N95 Respirator With the Formlabs Form 2

I had a little fun with being a product development engineer (for +20-years) in the medical device market… and have a little free time now(big sigh…) so why not design a bespoke respirator for me and my wife. My wife, after all, is still seeing patients, as a chiropractor, they are considered essential for acute cases, here in MN.

Knocked off an N95 respirator… custom fit using a 3D scan of the wearer, then sized and model it using the digital anatomy for a perfect fit. I own an Einscan Pro, for those that are curious. This scan was in Handheld Rapid mode at it’s highest resolution for this mode. It’s detailed enough for the original data to show a blemish that had reared its ugly head. So more than detailed enough.

  • Replaceable N95 filter Media (readily available, since it’s not intended for respirators).
  • Washable/sanitizable mask and filter frame (Formlabs Durable Material
  • Formlabs Elastic (Shore 60A) face gasket.


How is the elastic working out? soft enough?

Almost. As designed… it’s very thin and well supported by the underlying substrate… but 60A is pretty good. In the end it works well. Seals comfortably with little strap pressure, which makes long term comfort more likely. I’ve had it on for up to 4.5-hours now, and no sore spots.

I love projects like this–thank you for posting this one.

May I ask how you printed the gasket–flat to tray? And how much post processing time did you invest to remove the residual touchpoint materials left on the mask and frame?


The gasket was nearly flat-to-tray, but did have a slight tilt to it… to follow standard orientation convention (though that tilt angle was shallow. 8-10 degrees). As you know more tilt is better, and I’ll be using more in any subsequent builds… though too much with elastic or flexible is REALLY bad. When compared to stiff materials, those two need to be flat with 1000 support columns (but small contact points).

The Durable parts were quite fast. Simple file, light sand with 220-grit wet-dry, and a short visit to the media blaster with glass bead. If you don’t know media blasting is the secret to awesome, fast, perfect finishes with SLA.

The Elastic part was not finished until adhered to the mask(so to be supported). CYA seems to work the best for adhesive. Once bonded, the mask was frozen, then the gasket… now hard and supported by the mask, was filed down. Frozen again, sanded… then frozen one last time and media blasted. All in All… I’d say 30-minutes total finish time including bonding the Elastic to the Durable.

The secret here… don’t try to finish Elastic or Flexible without freezing them first. It’s friggin magic! #yourewelcome #SchemeIncTradeSecret


What exact product is the filter media? Is it a round cartridge, or do you cut it from bulk fabric?

I love the freezing technique you described! It’s the first time I’ve learned of this technique, but it makes complete sense.

Am I seeing it right that the touchpoints on the grommet face away from the edge that comes into contact with the face? So is sanding that area a matter of esthetics more than function?

And a couple of geeky questions: do you think it would make sense to curve the front of the mask a bit to match (very broadly) the curve of the face rather than presenting a flat, circular front? And, if so, do you think the slight distortion affects the efficacy of the filter?


1 Like

@formlabs new video I came across from an active duty naval officer discussing printing masks via formlabs and some improvements he is looking for date as of 4/16/2020;
source: https://twitter.com/MakerSpaceUSNA

-also be mindful that when wearing a mask, one should be
clean shaven to create a proper seal.


I’m not sure what you mean by “the touch-points” but every surface was sanded smooth for cleaning purposes. Proper sterile protocol would demand you clean sanitize the mask when finished wearing. Toss the filter media, sanitize the mask reload the filter with a clean one. So in order to make cleaning easy… all surfaces are finished.

The front of the mask needs to be flat. It most definitely can not be a compoud-curved surface, because the filter media can not follow that shape. A planar curve would be fine, but the attachment technique for the cage would need to be changed from what I did. The technique I used for this, was using a bayonette fitting with a sloped slot, so the cage compresses the filter media to assure a seal.

@SchemeInc Can you share any info on the filter media?

I love the design, its a very nice bit of work. However are you certain the use of the resins that you are using in this application is safe for contact with the skin over a period of hours?

I see Formlabs does not claim anywhere that it is. From your post you talk of Formlabs Durable and Formlabs Elastic materials being used.

The safety data sheet of those resins shows this:



It would seem that the resins do in fact contain monomers. The fact is that the resin NEVER cures 100% when printed, even with aggressive post cure procedures there is a risk of uncured material being held in the print, hence the need for very good testing to establish the real risk in using those materials for the face mask.

These articles contains some information that seems relevant:

This is using two layers of 3M (I am from Minnesota after all) Filtrete 2200 furnace filter media. From what I can determine. The MPR 2200 media is equivalent to N77 (at a minimum)… so two layers is N95. The flow resistance of this setup is very comfortable. If you pant/breathe fast, you can feel resistance, but normal respiration is fine.

Of course I am not… anecdotally, I’ve had this on for 3-6 hours a day for two-weeks with no problems (in addition to four other people).

The Durable material does not touch the skin… just the Elastic. That said. It may be possible to find an approved (USP Class IV) low durometer (I’d actaully aim as low as 30A) U-shaped gasket to replace the custom one.

It’s a fine line between Devils-Advocate and Troll… ain’t it.

It’s a fine line between Devils-Advocate and Troll… ain’t it.

It certainly is - we both do work for our day job in this area so I guess we might have some ideas. So a question for you:

Interesting comments that you have posted about the filtration - From what I understand the N95 specification shows the filtration should remove at least 95% of particles of a size between 0.1 and 0.3 microns. If I have that wrong, I am sorry and to help my own understanding please show me where in the N95 specification it says otherwise.

The MPR 2200 filter specs state it should remove 69% of E1 particles. Now E1 particle size is 0.3 to 1.0 microns, so it removes 69% (in a single layer) of particles that are BIGGER than the N95 size.

It would seem fair to assume that in a single layer its not as efficent as an N95 spec filter - but how did you arrive at your thoughts that two layers would equal the same as an N95 spec filter medium.

Just a thought though - A HEPA filter medium is normally said to filter 99.97% of particles down to 0.3um size - why not use that as an alternative…

It’s actually 0.3-micron and above. This goes for N100 (or true-hepa) also. You need to go to ULPA (better than HEPA) to get down to 0.12-microns.

As far as the math. 77% leaves up to 23% of the particulate available for pass two. 77% of 23 is ~18… 77+18 = 95. If your 69% is correct (the number I found was 77%)… then you end up at N90.

HEPA was attampted. WAAAYYYY too much resistance given the surface area available. :smiley: Not old did I sound like BANE… I passed out (j/k)

I pondered playing with one that could use the HEPA media, but I’m not completely unemployed at the moment… some clients are still moving on projects. :wink:

I agree with your thoughts regarding flow rates, You might be able to get a Glass media HEPA filter with
borosilicate microfibre, which does have a good flow rate - is that what you tried?

With regards clients - its gone crazy busy here!!

I love the maths - but wish it was that simple, Perhaps this guy has other ideas - To quote from his book on the subject

I don’t argue with his maths, nor his conclusion at the end… but nothing there supports, or contrdicts my calculation, since he is talking about changing a single medias properties to improve efficiency. There’s no reference to multiple layers of media. I’ll stand by my calculation as accurate.

I do not know the specific media construction I tried, since it was repurposed from a commercial filter.

If you are getting crushed… I have some bandwidth. Maybe we could talk! :smiley:

“Getting Crushed” How funny, that kind of statement states more about you than me…

"I don’t argue with his maths, nor his conclusion at the end… but nothing there supports, or contradicts my calculation, since he is talking about changing a single medias properties to improve efficiency. There’s no reference to multiple layers of media. I’ll stand by my calculation as accurate"

So, from what you say, if you double up on the thickness of filtration medium of a set filtration size then you can reduce the size of particles it will filter , so in turn it will filter smaller particles than the minimum mesh size of a single layer of the filtration medium.

Perhaps your source of technical information was the famous author Douglas Adams, but I cant find any references of his in regards the science of filtration.

But I do seem to remember one of my schooldays teachers mentioning that if you double up the thickness of a filter medium you don’t alter the size of particles that it filters but you do increase the pressure differential from one side of the filter to the other. On the other hand that could just be a memory from a moment in dementia, but who knows…

Hopefully this paper will give you a better insight into the science of filtration: http://aaqr.org/files/article/781/15_AAQR-12-07-OA-0179_162-171.pdf