key: cord-0951095-27gt6nmi
authors: Chen, T. X.; Pinharanda, A.; Yasuma-Mitobe, K.; Lee, E.; Steinemann, N. A.; Hahn, J.; Wu, L.; Fanourakis, S.; Peterka, D. S.; Hillman, E. M. C.
title: Evaluation of at-home methods for N95 filtering facepiece respirator decontamination
date: 2021-01-06
journal: nan
DOI: 10.1101/2021.01.05.20248590
sha: 30348fdc2019e35f231efe31233b5daa14666357
doc_id: 951095
cord_uid: 27gt6nmi

Shortages in N95 filtering facepiece respirators (FFRs) are significant as FFRs are essential for the protection of healthcare professionals and other high-risk groups against Coronavirus Disease of 2019 (COVID-19), a disease caused by severe acute respiratory syndrome coronavirus 2. In response to these shortages during the ongoing COVID-19 pandemic, the Food and Drug Administration issued an Emergency Use Authorization (EUA) permitting FFR decontamination and reuse. However, although industrial decontamination services are available at some large institutions, FFR decontamination is not widely available. Effective FFR decontamination must 1) deactivate the virus; 2) preserve FFR integrity, specifically fit and filtering capability; and 3) be non-toxic and safe. Here we identify and compare at-home methods for heat-based FFR decontamination that meet these requirements, but utilize common household appliances. Our results identify viable protocols for simple and accessible FFR decontamination, while also highlighting methods that may jeopardize FFR integrity and should be avoided.

Coronavirus Disease of 2019 , caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), is thought to be transmitted predominantly through exposure to short-range droplets and aerosols 1 . To protect against infectious airborne particles, possibly including sub-micrometer aerosols carrying SARS-CoV-2 2 , filtering facepiece respirators (FFRs) such as the N95 3 are recommended by the United States Center for Disease Control and Prevention (CDC) as the PPE severe irritant that can be hazardous to the lungs and eyes. Most UV-C light sources have spectral variation, including wavelengths that can damage the skin or eyes, and proper decontamination by UV-C light requires every surface to receive a sufficient dose of radiant energy, which is hard to ensure for masks due to their three-dimensional shape and metal inserts 11 . Another significant obstacle is that methods that provide sterilization, the complete elimination of all forms of microbial life, using heat (e.g., autoclaving) are highly likely to damage mask integrity. As a result, existing FFR decontamination methods, even though effective against SARS-CoV-2, do not typically remove all other forms of pathogens and contaminants. This constraint necessitates returning decontaminated masks to the original user, a major logistical challenge on an institutional scale, particularly during a healthcare crisis. Therefore, establishing safe and economical decontamination methods for FFR reuse, that could be implemented in a personal, household or resource-limited setting, is both necessary and urgent, considering the current global strain on the supply of N95 FFRs and the vast number of individuals vulnerable to COVID-19 and other respirable infectious agents worldwide.

In this study, we recognized the potential of widely available, commercial household heat-generating appliances to be used for FFR decontamination. We performed quantitative testing of a series of appliances and decontamination methods, evaluating both their ability to reliably establish and maintain decontamination conditions, and the effects of repeated treatments on mask fit and material filtration performance. We identify rice cookers with thermostatically-controlled warming functions as being the best of our tested heat-treatment methods at achieving decontamination conditions for N95 FFRs reuse by retaining FFR fit and filtration integrity for up to five repeated treatments. We also confirm that boiling FFRs is damaging to FFR fit.

Based on a review of previous studies that successfully used heat to deactivate enveloped viruses, we identified parameters that were likely to inactivate SARS-CoV-2 within FFR fabric using heat, while also preserving the fit and filtering properties of the mask. As a result of the COVID-19 response, extensive non-regulatory agency (e.g., National Institute for Occupational Safety and Health (NIOSH), CDC) testing and review of N95 decontamination strategies have been done 12 . Table 1 summarizes 1) temperature, humidity, exposure times and substrates that achieve viral inactivation; and 2) how these different conditions were found to affect FFR fit and filtration. In conjunction, these studies indicate that sustaining a temperature of 70-85°C for 30 minutes at 50% humidity 17, 19 or 60 minutes (with dry heat 15 ) should be sufficient to inactivate SARS-CoV-2. Based on prior studies 10 , we hypothesized that both of these conditions should preserve fit and filtration efficiency of FFRs for up to five rounds of treatment, in line with the CDC's recommendation of a maximum of five donning and doffing cycles for FFRs 10 . Additionally, treatment with steam autoclaves or immersion boiling should inactivate SARS-CoV-2 from FFRs, but previous studies indicated that these types of treatments may reduce FFR filtration efficiency and impact fit after as few as one round of treatment 14, 20 .

Identifying home appliances capable of accurately attaining decontamination conditions A wide range of commercial heat-generating appliances, including microwaves, rice cookers, sous vide machines, tumble driers, hair dryers, and ovens, were considered for achieving FFR-compatible heat conditions of 70°C for one hour. Appliances with minimum temperatures nominally above 70ºC or maximum temperatures nominally below 70ºC were first excluded. For example, standard gas ovens cannot be programmed to maintain temperatures lower than 76.7°C 21 and commercial tumble dryers and hair dryers do not exceed 60ºC during operation at the hottest setting 22 . Appliances that posed a potential safety risk to the user were also deemed unsuitable. Microwave-generated heat may cause FFRs with metallic components, such as the nosepiece, to spark or the plastic from the straps to melt 23, 24 .

Based on these temperature and safety criteria, we selected rice cookers and sous vide machines for further evaluation 25, 26 . Although many types of rice cookers with varying quality and accuracy exist, (Fig S1) 10, 20 . With a process temperature of 67.5ºC ± 7.8 ºC across the hour-long treatment, the Keep Warm setting of the Aroma rice cooker did not allow for precise temperature control. As it was impossible to maintain a stable temperature within the required range, the Aroma rice cooker was excluded from all subsequent tests. This result suggests that, even among rice cookers with Keep Warm . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted January 6, 2021. ; https://doi.org/10.1101/2021.01.05.20248590 doi: medRxiv preprint functions, not all may be suitable for FFR decontamination, and users should consult appliance manuals and specifications to confirm the device's ability to maintain the specified temperature range.

Evaluating the effects of appliance-based heat treatment on quantitative FFR fit Based on the temperature measurements, we decided to test and compare the effects of three different decontamination protocols on mask fit and function: 1) boiling at 100ºC for 10 minutes, 2) maintaining 70ºC for 60 minutes using the Cuckoo rice cooker and 3) maintaining 70ºC for 60 minutes using the sous vide machine. Boiling was included as it is simple to perform at home, but was expected to impair FFR function 9,14 (Fig 1) . assigned to either (B) x round(s) of appliance-based heat treatment OR to a control group. We tested three appliance-based heat treatment methods: Cuckoo rice cooker, sous vide machine and boiling. Boiled FFRs were treated up to three times. FFRs heated with sous vide machine and Cuckoo rice cooker were treated up to five times. Control FFRs were not exposed to any treatment. (C) Treatmentexposed and control FFRs were fit tested using the built-in N95 setting of a TSI PortaCount Pro+ Respirator Fit Tester 8038. Three technical replicate measurements were taken for each FFR and overall Fit Factors (Fig 2) and fit values for the different testing categories (Fig 3) were recorded. (D) The effect of appliance-based heat treatment on FFR filtration performance was evaluated for treatment and control FFRs using a modified version of the standard NIOSH test (Fig 4) . Filtration efficiency was tested on two different FFR areas. For the Uniair San Huei SH3500 (TC-84A-4313) model, FFRs were only exposed to heat using the Cuckoo rice cooker and the sous vide machine. They were treated up to three times. Fit was estimated for both treatment and control FFRs as it is described in (C), the masks were not tested for filtration performance.

FFRs assigned to be heated by the rice cooker were placed in breathable paper bags during the treatment in order to reproduce how a home user would be advised to handle contaminated FFRs (Fig   2, bottom center picture) 10 . For heat treatment with the sous vide machine, FFRs were sealed securely in a polypropylene plastic bag (double protection Ziploc© Freezer Bags) before being placed in water.

In order to completely submerge the FFR within water, two 140 g weights were placed on top of the plastic bag (Fig 2, bottom right picture). For boiled FFRs, as described in a previous study 9 , the N95 mask material was directly immersed in the water while most of the elastic components were kept out of the water to prevent altering the elasticity (Fig 2, bottom left picture) . For all three conditions, extra care was taken to ensure that the masks were positioned flat for the duration of the treatment in order to avoid deforming the mask.

Experiments were performed using two models of donated FFRs: the Halyard Fluidshield 3 (TC-84A-7521) (Fig 2, top) and the Uniair San Huei SH3500 (TC-84A-4313) ( Fig S2) . Care was taken to minimize the total number of masks used in these experiments considering that this research was conducted during the pandemic when PPE supplies were limited. Individual FFRs underwent one to five rounds of heat treatment followed by quantitative fit testing, and material filtration assessment in comparison to an untreated control (IRB reference AAAT5503) (Fig 1, Fig S2 and Supplemental Methods for full details of treatments and fit testing protocols). After exposing FFRs to different methods and repetitions of heat-treatment, the fit factor was estimated with the PortaCountPro+. The fit factor was estimated three times for each FFR (smaller dots) and average fit factor is reported to the nearest integer (red bar). The control FFR was not subjected to any rounds of heat treatment. Grey regression line is overlaid for each treatment method (y ~ mx + b). Analysis of variance p-values is shown when there is significant variation in means among treatment group and control. Significance (*) p < 0.05. Photographs of the heat generation appliances shown in the panel below their respective fit factor profile.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted January 6, 2021. ; https://doi.org/10.1101/2021.01.05.20248590 doi: medRxiv preprint As shown in Fig 2 (top) , exposing Halyard Fluidshield 3 FFRs to 70°C of dry heat for 60 minutes in the Cuckoo rice cooker up to five times did not decrease their fit quality (no significant difference among test group and the control FFR; p > 0.05). Halyard Fluidshield 3 FFRs that were treated with the sous vide machine showed a significant decrease in fit quality, and failed fit testing after four rounds of treatment (p < 0.01). Halyard Fluidshield 3 FFRs boiled for ten minutes failed fit testing after the second round (p < 0.01) (Fig 2 and Table S1 ). We conclude from these results that, out of the three tested methods, the Cuckoo rice cooker, and by extension, dry heat, is best at preserving Halyard Fluidshield 3 FFR fit.

Fig S3 shows equivalent analyses for the Uniair FFR model. However, overall fit quality of this FFR on our testers was found to be poor, and even the control FFR failed two out of three technical replicate fit test measurements ( Fig S3, Fig S4, Table S2 and Table S3 ). Based on our recorded poor fit performance and the lack of trends in fit quality with treatment conditions (Fig S4) , we cannot meaningfully draw conclusions on how the number of treatment repetitions affect the Uniair FFR fit.

These results underscore the importance of performing regular quantitative fit testing of specific PPE to specific subjects to ensure proper FFR fit under a respiratory protection plan 27 . The conclusions of this paper regarding FFR fit and filtration efficacy after treatment thus pertain primarily to the Halyard Fluidshield 3 (TC-84A-7521). It is, however, possible that trends in fit following heat-treatment holds for similarly manufactured FFRs.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted January 6, 2021. ; https://doi.org/10.1101/2021.01.05.20248590 doi: medRxiv preprint

Quantitative fit testing of FFRs involves a series of different test conditions for a specific wearer (normal breathing, deep breathing, head side to side, head up and down, talking, bending over, and normal breathing). These results are often only examined using the overall fit factor, a geometric mean of the individual categories. However, more detailed analysis (Fig 3) of results for each part of the test can provide valuable insights. For example, while FFRs exposed to 70°C in the Cuckoo rice cooker passed . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted January 6, 2021. ; https://doi.org/10.1101/2021.01.05.20248590 doi: medRxiv preprint fit testing after five rounds of treatment, and sous vide machine FFRs passed after four rounds of treatment, they did not score equally in all the test's categories. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted January 6, 2021. ; https://doi.org/10.1101/2021.01.05.20248590 doi: medRxiv preprint treatment method and category (y ~ mx + b) (Table S5 and S6). Analysis of variance p-values shown when there is significant variation in means among treatment category groups and control. Significance (*) p < 0.05.

For FFRs decontaminated in the Cuckoo rice cooker, there was a progressive decrease of the fit score (y = 212−16.8x, r2 = 0.24) when the subject was bending over (Fig 3) . For FFRs that were boiled (y = 131−31x, r2 = 0.99) or heated using a sous vide machine (y = 206−22.8x, r2 = 0.66) there was a progressive decrease in fit score for all categories. For sous vide machine treatment, this trend was strongest when the subject was bending over (y = 212−27.9x, r2 =0.34) or turning their head side to side (y = 224 −28.3x, r2 = 0.38). After three rounds of sous vide machine treatment bending over factors were significantly lower than control (p < 0.05) but the FFR still passed (i.e., overall fit factor of 100 or greater) the fit test with an overall fit factor of 123. After four rounds of sous vide machine treatment, the FFR failed the fit test with an overall fit factor of 69. The FFR performed significantly worse than the control (p < 0.05) in all fit categories but one (normal breathing) (Fig 3, Table S4, Table S5, Table   S6 ).

Evaluating the effects of appliance-based heat treatment on FFR filtration performance

To determine whether each of our three heat treatment methods affected the filtration performance of Halyard Fluidshield 3 FFRs, we performed filtration efficiency (FE) analyses on samples of material from each mask that underwent quantitative fit testing. For this analysis, we developed a modified version of the standard NIOSH test (TEB-APR-STP-0059) assessing the penetration of non-neutralized saline aerosols. Particle counts were measured in six different size bins, by a calibrated particle counter (bins of 0.3 μm, 0.5 μm, 1 μm, 2.5 μm, 5 μm, and 10 μm size classes). While this is not the first of such . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted January 6, 2021. ; https://doi.org/10.1101/2021.01.05.20248590 doi: medRxiv preprint pipelines 26 , our apparatus has all the major design elements needed to normalize for constant flow rates regardless of filter impedance (Fig 4A, Table S7 ). Furthermore, our experiments were conducted with a steady flow rate per unit area of the filter that exceeds the rates expected under standard NIOSH testing, which generally decreases filtering efficiency compared to lower flow rates 28-30 . Testing was performed on two different central sites on each treated and control mask, with one site directing flow through the fabric to mimic inhalation, and the other site testing flow in the exhalation direction. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted January 6, 2021.

as an aerosol generator. FFRs were attached to a sample holder and air and any suspended particles were drawn through the FFR by an evacuator suction pump and diverted to a particle counter. FFR impedance was measured through a pressure sensor attached to the sample holder. (B) The filtration efficiency for different particle sizes (bins of 0.3 µm, 0.5 µm, 1 µm, 2.5 µm, 5 µm, and 10 µm size classes) were determined for two mask regions of FFRs that were exposed to different methods and repetitions of heat-treatment. None of the treatment methods tested reduces Halyard Fluidshield 3 FFR filtration performance below 95%.

Comparisons between filtration efficiencies in the control and heat-treated FFRs indicate that none of our possible decontamination strategies reduced FFR FE below 95%, meaning that the required performance of 'N-95' FFRs was met by all of our treated masks under our test conditions (Fig 4B, Table   S7 ). We note further that a passing score during any of the quantitative fit testing procedures supports these results, as any damage to the N95 mask material would compromise the overall filterability of the mask and result in the FFR failing the fit test. On the other hand, FE testing also demonstrates that flow and filtration properties of the mask material are not the only important consideration, as proper mask fit is critical to maintain a high level of protection. As both a check of our novel filter test rig and a demonstration of very small leaks significantly affecting effective filter performance, we used an 18gauge needle to introduce small holes in the filter after normal testing. FE decreased as a function of the number of pinholes introduced into the mask material ( Fig S5) and pressure across the N95 FFR material decreased at a flow rate of 10 lpm (Table S8) .

Pasteurization of pathogens in food and vaccines occurs between 60°C and 70°C for 30 minutes 25 and SARS-CoV-2 has been experimentally inactivated in FFRs after one hour at 70°C 15 . Under these temperature conditions, some FFRs have been shown to preserve fit and filtration efficiency from three to over ten rounds of decontamination using dry heat 15, 31 . Here we assessed different heat-based . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted January 6, 2021. ; https://doi.org/10.1101/2021.01.05.20248590 doi: medRxiv preprint methods that are likely to deactivate SARS-CoV-2 (Table 1) using common household appliances (Table   2) .

We tested two FFR models, the Halyard Fluidshield 3 and the Uniair San Huei, after sequential rounds of heat treatment and performed blinded comparisons of their fit and filtration properties to control, untreated FFRs. The control Uniair San Huei FFR did not fit any of our volunteers and exhibited widely varying fit test results, and was thus excluded from our final analysis of potential decontamination methods. This result emphasizes the importance of regular fit testing, as proper fit is paramount for appropriate protection. For the Halyard Fluidshield 3 FFRs, we measured fit and filtration efficiency after each round of treatment. We found that bending over reduces fit factor, and therefore protection, the most (Fig 3) . We propose that paying special attention to movement-specific fit testing results when fitting a respirator could aid the end user to adjust their behavior while wearing FFRs (e.g., kneeling vs. bending).

We show that the Halyard Fluidshield 3 fit and filtration efficiency are preserved for up to five rounds of treatment using the Keep warm function of a rice cooker with precise temperature control. The CDC does not recommend re-using FFRs more than five times, so we did not subject the respirators to more than five rounds of treatment 10 . Sous vide machine treated masks preserved fit only up to four rounds, possibly because the respirator was slightly deformed by the water and weights. Flow and filtration analysis of treated FFRs demonstrated that none of the heat treatment methods lowered filtration efficiency below the required 95%. Moreover, we found that even the boiled Halyard Fluidshield 3

FFRs tested for this study preserved FE after three rounds of treatment, which is greater than what has . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted January 6, 2021. ; https://doi.org/10.1101/2021.01.05.20248590 doi: medRxiv preprint been previously reported 32 . It should be noted that this particular model of FFR, which was designed to protect against fluid splashes in the operating room, is rated as a surgical N95 and is thus considered water-resistant. Further research should be done to determine whether splash-protective respirators are more resistant to changes in FE during humid or water-immersive treatment conditions. Overall, while boiling may be a suitable way to decontaminate FFRs, repeated treatment reduced fit below the passing score much faster than 'dry' heat 9,31 .

Our results suggest that at-home appliances are a viable option for decontamination of masks. Since only a limited number of appliances were tested, it is essential for the user to test the selected appliance to guarantee it can reach and maintain 70°C for 60 minutes before attempting decontamination. Such testing can be done in a home environment, for example, by using a precise meat thermometer. As is also the case for commercial decontamination methods, it is essential to carefully examine the FFR after each round of heat exposure to ensure that it has no material damage, burns or tears. As indicated by differences in the goodness of fit between FFR models and between individual FFRs exposed to the same decontamination treatment, it is imperative to carefully examine the FFR before the first use and after each treatment and ensure close fit of the mask all around the face.

It is also important to note that FFRs treated with a dry heat at 70°C decontamination procedure as demonstrated here, will not be sterilized. Pathogens such as Clostridium difficile will not be killed and could still pose a risk to public health. Therefore, appliance-heat treated FFRs using methods like the ones described here should always be re-used by the same person 33 .

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted January 6, 2021. ;

While not recommended for clinical or routine use, the potential decontamination methods evaluated here demonstrate that household appliances could constitute a low-cost strategy for N95 decontamination that could be performed in resource-constrained settings or situations where no viable alternative exists. We note also that our analysis of decontamination conditions and test results on medical-grade FFRs suggest that similar at-home heat-treatment methods would likely be effective for decontaminating other COVID-exposed materials, such as cloth face masks, as long as the same provisions are taken.

Testing was performed on two N95 FFR models, the standard Halyard Fluidshield 3 (TC-84A-7521) and the Uniair San Huei SH3500 (TC-84A-4313) (Fig S2) . FFRs were randomly assigned to either a treatment group or a control group of untreated masks 15, 34 . Potential decontamination protocols were repeated up to five times for masks treated at 70ºC and up to three times for masks that were boiled (Fig 1) . The maximum of five rounds of treatment follows the current CDC safety recommendations for FFR reuse as part of crisis capacity and conservation strategies 27 . The Columbia University

Institutional Review Board reviewed this study's protocol and determined that ethical approval could be waived (IRB reference AAAT5503).

To heat the FFRs at 70ºC for one hour, we used two different types of home appliances: a rice cooker and a sous vide machine. Temperature within each appliance was recorded for the duration of the treatment using a wireless sensor (Inkbird Thermometer IBS-TH1) for the rice cooker and MeatStick . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted January 6, 2021. ; https://doi.org/10.1101/2021.01.05.20248590 doi: medRxiv preprint Wireless Thermometer 4335995327 for the sous vide machine. Both thermometers were benchmarked against a calibrated lab thermometer (Oakton Acorn Series pH 5 Meter and Thermometer) for accuracy.

FFRs that were assigned to be decontaminated in the rice cooker were placed in breathable paper bags and the rice cooker was preheated to 70ºC using the Keep Warm mode. After the rice cooker reached the specified temperature, the paper bag containing the mask was inserted and the temperature held at 70º C for 60 minutes. For treatment with the sous vide machine, FFRs were sealed in a polypropylene plastic bag (double protection Ziploc© Freezer Bags) before being placed in the water preheated to 70º C. In order to completely submerge the FFR, a 140 g weight was placed on top of the plastic bag. The temperature of the sous vide machine was set to 70ºC for 60 minutes. FFRs subjected to boiling were submerged directly in 1.3 L of boiling water at 100 ºC within a 1.4 L pot and were drained and allowed to air-dry overnight post-treatment (Fig 2, photo panels) .

Quantitative fit tests were performed using the manufacturer provided "N95" setting of a calibrated TSI PortaCount Pro+ Respirator Fit Tester 8038 with ambient particulates and aqueous aerosols produced by an ultrasonic humidifier (NYC tap water). Adherent to the Occupational Safety and Health

Nuclei Counter (CNC) Quantitative Fit Test for Filtering Facepiece Respirators was used 27 . This test reports a quantitative fit factor score for seven sequential test components (normal breathing, deep breathing, head side to side, head up and down, talking, bending over, and normal breathing) as well as an overall fit factor that is the geometric mean of the seven individual scores. A fit score of ≥ 100 is considered passing in each individual category. Based on the OSHA Respiratory Protection Standard, . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted January 6, 2021.

an overall fit factor of ≥ 100 is required for passing, regardless of performance (i.e., pass or fail) on each test component 27 . It should be noted that the system gives a maximum possible score of 200+ reported as 201 34 .

The CNC quantitative fit test was performed three times on all FFRs used in the experiment. We obtained a pre-treatment baseline fit factor (blind control) and post-treatment fit tests (also blind) to quantify fit factors and variance. All tests were performed on the same test subject to control for variability in facial features. To minimize user bias, the experimenter and subject were both blinded to which (if any) treatment each mask had undergone. Each mask was punctured and fitted with the commercial metal hose adapter designed for use with the PortaCount Pro+ Respirator Fit Tester 8038 and then coupled via 3/16" vinyl tubing to the Portacount unit. As per CDC recommendations, a user seal check was performed prior to each fit test measurement and adjustments were made until the user deemed that there were no detectable air leaks during forceful breathing 10 .

Due to the limited availability of NIOSH test apparatus, a modified version of the standard NIOSH procedure (TEB-APR-STP-0059) was developed to determine the particulate filter efficiency of N95 FFRs (41). Respirator filters were challenged by a NaCl aerosol generated using a humidifier (Vicks© Filter-Free Cool Mist Humidifier) filled with 2% saline solution. Non-neutralized particles were released into a polycarbonate test chamber (~150 L volume), which was maintained at a minimum 0.3μm particle count of 5000 particles per ft 3 . Each FFR was manually secured over an iso-KF 25 fitting.

After ~10 cm of ~25 mm ID stainless steel tubing, another KF-25 to barbed fitting joined via 3/8" ID . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted January 6, 2021. ; https://doi.org/10.1101/2021.01.05.20248590 doi: medRxiv preprint silicone tubing connected to the both the calibrated particle counter (Extech Instruments VCP300) and a high-performance surgical smoke removal device used as an evacuator (Buffalo Filter VisiClear) through a "wye" fitting ( Fig 4A) . Pressure drop across the filter is monitored with a Sensirion SDP8-500 differential pressure sensor, and the air flow volume is monitored by a Sensirion Mass Flow Meter (part number SFM3300-D). Both devices are read through their I2C communications channel through a custom program running on an Arduino Uno and then recorded on a PC at rates over 10 Hz.

Tests were conducted at a flow rate of 10 lpm with a 4.54 cm 2 area of effective filtration, which exceeds the rate of 85 lpm used during NIOSH testing when our filter patch is scaled proportionally to the area of the entire mask 35 . We compensated for any variation in the impedance of the mask material by changing the strength of the suction of the evacuator to obtain a flow rate of 10 lpm through each mask or blank (nozzle open to allow for unimpeded ambient air flow) prior to each filtration test.

Before measuring the filtration efficiency of each decontaminated or control FFR, the baseline ambient particle count within the aerosol shield chamber was measured as the particle count downstream of an open nozzle. First, particles were suctioned through either a blank or mask at a flow rate of 10 lpm for 20 seconds, then the suction was switched off. Then, following a 5 second pre-counting sample draw, the cumulative number of same-sized particles downstream of each mask or blank were counted over a 20 second sampling period at a 2.83 lpm sampling rate controlled by the particle counter's internal pump. The total volume of the internal tubing, sensors, and stainless port was ~1.35 liters, and designed such that the bulk of the air sampled by the particle counter during its total 25 second counter operation was air drawn through the filter unit at the 10 lpm volume flow rate, not air drawn through the filter . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted January 6, 2021. ;  by the pump action of the particle counter. Particles binned into 0.3 μm, 0.5 μm, 1 μm, 2.5 μm, 5 μm, and 10 μm size classes were measured. The filtration efficiency (FE) was calculated using the equation,

where and ℎ denote the concentrations of particles of the same size downstream of the mask and within the aerosol chamber, respectively. For each FFR, flow test measurements of two different central N95 material areas were obtained to sample the variance.

To assess the sensitivity of our modified flow test, the filtration efficiency of an untreated control FFR before and after introducing damage was measured using the same experimental setup and methodology as described above. Filtration efficiency test sensitivity was determined in two mask areas that were punctured with an 18-gauge needle one to three times ( Fig S5, Table S8 ). Each puncture was at most 0.3% of the filter area, yet reduced filtering efficiency of the 0.3 um particles by roughly 25% (e.g., from >95% to 75%). This large reduction in efficiency from such a small leak highlights the importance of proper fit, with no voids, for effective protection. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted January 6, 2021. assigned to either (B) x round(s) of appliance-based heat treatment OR to a control group. We tested three appliance-based heat treatment methods: Cuckoo rice cooker, sous vide machine and boiling. Boiled FFRs were treated up to three times. FFRs heated with sous vide machine and Cuckoo rice cooker were treated up to five times. Control FFRs were not exposed to any treatment. (C) Treatmentexposed and control FFRs were fit tested using the built-in N95 setting of a TSI PortaCount Pro+ Respirator Fit Tester 8038. Three technical replicate measurements were taken for each FFR and overall Fit Factors (Fig 2) and fit values for the different testing categories (Fig 3) were recorded. (D) The effect of appliance-based heat treatment on FFR filtration performance was evaluated for treatment and control FFRs using a modified version of the standard NIOSH test (Fig 4) . Filtration efficiency was tested on two different FFR areas. For the Uniair San Huei SH3500 (TC-84A-4313) model, FFRs were only exposed to heat using the Cuckoo rice cooker and the sous vide machine. They were treated up to three times. Fit was estimated for both treatment and control FFRs as it is described in (C), the masks were not tested for filtration performance. After exposing FFRs to different methods and repetitions of heat-treatment, the fit factor was estimated with the PortaCountPro+. The fit factor was estimated three times for each FFR (smaller dots) and average fit factor is reported to the nearest integer (red bar). The control FFR was not subjected to any rounds of heat treatment. Grey regression line is overlaid for each treatment method (y ~ mx + b). Analysis of variance p-values is shown when there is significant variation in means among treatment group and control. Significance (*) p < 0.05. Photographs of the heat generation appliances shown in the panel below their respective fit factor profile. (Table S5 and S6). Analysis of variance p-values shown when there is significant variation in means among treatment category groups and control. Significance (*) p < 0.05. A humidifier confined by an aerosol shield test chamber was used as an aerosol generator. FFRs were attached to a sample holder and air and any suspended particles were drawn through the FFR by an evacuator suction pump and diverted to a particle counter. FFR impedance was measured through a pressure sensor attached to the sample holder. (B) The filtration efficiency for different particle sizes (bins of 0.3 µm, 0.5 µm, 1 µm, 2.5 µm, 5 µm, and 10 µm size classes) were determined for two mask regions of FFRs that were exposed to different methods and repetitions . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted January 6, 2021. ; https://doi.org/10.1101/2021.01.05.20248590 doi: medRxiv preprint of heat-treatment. None of the treatment methods tested reduces Halyard Fluidshield 3 FFR filtration performance below 95%. MAIN TABLE LEGENDS:  MAIN TABLE LEGENDS:   Table 1 . Effect of heat and humidity on enveloped viruses on N95 FFRs. Log viral reduction >3 is sufficient to consider the virus inactivated. N/A: Not Applicable, the parameter was not tested. DMEM refers to Dulbecco's Modified Eagle Medium. 

The Zuckerman Mind Brain Behavior Institute, Columbia University provided funding to support this project.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted January 6, 2021. The authors declare no competing interests.

All data is available in the main text or the supplementary materials.

• Supplementary Figs S1 -S5

• Supplementary Tables S1 -S8

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted January 6, 2021. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted January 6, 2021. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted January 6, 2021. ; Fig S3. Average FFR fit factor Uniair San Huei. After exposing FFRs to several methods and rounds of treatment, the fit factor was estimated with the PortaCountPro+. The fit factor was estimated three times for each FFR (smaller dots) and the average fit factor is reported to the nearest integer (red bar). The control FFR was not subjected to any round of treatment. Grey regression line is overlaid for each treatment method (y ~ mx + b) ( Table S2 ). Analysis of variance p-values is shown when there is significant variation in means among treatment group and control. Significance (*) p < 0.05.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted January 6, 2021. (Table S3 ). (B) Analysis of variance p-values shown when there is significant variation in means among treatment groups and control. Significance (*) p < 0.05.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted January 6, 2021. ; The filtration efficiency of two N95 mask areas for different particle sizes (bins of 0.3 µm, 0.5 µm, 1 µm, 2.5 µm, 5 µm, and 10 µm size classes) were determined for an untreated FFR before and after puncturing with an 18-gauge needle once, twice, and three times. (B) The front view of two FFR mask regions subject to punctures, fitted to the sample holder, and tested for filtration efficiency is shown.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted January 6, 2021. Table S1 . Average Halyard Fluidshield 3 FFR fit factor and technical replicates. After exposing FFRs to several methods and rounds of treatment, the fit factor was estimated with the PortaCountPro+. The fit factor was estimated three times for each FFRs and the average fit factor is reported to the nearest integer. Control FFR was not subjected to any round of treatment. Analysis of variance p-values shown for variation in means among treatment group and control. Values rounded to two decimal places. Grey regression line overlaid for each treatment method and category (y ~ mx + b) in Fig 2. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted January 6, 2021. Table S2 . Average FFR fit factor and technical replicates Uniair San Huei After exposing FFRs to several methods and rounds of treatment, the fit factor was estimated with the PortaCountPro+. The fit factor was estimated three times for each FFRs and the average fit factor is reported to the nearest integer. The control FFR was not subjected to any round of treatment. Analysis of variance p-values shown for variation in means among treatment group and control. Values rounded to two decimal places. Grey regression line overlaid for each treatment method and category (y ~ mx + b) in Fig S3. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted January 6, 2021. 2 +200 +200 +200 +200 +200 +200 +200

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted January 6, 2021. 3   +200  +200  78  +200  +200  75  +200   4   1  32  110  99  72  63  46  114   2  76  116  99  92  108  99  110   3  159  +200  +200  +200  +200  197  +200   5   1  68  45  77  123  99  199  43   2  37  135  56  77  133  110  108   3  57  62  51  57  71  59  63   Table S4 . Halyard Fluidshield 3 FFR fit for different testing categories. After exposing FFRs to several methods and rounds of treatment, the fit for different testing categories was estimated with the PortaCountPro+. The fit was estimated three times for each category. Control FFRs was not subjected to any round of treatment.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted January 6, 2021. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted January 6, 2021. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted January 6, 2021. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted January 6, 2021. Table S6 . Analysis of variance among treatment category groups and control Halyard Fluidshield 3 FFRs. Analysis of variance p-values variation in means among treatment groups and control.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted January 6, 2021. ; https://doi.org/10.1101/2021.01.05.20248590 doi: medRxiv preprint

Personal protective equipment during the coronavirus disease (COVID) 2019 pandemic -a narrative review

The Infectious Nature of Patient-Generated SARS-CoV-2 Aerosol

Personal protective equipment and intensive care unit healthcare worker safety in the COVID-19 era (PPE-SAFE): An international survey

Infection Control Guidance for Healthcare Professionals about Coronavirus (COVID-19)

Association between 2019-nCoV transmission and N95 respirator use

Prevention and Control Strategies for SARS-CoV-2 Infection

Coronavirus Disease 2019 (COVID-19): Epidemiology, Pathogenesis, Diagnosis, and Therapeutics

Recommended Guidance for Extended Use and Limited Reuse of N95 Filtering Facepiece Respirators in Healthcare Settings

Enforcement Policy for Face Masks and Respirators During the Coronavirus Disease (COVID-19) Public Health Emergency (Revised)

N95 Respirator Cleaning and Reuse Methods Proposed by the Inventor of the N95 Mask Material

COVID-19 Decontamination and Reuse of Filtering Facepiece Respirators

Comparison of UV C Light and Chemicals for Disinfection of Surfaces in Hospital Isolation Units

Heat and Humidity for Bioburden Reduction of N95 Filtering Facepiece Respirators

N95 Mask Decontamination using Standard Hospital Sterilization Technologies

Assessment of N95 respirator decontamination and re-use for SARS-CoV-2

Humidity and Deposition Solution Play a Critical Role in Virus Inactivation by Heat Treatment of N95 Respirators

Quantitative form and fit of N95 filtering facepiece respirators are retained and coronavirus surrogate is inactivated after heat treatments

Effectiveness of Three Decontamination Treatments against Influenza Virus Applied to Filtering Facepiece Respirators

A scalable method of applying heat and humidity for decontamination of N95 respirators during the COVID-19 crisis

Sterilization of disposable face masks by means of standardized dry and steam sterilization processes; an alternative in the fight against mask shortages due to COVID-19

Wall ovens and ranges

Residential Clothes Dryer Performance Under Timed and Automatic Cycle Termination Test Procedures

Dry heat and microwave generated steam protocols for the rapid decontamination of respiratory personal protective equipment in response to COVID-19-related shortages

Evaluation of Five Decontamination Methods for Filtering Facepiece Respirators

An Exploration into the Bacterial Community under Different Pasteurization Conditions during Substrate Preparation (Composting-Phase II) for Agaricus bisporus Cultivation

Aerosol Filtration Efficiency of Common Fabrics Used in Respiratory Cloth Masks