We tried to make an adapter for both 3M and Honeywell P100 filters. The 3M bayonet adapter is a simpler design that was fairly easy to print with FDM or SLA printers. However, the threaded Honeywell part has proven far more difficult. In PLA we basically gave up on this working. The threading precision is critical or there can be leaks between the adapter and filter. We are attempting to print it with an SLA resin printer (photo below), but still had issues passing the fit test.
The testing equipment consisted of a PortaCount 8030 Respirator Fit Tester (TSI, Minnesota, USA), Particle Generator 8026 Tester (TSI, Minnesota, USA), Surface Pro (Microsoft, Redmond, WA), Nellcor OxiMax N-65 Pulse Oximeter (Medtronic, Minnesota, USA) (Figure 1), and a gauge manometer (Instrumentation Industries, Inc., Pennsylvania, USA). The PortaCount 8030 was calibrated March 6, 2020 by TSI.
Figure 1. From left to right: A) pulse oximeter, B) SurfacePro, C) PortaCount 8030, D) ARIA QR+ snorkel mask, and E) particle generator 8026.
Testing was done in an approximately 9 ft. x 9 ft. room with the particle generator on. A 3M 6800 series (3M, St. Paul, MN) full-faced mask served as the benchmark.
Full face snorkel-mask
An Aria QR+ (Ocean Reef, Inc., California, USA) Medium/Large full-faced snorkel mask was tested since the design minimizes fogging problems and CO2 rebreathing.
A hole was drilled into the mask and a PortaCount grommet for testing N95 masks was inserted. Silicone was added to seal the external and internal surfaces. Crazy glue was used to secure and seal the contact between the metal grommet and plastic tube to reduce air leaks that can impair accurate quantification of the fit factor.
Filter
For filtration we selected a 3M P100 particulate filter with 99.97 % filter efficiency meeting NIOSH P100-series test criteria.
3D printed adapter
The adapter was a modification of the APA – Aria Protection Adapter available at the Ocean Reef website (https://oceanreefgroup.com/covid19/). The original adapter was designed to accept a 40 mm particulate air filter with 1/7 inch thread. The design was modified to accept two 3M P100 filters. We made an earlier prototype with a single filter but found it required too much breathing effort to use for long durations. The CAD design is available here: https://www.tinkercad.com/things/7dZnjwUKZRr The design reduced the printing material required compared to a vertical alignment of the filter.
The adapter prototype was a black polylactic acid (PLA) print. PLA prints are porous and without an airtight coating they will not work in an air filtration application. Two coats of XTC-3D (Smooth-On, Inc., Pennsylvania, USA) were applied to make it airtight. One 3M 3PRG7 (3M, St. Paul, MN) inhalation port gasket was placed at each of the two inhalation ports. Finally, one P100 filter was attached at each of the two inhalation ports. Sticky putty (Alcolin, Cape Town, South Africa) was used to ensure an airtight seal between the mask and the snorkel connector.
Fit testing
Fit testing was performed using the OSHA 29CFR1910. 134 protocol in the PortaCount 8030. The fit test exercises include normal breathing, deep breathing, turning the head side to side, moving the head up and down, talking, grimace, bending over, and normal breathing. Each test exercise lasts one minute except for the grimace exercise which is 15 seconds. A passing fit factor for full face masks like this design is 500, while for half face masks is 100. The fit factor is expressed as the challenge aerosol concentration outside the respirator divided by the challenge aerosol concentration that leaks inside the respirator during a fit test. Therefore, the higher the number the better.
Test scenarios
Daily QA for the PortaCount with the particle generator active was performed and passed. We tested the 6600 series 3M mask as our benchmark and three experimental configurations of the snorkel mask. These four experimental setups were named as follows:
Snorkel mask no modifications: This setup preserves the original configuration of the snorkel mask including the mushroom valve and protective cover in front of it.
Snorkel mask mouth cover removed: This setup preserves the mushroom valve, but removes the plastic protective cover in front of it as shown in the previous video.
The mushroom valve by the mouth was also removed. Two square pieces of duct tape covered the resulting opening to seal the port’s opening.
Results
3M 6800 series mask
The 6800 series 3M mask served as our benchmark and reached its maximum fit factor in 1:15 minutes. O2 Saturation remained stable. Real time test results can be seen here:
The 3M mask passed the fit test results with a fit factor of 333867 (see Table 1). Fogging or humidity were not an issue.
Snorkel mask duct tape
This configuration reached its maximum fit factor in approximately 2:11 minutes. O2 Saturation remained stable. Real time test results can be seen here:
This configuration passed the fit test results with a fit factor of 32281 (see Table 1). Increased humidity decreased the comfort of the mask although fogging was minor.
Snorkel mask no modifications
This configuration reached its maximum fit factor in approximately 1:56 minutes. O2 Saturation remained stable. Real time test results can be seen here:
This configuration passed the fit test results with a fit factor of 15448 (see Table 1). Fogging or humidity were not an issue. Further evaluation of this configuration generated during inspiration a negative pressure of 2 cm of water and no positive pressure during exhalation.
Snorkel mask mouth cover removed
This configuration reached its maximum fit factor in approximately 1:08 minutes. O2 Saturation remained stable. Real time test results can be seen here:
This configuration passed the fit test results with a fit factor of 1105 (see Table 1). Fogging or humidity were not an issue.
Table 1. Fit testing results for the different mask configurations
3M 6800 series mask
Snorkel mask duct tape
Snorkel mask no modifications
Snorkel mask mouth cover removed
Normal breathing
297913
44910
18511
1123
Deep breathing
288127
40394
27612
1412
Head side to side
442017
63628
38360
1541
Head up and down
349546
66749
63259
1234
Talking
266729
21200
7933
1953
Bending over
326150
12676
7534
929
Normal breathing
454406
76511
16065
593
Overall fit factor
333867
32281
15448
1105
Fit Test*
Pass
Pass
Pass
Pass
*OSHA fit factor passing value is 500 or greater
Subjective user experience
A radiation therapist wore the non-modified snorkel mask from 9 AM to 3 PM while performing daily work activities which require an increased level of exertion while positioning and moving patients to the treatment couch of the linear accelerator. She took off the mask for lunch and for a break to have a drink. She treated between 25 and 30 patients that day. Visibility was great and comfort better than other PPE she had used. Near the end of the daily with increased physical activity she felt some breathing discomfort. She measured the O2 sat at the time which was 100 %. Overall her feedback was that it was a comfortable option she could tolerate for prolonged periods of time. The only negative feedback was that the patients and the other therapists had a hard time hearing her, requiring speaking up or using hand gestures. Talking on the phone revealed no issues.
We adapted the dual design for two Honeywell North filters, or one 22mm ventilator filter. We have only performed quantitative testing with the 3M design yet but will update this log with the results once we've done the others. All three designs are here. We tested the 3M design printed in PLA and coated in XTC 3D coating. We're also printing in e-Rigid PU resin on an EnvisionONE printer (EnvisionTEC). This is a semi-flexible airtight material that should be able to be used without coating. We will confirm and post back after quantitative testing, but prints look very nice!
Here is 3M P100 adapter: And the threaded Honeywell. Unfortunately the threading on this has not worked out for us - we will keep tweaking the design but we think it may be difficult to print the threading precise enough. One side is working but the other is not threading onto the filter well. This was printed on the Envision ONE which has pretty high resolution. We also tried in PLA on an FDM printer with abysmal results :)
We next made a dual filter design to increase air-flow and enable easier use of the mask. We chose a 3M P100 particulate filter with 99.97 % filter efficiency meeting NIOSH P100-series test criteria. While this is an excellent filter, it has become hard to find so we will look into adapting our design for other filters.
We modified the OceanReef adapter (found here) in TinkerCAD to adapt it to two 3M P100 filters. Editable file is here. We designed this modification with a flat base to enable printing with almost zero supports. We printed in both PLA with an FDM printer and resin with a SLA printer and in both cases the flat base helped produce high quality prints. For PLA, we set the infill density to 100% but the print was still porous and needed to be coated (we used Smooth-On XTC) to achieve an air-tight seal. Breathing through this adapter with both 3M P100 filters installed was much better than our single filter model so we moved forward with quantitative testing. For quantitative testing we printed the adapter in black PLA. NOTE: PLA prints are porous and without an airtight coating they will not work for this application! We coated the print in 2 coats of XTC-3D (Smooth-On, Inc., Pennsylvania, USA) to make it airtight. One 3M 3PRG7 (3M, St. Paul, MN) inhalation port gasket was placed at each of the two inhalation ports. Finally, one P100 filter was attached at each of the two inhalation ports. Sticky putty (Alcolin, Cape Town, South Africa) was also used to ensure an airtight seal between the mask and the snorkel connector.
Ideally, we would use a different process to create this adapter in a material that can form a perfect seal without the putty. However, given the ubiquitousness of FDM 3D printers and ease of printing in PLA, we thought it useful to present results from our testing in this material, with putty to achieve the air-tight seal. Here is a photo of the coated test adapter installed on a mask with putty:
Fit testing was performed using the OSHA 29CFR1910. 134 protocol in the PortaCount 8030. Here is the equipment used in the test:
The fit test exercises include normal breathing, deep breathing, turning the head side to side, moving the head up and down, talking, grimace, bending over, and normal breathing. Each test exercise lasts one minute except for the grimace exercise which is 15 seconds. A passing fit factor for full face masks like this design is 500, while for half face masks is 100.
The 6800 series 3M mask served as our benchmark and reached its maximum fit factor in 1:15 minutes. O2 Saturation remained stable. Real time test results can be seen here:
The 3M mask passed the fit test results with a fit factor of 333867. Fogging or humidity were not an issue.
In our test condition we removed the exhalation mushroom valve and plastic cover from the mask and covered the resulting hole with duct tape. This makes the mask less comfortable but removes the possibility of an infected mask user inadvertently exhaling viral particles.
This configuration reached its maximum fit factor in approximately 2:11 minutes. O2 Saturation remained stable. Real time test results can be seen here:
This configuration passed the fit test results with a fit factor of 32281. Increased humidity decreased the comfort of the mask although fogging was minor.
Subjective user experience
A radiation therapist wore the mask from 9 AM to 3 PM while performing daily work activities which require an increased level of exertion while positioning and moving patients to the treatment couch of the linear accelerator. She took off the mask for lunch and for a break to have a drink. Visibility was great and comfort better than other PPE she had used. Near the end of the daily with increased physical activity she felt some breathing discomfort. She measured the O2 sat at the time which was 100 %. Overall her feedback was that it was a comfortable option she could tolerate for prolonged periods of time.
Conclusion:
The modified full-faced snorkel mask tested solved two critical PPE problems in the current COVID-19 crisis: eye and face protection, and high quality air filtration to protect against SARS-CoV-2. The solution exceeded the OSHA requirements for a full faced mask in quantitative testing, and can be another alternative in the current climate of PPE shortage. We release the adapter design as open-source so users are able to modify this design to fit filters from different manufacturers (as we did from the Ocean Reef design), depending on the current supply.
We next modified the adapter provided by the Ocean Reef group (more info on that adapter can be found here) to include a single 3M P100 filter:
We printed this in PLA to test the fit.
While the fit was good, a test user found it took too much effort to breathe through a single filter (this is likely why most commercially respirators use two filters). We decided to adopt a 2 filter design for our next test.
Our first attempt to adapt a filter to the Ocean Reef mask was to use a HEPA filter from a vacuum cleaner. This was attractive as these filters are cheap, widely available, and supply does not appear limited at the moment. We modified the adapter provided by the Ocean Reef group, more info on that adapter can be found here.
We provide our design notes below for the benefit of documentation, but unfortunately we were not able to achieve adequate filtration with this approach. We believe the issue may have been the leaks between the walls of the filter and our adapter, although it is hard to be certain.
Photos of filter:
First design (printed in eRigidForm resin on EnvisionONE printer)
Non-quantitative N95 quality testing (it passed this "smell" test)
2nd smaller design:
This design was tested at Washington University in St Louis and unfortunately it failed to achieve the filtration level of a P95 mask. We abandoned this approach at this point and moved to testing 3M P100 filters (below). That said, there are supply issues with obtaining the 3M filters at the moment so we may revisit this design in the future.