The efficacy of different face masks in reducing droplet emission

What kind of mask should you wear during the pandemic? (Yes, you should always wear one if you’re around people.) A new paper in Science Advances (click on screenshot below, free pdf here, reference at bottom) gives a tentative answer to that question, assuming that you’re wearing the mask to avoid infecting other people. And that presumes that you’re carrying the virus, symptomatic or asymptomatic. (If you’re symptomatic, you shouldn’t be going out anyway.)

Note that the question here is probably not the one many people have, which is “Which kind of mask should I wear if I don’t want to get infected by others?” This is not the same question, as some of the features of the masks that decrease their efficacy (i.e., breaking up expelled droplets into smaller droplets), might not work in reverse. In general, though, those masks that reduce the number of droplets expelled when you’re speaking should also reduce the number of droplets coming in when someone’s speaking to you. The short answer is that fitted N95 masks (which most of us can’t get) are the best at keeping your droplets in,, followed by surgical masks and then two-layer masks of polyproplylene and cotton (multiple layers, as you might guess, are important). N95 masks with valves to allow you to exhale aren’t that great, as you might also expect, and knitted or fleece masks, particularly with one layer, are pretty useless. Bandanas are dreadful—barely better than controls, yet I’ve seen many people wear them.

If you can see through your mask, it’s a clue that it’s not a good one.

Click to see the article, which I’ve summarized below:

The authors used an inexpensive setup (roughly $200) to measure expelled droplets when individuals spoke through a mask for 40 seconds, repeating the phrase, “Stay healthy, people” five times, with the protocol repeated ten times for each mask and the control (no mask) trial. Droplet size and number were measured in a dark chamber using a laser and a cellphone camera.

I’m not sure why the authors are so keen on the inexpensiveness of the apparatus, as individuals aren’t going to do this at home, and a more professional setup in a lab would surely reveal general results that wouldn’t need to be replicated on this inexpensive apparatus. At any rate, the apparatus is diagrammed below, with the diagram and caption from the paper:

(From paper): Fig. 1 Schematic of the experimental setup. A laser beam is expanded vertically by a cylindrical lens and shined through slits in the enclosure. The camera is located at the back of the box, a hole for the speaker in the front. The inset shows scattering for water particles from a spray bottle with the front of the box removed. Photo Credit: Martin Fischer, Duke University.

They used a single speaker for most of the tests to reduce variation, but also used four other speakers on 3 masks and the no-mask control (also ten replicates each) to test the replicability of the results. And they tested fourteen masks, shown below. I’ll describe them as they do in the paper, for the table they give is hard to read.

Masks marked with an asterisks were tested by four different speakers saying the same thing (one different speaker for each of three masks and the control), while the rest of the masks were tested with a single speaker. Numbers below correspond to the diagram above except for the control (no mask) and “swath” mask, which I assume is like a turtleneck pulled up.

  1. “Surgical” mask*.  3 layer
  2. Valved N95
  3. Knitted
  4. “Polyprop”: 2 layer polypropylene apron mask
  5. “Poly/cotton: 3 later cottong/polypropylene/cotton mask
  6. MaxAT mask: 1 layer Maxima AT mask
  7. “Cotton2” 2-layer cotton pleated mask
  8. “Cotton4”: 2-layer cotton, Olson-style mask
  9. “Cotton3”: 2-layer cotton pleated style mask
  10. “Cotton1” 1-layer cotton, pleated style mask
  11. Fleece: Gaiter type neck fleece
  12. “Bandana”*: Double layer bandana
  13. Cotton5*: 2-layer cotton, pleated-style mask
  14. Fitted n95 mask: no exhalation valve and fitted
  15. “Swath” mask: swath of polypropylene mask material (not shown)
  16. “None”: control 

And the results in short are below, showing the number of droplet counts, relative to the control (green dot to the right). 1 means as many droplets expelled as with no mask, while close to zero means almost no droplets expelled relative to no mask. All black dots represent the means of ten replicates with a single (and the same0 speaker (lines are standard deviations), while the four colored dots represent the means for four other speakers. These are close to the single speaker using all three tested, giving confidence that the results may be general. Have a look:

(From paper): Droplet transmission through face masks. (A) Relative droplet transmission through the corresponding mask. Each solid data point represents the mean and standard deviation over 10 trials for the same mask, normalized to the control trial (no mask), and tested by one speaker. The hollow data points are the mean and standard deviations of the relative counts over four speakers. A plot with a logarithmic scale is shown in Supplementary Fig. S1.


Swath masks, which did well (fifth best) but aren’t shown, involve wearing swaths of polypropylene mask material, as shown below (source of photo is here, which gives a simpler summary of the recommendations):
Even cotton masks are okay, with 1 layer being better than pleated, but when you get to the knitted masks, bandanas, and fleece masks (which produced more droplets, probably by breaking up the big ones into smaller ones), you’d best avoid them.

So what does this mean for you? Assuming that you can’t get a fitted N95 mask, your best bets are surgical masks, which are available (remember, these are designed to keep medical professionals from exhaling microbes into patients’ wounds and bodies), multilayered poly/cotton and poly/propylene (masks 4 and 5 above), or a “swath” of mask material (polypropylene), shown above. If you have a choice, get a surgical mask, or wear masks 4 or 5. But, except for fleece, some mask is better than no mask.

Caveats: Remember first that these masks are tested for EXHALED DROPLETS, not droplets inhaled, which would be harder to test. So sites that imply that these are the “best masks for you to wear” are leaving out that crucial information. I suspect there will be a correlation, but perhaps not a perfect one.

The authors offer other caveats, like the inability to measure total droplets in the chamber, and their use of a cellphone camera, which reduces sensitivity to detecting laser-reflecting particles. Even so, the droplets that could be detected in this method are half a micron: 0.001 mm or 0.00004 inches, which are small. They also emphasize the small number of speakers (five, with most masks tested using a single—and the same—speaker), and that would warrant replication, since some people speak more or less forcefully than the speaker they tested.

These are the necessary reservations, but these are valuable data nonetheless. But again, remember that these data tell you how to protect other people from your exhaled droplets. Even so, I’d suggest getting yourself some 3-layer surgical masks if you can. I believe you can re-use them if you let them disinfect for a week or so before you wear them again, but I’m not an expert here, so take that with a grain of salt and do your own checking.

h/t: SImon


Fischer, E. P., M. C. Fischer, D. Grass, I. Henrion, W. S. Warren, and E. Westman. 2020. Low-cost measurement of facemask efficacy for filtering expelled droplets during speech. Science Advances:eabd3083.


  1. Posted August 11, 2020 at 8:50 am | Permalink

    We can pick up non-fitted N95s locally easily now. Also surgical masks over AMZ.

  2. Posted August 11, 2020 at 9:26 am | Permalink

    I’ve been using surgical masks since February. I always thought the cotton would be better, but multi-layers helps.

    Thermodynamically the masks may behave differently for inhalation as the order of the filters is probably important. Ballistic vs. diffusive gas behave very differently too. Breathing out typically pushes matter directly into the mask. Ambient air can easily diffuse around corners.

    One need only stand within twenty feet of a smoker or heavily cologned man to know how ineffective masks can be at keeping air out.

  3. Posted August 11, 2020 at 9:29 am | Permalink

    Thank you for posting this. I’m not sure what my cloth masks are made from, but I know I can’t blow a bee off my arm while wearing them. In the meantime, I have ordered some polypro surgical masks. Great article. Thanks muchly.

  4. Posted August 11, 2020 at 9:48 am | Permalink

    Thanks for this. I think this is the clearest and easiest to understand study I’ve seen regarding the efficacy of masks. I wear mask number 1 which I’m pleased to see performs well. Although, when I think about it, I’m pretty certain I do not have COVID19 due to not interacting with people outside of my household for more than a week.

    I wouldn’t be at all surprised if no mask provides adequate protection to the wearer. As I understand it, a lot of the transmission comes from touching surfaces that other people have breathed all over and then touching your face. I’ll put my faith in frequent hand washing on that count.

  5. Posted August 11, 2020 at 9:52 am | Permalink

    Saw an interesting interview last night on CNN with a healthcare expert (sorry, don’t remember which one). She thought that the justification for wearing a mask in order to protect others was a bit overstated. It also protects the wearer because, even if one gets infected, it reduces the size of the inoculant, the infecting virus dose, which reduces the severity of the resulting illness. It makes sense to me as a smaller dose would give one’s immune system more time to mount a response.

  6. rickflick
    Posted August 11, 2020 at 9:55 am | Permalink

    I have an N95 I use when there’s a larger risk and a simple mask for other occasions. I only leave the house about once a month, and rarely near many people. I believe they prevent inhaling particles to some extent, and that might be very important. Some studies have shown that low viral loads can produce milder cases and build immunity. So, while a mask can’t protect the wearer 100% it might be very beneficial.

  7. nay
    Posted August 11, 2020 at 10:43 am | Permalink

    Thank you, Professor! Very clear, concise and complete. I usually use the surgical mask, but was given a couple of printed polyester ones which I find hard to breathe through – so I’m not inhaling particles because I can’t inhale? Hahahahah; pretty, but not useful.

  8. Kevin
    Posted August 11, 2020 at 10:49 am | Permalink

    Choosing the wrong mask can be quite drastic:

    In this case I believe the problem was that the masks attach around the ears rather than the whole head. The latter apparently closes the mask better against the face.

    This case is mainly an issue for professional healthcare staff and not the general population.

    Pending possible enquiry by
    National Audit Office into “alleged mishandling of PPE contracts”.

  9. Posted August 11, 2020 at 11:08 am | Permalink

    Still trying to figure out how a fleece gaiter can be worse than nothing. Measurement error, I suppose.

    • Posted August 11, 2020 at 1:02 pm | Permalink

      the effect of the fibers in those masks is to break up droplets into smaller droplets, while doing a crappy job at absorbing them. So the aerosol cloud reads as larger.

      • Posted August 11, 2020 at 4:05 pm | Permalink

        Basically yes. This is from the paper:

        “This insight is important to in- terpret the result of the neck fleece. The neck fleece has a larger transmission (110%, see Fig. 3 (A)) than the control trial. We attribute this increase to the neck fleece dispersing larger droplets into several smaller droplets, therefore in- creasing the droplet count. The histogram of the binary di- ameter for the neck fleece supports this theory (see Suppl. Fig. S5).”

    • Nicolaas Stempels
      Posted August 11, 2020 at 1:16 pm | Permalink

      I use N 95’s because of my professional activities. I’m provided with 5 masks for one month.
      After one day I put it in a paper bag and on to the next. Every mask is hence used 6 times.
      When outside I wear it under my chin, pulling it over nose and mouth when in closer contact with others. Inside or in a car with several people, masks should be worn ‘religiously’.
      The N 95, if fitted properly, does not only prevent you from infecting others, but offers personal protection too, I’m sure the other masks also do so , albeit to a lesser degree.
      It is important to realise is that masks do not just protect others from direct infection, it also protects surfaces from getting contaminated. The surfaces you touch with your hands.
      If everybody would wear a mask (except for the bandana and the fleece), would disinfect hands regularly, and avoid meetings with more than, say, 10 people, the epidemic would disappear.
      On a side note, the way so many people wear their masks, makes us understand why condoms are not a waterproof way to prevent pregnancy and VD’s.

  10. eric
    Posted August 11, 2020 at 12:10 pm | Permalink

    We’re using #8 (“Cotton #4”) equivalents. Glad to see they do okay. I’ll go looking for some N95s, but if I can’t order any, I’m not going to get too angsty about it.

  11. Jim batterson
    Posted August 11, 2020 at 12:38 pm | Permalink

    Very good information. Looks like any reasonable barrier is helpful in reducing ones cross-section with droplets. Walking through an exhaled cloud cannot be that much different from blocking your own exhaled droplets. Thanks!

  12. Mark R.
    Posted August 11, 2020 at 12:54 pm | Permalink

    I wonder how effective plexi face shields are. I see a lot of grocery store workers wearing them.
    I’ve been using KN95 masks which are similar to the N95. I did some research on how to clean them for multiple use and it seems UV light kits work, but the data wasn’t definitive. I just use them a couple times and throw them out. At $2 a mask, it adds up quickly.

    • Posted August 11, 2020 at 1:07 pm | Permalink

      I think they are thought to be very effective at protecting the wearer in terms of reducing the droplets you breathe in. Aerosol droplets tend to move along the currents of air, and the face shields create a good laminar air flow effect so the droplets just go around you.
      So masks are best at protecting other people, presumably with some effect at protecting the wearer. But face shields work to protect the wearer.
      I wear both when I go to a store.

    • Nicolaas Stempels
      Posted August 11, 2020 at 1:22 pm | Permalink

      Those face shields are very good at protecting you -and others, if worn in combination with a mask.
      In the supermarkets here, most cashiers (a high risk profession) wear both.

    • Mark R.
      Posted August 12, 2020 at 12:22 am | Permalink

      Thanks for the info Mark and Nicolaas. Helps a lot.

    • Posted August 12, 2020 at 6:04 pm | Permalink

      My understanding is that the KN95 mask is the Chinese version of 3M’s N95 mask. Lots of different companies in China were making them and they weren’t all well made. Particularly when everyone in the world was scrambling for masks, many states and countries received shipments of Chinese KN95 masks that they wouldn’t use and/or sent back. I hope you’ve got a good one.

  13. Nicolaas Stempels
    Posted August 11, 2020 at 1:24 pm | Permalink

    Sex in the times of Covid: Condom, Mask, Gloves and only doggie style 🙂

    • Ken Kukec
      Posted August 11, 2020 at 2:02 pm | Permalink

      Works best if you begin foreplay in full PPE, then strip down to the above immediately before commencing coitus.

      • Posted August 12, 2020 at 5:40 am | Permalink

        Does a nurse’s uniform count as PPE? Asking for a friend.

        • Kevin
          Posted August 12, 2020 at 8:02 am | Permalink

          @Colin McLachlan
          No it doesn’t. And neither does latex underwear. Best regards to your friend

  14. Posted August 11, 2020 at 3:16 pm | Permalink

    Lots of people pay lip-service to mask wearing but have it off the nose making it pointless!

  15. Eric Shumard
    Posted August 11, 2020 at 4:26 pm | Permalink

    If one examines the graph of droplets over time (Fig 3B) it appears that the best way to minimize droplets is to minimize speaking. That’s more effective at reducing droplets than wearing most types of masks while speaking.

  16. Diana MacPherson
    Posted August 11, 2020 at 5:14 pm | Permalink

    I’ve been wearing the 3 layer polyester/cotton-in between stuff -cotton

  17. Seth
    Posted August 12, 2020 at 5:48 am | Permalink

    I’d say the inexpensive cost of the apparatus is notable since it’s within reach of an interested hobbyist. With so many people staying home so much now, it’s an appealing type of project for the sort of person who likes to build things as a hobby. It’s very neat “citizen science”.

  18. Kevin
    Posted August 12, 2020 at 7:47 am | Permalink

    Eye goggles are also an option instead of plastic visors. I have the transparent type used for woodwork and lathe work. I had them already since I do actually have a lathe. The N95 and the equivalent EU 149 standard are masks also sold for industrial/non-medical uses (the technology is apparently an offshoot from bra-cup manufacture!!!)

    I keep goggles in my shoulder bag with the masks. The visors are awkward to carry.

    It is unlikely that a mask alone will protect you from an aerosol because the virus will get at your eyes. The mask is supposed to protect others from the wearer’s aerosol. It also slows down the SPEED and spread of the exhaled air, which may be more important than droplet size. Also important as a block to sneezing and coughing.

    Some friends of mine converted a FabLab for visor manufacture (using ranks of 3D printers: to their credit they produced +100,000 for hospital use). They tell me that you can also have sealed visors with positive air pressure and filters (these make breathing more comfortable and should more completely protect the wearer).

    A major problem with normal masks is that they are difficult to wear all day: many of the food shops here enforce masks for customers, but the staff mostly don’t seem to wear the masks OR visors, and some are protected only by perspex screening at the till (occasionally visors). The staff are in continuous contact with sales items and customers.

  19. not important
    Posted August 13, 2020 at 8:19 am | Permalink

    Can you please also provide details on the scientific basis for your claim that you should always wear a mask when you are around people?

    I hope that you are not making this claim lightly, given the current political climate in which those of us who choose not to wear masks are being demonized by the media and politicians in their attempt to keep the coronavirus hysteria going until at least the November elections.

    Like it or not, at a political level, mask wearing outside settings such as nursing homes and hospitals sends a clear signal of subordination to and acceptance of authoritarian and xenophobic policies like lockdowns and travel bans that have been implemented by our scientifically illiterate political establishment at the behest of our scientifically illiterate media.

    These policies, which are driven by panic, pseudoscientific speculation and political partisantship, have already done immeasurable harm to society and human life, so I think you should have a very decisive scientific justification before you go around telling people that they should always wear a mask.

    • Posted August 13, 2020 at 8:23 am | Permalink

      I really dislike people like you who haven’t done your homework about the efficacy of masks in preventing the exhalation and inhalation of respiratory droplets, which are the vectors for coronavirus. Your political ranting merely serves to justify the spread of the pandemic, as well as more illnesses and deaths.

      Why don’t you try Googling before you come over here with your dangerous suggestions?

      A few sources (you can look up the rest for yourself):,studies%20of%20real%2Dworld%20scenarios.

      Tons of reference here:

      Now go away. I don’t want posters endangering public safety.

    • Posted August 13, 2020 at 3:22 pm | Permalink

      The professor tells you the truth in accordance with current scientific understanding of the methods we can take to protect ourselves and others (which changes as more is learned) to the best of our ability during the pandemic. There are any number of sources available online from Dr. Fauci, the CDC, other federal agencies, and most of the states. I don’t think you can go wrong if you pay close attention to what New York has done and, is doing, to limit the spread. Governor Cuomo has been an ongoing source of useful information. His state has had much better “luck” in controlling the disease than the states that aren’t wearing masks, distancing, staying at home if needed, etc.

      I routinely wear either one of the blue masks or cloth masks when leaving my house, and as I’m almost 80 and I seldom leave my house. My son does most of the shopping. I see a great many people walking or cycling if they get cabin fever. I infrequently get in my little car by myself and take a trip along the beautiful backroads of Oregon and then return to my nest. I have another 3M mask with filters which I haven’t felt the need to wear yet. Not to advertise for 3M, but I think you can trust any of the various types of masks made by 3M. They’ve established a fine reputation in all the products they manufacture. You can’t trust the KN95 masks made in China.

      I would prefer that no-one die from this virulent disease, but that’s impossible. The next best thing is to protect ourselves and those around us as best we can. If individuals are so adamant about their so-called rights as not to wear a mask and distance, then I want them to stay away from others whom they endanger and endanger only themselves, their families and their cohorts.
      I think if you do otherwise and cause the death of innocent people around you, you’re guilty of murder and I wish you could be
      held accountable.

    Posted August 13, 2020 at 1:17 pm | Permalink

    Thanks! I see a lot of the “surgical type” – light blue with folds sold at retail stores. They seem to all be made in China, several brands – are they all as good? Which brand did you test? Would you please mention some of the other brands/styles that were highest rated? There are a lot of choices and articles like this certainly help us all out!

    PS NOT Google pleeze… I use DuckDuckGo.

%d bloggers like this: