Particulate matter (PM) and volatile organic compounds (VOCs) are hazardous air pollutants in outdoor and indoor air. The impact of these pollutants on human mortality is well-documented, in which PM is a leading cause of death1, and many VOCs are known carcinogens2. The general approach to controlling air pollution has been to decrease emissions. Air pollution from anthropogenic emissions is not just a local and acute problem; it is also a global and chronic problem as evidenced through terrible events such as the Great smog in London in 1952 caused by sulphates from coal fuel combustion and the Los Angeles smog in 1943 caused by nitrates from automotive emission. However, the removal of pollutants through air filtration can also be an effective approach both indoors and outdoors4,5.
Currently, the high-efficiency particulate air (HEPA) polymers and fibre filtration that utilises electrostatic force is considered to be the best solution for the removal of PMs6,7. However, the replacement or discarding of filters pose a serious issue leading to additional waste, and filter regeneration is useless owing to the decrease in electrostatic forces required for filtration. To overcome this disadvantage, ceramic filters (CFs) used in automotive applications are considered as a suitable candidate for the next-generation filter supporting facile regeneration. CFs have generally been used to remove PM from the exhaust gases emitted from diesel engines, which involves two types of filtration mechanisms, namely deep-bed and cake filtrations. The dominant mechanism in a clean filter is initially deep-bed filtration during which the particulates are deposited inside the porous wall. As the dust load increases, a particulate layer (i.e. the “cake”) is formed along the wall surface in the inlet channels and cake filtration becomes the prevailing mechanism8. Soot, which is a PM, is initially captured through the deep-bed filtration and later deposited as a cake. However, the capture of PM via cake filtration would be more preferable because PM captured by deep-bed filtration cannot be easily removed from the CF during regeneration, which requires hot airflow and plasma flame at high temperatures8,9. Regarding the removal of VOC, activated carbons have generally been used with a sorption method under ambient conditions. However, they have similar problems to PM filters, which are disposable, have a low adsorption capacity, and cannot be used for long periods. To address these problems, a catalytic honeycomb CFs have been utilised. However, the automotive catalysts coated on the CF have been shown to be easily sintered and degraded during regeneration with high temperature10–12. Therefore, it is necessary to develop a filter that can be easily regenerated and simultaneously removes PM and VOC operated under ambient conditions. Thus, we considered photocatalysts as one of the most promising candidates owing to their unique properties such as strong oxidation ability, biological and chemical inactivity, and operation at room temperature13. UV photocatalytic oxidation (PCO) reactions using reactive oxygen species (ROS), such as hydroxyl radical (∙OH) and superoxide anion (∙O2−) generated from the photocatalyst under UV light irradiation, not only effectively remove and decompose VOCs14,15 but also disinfect bacteria and viruses16. Although their photocatalytic capability is limited, and their reaction rate is low due to their wide band–gap energy and fast recombination of the photo–induced electron–hole pairs17,18, they still have a great advantage of operating under ambient conditions such as room temperature, unlike general thermal catalysts.
In this study, we introduce a ceramic catalyst filter (CCF) as a new class of filter that simultaneously removes PMs and VOCs as primary air pollutants, and can be regenerated and used for long periods by simple water washing. The design of the CCF is based on a novel concept, wherein PMs are collected in the inlet channel of the ceramic filter, and VOC gases are decomposed by a photocatalyst coated in the outlet channel, under ambient conditions. The ceramic filter possesses an inlet channel with plugs and a porous inner wall, through which the air flows toward the outlet channel after penetrating the wall (Fig. 1). First, we selected a commercial CF designed with a conjugated plug to allow wall-flow through the porous wall for in-flowing air19,20. The filtration mechanism of CF to remove PM is a combination of deep-bed and cake filtrations, as shown in Fig. 18. The greatest advantage of the PM filter is that the surface area over volume of the filter porous wall (m2/m3) is much larger than that of other filters prepared by polymers and fibres, which enables it to be used for longer durations20. Concerning regeneration, ceramics fired at high temperatures exhibit strong heat- and water-resistant properties and are also used as water-treatment membranes21. In addition to PM removal, we used it as a support for photocatalysts to oxidise VOCs to CO2 under ambient conditions without thermal sources, as shown in Fig. 1. The photocatalysts can remove VOC by the PCO reaction using ROS such as hydroxyl radical (∙OH) and superoxide anion (∙O2−) generated from the photocatalyst under UV light irradiation at room temperature. Our novel design concept entails that, along sequential air flow, PM are first captured by the ceramic porous wall at the inlet channel with initial deep-bed and following cake filtration mechanism, and then VOCs exiting the porous wall are decomposed using the PCO reaction by the UV–photocatalyst system at the outlet channel. Finally, to realise this concept, we also attempted an uncommon method of catalyst coating over the outlet channel surface for preventing catalyst deactivation by PM.
For the rational design of CCF, we prepared commercial ceramic filters with 100, 200, and 300 CPSI, where CPSI implies cells per square inch of the filter. To determine the optimal length of CCF, we investigated the pressure drop of the filters along the CF length using a pressure drop model developed by Masoudi et al.8 and Konstandopoulos & Johnson22. The flow characteristics such as distributions of velocity and pressure of CFs inside each CF channel were investigated using computational fluid dynamics (CFD) simulations (Extended Data Fig. 1a). Using this developed model, the porous-wall lengths were optimised to 60–120 mm, which indicate the ranges of minimum pressure drop (Extended Data Fig. 1b). The predicted pressure drop was estimated to be 71 Pa at an optimal porous-wall length of 105 mm (Extended Data Table 1). Based on these results, commercial CFs with a filter length of 115 mm with each plug of 5 mm were procured from Corning Co. with 200 CPSI and 8.0 mil wall thickness (0.203 mm). Thus, the CF having an optimal length was carefully employed to prepare the CCF.
Fig. 2a illustrates the internal cross-section morphology along the filter length describing a specific feature of the CCF. According to the filtration mechanism of the CF, an air flow with PMs and gaseous air pollutants that entered into the filter would penetrate through the porous wall, where only PMs can be captured on and/or in the pores of the wall. The core of the CCF that removes PM is obtained by an additional surface coating treatment of metal oxide, known as the “membrane”. The morphology of the porous surface with microstructures before and after the membrane coating on the bare-CF can be seen in the scanning electron microscope (SEM) images (Fig. 2b, Extended Data Fig. 2). The membrane with a net-type shape was uniformly spread over the wall surface in the filter channel. The vertical cross-section of the CF shows well-coated membrane layers on the inner wall surface (Fig. 2c), and the characteristic membrane component (Bi) is observed along the wall, which is mainly composed of Mg in cordierite (Fig. 2d). The mean pore size after membrane coating was reduced from 11.6 to 8.2 μm while maintaining the porosity of CF (Extended Data Fig. 2c). It is well known that the pore size reduction could affect permeability related to filter efficiency. In particular, the control of mean (average) pore size plays a critical role in determining the quality factor, including filter efficiency (FE) and pressure drop23,24. We evaluated the filter efficiency (FE, %) and pressure drop (Δp, Pa) under a single-pass test condition of an air flow of 1 m/s by using a customised aerodynamic equipment. Compared to the bare-CF, the membrane coated filter achieved an enhanced PM10 (particulate matter, size below 10 μm) FE of 98% and PM2.5 (particulate matter, size below 2.5 μm) FE of 97.7%, respectively (Extended Data Table 2). It plays an important role in successfully removing PMs by cake filtration mechanism beyond the initial deep-bed filtration through the CF8. Here, we improved the filtration mechanism by membrane coating, which resulted in a significant increase in the FE of PM1 (particulate matter, size below 1 μm) as detailed in Extended Data Table 2.
To remove VOC by the CCF, we developed a Cu2O-TiO2 catalyst based on a TiO2 (anatase) photocatalyst. Cu2O having Cu1+ can act as a co-catalyst in the conduction band as an electron acceptor, further producing ROS such as superoxide anion (∙O2−) (Fig. 1)25,26. The oxidation state of Cu on the surface of TiO2 was confirmed to be Cu1+ by X-ray photoelectron spectroscopy (XPS) analysis and Raman spectroscopy (Extended Data Fig. 3a, b). The basic characterisations, such as UV–visible absorbance and electrochemical impedance spectroscopy (EIS), were measured (Extended Data Fig. 3c, d). The PCO reaction activity of Cu2O/TiO2 catalyst can be improved because Cu2O behaves as an electron acceptor, leading to easy separation of hole–electron pairs and reduction of hole–electron recombination27. To verify this mechanism, we evaluated the electrochemical characteristics by measuring the photocurrent density, Mott–Schottky plot, and photoluminescence (PL) (Extended Data Fig. 4a–c). With these enhanced characteristics of the Cu2O/TiO2 catalyst, we measured the intrinsic formaldehyde decomposition activity of powder catalysts (TiO2, CuO/TiO2, and Cu2O/TiO2) under relative humidity (RH) 0% and RH 50% conditions, resulting in 93% @ Cu2O/TiO2 > 84% @ CuO/TiO2 > 78% @ TiO2 at RH 50%, which was higher than that at RH 0% owing to the formation of hydroxyl radical (∙OH) by the oxidation reaction with water (Extended Data Fig. 4d)13,15. Furthermore, we assessed the removal efficiencies with five representative VOCs, namely formaldehyde, ammonia, acetaldehyde, acetic acid, and toluene gas, over 1 g of Cu2O/TiO2 catalyst coated on honeycomb (Extended Data Fig. 4e). The average removal efficiency was 82.4%, with the highest being 93% for acetic acid and the lowest being 62% for toluene. Therefore, we confirmed the development of a new TiO2-based catalyst via Cu2O as co-catalyst, which improves the photocatalytic activity through facile charge separation and high charge carrier density. For the Cu2O/TiO2 catalyst, we coated similar amounts of catalysts for fabricating CCFs (38–40 g of Cu2O/TiO2 or TiO2 catalysts per apparent volume of the filter (L)), where we initially used the CF without the membrane to confirm the catalyst performance. Enhanced reaction efficiencies (REs: VOC removal and CO2 production) of nearly 90% using the Cu2O/TiO2 catalyst were obtained with CO2 as the product gas, while those using TiO2 were 76% (Extended Data Fig. 4f). In particular, we extensively monitored the feasible side gas-phased products of PCO reaction, including CO by Fourier transform infrared spectroscopy (FT-IR) along the reaction time, but these were not detected (Extended Data Fig. 4g). This implies that CO2 was the only product resulting from the PCO reaction. Thus, we concluded that all the reacted HCHO gases were perfectly decomposed into CO2 gas by the PCO reactions over the Cu2O/TiO2 catalyst. We used formaldehyde (HCHO), which is a well-known as highly carcinogenic gas2, as a representative VOC for the removal test (see Methods). The light intensity test and ray tracing simulation (Zemax) were conducted to optimise the UVA-LED system (Extended Data Figs. 5 and 6). We set up the UV-activated system optimally with a light source using a 2 × 2 UVA LED array with a suitable light intensity of 38.1 (centre) and 40.8 mW/cm2 (side), and a maximum light propagation distance of approximately 30 mm inside the CCF.
Fig. 3a shows a snapshot of the dip-coating of the catalyst with membrane-coated CF for CCF fabrication. When the membrane coated CF was dipped in the catalyst coating slurry, the slurry is not absorbed into the cell, implying no penetration beyond the wall. It is an important design rule that the catalyst should be coated on the outlet channel surface against air flow to prevent deactivation by PM. To observe the inside of the cell (channel), we conducted SEM analysis along with Electron Probe X-ray Microanalysis (EPMA) to confirm the spatial distribution of the main elements in the cell. Fig. 3b shows the SEM image of cross-sectional CCF and the element mapping of Ti, which indicates a well-coated catalyst layer on the inner channel surface of CCF. The SEM image shows the wall centre of CCF, which indicates the inside pores (black), coated catalysts (grey), and membranes (white). The mapping image of Ti demonstrates that the catalyst was coated on a single side including the inside wall, implying that no penetration of the coating occurred to the other side (membrane coating zone). Although the inability of the slurry to penetrate the membrane was not investigated in this study, it may be caused by the micro/nano structures generated by the surface treatment (membrane coating) as per Cassie-Baxter28. For practical CCF fabrication, we confirmed the relationship among the pressure drop, FE, and RE (or removal efficiency) along the catalyst coating lengths (Fig. 3c, d). The pressure drop increased exponentially along the coating length; however, it was stable up to the 50 mm coating zone. FE decreased gradually as the coating length increased. However, RE achieved an optimum value at the 50 mm coating zone. Thus, we obtained the optimum specifications to fabricate the CCF for practical applications from systematic testing along the catalyst coating length (Fig. 3d, Extended Data Table 3). In addition, to better understand the characterisation along the coating zone length, we investigated the wall-normal velocity along the CF length using CFD simulations (Fig. 3e). The wall-normal velocity passing through the inner wall of the CF increases from above 30% of the total CF length to the maximum at the outlet channel, which implies the photocatalytic reaction might be the most effective within 70% of the total CF length from the outlet.
Among the various regeneration methods such as water washing by water flow and sonication, we found that simple water washing in the direction against PM capture in CF or CCF is the most effective way to clean dust and regenerate the initial pressure drop of the filter (Extended Data Fig. 7a and b). The maximum dust loading capacity of CF up to final pressure drop (~ 250 Pa) was evaluated to be approximately 20–30 g/L, while the normal disposable HEPA or medium filters (MFs) for six months have a maximum capacity of 5 g/L (Extended Data Fig. 7c)3. For ten regenerations by water washing, the CF achieved an accumulated dust loading capacity exceeding 216 g/L, while maintaining an efficiency loss of barely 9% compared to the high initial efficiency of 98% at a high linear velocity of 1 m/s (Fig. 4a). However, as the regeneration cycle progressed, the commercial HEPA filter (SHLNP-6065AH, SOLTI Co, LTD) rapidly decreased in its dust loading capacity (48.4 g/L for ten regenerations) and could no longer be operated (Extended Data Fig. 8). The CF may possibly indicate a usage of two years without regeneration exceeding the maximum dust loading capacity four times that of the normal HEPA filters. Thus, it implies that we can use the CF for 20 years through ten regenerations of simple water washing. Finally, we simultaneously achieved PM and HCHO removal using CCF (Fig. 4b). For the CCF with Cu2O/TiO2 catalyst, the PM10 FE, HCHO removal efficiency, and pressure drop initially were about 95%, 82%, and 20 Pa at 10 L/min (linear velocity of 0.12 m/s), respectively. Unprecedentedly, these high performances are maintained even after ten regenerations with simple water washing. This study is the first to demonstrate the CCF as a new class of filter realised using our main concept suggested in this study. Thus, with simple water washing, we achieved the facile regeneration of the CCF.
Prior to using the CCF as a prototype for various applications, we first applied the CF panel system to purify the air entering a building to replace current commercial MF (Ultra Low Pressure Drop (ULPD), Dae Young Air Filter) in heating, ventilation, and air conditioning (HVAC) systems (Extended Data Fig. 9a). Through this, we confirmed practically that PM2.5 FE remains higher than 98% for 30 months without replacement and regeneration of the CF panel system, while the MF showed low FE (62%) and required replacement every 3–6 months (Extended Data Fig. 9b). In addition, we developed a free-standing air purification system by applying CF as another type of proto system, and UV light system for CCF is now being installed to purify the air in the underground parking lot (Extended Data Fig. 9c). The FEs of PM10 and PM2.5 of the system are still observed above 90% for about 12 months for a flow rate of 4,000 m3/h (Extended Data Fig. 9d). Note that all the evaluations are in-operation by now. Furthermore, we conducted a CFD simulation to predict the air flow and PM concentrations in huge underground parking lot (volume of 96,000 m3 for 600 cars). After 21 CCF proto systems were located, air flow was more uniformly distributed without a quiescent zone, and the overall volume averaged PM10 concentration showed a surprising reduction of 32% (Extended Data Fig. 10). To confirm the prediction result of the CFD simulations, we practically test the spatial efficiency using 21 installed CCF proto systems.
In summary, we first designed a new class of facile regenerable CCF for the simultaneous removal of PM and VOC under ambient conditions using membraned porous wall flow filtration and UV-activated photocatalyst in the porous CF for air purification. The new filter system based on our proposed concept exhibited a PM FE of about 95% and VOC RE of about 82% with perfect PCO reaction to form CO2 under single-pass air flow condition. The Cu2O/TiO2 catalyst can improve the VOC (HCHO) RE by trapping electrons from TiO2 as well as by preventing recombination of electron–hole pairs on the catalyst. Furthermore, we demonstrated that the CCF filter can be fully regenerated by simple water washing, and the high removal performance of the filter is sustained even after ten regeneration cycles. The CCF can possibly indicate a usage of two years without regeneration, with nearly four times the maximum dust loading capacity than that of conventional HEPA PM filters, suggesting long-lifetime for 20 years through ten regeneration cycles via simple water washing. The facile regenerable, long-lifetime filter system proposed in this study is expected to provide a new paradigm for indoor and outdoor air purification through sustainable technology.
Online Content
The Methods, along with the Extended Data display items, are available in the online version of the paper; references unique to these sections appear only in the online paper.