. During retraction on bare glass, the force precipitously drops to zero after a short distance, as the van der Waals interaction is confined to a few nanometers. Conversely, the PRO glass holds onto the particle until the liquid meniscus ruptures. As indicated by the yellow arrow in Fig. 2d, the liquid meniscus ruptures ~2.5 μm from the substrate. Hence, this capillary adhesion mechanism is effective over micrometer-range distances, resulting in an adhesion energy that is 3~4 orders of magnitude higher than bare glass. Furthermore, the enhancement in adhesion due to the thin liquid layer is observed for both 15 μm silica and polyethylene (PE) spheres (Fig. 2f and Extended Data Fig. 5). Larger particles have an increased adhesion force on both bare and PRO glass because both van der Waals and capillary forces are proportional to the particle size. The hydrophobicity of the particles has a modest influence on adhesion force, as evidenced by a comparison between hydrophilic silica and hydrophobic PE particles. The shape and size of the meniscus can be varied by the surface energy of the particles, leading to differences in adhesion forces38. Nevertheless, it is noteworthy that all tested particles interacting with the thin liquid layer display micronewton scale adhesion forces, which is sufficiently high for effective capture of PMs.
Filtration at rapid airflow
Due to its outstanding adhesion for PMs, the PRO filter offers unique filtration capabilities, including efficient filtration under high-speed airflow. As the face velocity of the airflow increases, the probability of PMs coming into contact with filter fibers rises, potentially elevating the FE39,40. However, in practical applications, the FE of a conventional filter generally diminishes beyond a certain airflow speed due to PMs detaching from filter fibers41,42 (Fig. 3a). This occurs when the drag force surpasses the adhesion force. Since drag force increases with airflow speed, PMs are more easily detached under faster airflows. As such, conventional filters are less efficient at fast airflows, demonstrated by the FE measurements using a conventional polyester filter (Fig. 3b-d). For example, the FEs for PM1.0 show an increase up to a face velocity of 4 m/s (orange line in Fig. 3b) while the FEs for PM2.5 start to decline above a velocity of 2.4 m/s (orange line in Fig. 3c). Larger particles have a stronger drag force under the same airflow, resulting in a greater tendency for PMs to detach; the FEs for PM10 decrease just above a velocity of 0.8 m/s (orange line in Fig. 3d). Numerical calculations corroborate that drag forces surpass the adhesion forces for all PMs, ranging from 0.3 to 10 µm, on conventional filters at airflows slower than 3 m/s as shown in Extended Data Fig. 6; it highlights limitations of conventional air filters in rapid filtration systems like those in clean rooms, data centers. In contrast, the FEs of PRO polyester filter increase for all PM sizes with increasing airflow velocity. Due to the elevated PM adhesion, PRO filter overcomes the issue of particle detachment (green lines in Fig. 3b–d). The enhanced filtration performance with high air permeability allows the PRO filter to be used as a suitable candidate for applications requiring rapid and large-scale filtration.
PM resuspension
Moreover, strong capillary adhesion suppresses the resuspension of captured PMs. In conventional filtration systems, PMs that have already been captured are often detached by external airflow or sudden impact, resulting in PM resuspension into air. This phenomenon is experimentally demonstrated here; PMs initially trapped on a conventional filter are detached when air is blown through the filter by an air gun, as observed by a high-speed camera (Fig. 3e and Supplementary Video 2). A hanging piece of paper is placed to demonstrate air flow. After blowing air, the detachment of PMs was clearly observed by photography and SEM (Fig. 3f). In contrast, the PRO filter retains PMs when subjected to the same airflow (Fig. 3g and Supplementary Video 2), without a change in the PRO filter media (Fig. 3h).
This unique characteristic allows PRO filters to function under multidirectional airflows. Due to issues of resuspension, so far, conventional filtration systems operate only under unidirectional airflow. To illustrate multidirectional airflow functionality, we measured the weight of filtered PMs after filtration tests in opposite directions (Fig. 3i). Following an initial filtration step, the PRO filter is inverted and exposed to an equivalent quantity of PMs introduced from the reverse side. Nearly the same weight increase was observed, caused by PM filtering without resuspension; this filtration process continues to function for more than five cycles. This study presents the first instance of a multidirectional filtration strategy employing filter media, further attesting the versatility and efficacy of the PRO filter concept.
Pressure drop increase by PM filtration
The thin liquid film in the PRO filter provides advantages in the i) lifetime of the filter and ii) energy consumption of the filtration system. Both are related to pore clogging. Accumulation of PM within the pore leads to an increase in the pressure drop (ΔP) across the filter. This pressure increase is particularly profound in filters with smaller pore sizes, such as high efficiency particulate air (HEPA) filters. Interestingly, PRO HEPA filter exhibits approximately 30% slower rate of ΔP increase compared to a bare HEPA filter, shown as gray and black lines in Fig. 4a, respectively. The underlying reason for this reduced rate of ΔP increase in the PRO filter can be attributed to the distinctive structure in which PMs are captured on its thin liquid layer (Supplementary Video 3). On conventional filters, PMs form dendrites during filtration. The PM dendrites have a large effective thickness (teff) because of its loosely packed structure (Fig. 4b), leading to effective pore clogging43. An image of a planar metal mesh shows a decrease in pore size due to captured PMs (Fig. 4c and Extended Data Fig. 7); analysis reveals a reduction in the projected pore area fraction (i.e., projection pore area/unit area) from 0.35 to 0.21 mm2/mm2. On the other hand, PMs captured on the PRO filter are densely packed by capillary forces (Fig. 4d). This dense arrangement minimizes the effective thickness of the filter medium. Observations on a PRO metal mesh confirms a slower rate of increase in the effective thickness (Fig. 4e and Extended Data Fig. 7). The projected pore area fraction is reduced only to 0.30 mm2/mm2, keeping pores approximately ~40% more open compared to the bare metal mesh. This compact arrangement of captured PMs effectively mitigates the rate of ΔP increase for the PRO filter.
With increasing ΔP across the filter, the fan of a filtration system needs to rotate faster, consuming more energy. To reduce the rate of ΔP increase, filtration systems commonly employ a dual-filter (or sometimes multiple-filter) strategy, involving a combination of low- and high-efficiency filters. The low-efficiency filter, commonly known as a pre-filter, has relatively large pores capable of capturing large PMs. The pre-capturing of large PMs prevents clogging of small pores on high-efficiency filters, such as HEPA filters. The incorporation of a bare pre-filter in front of a HEPA filter exhibits a reduced rate of ΔP increase by approximately 1.8 times compared to a stand-alone HEPA filter, indicated by the orange and black lines in Fig. 4a, respectively. We hypothesized that effective by a PRO pre-filter would suppress clogging of HEPA filter, as illustrated in Fig. 4f. Indeed, a PRO pre-filter offers an even slower rate of DP increase, as indicated by blue line in Fig. 4a. Furthermore, we conducted a comparative analysis of the mass of captured PMs using different filter combinations. Although the total mass of captured PMs by both the PRO pre- / bare HEPA filter and the bare pre- / bare HEPA filter were nearly identical due to the almost 99.9% FE of the HEPA filter, the distribution of captured PMs differed substantially. When used with the PRO pre-filter, the HEPA filter contained ~60% less captured PMs compared to the bare pre-filter; this is because the PRO pre-filter captures ~30% more PMs than the bare pre-filter, as indicated by the blue and orange bars in Fig. 4g, respectively. The images on the right side of Fig. 4g show the bare and PRO pre-filters following the filtration process. The darker color of the PRO pre-filter indicates its enhanced PMs capturing capability.
By integrating effective pre-capturing and capillary-driven densification of PMs, the combination of a PRO pre- / PRO HEPA filter (green line in Fig. 4a) displays the longest time of use. For example, this PRO combination delays the time needed to double ΔP by three times of the bare combination (orange line in Fig. 4a). This result confirms that PRO filters have three times prolonged lifetimes compared to conventional filters. In addition, the gradual rate of ΔP increase in the PRO filters contributes to sustaining a relatively high level of air permeability, which decreases energy consumption of the ventilation system. In comparative tests involving both types of pre-filters and HEPA filters, approximately 20% less electricity was consumed when the PRO filters are employed compared to their bare counterparts (Extended Data Fig. 8).