I'm writing up an academic paper (medical) on in-car exhaust levels and their effect on health. So as to ensure detection of leak-points that are located further along than the catalytic converter, in the discussion I would like to recommend the testing of an exhaust gas that is not affected by passing through the catalytic converter. Would I be correct in thinking that toluene is one such gas, and if so, would it be a good marker gas to use? Many thanks!

  • $\begingroup$ If you are measuring the actual outputs from a real engine, then carbon monoxide would be a good candidate as it is a key target for catalytic converters and is easy to measure. Toluene would not be as the amount will vary hugely by fuel composition and is probably not a major feature in exhausts from modern engines anyway. $\endgroup$
    – matt_black
    Feb 28 at 17:03
  • $\begingroup$ Hi matt_black, many thanks for your comment. As it happens the study measured precisely that - I put CO dataloggers in in a bunch of cars driven by laypeople - and the results were pretty worrying! Some of the highest readings, interestingly, fell very quickly as the engines warmed up. This made me think that the CO may have originated from leaks distal to the catalytic converter - which is why a dual test, of CO and also of a non-catalysed gas, would be necessary to detect all in-cabin leaks. You can read about the project here - airsafe.london. $\endgroup$ Feb 29 at 13:25
  • $\begingroup$ While the possibility of leaks from the exhaust is worth exploring, it would not be my first or most likely hypothesis explaining the presence of CO inside the car. Just normal atmospheric circulation will lead to that. And while cold engines might produce more CO, cold catalytic converters certainly do so the profile of the gas over time inside the care doesn't need leaks to explain the pattern observed. $\endgroup$
    – matt_black
    Feb 29 at 13:33
  • $\begingroup$ If you're interested in learning more about the many factors that influence in-cabin CO, Kaur, Niewenhausen et al 2009 is a good place to start - from there, read Alameddine et al 2016, Hachem et al 2021 and Lim J 2019. Then look at the field studies - Moreno T et al 2019, Hachem M et al 2021, and round off with the Lancet meta-analysis (Cepeda et al 2016). And then once my paper is out, you could even read that too. TLDR, cars do often leak. $\endgroup$ Feb 29 at 20:03
  • $\begingroup$ (and much thanks for the practice in thesis defence - always helpful! have a good one) $\endgroup$ Feb 29 at 20:04

1 Answer 1


A much better tracer for leaks than toluene would be helium.

Compared to toluene, helium is a noble gas and hence unlikely to react at the temperatures and presence of the catalysts in a converter. (Different to toluene, it plausibly would pass unchanged the combustion process of an Otto, or Diesel engine, too.) It is inflammable and non-toxic. Its monoatomic molecules are considerably smaller in size, than a toluene molecule and hence can pass smaller cracks and openings. At ambient conditions, helium already is a gas .

Helium sniffing already is a technique introduced in the industry to check e.g. joined gas pipes and vacuum installation for leaks: an external probe aspires air for instance next to a fitting, while the tubing is flushed internally with some of helium. The air aspired outside is submitted to a small mass spectrometer attuned to the very signal of helium and of high (often adjustable) sensitivity. The technique equally benefits from the the low terrestrial concentration of helium (except a few known natural sources and artificial processes) and hence low background.

enter image description here

(image credit to TQC Automation & Test Solutions)

A little cart can house both the gas cylinder with the tracer gas and the necessary equipment:

enter image description here

(image credit to Leybold, which provides some additional information here)

  • $\begingroup$ Hi there Buttonwood, this is really helpful: very much appreciated. $\endgroup$ Feb 29 at 13:27

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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