Air pollution is one of those things, like weather and climate, that because it affects everyone we all have an opinion about it. This makes it an ideal area for “citizen” and “traditional” scientists to collaborate and work towards improving our lives. However, air pollution is not easy (I’ve been at it for 15 years and I still can’t claim to get it) and clarity on the concepts used is key to maintain the communication and collaboration.
One of those concepts that is very easy to get confused about is particulate matter (PM).
Gaseous pollutants such as ozone (O3), carbon monoxide (CO) or nitrogen dioxide (NO2) are easily identifiable chemical species. This means that we can have standard samples of these gases of a defined concentration and they will all be identical. I can use these standards to calibrate my instruments and compare the response of new sensors to these standards.
The problem with PMX
Unfortunately, particulate matter is not that civilized. For starters, PM is not a species but a rather complicated mixture of many things whose physical and chemical characteristics change in time and in space. So instead of “measuring PM” we’ve settled for quantifying some of PM’s characteristics. That’s why we have several metrics related to particulate matter:
- PMX : Mass of particles smaller than X μm per volume of air. Typical values for X are 10 and 5. However, more recently, X has been seen to take values as small as 0.1.
- NX : Number of particles larger than Xnm per volume of air. In this case, X is usually 3, 10, 20 and occasionally around 200.
It should be fairly obvious from those definitions that, for instance PM2.5 is a subset of PM10, i.e. those particles measured as PM2.5 are also measured as PM10 . In fact, looking at a microscope image of a typical PM10 filter we can see particles of all sorts of sizes and shapes.
Then why people think of PM10 as a pollutant different from PM2.5?
In a nutshell … because we (scientists) have been bad at communicating these subtleties to the general public and we have allowed sloppy language to misrepresent what PM and their metrics are.
Let’s see if an analogy helps.
Let’s think of a party where 100 guests. Conceptually, PM is the whole set of guests and every PMX metric is a different property of this group of people. For instance, we can define GuestX as the number of guests whose’s names have X letters or less. Then it would be clear that any guest counted for Guest5 must also count for Guest10 and it is meaningless treating those two sets as independent entities. Furthermore, what if instead of counting people we weigh them … or we count their teeth, or their hairs … or whatever. It doesn’t matter, none of those values actually represent the set of guests, all they do is give you information about that set.
Now, why is this relevant for low cost environmental monitoring?
I’ve seen in some of these low cost sensing platforms fora people arguing about calibration of the sensors and comparability with regulatory technologies and these discussions often feel like at some point people get tired of trying to find the real calibration curve and instead drop the whole system. At the moment, low cost PM sensors are optical sensors that use light scatter to get information about the aerosol and it is very important to understand what kind of information we’re getting out of them.
Particularly important is to acknowledge what they do and don’t measure. First, none of the optical methods actually measure PMX because none of them weighs any particle. All of them give something not entirely unlike PMX but not exactly PMX.
A tricky claim from some sensors is that they give particle number concentration. It kind of makes sense, an optical sensor whose operating principle is based on individual particles scattering light must be able to relate the scattering “event” to a particle and hence give the number of particles. Well … no. None of the low cost optical sensors actually count particles, what they do is measure the accumulated scattering of a “bunch” (and that’s the technical term) of particles in their sensing volume (see here for a very nicely done test of one of these sensors).
Let me expand a little bit. In theory, if particles are going, in an orderly fashion, one by one in a single line passing through the sensing volume, then yes, one peak on the light sensor would mean one particle and the intensity of that peak is related to its size. There are instruments that do that but are expensive and the reason is not the optics but the plumbing … getting the damned particles to line up is quite hard! In low cost sensors that have uncontrolled flow (moving the air through diffusion and convection is not controlled flow) all you can hope for is to have enough particles in the sensing volume that the aggregated scattering is enough to be seen by your sensor.
Now, I know that a lot of people have tested these little things against “standard” technologies (I’ve done it as well and I love these sensors!) but please don’t believe that you have a “universal” calibration for your sensor or that you can put it out and automatically get number concentration or PMX. If you’ve done your homework you can get something that correlates with what you want to measure and you should be happy with that … after all, you didn’t really wanted to measure PMX because of its inherent beauty but because of the health impacts of air pollution and for those PMX, as well as other pollutants are as much a hazard on themselves as tracers for complex sources like combustion processes.
So in conclusion … PM is complicated, low cost sensors are very useful but they should not be expected to deliver the same kind and level of information as their more expensive cousins.
