Effectiveness of Common Fabrics to Block Aqueous Aerosols of COVID Virus-like Nanoparticles

Layered systems of commonly available fabric materials can be used by the public and healthcare providers in face masks to reduce the risk of inhaling viruses with protection about equivalent or better than the filtration and adsorption offered by 5-layer N95 respirators. Over 70 different common fabric combinations and masks were evaluated under steady state, forced convection air flux with pulsed aerosols that simulate forceful respiration. The aerosols contain fluorescent virus-like nanoparticles to track transmission through materials that greatly assist the accuracy of detection, thus avoiding artifacts including pore flooding and the loss of aerosol due to evaporation and droplet break-up. Effective materials comprise both absorbent, hydrophilic layers and barrier, hydrophobic layers. Although the hydrophobic layers can adhere virus-like nanoparticles, they may also repel droplets from adjacent absorbent layers and prevent wicking transport across the fabric system. Effective designs are noted with absorbent layers comprising terry cloth towel, quilting cotton and flannel. Effective designs are noted with barrier layers comprising non-woven polypropylene, polyester and polyaramid.

propagation rate, and skin reactivity as mandated by the National Institute for Occupational Safety and Health (NIOSH). 11 The first two are directly related to the effectiveness of the material to serve as a barrier to aqueous viral aerosols. ASTM F2100-19E1 specifies assessing filtration using 100 nm sized particles of salt aerosol. 12 The N95 certification indicates that 95% of the total particles in a salt aerosol with an average particle size of 300nm are blocked by a material under standard conditions. N95 respirators are typically used by physicians and surgeons. ASTM F2299/F2299M-03(17) uses light scattering particle counting of latex spheres between 100 nm and 5 µm in diameter. 13 Bacterial efficiency standards are described in ASTM F2101 which requires aqueous bacterial aerosols having 3 µm diameter droplets. 14 Meanwhile, aqueous aerosols from speech, sneezing and coughing have size distributions spanning several orders of magnitude 7,15,16 up to thousands of microns. The SARS-CoV-2 virus itself is found in various shapes with polydisperse diameters ranging between 60 and 140 nm. 17 Salt particulates, latex spheres, bacteria and viruses are widely diverse in size, shape, surface chemistry and interfacial properties.
These properties can affect the transport and adhesion within the complex surfaces of materials used in PPE face masks.
Comparative studies of common, household fabrics generally indicate these materials are more permeable than medical grade PPE and widely variable in their filtration efficiencies. For example, Rengasamy 18 provides caution that fabrics can exhibit a range of filtration efficiencies. Examples of sweatshirts, t-shirts, towels, and scarfs from different manufacturers were tested with polydisperse (75 ± 20 nm) salt aerosols and 13 sizes of monodisperse salt aerosols (20 nm-1 µm) at face velocities of 5.5 and 16 cm/s. Particulate transmission through the materials was determined by measuring the particle count upstream and downstream of the filter media using a scanning mobility particle sizer. Fractional transmissions ranged from 40-90% for the polydisperse aerosol and 40-97% for the monodisperse aerosols at 5.5 cm/s. Davies 19 investigated the filtration performance of common household fabrics to remove airborne viruses and bacteria. Fabrics were exposed to aerosols containing either Bacillus atrophaeus (0.95-1.25 µm) or Bacteriophage MS2 (23 nm). Aerosols were delivered in a closed chamber at 30 L/min and particle counts were measured upstream and downstream of the filter media. Based on a combination of filtration efficiency and pressure drop, the highest performing fabrics were 100% cotton t-shirt and pillowcase. The surgical mask had a 96% mean filtration efficiency for the 1 µm particles and 90% for the 23 nm particles. In comparison, the 100% cotton t-shirt had a 69% mean filtration efficiency for the 1 µm particles and 51% for the 23 nm particles. The pillowcase had a 61% mean filtration efficiency for the 1 µm particles and 57% for the 23 nm particles. Hence these fabrics are far more permeable than N95 respirators. These investigations did not attempt to combine multiple types of fabric layers to achieve comparable performance as the NIOSH certified medical respirators and masks, such as the N95 respirators. Recently, Konda 20 studied several common fabrics such as cotton, silk, chiffon, flannel, polyester blends with up to two layers. Cotton quilt and cotton/chiffon performed about as well as an N95 respirator at filtering saline aerosols. Although the methodology is in compliance with NIOSH 42 CFR Part 84 test protocol, the instruments are noted to have poorer counting efficiencies for particles smaller than about 300nm. Furthermore, unknown fractions of the aqueous aerosol particles are lost by evaporation as well as break-up into undetectable, smaller droplets. These aerosols did not contain virus nanoparticles that could be independently identified when transported through the materials. Blocking the transport of virus particles is a prime function of mask fabric.
In response to the time-sensitive need for alternative PPE, we identify commonly available fabric materials that the public and healthcare providers can use in face masks to reduce the risk of viral aerosol inhalation. Over 70 different common fabric multi-layer designs are compared to NIOSH certified medical respirators and ASTM certified masks for filtration efficiency using protocol conditions similar to those of ASTM standards. A common design theme emerges for many layered fabric designs that may reduce the risk of viral inhalation from aerosolized contamination directly striking the mask in both healthcarepatient interactions as well as public interactions with limited physical distancing.

RESULTS AND DISCUSSION
Fluorescent, virus-like nanoparticles emulate the size and surface character of SARS-CoV-2 virus particles and are readily detected and counted. Rhodamine 6G is incorporated into nanoparticles as it is highly photostable and fluoresces with high quantum yield efficiency. It remains well partitioned within the nanoparticle matrix of poly(lactic-co-glycolic acid). Figure 2 is a scanning electron microscope image of a small cluster of primary nanoparticles. Most of the encapsulated nanoparticles have spheroidal shape with some shallow wrinkles. Wrinkles may be due to the sheer stress present during the formation of the core-shell structure. The measured primary particle sizes of the nanoparticles range between 10 and 200 nm, which is the same range as SARS-CoV-2 virus particles 17 , see Figure S1. Zeta potential measurements indicate neutral surface charge over six decades of concentration, see Figure S2 This result is independent of the aerosol pressure that could be applied. Similar pore flooding can occur in salt solution aerosol testing. Nanoparticle transmission through porous materials begins to occur without pore flooding as the steady-state volumetric flow rate of air exceeds the incident volumetric flow rate of aqueous aerosol. Partial flooding decreases the effective material porosity and leads to exaggerated filtration efficiency. In practical terms dense fabric masks do not transmit nanoparticles into a mask, such as virus particles, without active respiration or permeating air convection.
The transmission measurement of nanoparticles through mask materials is based on test conditions that emulate ASTM methods, enable high precision and repeatability, and reproduce sensibly physiological conditions. The rate of human ventilation at rest is nominally 6 L/min 21 and can increase several-fold upon active exertion. Our testing establishes a baseline steady-state air flow of 14 L/min through each test material. Each test is subjected to a total threat of 2 mL aqueous solution containing the fluorescent virus-like nanoparticles at 0.5 mg/mL. This total threat volume is delivered by 26 pulses of aerosol, each lasting one second. The duration and overpressure of the pulses emulates forceful expiration, i.e. a spray resulting from a sneeze, cough or speech from an infected individual. The steady-state air flow being in excess of restful ventilation replicates a slightly elevated ventilation rate as a safety margin, prevents pore flooding, and enables improved statistical repeatability in the nanoparticle count measurements. The pulsed aerosol droplets are polydisperse in size and closely match the size range from forceful expiration. Nanoparticles transmitted through the test material are collected at a distance of 1 mm on a glass slide. The gas flow and slide placement configure the system to be well within the estimated collection regime, i.e. particle capture limit. 22 After a nanoparticle collides with the glass, the rebounded kinetic energy is insufficient to escape the attractive potential energy. Specific details about the aerosol transmission testing are described in the Methods Section, also see Figure 1. Over 70 different common material arrays were evaluated under steady-state air permeation against pulsed aerosols that simulate forceful expiration. A list of materials is provided in the Supporting Information, see Table S1. Table 1 summarizes our most notable transmission results and comparative statistics. Data for the 5-layer N95 respirator by 3M™ are provided in the first row. This is the standard PPE recommended by the Centers for Disease Control and Prevention (CDC) when caring for SARS-CoV-2 patients undergoing an aerosolizing procedure. Several 30 mm diameter samples were cut from around the respirator. The fractional transmission is the nanoparticle count transmitted through the material, normalized by the incident nanoparticle count. The fractional transmission standard deviation across all sampled locations exceeds the typical standard deviation for the nanoparticle counting measurement. This suggests that the filtration efficiency is dependent on the location of the mask. This is reasonable for a stack of non-woven layers that are pressed heterogeneously into a shape comprising highly varying curvature and thickness. Nonetheless the overall average and standard deviation of nanoparticle counts over 69 independent measurements are provided across the entire respirator.
Remaining materials shown in Table 1  respirator. Thus, the permeability index for the 5-layer N95 respirator is unity. This index is included with the p-value indicating that the fractional transmission of the material is indistinguishable from the fractional transmission of the 5-layer N95 respirator. Here p < 0.05 represents 95% confidence that the two materials are distinguishable. This is a double tailed test because materials may be distinguishable by having significantly higher fractional transmission or lower fractional transmission than the 5-layer N95 respirator. The p-value is computed using Welch's t-test as the variances of the material and 5-layer N95 respirator are unequal and must be estimated separately.
Several layered systems exhibit fractional transmission statistically lower than or equivalent to the 5-layer N95 respirator. Specifically, Sheldon G mask with cellulose filter and combination masks, combining two outer layers of white denim with two inner layers of OLY-FUN nonwoven polypropylene and two-layer of Kona quilting cotton with four layers of OLY-FUN exhibit fractional transmissions of 0.16 ± 0.06 ppt, 0.31 ± 0.07 ppt, and 0.40 ± 0.18 ppt, respectively. These mask designs achieve 72%, 55%, and 28% lower fractional transmission than the 5-layer N95 respirator, respectively. Effective materials comprise both absorbent, hydrophilic layers and barrier, hydrophobic layers. Although the hydrophobic layers can adhere virus-like nanoparticles, they may also repel droplets from adjacent absorbent layers and prevent wicking transport. High fiber density and tortuosity increases the probability of collision with aerosol droplets. Effective designs are noted with absorbent layers comprising terry cloth towel, quilting cotton and flannel. For example, two layers of terry cloth, two layers of white flannel, and four layers of Kona quilting cotton exhibit fractional transmissions of 0.50 ± 0.12, 0.51 ± 0.24, and 0.62 ± 0.17, respectively. These commonly available mask materials exhibit fractional transmissions within 10% of the five-layer N95 respirator. Effective designs are noted with barrier layers comprising OLY-FUN (nonwoven polypropylene), lab coat (polyester/polyaramid), cotton coated with spray-on fabric protector as well as traditional synthetic aliphatic and aromatic polymer fibers. Although some terry cloth and cotton multi-layers are effective alone, inclusion of an additional hydrophobic repelling layer is recommended to prevent wicking transport for higher volume threats. Sole use of denim is not effective-in general the yarn bundles are very dense but spaced with wide interweave gaps to promote breathability in jeans. This is demonstrated by high fractional transmission by two layers each of 4oz light weight blue denim, 7oz midweight blue denim, and 11oz heavy weight stretch black denim of 3.91 ± 1.82 ppt, 7.61 ± 0.63 ppt, and 9.43 ± 0.99 ppt, respectively. The two-layer denims exhibit 698%, 1359%, and 1684% higher fractional transmission than the 5-layer N95 respirator, respectively. The fusible polyesters considered are also highly porous. Several additional layered systems exhibit fractional transmission statistically equivalent to the duckbill surgical mask. These may be effective in conjunction with additional safeguards, such as social distancing, and smaller threat volumes.

CONCLUSIONS
Commonly available fabric materials can be used by the public and healthcare providers in face masks to reduce the risk of inhaling viruses from aerosols generated by coughs, sneezes and speech from infected individuals. The protection by some layered designs offer protection about equivalent or better than the filtration and adsorption offered by 5-layer N95 masks. Effective materials comprise both absorbent, hydrophilic layers and barrier, hydrophobic layers. Although the hydrophobic layers can adhere virus-like nanoparticles, they may also repel droplets from adjacent absorbent layers and prevent wicking transport. Effective designs are noted with absorbent layers comprising terry cloth towel, quilting cotton and flannel. Effective designs are noted with barrier layers comprising non-woven polypropylene, polyester, polyaramid.
This work responds to the time-sensitive need for alternative personal protective equipment for healthcare workers as well as face masks for the public. Considering the results of this work and prior work, recommended mask designs include those multi-layered combinations in Table 1 that exhibit transmission either equivalent or lower than the transmission offered by 5-layer N95 masks. It is critical that the materials edges conform snugly to the face to prevent aerosol from entering gaps between the face and mask. The mask must not enable viral imbibition by the lips, tongue and saliva. Ideally the mask does not contact the lips, or there is at least one hydrophobic layer fabric in contact with the face, so aerosol trapped from the exterior does not wick through the mask and become transported by the mouth. Since aerosol transport through a mask is predicated on forced convection air flux, it is recommended that individuals wearing masks reduce inhalation intensity when placed in contact of an unsafe aerosol.

PLGA Nanoparticle Preparation:
Nanoparticles (NP) were prepared by mixing 100 mg PLGA pellets with 1 mL ethyl acetate, 20 μg rhodamine 6G, and 12 mg eicosane. The resulting mixture was vortexed for 5-10 minutes until homogenized. 2 mL of 5 wt% PVA was added and sonicated for 2 minutes using an ice water bath to prevent evaporation of ethyl acetate. This solution was mixed to 50 mL of 3 wt% PVA solution immediately after sonication and stirred at 800 RPM for 2 hours until the ethyl acetate evaporated. The resulting solution was split into two centrifuge tubes and centrifuged at 6000 RPM for 5 minutes followed by the removal of the supernatant. The remaining precipitate was diluted with deionized water and vortexed for another 5 minutes. The centrifugation and rinse were repeated 3 times. The final precipitate was diluted with 30 mL water to obtain a final experimental concentration of ca. 7 mg/mL. A small aliquot of dispersion was weighed both wet and dry to determine accurately the actual NP concentration. This stock solution was further diluted to 0.5 mg/mL for experimentation. This concentration was chosen after a series of experiments to determine optimal NP concentration such that NP do not aggregate, did not clog the fabrics and did not clog the aerosol generator. NP size distribution, shown in Figure S1, and zeta potential tests, shown in Figure S2

Aerosol Transmission Testing
Test apparatus: A test apparatus was designed to analyze the degree of transmission of aerosols through various materials. Figure 3 illustrates a schematic of the test apparatus and identifies the components. A labeled photograph shows the actual components in Figure S3. Design parameters for this system were informed by ASTM procedures that involve testing the performance of surgical masks in filtering aerosols. 23  Aerosol Droplet Size Distribution: Droplet size distribution was determined by using the spray apparatus and spraying directly onto a 0.5" x 0.5" glass slide. The spray collected from one aerosol burst was then evaluated under a Keyence VHX-970F Optical Microscope from Keyence Corporation (Itasca, IL, USA). Images were captured at 20X magnification for large droplets and aerosols and 100X magnification for all droplets to understand the full droplet size distributions. A total of 64 images were taken. The raw images were further processed using ImageJ 26 to subtract the background with a 50 pixel rolling ball radius and a dark background. A scale of 26 pixels was identified as the equivalent of 10 μm. The images were also cropped from the bottom by 50 pixels to remove the magnification and scale bar texts to remove any erroneous particles being counted due to the text. The image was then converted to an 8-bit image format to which a minimum and maximum contrast threshold was set to 0 and 225, respectively. This resulted in black (droplets) and white (background) images. These black and white images were then counted and measured using the counting function of ImageJ, within the Analyze feature, using an ellipse outline method. An example of the subsequent image alterations is located in Figure S4. Figure  Microscope, containing an Olympus U-TV1XC center and Olympus XM10 camera. An X-CITE 120LED Boost laser controller from Excelitas Technology was used for a fluorescent laser source run at 45% power for fluorophore excitation. At least 9 images were taken at 20X optical zoom per fabric to determine particle concentration per area and multiple experiments were conducted per fabric using the accompanying Olympus cellSense Standard 1.16 software. A constant gain and exposure were chosen of 18 dB and 1.109 s, respectively, and a fixed scale contrast was applied between 0 and 5000. Individual images were post-processed in ImageJ, similar to the Droplet Size Distribution protocol. The raw images were further processed using ImageJ to subtract the background with a 500 pixel rolling ball radius and a dark background. A scale of 160 pixels was identified as the equivalent of 50 µm. The images were cropped from the bottom by 50 pixels to remove the magnification and scale bar texts to remove any erroneous particles being counted due to the text. The image was then converted to an 8-bit image format to which a minimum and maximum contrast threshold was set to 15 and 250, respectively. This resulted in black (droplets) and white (background) images. These black and white images were then counted and measured via the ImageJ counting feature, within the Analyze feature, using an ellipse outline method.
An example of the subsequent image alterations is located in Figure S4. The ellipses are counted, and diameter measured to obtain the total distribution of droplets by size and frequency. The aforementioned procedure was automated by creating a custom Plugin using ImageJ batch scripting language, to remove human bias during image analysis and to exponentially speed up analysis. The veracity of the script was confirmed by manual analysis of each step. The nanoparticle count and size distribution are included in Table 1.