I’m not a chemist, but a software engineer. I’m trying to build a device that could detect if someone is burning trash (mostly plastics and similar materials) vs wood or diesel. At my disposal I have the following sensors: https://www.mysensors.org/build/gas

What sensor or combination of them would be most reliable to detect smoke resulting from burning illegal waste?

  • 1
    $\begingroup$ This is not trivial (especially in the sense of what is illegal - is breakdown product A of a tire "legal" versus breakdown product B). You'll find a lot of pertinent research under the airport/national security sector (John Kauer's research on the "electronic nose" might be a good jumping off point). $\endgroup$ – jonsca Jul 15 '19 at 0:11
  • 3
    $\begingroup$ Even though this is a good question, I have a feeling that those gas sensors are too primitive and probably won't consistently detect "good" or "bad" smoke as it consists of numerous complex organic molecules (which are actually making it hazardous or add certain smell). Gas chromatography is what would give a definitive answer, but it requires a bulky complex device. $\endgroup$ – andselisk Jul 15 '19 at 0:15
  • 2
    $\begingroup$ It doesn’t have to be 100% accurate. It only has to give enough probability to merit sending a physical patrol with more accurate equipment for suspicious smoke in chimneys. $\endgroup$ – Karim Agha Jul 15 '19 at 1:35

The typical gas sensor marketed to hobbyists and enthusiasts (except for $\ce{CO2}$) are of the metal oxide semiconductor type, not to be confused with MOS as in MOSFET.

How do these work?

In these detectors, a semiconductor material made of the oxide of a metal, usually tin.

I'll quote from this question:

...sintered composite based on the semiconductor material $\ce{SnO2}$, which is the metal oxide. The different sensors have different admixtures and operating conditions to achieve different sensitivities to different gasses.

The principle of operation of MOS-type gas sensors is explained in Figaro sensors' Operating principle; MOS type.

The Figaro link shows the operating principle in the GIF below:

enter image description here GIF (Source: Figaro)

When the $\ce{SnO2}$ is heated by the (usually 5V but not always) heater supply, oxygen in the air will dissociate at the surface and oxygen ions will stick to the surface, depleting the semiconductor and cutting off the flow of current. When a carbon-containing gas (e.g. $\ce{CO}$ or $\ce{C_xH_y}$ (but not $\ce{CO2}$)) comes along, it will combine with the adsorbed oxygen and remove it from the surface. The more gas, the lower the equilibrium concentration of adsorbed oxygen and the higher resulting current flow through the semiconductor.

Not as gas-specific as you might think

By adjusting the amount and types of additives to the sintered $\ce{SnO2}$ and adding catalyzing metal clusters to its surface, the manufacturer can enhance sensitivity to some gases and reduce it to others, but there is no single-gas specificity with these sensors.

Since the carbon in $\ce{CO2}$ is already fully oxidized, it won't react with the adsorbed oxygen. Those sensors (e.g. MG-811) use a chemical cell sensor based on a solid electrolyte. They need very high impedance amplifiers that measure millivolts, a bit like pH sensors, in fact, a lot like pO2 sensors!

CO and VOCs

I would recommend you use a $\ce{CO}$ sensor (e.g. MQ-7, MQ-9, MQ-307A or MQ-309A) since burning garbage will likely be incomplete (not in an industrial incinerator) and a "Air Quality" VOC sensor like MQ-135.

You can experiment with others, some people like to build an "electronic nose" which is an array of a bunch of these, then they plot the data for different kinds of smoke and environments to see what happens, and try to use the pattern to identify the likely problem.


Inefficient combustion can also produce particles as soot, and there are more expensive "dust sensors" or "PM2.5 sensors" which measure small particles (2.5 microns or larger) in the air. Soot is a hallmark of incomplete combustion (think of the smoke from a badly tuned engine) and a huge problem from incomplete or low-temperature combustion from a garbage fire.

None of these will be conclusive, you'll have to look at your data and look at when there is a fire, and learn how the signals correlate to the events.

Ultraviolet light

There are UV light based fire detectors as well, but you'd need direct line-of sight access to the flames. If the sensor is designed to be solar-blind by adding a filter that removes all wavelengths from the Sun that make it to the ground, then it might be helpful here. However, they are usually designed to work indoors where indoor lighting may have less UV than outdoor sunlight. This would take some more investigation.

read the data sheets carefully

Finally, read each sensor's data sheet carefully and note carefully the specific voltages that each sensor needs. The "Air quality" sensors need you to run for a certain amount of time (about a minute) at one voltage, then a different amount of time at another voltage because the surface chemistry is a bit more complex.

Sensors can also be found here:

| improve this answer | |

Not the answer you're looking for? Browse other questions tagged or ask your own question.