Detection of ground level ozone with light

Hi there,

I'm working on a project where I'm trying create a sensor to measure the relatively lower concentrations of O3 at ground level. The main idea is to use the unique spectral characteristics of O3, shine a UVC light (at 254nm, where the absorption cross section is the highest) through an air sample, and use the Beer-Lambert law to calculate the Concentration. With separate temperature and pressure sensors, the data can be converted into ppb (=particle per billion). Because of the expected environment, a constant airflow of 10m/s is expected.

After researching and planning for weeks, I have two possible solutions that would be whithin size limit and my budget:

1st,
Use a measuring tube, with the light at one end and a light sensor at the other. First a normal air sample is passed through the tube, while measuring the light intensity. Then another air sample is passed through an activated carbon filter which removes O3 and then passed through the measuring tube. The assumption is that at the second sample, the measured light intensity is higher by a few percentages, enough to calculate concentration.

To make the result more accurate, the first sample probably would have to be filtered from dust and other fine particles.
One possible problem is with the unreliability of the carbon filters. Commercial ones usually aren't designed with O3 filtering in mind, and their efficiencies vary with time, humidity, temperature, etc. Buying a specially designed ozone scrubber/catalyst is not a possibility.

2nd,
Use two measuring tubes with the same setup as above, except that one has an UVA light at ~375nm where O3 doesn't block light at any significant amount. Then after comparing the different light intensities (like a dobson spectrophotometer) we can calculate the concentration.
In theory, the two intensity values are equal if no O3 is present.

I came across this method a few days ago, and not having to use filters seem promising.

Do you have by chance any suggestions which one I should use?

Thank you!

A good general rule of thumb is that directly measuring a ratio (your second approach) is usually better than taking the difference of separate readings. Note my use of the weasel word usually.

Can you put both light sources and sensors in the same tube, with one downstream from the other? If so, you reduce your error sources.

Have you done any research to estimate the range of ##O_3## concentrations that you need to measure? And the differences in light intensity? Will you be able to measure those differences accurately enough to meet your needs?

Didn't Forrest Mims famously devise a simple 'Citizen Science' sensor circuit to detect and measure UV ?

IIRC, these sensors soon confirmed that 'bizarre' BAS reports of an ozone 'hole' were valid, and eg NASA satellites' data reduction streams had simply discarded such Antarctic readings as 'anomalous'...
Oops...

Fall-out led to global policy changes on use of CFCs and their ozone-eating ilk...
https://en.wikipedia.org/wiki/Forrest_Mims

There are many commercially available units for this purpose. They all (I think) do some version of comparing a 'scrubbed' vs. 'unscrubbed' gas sample using the same UV Photometer. At the levels you're talking about (PPB), stability and noise reduction are significant design drivers.

jrmichler said:

A good general rule of thumb is that directly measuring a ratio (your second approach) is usually better than taking the difference of separate readings. Note my use of the weasel word usually.

Can you put both light sources and sensors in the same tube, with one downstream from the other? If so, you reduce your error sources.

Have you done any research to estimate the range of ##O_3## concentrations that you need to measure? And the differences in light intensity? Will you be able to measure those differences accurately enough to meet your needs?

Putting the light sources at the intake part would probably heat the measuring tube up, which would distort the measurement. I was expecting a reading anywhere between 10 and 50ppb, however I've just come across a HUGE miscalculation... The equation I used was the following:

ppb=R*T*10⁹*ln(I₀/I₁) / P*N_A*σ*L

where
T is given in K
P in atm
L is the length of light path in cm
I₀ is light intensity without O₃
I₁ is light intensity with O₃
and σ is the absorption cross section of O₃ at the measured wavelength (0.4*10^-17)

The mistake I made was that instead of atmospheres, I calculated with pascals. Now after doing it the right way, the intensity difference of light at 5ppb is 10^-4%. The sensor I currently have gives an analog value of 1V, thus the difference I'd have to measure is 1/10^th of a milivolt. Maybe I can amplify that signal the way it is done in speakers? I'm not sure though. From what I've read this range is not that difficult to amplify and circuitry for it on breakout boards is sold relatively cheaply.

Another solution could be that I direct the light from the LED, so that effectively all of its 20mW power reaches the sensor.
If that is not possible, than I just reduce the measurin tube lenght from 10cm to 1cm which would increase the measured difference.

Dullard said:

There are many commercially available units for this purpose. They all (I think) do some version of comparing a 'scrubbed' vs. 'unscrubbed' gas sample using the same UV Photometer. At the levels you're talking about (PPB), stability and noise reduction are significant design drivers.

with the second desing (2 different spectra UV leds) stability would not be a problem, as the leds light stay constant after reaching operating temperature. However with the low signal difference, I'll probably have to look into signal amplifyers and other circuitry to cancel electrical noise

Welcome to PF.

El_Burnie said:

Maybe I can amplify that signal the way it is done in speakers? I'm not sure though

For gas absorption measurements, the optical path can be lengthened, by using mirrors to pass the light many times through the gas chamber, or by folding the optical tube.

Sensitive instruments often chop between two signals, then amplify the AC signal, before synchronously detecting the differential signal amplitude.

You might use one stable UV source and detector, but alternate physically between the measurement optical path, and the reference optical path.

Baluncore said:

Sensitive instruments often chop between two signals, then amplify the AC signal, before synchronously detecting the differential signal amplitude.

Yes, beat me to it! I would definitely consider this (a lock-in amplifier, etc.) to improve sensitivity. Great suggestion! If you do use two sources you may be able to share one detector.

DaveE said:

Yes, beat me to it! I would definitely consider this (a lock-in amplifier, etc.) to improve sensitivity. Great suggestion! If you do use two sources you may be able to share one detector.

Thanks, you don't know how big relief this is for me. For a moment I thought the whole project was lost...
So if I switch the 2 lights lets say every quarter seconds, than I will get a wave signal at the output of the sensor.
Does that mean that in a simplified model, the wave's peak is at a constant and the trough will change with O₃ concentration?

El_Burnie said:

Thanks, you don't know how big relief this is for me. For a moment I thought the whole project was lost...
So if I switch the 2 lights lets say every quarter seconds, than I will get a wave signal at the output of the sensor.
Does that mean that in a simplified model, the wave's peak is at a constant and the trough will change with O₃ concentration?

I'm not sure what your saying here, but no, I think,

Study lock-in amplifiers or synchronous detection a bit.
https://www.zhinst.com/americas/en/resources/principles-of-lock-in-detection

You can think of this (sort of) like your light sources are like two different radio stations transmitting at different frequencies and your detector is an antenna for a receiver that can tune in one and ignore everything else. It's not a great analogy if you dig too deep, but that's how analogies are.

PS: Ok, I see what your saying now. Yes that can work too but probably not as sensitive. The issue is how you would measure peaks and troughs without additional errors. In either case it's really a filtering problem; how do you only see the signal you've sent into the gas and not other stuff.

Maybe with this? IDK if it's any good, but it's cheap to try it.
https://www.ebay.com/itm/3549522809...Ik2pffgyE3XE6FaEZJeV5RfSDG|tkp:Bk9SR7yF6OqqYw
https://www.analog.com/media/en/technical-documentation/data-sheets/ad630.pdf

To be clear:
You're re-inventing the wheel. Exactly what you describe is available from Teledyne, 2BTech, BMT, and several others. I won't go into great detail, but you're probably not going to 'whip out' an accurate, reliable, UV-Photometric atmospheric ozone sensor very quickly. You're certainly not going to do it for less than you could just buy one. Aside from the general stability requirements, ozone is a very reactive substance - It is the Schrodinger's cat of atmospheric gases. You can design a sensor (I have), but don't under-estimate the required effort.

Dullard said:

To be clear:
You're re-inventing the wheel. Exactly what you describe is available from Teledyne, 2BTech, BMT, and several others. I won't go into great detail, but you're probably not going to 'whip out' an accurate, reliable, UV-Photometric atmospheric ozone sensor very quickly. You're certainly not going to do it for less than you could just buy one. Aside from the general stability requirements, ozone is a very reactive substance - It is the Schrodinger's cat of atmospheric gases. You can design a sensor (I have), but don't under-estimate the required effort.

Yes, this could be a fun DIY thing or school project, but you won't beat established instrument manufacturers or low cost "eBay" sellers. Go for it, but don't expect to make any money.