Thomas W. Bement1, David J. Downey1, Ania Mitros3, Rebecca Lau3, Timothy A. Sipkens1,2, Jocelyn Songer3, Heidi Alexander1, Devon Ostrom4, Hamed Nikookar1, Steven N. Rogak This email address is being protected from spambots. You need JavaScript enabled to view it.1 1 Department of Mechanical Engineering, University of British Columbia, Vancouver, BC V6T 1Z4, Canada
2 Metrology Research Center, National Research Council of Canada, Ottawa, Ontario K1A 0R6, Canada
3 MakerMask, Orange, Massachusetts, USA
4 Artist/Researcher, Toronto, Ontario, Canada
Received:
January 26, 2022
Copyright The Author(s). This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are cited.
Revised:
July 17, 2022
Accepted:
July 18, 2022
Download Citation:
||https://doi.org/10.4209/aaqr.220044
Bement, T.W., Downey, D.J., Mitros, A., Lau, R., Sipkens, T.A., Songer, J., Alexander, H., Ostrom, D., Nikookar, H., Rogak, S.N. (2022). Filtration and Breathability of Nonwoven Fabrics Used in Washable Masks. Aerosol Air Qual. Res. 22, 220044. https://doi.org/10.4209/aaqr.220044
Cite this article:
We consider fabrics that can improve upon the performance of the widespread all-cotton mask, and examines the effect of layering, machine washing and drying on their filtration and breathability. Individual materials were evaluated for their quality factor, Q, which combines filtration efficiency and breathability. Filtration was tested against particles 0.5 µm to 5 µm aerodynamic diameter. Nonwoven polyester and nonwoven polypropylene (craft fabrics, medical masks, and medical wraps) showed higher quality factors than woven materials (flannel cotton, Kona cotton, sateen cotton). Materials with meltblown nonwoven polypropylene filtered best, especially against submicron particles. Subsequently, we combined high performing fabrics into multi-layer sets, evaluating the sets’ quality factors before and after our washing protocol, which included machine washing, machine drying, and isopropanol soak. Sets incorporating meltblown nonwoven polypropylene designed for filtration degraded significantly post-wash in the submicron range where they excelled prior to washing (Q > 50 kPa–1 at 1 µm, respectively, degraded to Q < 10 post-wash). Washing caused lesser quality degradation in sets incorporating spunbond non-woven polypropylene or medical wraps (Q = 12 to 24 pre-wash, Q = 8 to 10 post-wash). Post-wash quality factors are similar for all multi-layer sets in this study, and higher than Kona quilting cotton (Q = 6). Washed multi-layer sets filtered 12% to 42% of 0.5 µm, 27% to 76% of 1 µm, 58% to 96% of 2.8 µm, and 72% to 100% of 4.2 µm. The measured filtration and pressure drop of both the homogeneous and heterogeneous multi-layer fabric combinations agreed with the estimations from a model assuming layers filter independently. Further examination of selective nonwovens showed that IPA degraded their filtration, while washing and drying produced variable effects on their filtration. Variability in filtration and pressure drop was observed in and across Filti samples.HIGHLIGHTS
ABSTRACT
Keywords:
Particle filtration, Reusable, Washable, Nonwoven fabrics, Mask
Airborne respiratory pathogens, including the SARS-CoV-2 virus responsible for the COVID-19 pandemic that has killed millions of people worldwide annually (Lewis, 2020; Morawska and Cao, 2020; CDC, 2021a). Airborne transmission occurs via virus-laden particles generated from breathing, talking, coughing, and sneezing. Research during the 1918 pandemic (Kellogg and MacMillan, 1920) established that masking, including non-medical masks, reduces viral transmission rates (Chu et al., 2020; Eikenberry et al., 2020; Leffler et al., 2020; Lyu and Wehby, 2020; Stutt et al., 2020; Abboah-Offei et al., 2021; Gandhi and Marr, 2021). Current case studies suggest over a 70% reduction in SARS-CoV-2 transmission rates when masks are used effectively (Doung-ngern et al., 2020; Malone, 2020; Wang et al., 2020). Public health organizations widely implemented mask mandates to reduce viral transmission rates for the COVID-19 pandemic (WHO, 2020; CDC, 2021b). As a result of these mandates and broader public interest in masks, the mask market has expanded with buyers using selection criteria beyond filtration efficiency, including comfort, fashion, environmental impact, cost, access, and supply chain ease. Washable masks remain widely used, including for cost and environmental reasons, such that their effectiveness remains an important research topic. In this work, we evaluate candidate materials for reusable masks, targeting materials that outperform cotton alternatives in terms of reducing airborne disease transmission. While woven materials, including cotton, are commonly used in washable masks, nonwoven materials, especially meltblowns, have significantly higher quality factors (Maher et al., 2020; Pei et al., 2020; Wilson et al., 2020; Bagheri et al., 2021; Crilley et al., 2021; Drewnick et al., 2021; Morais et al., 2021, Rogak et al., 2021). Li et al. (2020) found comparable performance for cellulose, nonwoven materials, and surgical masks. Nonwoven materials are widely available and commonly used for sewing, crafting, medical sterilization wraps, scrubs, and medical masks. Many are hydrophobic and biocompatible. Here, particle filtration efficiency (PFE) is measured as a function of particle size. Particle size is a critical parameter in filtration testing, with different filtration mechanisms dominant at different sizes, as well as changes in the number of expired particles (Morawska et al., 2009; Johnson et al., 2011; Asadi et al., 2019; Bake et al., 2019), viable viral loading, (e.g., Coleman et al., 2022; Zhang et al., 2021), and settling times. We tested filtration of particles with aerodynamic diameters between 0.5 µm to 5 µm for several reasons. Firstly, most common fabrics can easily remove particles above > 5 µm (Leith et al., 2021), so evaluation at larger sizes does not help differentiate between good candidate materials. Secondly, tests at this size range can be done with a single particle spectrometer, simplifying the experiment. Finally, and most importantly, particles in this size range seem to carry highest risk of infection. The lower end of this range yields particles similar to that tested by existing standardized test methods (NIOSH, 2019), which test about mass median aerodynamic diameters around 0.34 µm (Sipkens et al., 2022). We focus on the effect of layering, machine washing, and drying of these materials, including composite masks formed from a combination of these materials. Studies on the effect of mask washing show varied results. Sankhyan et al. (2021) noted the deconstruction of cotton fibers, using electron microscopy, and an increase in inhalation resistance, but no change in filtration efficiency. Hao et al. (2021) found negligible effects of washing on the filtration efficiency of a number of woven materials and a synthetic microfiber cloth. By contrast, samples of N95 mask materials exhibited a reduction in the PFE from 98% to 50% for 0.3 µm particles, a likely consequence of a loss of their electret properties. Reutman et al. (2021) considered the effect of washing on a 3-layer mask prototype containing a layer of meltblown polypropylene and found that the filtration efficiency reduced from 85% to 70% for low face velocities and submicron particles. Everts et al. (2022) showed that high-quality medical masks reprocessed 10 times by water immersion methods maintained higher filtration efficiency than new, non-medical, 3-ply disposable masks as well as cotton and cotton-polyester mix fabrics, even when triple layered. In this work, we present results for single layers of candidate materials, before continuing on to consider the effect of washing and layering, to form composite masks. We use the interim guidance of the WHO (2020) for non-medical fabric masks (suggesting a minimum of three layers, including a hydrophilic material for the skin-touching layer, a filter layer, and a hydrophobic outer layer) to guide the formation of composite masks. Filtration of particles is influenced by four mechanisms: diffusion, impaction, interception, and electrostatic forces (Hinds, 1999). For small particles (< 0.1 µm), diffusion plays a prominent role. Larger particles (> 0.5 µm) are filtered mostly by impaction and interception. Electrostatic forces are especially influential in the 0.1–2 µm range and key to electret materials. Degradation of electret media by different cleaning methods reduces the quasi-static charge and, thereby, filtration (Xiao et al., 2014; Ou et al., 2020). Bioaerosols generated by breathing, coughing, sneezing, and talking show a particle number distribution that peaks around 0.8 µm diameter particles, with numbers of larger particles decreasing up to 1000 µm diameter (Morawska et al., 2009; Johnson et al., 2011; Asadi et al., 2019; Bake et al., 2019). However, viral concentration (copies per unit particle mass) varies with particle size and tends to be highest in particles below 5 µm aerodynamic diameter. A study of COVID-19 patients (Coleman et al., 2022) found that particles ≤ 5 µm contributed 85% of the total viral RNA load detected from 13 patients. A similar result was observed in monkeys, with 0.65–4.7 µm particles accounting for 77–79% of total virus shed by infected cynomolgus macaques (Zhang et al., 2021). In influenza patients, Milton et al. (2013) found 8.8 times more viral copies in ≤ 5 µm particles than in > 5 µm particles. In cough droplets of influenza patients, Lindsley et al. (2010) found that 42% of the influenza RNA was contained in particles < 1 µm aerodynamic diameter, 23% in particles 1–4 µm, and 35% in particles > 4 µm. Looking at transmission rather than viral counts, Zhou et al. (2018) found that droplets < 1 µm did not cause ferret-to-ferret influenza transmission, whereas droplets 1.5–15.3 µm did result in infection. Zhou et al. (2018) also showed that high particle counts may not imply high infectivity: while 76.8% of total airborne particles released from the ferrets had aerodynamic diameters of 0.52–1.54 µm, ferrets exposed to < 1.5 µm particles did not get sick and viral RNA was detectable only in particles > 4 µm. At the same time, where and whether particles settle also depends on particle size. Inhaled particles in the 0.1–10 µm range deposit in the lungs with higher deposition rate for > 1 µm particles than < 1 µm particles (Park and Wexler, 2008). According to Carvalho et al. (2011), 1–5 µm particles are deposited deep in the lungs, whilst those > 10 µm are generally deposited in the oropharyngeal region, and most particles 0.1–1 µm are exhaled. Sosnowski (2021) estimated 75% of particles between 0.2 and 0.8 µm are exhaled and therefore are not deposited in the lungs. The widely used NIOSH 42 CFR Part 84 standard (hereafter referred to as the NIOSH N95) focuses on filtration efficiency around 0.3 µm aerodynamic diameter (0.075 µm mobility diameter, see discussion in Sipkens et al. (2022), typically close to the most penetrating particle size. This standard aims to protect against workplace hazards, like dust, rather than specifically targeting transmission of airborne disease. Given that these materials show larger differences for larger particle sizes, there is a need to characterize these materials at these larger particle sizes. Medical masks are tested for bacterial filtration efficiency (BFE) with a mean particle size of 3 µm (given a density around that of water, the aerodynamic and mobility diameters are roughly equal), within our test range. Experimentally, particle filtration efficiency (PFE) at a given size is defined as where η is the PFE, P is the penetration, and Nf and Ns are the filtered and source concentrations, respectively. In terms of source control Ns corresponds to particles generated by an infected individual, while for personal protection Ns corresponds to particles from some external source. Uncertainties scale with the value of the penetration, such that high filtrations typically have lower uncertainties. Uncertainties also expand at the edges of the size distribution, with the measurements being most reliable about the peak of the size distribution. The pressure drop (Δp) across the mask affects both breathing effort and leakage around the sides. Breathing effort directly impacts comfort and thus may impact how consistently a person wears a mask. Quality factor (Q), which is a key metric that combines PFE and pressure drop (Hinds, 1999), defined as: Although we use a natural logarithm in defining Q, sometimes base-10 logarithm is used (Zangmeister et al., 2020). Within this work, we consider the effect of layering various materials. When filtration is not strongly influenced by particle charge, the filter layers are expected to act independently. Then, for each size class, the net penetration for a layered mask is given by the product of the penetration for the individual layers (a consequence of each layer seeing only the particles not filtered by the previous layers), where Pml and ηml are the penetration and PFE for the multilayer mask and Pj and ηj are the penetration and PFE for the jth layer. By contrast, the pressure drop is given by the sum of each layer, as an extension of Darcy’s law. As a result, any single material is expected to have the same quality factor, regardless of the number of layers. We note that integrated filtration efficiencies, such as the mass-based filtration efficiency measured by the TSI 8130A, are less likely to follow this trend, due to the nature of the integration step. Fig. 1 shows the system used for filtration testing. A TSI 3076 atomizer generated the challenge aerosol from a 20 g L–1 NaCl solution. The atomizer was fed with air by a mass flow controller (ALICAT MCS-10SLPM-TFT) supplied with pressurized room air at 2–3 SLPM. Total particle concentration was below 3000 # cm–3, according to the maximum concentration range of the TSI 3330, and was not adjusted during the tests. A Senserion SPS30 was used to monitor the upstream to ensure that aerosol concentrations were consistent. The particles were diluted with room air in an extraction duct (ambient particles were less than 1% of the total). The NaCl particles passed through an X-ray charge neutralizer (TSI model 3088) resulting in a quasi-equilibrium bipolar charge distribution (Johnson et al., 2011). The effect of this distribution was checked using a differential mobility analyzer (DMA) column as an electrostatic precipitator to remove all of the charged particles. Comparing the neutralized to uncharged case, we confirmed that the variation of net charge in the neutralized distribution was negligible (see Supplemental Information A). It is important to note that the natural charge distribution produced by a nebulizer will result in the measurement of a higher PFE than that measured for neutralized particles (Corbin et al., 2021), which can be a reason for differences in PFE between studies. The remaining tests were run without the DMA column. Following neutralization, the aerosol was diverted to either a bypass line or through flat filtration media that was clamped in a holder with a 21 mm diameter flow passage. Following the sample/bypass section, an Optical Particle Sizer (OPS, TSI Model 3330) measured the total particle counts in 16 size bins ranging from 0.3 to 10 µm (optical equivalent diameter). Filtration efficiency was determined by comparing the counts for the filter versus bypass flow paths. The OPS flow rate was 1 L min–1, resulting in a theoretical face velocity (i.e., normal to the fabric) of 4.9 cm s–1 through the sample material. Considering typical mask flow areas, this corresponds to an inhalation flow of ~50 L min–1, midway between a resting rate and that used for N95 testing (Caretti et al., 2004). Higher face velocities would yield larger PFEs for particles above several microns due to increased impaction, but lower PFE for the smallest particles captured by electrostatic interactions (Corbin et al., 2021). The OPS sizes particles by the magnitude of the scattered laser light from single particles, assuming a refractive index of 1.4 for NaCl. Using the mean particle size based on the bin limits, we can use this refractive index to convert the bin’s mean geometric diameter to the aerodynamic diameter, which is often used when reporting PFE. A table showing the conversion for the bins is included in Supplemental Information B. In what follows, we present PFE as a function of aerodynamic particle size, or for compactness, at an aerodynamic size of 1 micron. Nonwoven materials are produced by mechanical, chemical, thermal, or solvent treatments to hold fibre webs together in a disordered matrix. Nonwoven fabrics are either dry formed or wet laid. Dry formed materials are subdivided further into air laid, dry laid, spunbond (Fig. 2(a)), meltblown (Fig. 2(b)), and electrospun (Fig. 2(c)). The nonwoven materials tested in this paper are spunbond (SB), spunbond-meltblown-spunbond (SMS), and electrospun. The SMS materials utilize two SB layers as a substrate and support for the weaker but high filtration meltblown middle layer. Some SMS materials contain additives and coatings which make them ill-suited for masks. In contrast, woven materials (Figs. 2(d) and 2(e)) typically have larger gaps between fibers and lower quality factors than nonwovens. However, woven cottons or cotton blends are hydrophilic and are effective as the skin-touching layer of a composite mask. The individual materials tested in this study are listed, along with their shorthand names and key properties, in Table 1. Each material was tested 3 times with fresh samples in each test. For a subset of the materials, multiple layers were tested together. Based on the measured quality factors, seven composite material sets (A–G, Table 2) were selected as potential improvements over the cotton-only mask and tested. The impact of sanitizing the masks was investigated in terms of washing and isopropanol (IPA) soak. For washing, the composite samples (A–G) underwent ten cleaning cycles to simulate reuse of the masks. Cleaning cycles included machine washing and drying. Samples were washed with a Huebsch commercial front-load washer (HFNLYRSP111CW01) set for “normal loads”, “warm water”, and “light soil” using a Purex detergent (labeled as “Purex Dirt Lift Action Coldwater Laundry Liquid”). Samples were placed in a laundry bag prior to washing and stayed in the bag through drying. Samples were dried with a Huebsch commercial electric dryer (HDEY07WF1502), set for 60 minutes on “low temperature, rapid”. No dryer products were used. For the IPA soak, the samples were soaked in IPA for at least 6 hours (Xiao et al., 2014), then hung dry for at least 24 hours. A two-phase procedure (washed and IPA) was also considered, in which the transient static charge introduced by the dryer after the 10 cycles of the washing phase was removed using a subsequent IPA soak, noting that the IPA soak could also damage the mask. The materials fall into families with similar quality factors. Fig. 3 shows filtration efficiency for 1 µm particles as a function of pressure drop, overlaid with lines of constant quality. Measurements for other particle sizes are included in the online Supplemental Information Section C. Quality factors for the various materials are also given in Table 1. The highest performing materials were the N95 respirators and medical masks, even after IPA soak, consistent with Rogak et al. (2021), with quality factors from 27 to 119 kPa–1. This was followed by the SMS medical wraps (H400, H100, and Advancheck SMS), with quality factors from 13 to 36 kPa–1, and the craft-grade spunbond materials (Pellon 930, SmartFab, OlyFun), with quality factor from 8 to 11 kPa–1. The Filti mask material showed variability in performance, with quality factors of 46 kPa–1 (batch “Flti(2020)”, acquired in March 2020), 78 kPa–1 (batch “Flti(2021)”, acquired in March 2021), and 101 kPa–1 (batch “Flti(2022)”, acquired in May 2022). “Flti(2020)” performed poorly in our study, filtering 22% at 0.5 µm. Samples from later batches (”Flti(2021)” and “Flti(2022)”) showed better (average 93% at 0.5 µm and 98% at 4.2 µm) but inconsistent filtration. Of the 10 samples from batches “Flti(2021)” and “Flti(2022)”, 4 filtered at or above 95% at 0.5 µm, one ”Flti(2021)” sample only filtered 80% at 0.5 µm. This same sample was subsequently treated with IPA in Phase 2 of our study (Section 3.5), and continued to show lower filtration than the other 9 samples. Ballard et al. (2021) also observed variability in Filti’s single-layer filtration ranging from 80% to 95% at 0.3 µm. The cotton materials showed a broad range of quality factors depending on the weave. Flannel had the highest quality factor among the woven cotton materials, 8.3, consistent with Zangmeister et al. (2020), who noted higher filtrations in heavily napped cotton fabrics, including flannel. The flannel in this study has a relatively disordered and fluffy structure, visually similar to that of the non-woven materials in this study (see Fig. 2). The low nap woven cottons (Kona and sateen) exhibited the lowest quality factors, between 4 and 5 kPa–1. Particle charge did not affect filtration. Filtration was nearly identical for uncharged and neutralized particles (Appendix A), consistent with Zangmeister et al. (2020), who found that the effect of particle charge is minimal for cloth-based masks, at a smaller challenge aerosol size. For a subset of the materials, the effect of layering was considered. Results for each material showed consistent quality factors across different numbers of layers in the test sample (Supplemental Information Figs. C1, C2, C3 and Supplemental Information D). Some recent studies (Zangmeister et al., 2020; Zhao et al., 2020) deviated from this model, with higher measured filtration in the first layer than subsequent layers. Possibly this discrepancy results from their focus on smaller particles, which are more influenced by electrostatic effects. Irregular variations in pressure drop, PFE and quality factors (Supplemental Information Figs. C4 and D1) are largest in the craft-grade spunbond materials (Oly Fun, SmartFab, Pellon 930). Backlit microscopy showed variations of fibre density in localized regions (Supplemental Information Fig. D2) that likely underlie the measurement variations. All seven multilayer combinations had a hydrophilic skin-facing layer and all but G had a hydrophobic outer layer, as recommended by the WHO. Flannel’s high-quality factor and comfort make it a desirable hydrophilic layer, however, with a high pressure drop, it reduces the acceptable pressure drop for the filtering layer. Since polypropylene and polyester are also hydrophobic, all the nonwovens (SMS and SB) are appropriate for the hydrophobic layer(s). Fig. 4 indicates that the effect of layering is well described using Eqs. (3) and (4). This supports the idea that layers act independently for both homogeneous and heterogeneous sets. Fig. 5 summarizes the effect of machine washing, drying and IPA soak on the heterogeneous sets A-G (see Table 2). For larger particles (4.2 µm), all sets showed good filtration efficiency before and after washing (PFEs of 72% to 100%). After 10 cycles of the washing protocol, sets C and D had a statistically significant decrease in filtration efficiency (α = 0.1 in a two tailed t-test) at 4.2 µm, while all other sets showed no statistically significant difference. For 0.5 µm particles, sets D and E had the best filtration efficiency (92%–97%, 96%–98%, respectively) before washing. All sets decreased in filtration efficiency after the washing protocol, with sets C, D and E degrading most drastically (post-wash filtration 18%–20%, 36%–42%, 33%–40%, respectively). This filtration degradation may result from the wash altering the delicate fibers, and IPA soak neutralizing electrostatic charge in the meltblown cores of H100 and Type 2 Red Cross Surgical Mask and in the proprietary nanofiber technology in Filti. Sets A, B, F and G showed moderate decrease in filtration efficiency at 0.5 µm. Set F with spunbond nonwoven polypropylene was the most breathable set with a significantly lower pressure drop than all the other combinations. Sets C, D, and E showed a significant increase in pressure drop after washing, possibly due to the change in porosity and fiber density post-wash. The outer layers of sets D and E (Filti and Pellon 930, respectively) showed visual deterioration after wash. Sets A, B, and F showed small changes in pressure drop; the calendered spunbond nonwoven in these sets (Advancheck SMS and SmartFab) may have provided stability to the fibers. Set G (woven) decreased in pressure drop, perhaps due to an increased pore size after the washing protocol. Accounting for both filtration and pressure drop, the quality factor of sets C, D, and E started out significantly higher than the other sets, but this performance advantage largely disappeared after washing. The washing protocol for the above tests used an IPA soak to remove the static charges that might artificially (and temporarily) increase the filtration efficiency. However, this might also degrade the performance of electret materials beyond the effect of regular washing. Therefore, additional tests were performed to isolate the effect of the IPA soak in the next section. We tested additional samples of SmartFab (N = 3), Advancheck (N = 4), and Filti (March 2021 batch, N = 4) with the following treatments: Samples were sewn between two layers of quilting cotton to emulate similar mechanical encapsulation and protection as in common sewn mask construction for the 10 cycles of machine washing and drying. They were washed with an LG front-loading washer (model WM2487HRMA) on the default cycle, using Seventh Generation free & clear laundry detergent. They were dried with an LG gas dryer (model DLGX7188RM) on the default cycle. No dryer products were used. For IPA soak, samples were soaked in 99% IPA for 8 minutes, then air dried for 40 minutes. Fig. 6 shows the variable effects of IPA and washing on the different materials. Machine washing and drying affected the three materials differently. Filti’s filtration significantly decreased (on average, from 93% to 23% @ 0.5 µm and 98% to 76% @ 4.2 µm) while pressure drop decreased substantially, indicating some structural changes to the material. SmartFab filtration slightly decreased after washing (4-layer SmartFab: from 25% to 20% @ 0.5 µm and 65% to 53% @ 4.2 µm) while pressure drop remained stable (from 19 Pa to 18 Pa). In contrast, washing appeared to increase the Advancheck filtration (from 10% to 21% @ 0.5 µm and 72% to 82% @ 4.2 µm), with minimal effect on pressure drop (from 22 Pa to 19 Pa). IPA lowered the filtration of all three materials, both on untreated samples (from black to yellow dashed line in Fig. 6) and washed samples (from green to pink dashed line). By itself, IPA only slightly lowered Filti’s filtration (from 93% to 90% at 0.5 µm). For Advancheck, IPA decreased the filtration of washed samples (from 21% to 8% at 0.5 µm) more significantly than untreated samples (from 10% to 9% at 0.5 µm). In contrast, all SmartFab samples experienced a significant filtration degradation after IPA soak (unwashed from 25% to 7%, and washed from 20% to 8% at 0.5 µm). The reasons for the varied post-treatment results between Filti, Advancheck, and SmartFab are unknown. Mechanical changes to the fibers and degradation of electrostatic charge probably differed among the materials. Washable fabric masks remain in widespread use in many parts of the world and offer a sustainable alternative to disposable masks. We evaluated several families of candidate materials before and after machine washing and drying, measuring the pressure drop and the filtration of particles between 0.5 and 5 µm aerodynamic diameter, and compared materials based on quality factor, a metric combining filtration with pressure drop. Cottons (Kona, sateen) had the lowest quality factor, with cotton flannel exhibiting a better quality. However, the non-woven materials had the highest performance. Before washing, samples containing meltblown polypropylene (SMS surgical wraps, N95s, and surgical masks) and most Filti outperformed spunbond polypropylene (OlyFun, SmartFab) and polyester (Pellon 930), especially for submicron particle filtration. For the nonwoven craft materials, we observed fiber density variations in localized regions, resulting in variation in the measured pressure drop and filtration efficiency between samples. Filti materials varied in pressure drop and filtration efficiency. After machine washing and drying, and an IPA soak, the samples with meltblown materials degraded to similar quality factors as the spunbond materials. Degradation was greatest for filtration of the smallest particles (0.5 µm), where electrostatic forces are of key importance in electret-based filters such as the meltblown layers of an N95. Possible degradation mechanisms include neutralizing the quasi-static electric charge of the meltblown in the first wash, with subsequent washes removing or melting fine fibers. IPA further removed any electric charge of the materials. Encasement between layers of spunbond did not protect the meltblowns enough to maintain filtration excellence through the machine wash and dry cycles used in our study. Gentler wash methods can reduce degradation (Everts et al., 2022). The higher quality factor of flannel, relative to woven cotton, may be due to its relatively disordered and fluffy structure, visually like that of the non-woven materials in this study. The measured penetration and pressure drop of the homogeneous and heterogenous multi-layer sets were consistent with the simple theory that the net penetration is the product of constituent layer penetrations and the net pressure drop is the sum of constituent layer pressure drops. On selected materials, the effect of IPA was isolated from the effect of washing and drying. Washing and drying significantly decreased the filtration and pressure drop measured for of Filti, slightly decreased the filtration of SmartFab, and possibly improved the filtration of Advancheck. IPA degraded filtration of all three materials, suggesting that all three are electret. Overall, we have shown that spunbond non-wovens and cotton flannel offer a sustainable improvement over the widely-used, woven cotton masks for scenarios in which N95 respirators are not used. This work was partially supported by the NSERC Discovery Grant for S. Rogak. Stephen Salter of Farallon Consultants Limited provided guidance on the study formulation and direction. Timothy Sipkens was partially supported by the Public Health Agency of Canada (PHAC) and by Pillar 4 of the National Research Council Canada (NRC) Pandemic Response Challenge Program. MakerMask has received support from Helpful Engineering. Tests on Filti and tests of the effect of the IPA wash were performed by David Downey.1 INTRODUCTION
2 BACKGROUND AND METHODS
2.1 Filtration Mechanisms and Particle Size
2.2 Computing Particle Filtration Efficiency
2.3 ApparatusFig. 1. Filter test apparatus. The duct between the atomizer and the sample port was approximately 2 meters. The sample was drawn from near the midpoint of this duct.
2.4 Base Materials and Composite MaskFig. 2. Optical microscopy images of (a) spunbond (SmFb), (b) SMS (H100), (c) electrospun (MaT2), and (d–e) woven (Kona, Flan) materials. The SMS sample shows only the outer spunbond layer as the meltblown layer is hidden below it.
2.5 Cleaning Procedure
3 RESULTS AND DISCUSSION
3.1 Individual SamplesFig. 3. Particle filtration efficiency at 1.0 µm versus pressure drop for all sample materials. Filti results shown at the bottom to highlight differences in batches and treatments. In the legend of the top graph, “SMS” denotes spunbond-meltblown-spunbond, and “Standard” denotes standard-compliant disposable. Samples noted with “IPA” were treated with isopropanol. The number of layers is denoted by the suffix, for example, “×2” being 2 layers. The qualities (Q) from this figure are calculated using the average filtration efficiency and pressure drop.
3.2 Homogeneous Layering
3.3 Heterogeneous LayeringFig. 4. Measured filtration overlaid with theoretical filtration from homogenous layering testing. For each color the solid line represents measured values while the dotted lines represent the theoretical values. The Filti used in set D is from the March 2021 sample.
3.4 Effect of Washing, Drying and IPA SoakFig. 5. PFE for treated (washed/dry/ IPA, dashed lines) and untreated (solid) masks. Colors and symbols denote the different mask sets as described in Table 2. Pressure drop is indicated in the brackets in the legend. The Filti used for Set D is from the March 2021 sample.
3.5 Isolating the Impacts of IPA and Washing
Fig. 6. Average measured filtration of Filti, Advancheck, and SmartFab: untreated (black), IPA treated (yellow dash), washed (green), washed and IPA treated (pink dash). ΔP shows average pressure drop, with the range in parenthesis. [N = ] represents the number of samples. Standard deviations are used here to represent the variability of the ΔP and the PFE error bars.
4 CONCLUSIONS
ACKNOWLEDGMENTS
REFERENCES