Modified full-face snorkel masks as reusable personal protective equipment for hospital personnel

Here we adapt and evaluate a full-face snorkel mask for use as personal protective equipment (PPE) for health care workers, who lack appropriate alternatives during the COVID-19 crisis in the spring of 2020. The design (referred to as Pneumask) consists of a custom snorkel-specific adapter that couples the snorkel-port of the mask to a rated filter (either a medical-grade ventilator inline filter or an industrial filter). This design has been tested for the sealing capability of the mask, filter performance, CO2 buildup and clinical usability. These tests found the Pneumask capable of forming a seal that exceeds the standards required for half-face respirators or N95 respirators. Filter testing indicates a range of options with varying performance depending on the quality of filter selected, but with typical filter performance exceeding or comparable to the N95 standard. CO2 buildup was found to be roughly equivalent to levels found in half-face elastomeric respirators in literature. Clinical usability tests indicate sufficient visibility and, while speaking is somewhat muffled, this can be addressed via amplification (Bluetooth voice relay to cell phone speakers through an app) in noisy environments. We present guidance on the assembly, usage (donning and doffing) and decontamination protocols. The benefit of the Pneumask as PPE is that it is reusable for longer periods than typical disposable N95 respirators, as the snorkel mask can withstand rigorous decontamination protocols (that are standard to regular elastomeric respirators). With the dire worldwide shortage of PPE for medical personnel, our conclusions on the performance and efficacy of Pneumask as an N95-alternative technology are cautiously optimistic.

The primary function of respiratory PPE is to protect the wearer from exposure to pollutants present in air, specifically in this case from particles exhaled/sneezed/coughed by an infected individual. The residual exposure of the wearer depends on three independent additive components: the leak at the face, the penetration through the filter and the internal contamination.
Residual Exposure = leak at the face + penetration through the filter + internal contamination (1) The leak at the face depends on how well the mask forms a seal with the wearer face, or said differently, how well the mask fits the wearer face. The fit is clearly dependent on the mask shape and the morphology of the individual face, and should be determined systematically for each wearer with each of the mask models used. As the origin of the leaks is a breakthrough in the sealing, the fit is considered independent from the nature of the pollutant.
The penetration through the filter depends on the efficiency of the filtering material to remove particles from the air. The filtration efficiency is clearly a characteristic of the filter material and the nature of the particle, and is independent from the individual and from the mask shape. The filtration efficiency has to be measured under normalized conditions, or at least with well characterized particles corresponding to the pollutant the wearer has to be protected from. Filters with less than 99% filtration efficiency at inhalation flow rates will interfere with fit tests and can result in a false fit test failure. This problem can be addressed, in quantitative fit testing, by testing masks with lower filtration efficiencies in the N95 mode available in some TSI PortaCount machines.
The internal contamination is mainly due to an inadequate maintenance of the mask, but could be significantly reduced by wearer training, adapted maintenance, and storage protocols (such as sterilization).
Currently, the particles exhaled by the mask wearer have the same size range as those generated by the patients, and which the mask should protect against. The global protection factor of the mask cannot be determined while the mask is being worn, as the wearer-exhaled particles could be misinterpreted as a "leak".
Thus, for multiple scientific reasons, the fit and the filtration efficiency must be determined separately. The exposure of the wearer will be considered adequate when both fit and filtration efficiency criteria are respected.
For their intended purpose, the seal tests of snorkel masks are done by the manufacturing companies underwater. However, the sealing ability of the snorkel masks on dry skin is unknown. Per CDC and NIOSH regulations on the use of elastomeric respirators, a fit test can be performed in the same manner as N95 respirators to ensure seal and safety to use for an individual [19]. At this time, we recommend that all practitioners seeking to utilize these masks perform a fit test under standard N95 fit test conditions.
In addition to this recommendation, additional fit test experiments have been performed in our laboratories. Practically, two types of fit test can be conducted: • Qualitative fit test: a liquid aerosol with a sweaty or bitter taste are generated within a confinement around the head of the mask wearer. The result of the fit test is based on the detection of the taste under the mask.
• Quantitative fit test: the method is based on a particle counting outside and inside the mask in parallel using the TSI PortaCount device. The ratio out/in gives the fit factor.
As explained above, the fit test is only meant to measure the ability of the mask to form a seal with the wearer's face, and not the efficiency of the respiratory protection. The principle of this test consists of successive measurements of the particulates concentrations inside and outside of the mask during normalized exercises. The ratio between the external and the internal concentrations is called fit factor (FF). The relevancy of the results is dependent on a few assumption: • The efficiency of the filter is high enough to assure that the particle penetration rate is insignificant compared to the expected leak rate, within the range of the measured particle size (0.015 − 1 µm). P3, N100 or HEPA filters are generally required to reach this specification (filtration rate > 99.95 % at 0.3 µm), with a theoretical FF > 2000 in case of perfect fit. For filters with significant penetration rate (P2 or N95), the measure is based on a smaller subset of particle sizes (around 0.04 µm) using the N95 protocol to avoid the counting of the filter-penetrating particles.
• The range of the particle size measures by the PortaCount (0.015 − 1 µm) has been selected to stay mostly outside of the range of particulates generated by human exhalation, apart from smokers (it is inadvisable for smokers to perform these tests, but at a minimum, they should not have smoked within 30 minutes of testing). This is necessary because the particulates generated by mask wearer would otherwise be misinterpreted as a leak into the mask.
• The ambient particle count exterior to the mask, in the particle size range measured by the machine, must be significantly higher than the quantity of particles that could be generated by the wearer by any method. Quantitative fit testing units such as the TSI PortaCount are programmed to abort testing if the ambient particle count decreases under a minimum level.
Practically, the quantitative fit test will not measure only the leaks at the wearer's face, but also any leaks in relation with the connection after the filter or with the exhaust valve. In this way, high fit factors reflect both that the leakage at the wearer's face is acceptably low (wearer dependent), but also that the residual leaks at the level of the adapter and the chin valve are acceptable as well (wearer independent).
Here we present quantitative fit test results. We will first discuss the multiple tests run at Stanford University (through multiple groups) on the Dolfino Frontier Mask. We will then review the results both from Stanford and from EPFL on the Subea Decathlon mask. Both mask models passed quantitative fit testing with fit factors that meet the industry-standard threshold for typical half-face elastomeric respirators.

Stanford Prakash Lab Testing: Experiment Parameters
• Fit measured using a PortaCount Pro+ • 4 candles to provide a constant ambient particle source • Test Subject: 28 year old male, no facial hair, no history of smoking • HME HEPA filter (Pall Ultipor 25) • Snorkel Mask (Dolfino or Decathlon) • Modified adapter from Formlabs high temperature resin In order to collect a thorough measurement, each test was repeated in three locations, sampling from the mouth chamber, the eye chamber, and directly from the adapter, right after the filter. This required building two modified adapters. The adapter which sampled directly after the filter was modified by drilling a hole in the adapter between the filter connection port and the mask port. The drilled hole was cleaned and smoothed before a luer lock connector was press fit into the hole and sealed using epoxy. The epoxy was allowed to cure for 24 hours and the assembly was subsequently washed with isopropyl alcohol for 2 minutes. The adapter which sampled from the eye chamber and mouth chamber used a flexible tube to sample air from a desired location. This adapter was modified by drilling a hole large enough for the flexible tube to pass through. The hole was once again cleaned and a luer lock connector was pressed into the tube. The luer lock connector and tube were press fit into the drilled hole and sealed using epoxy. The epoxy was allowed to set for 24 hours before the entire assembly was washed for 2 minutes in isopropyl alcohol. Both adapters were allowed to dry completely before any tests were conducted.  Fit factor results when the PortaCount sample tube was connected to the mouth chamber, eye chamber, and inlet port of a Pneumask. This mask consisted of a Dolfino mask connected, via an adapter, to a HEPA rated mechanical HME filter (Pall Ultipor 25). These tests were completed in half-face respirator mode and the mask passed in all three cases.
Further tests were completed independently at Stanford Occupational Health and Safety which confirmed that the Dolfino mask passes the quantitative fit test. Two additional fit tests were conducted at Stanford Environmental, Health and Safety -one completed using the requirement of a half-face elastomeric respirator and another using the fit factor for a full-face tight fitting Air Purifying Respirator, and the results are shown in Figure 4. The participant (female) wearing Dolfino Frontier with a custom adapter, and a Pall Ultipor 25 breathing filter, who is typically a size M, was still able to pass the quantitative fit test well beyond the requirements on both tests. The minimum passing fit factor was 100 for half-face respirator and 500 for a full-face respirator. The activities that were tested while wearing the mask included bending over (50 seconds), jogging in place (30 seconds), moving head side to side (30 seconds), and moving head up and down (30 seconds), and the fit factor outcomes were all above 750. Please note that, although the test yielded positive results, this was conducted with a limited testing sample and does not yet indicate any certification/endorsement from EH&S of the product. Each entity should also conduct its own evaluation and testing before use.

5/27
Subea Decathlon Experimental Results Testing was undertaken both at Stanford University and at EPFL with quantitative fit testing for the Subea Decathlon Mask. Below are the results from our tests at Stanford.

Figure 4
Fit factor results when the PortaCount sample tube was connected to the mouth chamber and inlet port of a Pneumask. This mask consisted of a Decathlon mask connected, via an adapter, to a HEPA rated mechanical HME filter (Pall Ultipor 25). These tests were completed in half-face respirator mode and the mask passed in all cases. Important Note: The adapters used in the above test did not fit the Subea mask as well as the Dolfino mask. Tape had to be used to form a better seal. Thus these results should be interpreted as a lower bound on the sealing capabilities of the mask.
Validation of the fit of the Decathlon mask was completed in a separate series of experiments at EPFL, also using a PortaCount Pro+ in N95 mode and following the OSHA 29CFR1910.134 protocol. The Decathlon EasyBreath mask was connected to a medical grade HME filter (DAR adult-pediatric electrostatic filter HME, small) with a 3D-printed PLA connector. The mask was in Pneumask-G configuration (3 snorkel ports connected to the filter, chin valve non modified). The mask was connected through the silicon skirt of the eyes chamber, as indicated in Figure 1f, using the standard connector sold by the manufacturer of the particle counter. In N95 mode, the test was positive for the two individuals (men, freshly shaved) tested, with a fit factor of 200+, which is higher than the requirement for half-masks (100). Removing the chamber valves to connect permanently the eye chamber and the mouth chamber led to the same results. This test was run in N95 mode because the test was completed using an HME filter which was not HEPA rated (fit factors around 4 would have been obtained with the N100 normal protocol under the same testing conditions). The full results from EPFL testing are shown in Figure 5. Fit factor test results from EPFL (translated from French). Note that we have identified a common testing issue mistake -you cannot use the default N100 mode unless the respirator or HME filter is rated above 99% at respiratory flow rates.

7/27
Summary of Fit-Test Findings The successful results for the fit test with the different individuals used in this study indicate that the Dolfino and Decathlon masks both form acceptable seals, showing also that the custom adapter and the chin valve do not generate significant leaks. The measured fit factors correspond to the requirement for elastomeric half-mask. The remaining performance of the mask depends on the efficiency of the filter, which is attached to the mask.
The position of the sampling point gives similar results between the mouth and the eyes chambers. However, a sampling point directly connected on the custom adapter shows significant higher fit factors, which seem over evaluated, probably due to the proximity of the filter. In this case, the measured particle concentration should not be relevant of the real concentration in the breathed air.
The use of the N95 protocol of the PortaCount is important for the fit evaluation when HEPA filters are not available, especially with HME virus filters. The efficiency of the filters should be measured independently of the PortaCount system to assure safe working conditions.

Details about Filtration Efficiency Testing
We developed an simple experimental test rig and method for testing the particle filtration efficiency of various materials. Please note this setup is not the standard testing method which typically uses the TSI Automated filter tester 8130A. The setup pictured in Fig. 8 includes a LightHouse handheld particle counter (Model 3016 IAQ), Intex QuickFil 6C Battery Pump, a rubber stopper with 2 holes covered by 2 kim wipes to mitigate the airflow, Incense: Satya Sai Baba Nag Champa 100 Gram, connectors (universal cuff adaptor, teleflex multi-adaptor), and filters to test (Hudson RCI Main Flow Bacterial/Viral Filter, Romsons HME Disposable Bacterial Viral Filter, Pall Ultipor 25 filter). The pump with the rubber stopper, covered by 2 kim wipes, in it, provides an airflow within a range of 5.6 -11.32 l/min to mimic that of breathing. The incense produces particles of various sizes, including those in the range picked up by the detector (0.3 µm -10 µm). With the pump on, we measure the number of particles produced by the incense. Then we place the filter on the setup and run the particle counter to measure the number of unfiltered particles. To calculate the filtration efficiency, we calculate the ratio of unfiltered particles to the number of particles produced by the incense, and then subtract from one. The filter efficiencies for the 3 filters tested are reported in Figure 9. We constructed an experimental system for measuring the pressure drop across various materials, including N95 masks, during inhalation and exhalation. The setup in Figure 10 includes an Intex QuickFil 6C Battery Pump, a Honeywell AWM700 Airflow sensor, a Honeywell ABPDLNN100MG2A3 Pressure Sensor,a rubber stopper with 2 holes covered by 2 kim wipes to mitigate the airflow, connectors (universal cuff adaptor, teleflex multi-adaptor), and filters to test (Hudson RCI Main Flow Bacterial/Viral Filter, Romsons HME Disposable Bacterial Viral Filter, Pall Ultipor 25 filter). The pump with the rubber stopper, covered by 2 kim wipes, provides an inhalation or exhalation airflow within a range of 0.2-0.4 cfm to mimic that of breathing. With the pump on, we measure the airflow applied to the mask, and the differential pressure drop across the mask. The pressure drops for the 3 filters tested are reported in Figure 9.

Summary of Filtration Testing Findings
We have found that the Decathlon Subea masks and the Dolfino masks are both capable of forming a seal that exceeds the standards required for half-face respirators and N95 masks (fit factor >100). The masks must still be properly secured and sized appropriately for the wearer with the fit verified according to the standards of the institution where the PPE is being used. The sealing capabilities of these masks were tested using a TSI PortaCount Pro+ (in half-face mode using the OSHA standard) on a system that consisted of the mask, a custom adapter to connect the mask and filter, and a HEPA rated HME filter. The custom adapter was modified to include a sampling port to which the TSI PortaCount Pro+ could be connected, allowing measurements at three different places inside of the mask (next to the filter, in the eye space, and in the mouth space). The particle concentrations in all parts of both masks were found to be less than 1 part in 100 relative to the ambient particle concentration (Fit factor of >100). The Decathlon Subea mask was also tested at EPFL, again with a TSI PortaCount Pro+, on a system consisting of a Decathlon mask, a custom adapter, and an electrostatic HME filter. The test was run by connecting the PortaCount Pro+ to a port which was installed in the rubber siding of the mask next to the eye chamber. The test was run using an OSHA standard in N95 mode (as required by the PortaCount Pro+ for filters with <99% efficiency). The PortaCount Pro+ reported a fit factor of >200 (a particle count of less than 1 part in 200 relative to ambient conditions) for two different wearers. Repeating the test with the eye chamber and mouth chamber directly connected to allow bidirectional airflow between the two chamber resulted in the same fit factor. The sealing capability of both the Dolfino and Decathlon masks has been shown to exceed the standards for half-face respirators and N95 respirators. These tests were verified at multiple locations within Stanford and at EPFL using different masks, wearers, adapters, filters, PortaCount machines, and machine operators.

9/27
Figure 10 Setup for measuring pressure drop across filter.

Theoretical Estimation of Chin Valve Closure Time
Aside from the above testing, we have also done the following calculation to see how long the chin valve takes to close ( Figure  11A, Figure 12), assuming standard exhale -to assess the likelihood of localized backflow. The following figure ( Figure 11B) shows the schematic of a circular chin valve, which is pinned at the center. The valve is assumed to open from the bottom side, moving from vertical position to an angled location after exhalation of air. Forces acting on the valve include the force due to the gauge exhale pressure, (P ex,g = Pex − P atm ), and elastic forces. The elastic force occurs to return the valve to its original shape, this force is simply modeled by a linear spring formula. These forces should balance for a static valve at (2): where A and k denote the cross-section area and the elasticity constant of silicon, respectively. Now, let's assume inhale starts at time t=0 and the valve is in the angled position (2) shown in Figure 11B, α = P ex,g A kr At each time instant, in addition to the elastic force, the pressure forces due to the inhale, P in , and atmospheric pressure are acting on the valve: On the other hand: And therefore: This equation is a second-order ODE with boundary-conditions: θ = α and dθ dt = 0 at t = 0. This leads to the solution: Accordingly, the time it takes for the valve to reach θ = 0 equals: The maximum closure time occurs when we assume there is no force due to pressure during inhalation, P in = P atm , and equals: In order to determine a time associated with this dynamics, we need to directly measure some of the properties of materials used in elastomeric valves. For an order of magnitude calculation, we use known numbers; for a silicon valve with density ρ = 2.3290g / cm 3 , thickness l = 0.35mm, and radius 13.5mm, the mass of the moving section is m = 0.233g.Assuming elasticity constant k = 1N/m, this formula suggests that the valve closes after approximately t max = 0.024sec.

Exercise Test
Currently, 3 exercise tests have been conducted by volunteers using the Pneumask-G configuration. In the first 2 tests, the volunteer was a 38 years old male, ASA 1 status, with weight of 83kg for a height of 1.80 m (2.03 m2 total body surface area by Mosteller formula). Both tests were performed at FiO2 of 21% (ambient inspired oxygen fraction at a barometric pressure of 1011 hPa).
In the first test, a Decathlon FreeBreath mask was used with a 1.6 cmH2O pressure drop, bacterial/viral filter with no HME (Heat and Moisture exchanger). The mask has been worn on a treadmill at maximum inclination, for 10 minutes, to measure the CO2 level during intense activity. For a constant running speed of 6mph, the inhaled CO2 remained below 2mmHg at all times, while the exhaled CO2 rises up to 48mmHg on peak physical effort.
The second test was performed by the same volunteer subject, in the same treadmill machine, under same general conditions for 1 hour, with the Pneumask-G configuration but with a HMEF filter with a pressure drop of 4.5 cmH2O, which is almost 3 times higher than the filter used for the first test. For the second test, we monitored heart rate, SpO2, non-invasive blood pressure (NIBP), ECG, End-tidal CO2, Inspiratory CO2, FiO2, as well as a number of subjective measures including discomfort and stamina. The results of this test are summarized in Table S??. These results indicate that the change in Inspiratory CO2 throughout use of the device, in exertion that simulates that of most healthcare work, is negligible and in line with NIOSH standards [19]. Subjective comfort/discomfort was rated from 1 of complete discomfort to 10 of complete comfort. It is notable that this never fell below a rating of a 7. Further, the volunteer had an appropriate HR response for the level of exertion and no further alterations in physiological processes were noted. This indicates the device performs similar to elastomeric respirators under near identical conditions.

Donning and Doffing Procedures
We have developed suggested donning and doffing procedures based on the recommendation of UCSF [26] and from Stanford and UCSF feedback on our prototypes. A set of suggested procedures is below (Further review of this protocol by infection disease specialists or EHS officers in hospitals is required).

Donning Procedures
1. Eat, drink and go to the restroom. Check everything that you need to bring into the patient's room is available. 11. Put on the mask fully. Adjust and tighten the straps as needed.
12. Inhaling or exhaling while closing off the top of the filter with your hand can be done for another seal check. With a good facial seal, the mask should suck down on your face with a good seal with inspiration, and not lift off face with exhalation.
14. Place your gloved thumbs through the loops of the gown. Put on the second layer of gloves.

Doffing Procedures
1. In the anteroom (or the doorway if there is no anteroom), grasp the gown and pull away from your body. Carefully fold the gown into ball-shape and only touch the outer surface of the gown with gloved hands. Remove the gown and the outer gloves together. Dispose the gown and the gloves into a dedicated trash can.
2. Perform hand hygiene (you should still be wearing your inner layer of gloves).

12/27
5. Inspect the integrity of the inner glove. If the inner glove is intact without gross contamination, proceed through the next step. If the inner glove is torn, broken, or grossly contaminated, remove the inner glove. Perform hand hygiene, and put on a new pair of gloves.
6. Lean head forward into sniff position. Loosen the straps and remove the mask without touching your face.
7. Use EPA-approved alcohol-based wipe, bleach wipe or hydrogen peroxide wipe to wipe the outer surface of the mask, adapter and filter. Dispose the wipe. The required contact time for different EPA-approved wipes can be found in Appendix A Table S5.
8. Use EPA-approved alcohol-based wipe, bleach wipe or hydrogen peroxide wipe to wipe the inner surface edges of the mask. Dispose the wipe.
9. Gently disconnect the filter from the mask. Dispose the filter if it is designed for one-time usage. If you plan to reuse the filter, follow the instructions from the manufacturer of the filter to store it properly, and DO NOT WET the surface of the filter.
10. Put the mask into a clean dedicated box.
12. Remove the surgical cap.
14. Remove the inner gloves.
16. (Optional for Bluetooth user) Remove the ziplock bag which contains the Bluetooth microphone. Wipe the surface of the ziplock bag with EPA-approved wipe (Appendix A Table 5) and open the ziplock bag. Perform hand hygiene. Dump the microphone from the ziplock bag with one hand (dirty hand) onto the other hand (clean hand). Dispose the ziplock bag with your dirty hand. Place the microphone somewhere you will not forget with your clean hand. Perform hand hygiene.
17. By the end of the shift, bring the box and the mask to somewhere you can fully decontaminate it following our decontamination protocol, or hand it to hospital technicians per hospital policy. The recommendation of using disinfection wipe between patients and fully washing it after one shift is following the recommendation of NIOSH on elastomeric respirator.

Current Android app download instructions
1. Download the beta version of the app by visiting http://kylecombes.com/app-debug.apk on your phone.
2. Open the download. Your phone should prompt you to allow installations from unknown (i.e. non-Google Play) sources. If it does, simply follow the on-screen instructions. If no such prompt appears and the installation merely fails, do the following: • Allow app installations from unknown sources by going to your phone's Settings -> Security -> Unknown Sources and enable installations. This setting might have a different name, depending on your version of Android and device manufacturer.
• Try to install the app again by re-opening the download or revisiting the link above.

App usage instructions
1. Connect your phone to a Bluetooth device with a microphone. We recommend using cheap Bluetooth headphones, which can be put inside a sterilized plastic bag, and placed inside the face compartment of the snorkel mask.
2. (Optional, for extra amplification) Connect an external speaker to the phone's wired headphone jack using an aux cord. If you do not do this, the phone will use its internal speakers.

13/27
4. If the app detects a Bluetooth device capable of streaming audio, it will say "Bluetooth ready." If it does not, follow the on-screen instructions to ensure your device is connected properly. 5. Click "Start" to begin relaying audio from the microphone to the speaker.
6. You may now turn off the phone's screen or switch to another app if you would like. If you need to switch speakers at any point, click "Stop," do any plugging/unplugging, and then click "Start." 7. Click "Stop" end relaying audio.

Troubleshooting
If the sound is playing out of the Bluetooth speaker instead of through the wired speaker, simply unplug and replug in the wired speaker. Currently, the iOS version of this app is under development, and the Android app is pending publication in the Google Play Store. Please reach out to the provided contacts if you have any feedback, questions, or concerns.

Decontamination Protocol Testing
Due to the cost and design of the full-face snorkel mask, sustainable use requires that the mask itself be reused. Thus, it will need to survive common decontamination procedures such as autoclaving or immersion in a bath of bleach or ethanol. We have performed preliminary tests in which we subjected the Dolfino Frontier mask to the conditions involved in common decontamination procedures; the mask is stated by the manufacturer to consist of either silicone or thermoplastic rubber and polycarbonate lenses.
We have developed and tested our decontamination protocols based on recommendations from the CDC [21], OSHA decontamination protocols for respirators [7], and the consensus of National Academy of Science on reusable elastomeric respirators (p. 76) [8]. From these guidelines, a simple approach could potentially be the combined usage of detergent and bleach to achieve decontamination of the snorkel masks. Besides sodium hypochlorite, there are other hospital-used disinfectants that meet the EPA's criteria for use against SARS-CoV-2 [9] or CDC guidelines on chemical disinfectant use [10]. Among these, some hydrogen peroxide solutions, such as Accelerated Hydrogen Peroxide, offer the advantage of potentially being less harmful to the user and equipment, while only requiring a short contact time of just a few minutes. Ethylene Oxide sterilization is another commonly used method to disinfect heat sensitive equipment [11]; however, it requires specialized equipment and facilities, and whether access to such services, with the required turnaround time, is widely available to health institutions needs to be determined.

Autoclaving
We first performed a preliminary test to check whether mask functionality survives over the course of multiple autoclaving cycles. Before autoclaving the mask, we first took reference photos of its condition ( Figure S2). We then autoclaved the mask using a 30 minute gravity cycle at 121 deg C and 15 psi, with 10 minutes of warm-up before sterilization and 30 minutes of drying afterwards. Afterwards there was a mild "hot plastic smell". Small scratches were found upon visual inspection of the black plastic material. After letting the mask rest for at least 30 minutes to cool down, we again autoclaved the mask for another identical 30 minute gravity cycle. The mask survived both cycles of autoclaving without damage. Finally, we again let the mask rest for at least 30 minutes to cool down and then autoclaved the mask for another identical 30 minute gravity cycle. After this third round of autoclaving, with a cumulative autoclaving time of 90 minutes, the silicone rubber of the mask strap and mask seal appeared to remain elastic and functional. Mask was worn after autoclaving with no apparent loss of function.

Bleach Immersion
Besides autoclaving, the mask may be immersed in a bath of bleach for decontamination [12]. Thus, we tested whether a mask could survive the relatively harsh chemical conditions of immersion in a bath of bleach. We immersed a new snorkel mask for 10 hours in a bath of 10% bleach. There was no apparent damage afterwards despite some white coating which can be easily washed off ( Figure S3). We thus concluded that our mask should be able to survive most bleach disinfection protocols used in the hospital [7].

Ethanol Immersion
Besides autoclaving and immersion in bleach, the mask may be immersed in a bath of ethanol for decontamination [12]. Thus, we tested whether a new mask could survive immersion in a bath of 70% ethanol for 10 hours ( Figure S4). No apparent damage was noted afterwards. (Note that we should always use 95% ethanol to make the 70% ethanol solution, since 100% ethanol may contain trace amounts of benzene which is carcinogenic.) 14/27

Stretch Test
With our three snorkel masks treated under three different decontamination conditions (3 cycles of autoclaving, 10 hours of bleach immersion, and 10 hours of ethanol immersion, respectively), we then performed a simple stretch test on the elastomer bands of each mask by holding the strap with both hands such that the thumbs touched each other at the tips, then pulling and qualitatively observing the separation. The straps for the ethanol-treated mask appeared to have stretched the most, while the straps for the bleach-treated mask appeared to have stretched the least. Nonetheless, all the masks are functional and seal well after all the cleaning processes.

Dry Heat at 65C
Several reports indicate that dry heat at 65 degrees C is capable of killing any viral particles [13,14]. Although we did not explicitly test this protocol, given the fact that the masks survived 121 degrees C in the autoclave for 30 minutes, we can safely infer that our mask will also survive a dry heat disinfection protocol.

Decontamination Summary -Guidelines
Based on above testing results and the recommendation from OSHA [7], we developed suggested protocols for cleaning and decontaminating our snorkel mask (Dolfino Frontier), which is available below. Please note that this protocol is not formally approved, and each hospital should consult their EH&S officers or infection disease specialists for a standard operating procedure. If you are using disinfectants other than bleach, please also check this table compiled by EPA for recommended cleaning time.
Another important note is that not all snorkel masks can tolerate the decontamination. For example, snorkel masks from Animdive, Tinmiu and Keystand cannot withstand the temperature of autoclave or industrial washers commonly used in OR. If you are using snorkel masks other than Dolfino Frontier, please perform appropriate testing before usage.
2. Put on a gown, gloves, and a protective mask.
3. Wipe the surface of the filter with 70% ethanol or hydrogen peroxide, and carefully remove the filter.
4. Discard the wipe as biohazardous waste. Discard the filter if it is intended for single-use only. If you plan to reuse the filter, follow the instructions from the manufacturer of the filter to store it properly, and DO NOT WET the surface of the filter. 10. Dry the mask and adapter with a clean lint-free cloth or allow to air dry.
11. Reassemble the mask and store in a clean space. 15/27

Failure Modes and Effects Analysis (FMEA)
We have also performed a failure modes and effects analysis on our Pneumask-G design. In this analysis, we first decomposed the product into different components and listed the primary functions of each component. We then analyzed what would happen if each component fail to serve their primary functions and how much negative impact it would bring. By considering the severity, chance of occurrence, and chance of detection of that malfunctioning scenario, we can compute a semi-quantitative score. By comparing the semi-quantitative score of each possible failure mode, we can identify the most important failure modes that require immediate action or improvement in design. The analysis suggested several points that may be useful for anyone to further build upon our system: (1) Using surgical hood or any additional coverage to protect the filter surface, lateral side of the mask and the straps from gross contamination may provide additional protection (2) Separating the airflow pathway for inhaled air and exhaled air as much as possible can further improve performance for the Pneumask-G design. Modification or blocking of the chin valve would require complete redesign of airflow pathway. (3) Use a single-usage filter if available. For filters designed for repetitive usage, follow the instructions and regulatory-approved extended use claims (EUC) of the manufacturers (4) Avoid prolonged usage if possible (5) Long-term durability of the adapter when subjected to many cleaning process cycles will likely be dependent on the exact material and manufacturing process. Please perform appropriate further failure testing to characterize usage lifetime of these adapters in the material that is used. (6) Instructions on cleaning should specifically mention the necessity of cleaning the valve, as discussed in step 5 of the decontamination protocol in the subsection above. (7) Use of voice amplifying system may help to minimize risk due to jaw movements (8) Encourage face washing after doffing. Tables   Table S1: List of commercially available respiratory filters which are initially designed for ventilators. Table S2: List of commercially available filters which are not initially designed for medical usage. More characterizations of the filtering abilities are required before using in a hospital setting. The star marks next to the product name indicate the ones Prakash Lab is planning to test. (N/A = not available.)  Table S5: List of EPA-certified wipes. It is ordered by required contact time. All listed products are approved for healthcare use. If a product qualifies for the emerging viral pathogen claim, it is effective against a harder-to-kill virus than human coronavirus. All products on this list meet EPA's criteria for use against SARS-CoV-2, including those marked as "No" in this column.