I'm am doing research in creating an artificial nose. I need to know how this process works by someone with domain expertise; essentially, what technologies are available currently to detect a broader range of particles in a closed system. My understanding so far goes like this on a high-level overview: particle comes into contact with the olfactory cells and particles from the environment have a custom signature which we associate as smell of a banana or orange, now how to do this artificially or how is this done artificially?

  • 2
    $\begingroup$ Welcome to Chemistry.SE! Please take a minute to look over the help center and tour page to better understand our guidelines and question policies. $\endgroup$
    – A.K.
    Aug 29 '18 at 23:17

"What technologies are available currently to detect a broader range of particles in a closed system" a LOT of technologies allow this: spectrophotometry, NMR, GC-MS, etc. These technologies give a lot of information about the molecular structure (but not about the molecular function).

The hard part in your idea is trying to replicate the sensivity of a nose, which is, at the moment, impossible as far as I know.

Trying to deduce via a robotic nose the smell of a novel compound is a task that could, theorically, be made if that "nose" had exactly the same receptors and "circuits" of our nose, since receptors interact in a unique way with molecules, and that is the source of all those unique smells that we sense.

A more approximate approach, doable with current technologies, could be a database-based approach: a lot of "smelly" molecules could be categorized and described in terms of smell. The analyzer (one of the machines mentioned earlier, for instance) could identify the molecule(s) involved, and the database could provide information about the smell in that environment.

The problem? A sistem like this doesn't work like a nose at all! Any unknown molecule couldn't be "smelled".

Maybe in the future, with the improvement of technology, a molecular docking study will be performed in seconds, and that, when all the proteins associated with smell will be known, could allow us to know the exact way a molecule affects our receptors, our brain and our perception, hence creating a working "digital nose".

And finally, let me add a note: you said "particles from the environment have a custom signature which we associate as smell of a banana or orange". One might argue wether that "signature" is in the molecule or in the nose itself. While many compounds are sensed in a similar way within a population, there are genetic mutations between species and even individuals. Puppies are known to be appealed by cat feces, for instance: is that just a behavioural thing, or their perception of the smell of cat's feces is different from ours? Some people like the smell of truffles, others hate it: it's (at least partially) because of the individual ability of detecting molecules.

The point is not that of detecting a "signature", but in detecting molecules and interpreting them via our specific "signature". If olfactory proteins didn't exist, smell wouldn't exist: it's not a physical property of the molecule, but the result of an interaction with who "smells".

Don't be fooled: that's not a philosophical sentence: light wouldn't cease to exist if we had no visual receptors. But for sure colours wouldn't exist, since also colours exist only because we are evolved to detect light as a mixture of three different chemical responses (each associated with one of the RGB colours).

  • $\begingroup$ That was a response I was really hoping to hear, thank you. I'm going to go through your points and attempt to address and add some thoughts to your statements.. I agree with you about the database approach it would be more efficent then having all these stored on the device but something like you have the system and it's sensors that can detect the molecules and send that via RF to the other computer to be processed. Essentially these molecular structures would have a dictionary associated with the structure in theory. $\endgroup$ Aug 31 '18 at 3:07
  • $\begingroup$ --- combine the use of Machine Learning we could detect these molecular structures and associate them with their "smell" the hard part or rather time consuming part would be the collections of all these unique smells; this part is my background but i'm more or less void of knowledge in biological systems and biochem. I think smell(s) are subjective for sure and there are mutations among different species that have no doubt had a enviromental motivator for survival or perhaps no explanation like the puppies but leaving of the subjectivness of the smells and subcouncious biases or experiences $\endgroup$ Aug 31 '18 at 3:13
  • $\begingroup$ But you're confident these sensors could react with a more broad range or particles ? not just a specific target? again, I'm new to some of these technolgies you meantioned i'm just trying to get a better understanding if the current technology can be manipulated in a way that could detect broad range or particles and have a unique reaction producing dynamic results given the input of the particles. $\endgroup$ Aug 31 '18 at 3:17
  • $\begingroup$ I missed a very important point as well. "replicating the sensitivity of the nose" may be a naive statement but couldn't this be solved by a greater density of sensors in the system? the problem that comes to my mind is the physical space of the device but this isn't an issue for a prototype given it's not > 100 lbs or X of course the technology in the future or a better understanding of the system would directly contribute to more efficient sensors kind of like the transistor perhaps $\endgroup$ Aug 31 '18 at 3:23
  • $\begingroup$ Basically any liquid or gas can be analyzed, so there are no limits to the analyzed compounds. What I meant by "replicating the sensitivity of the nose" was intended in the distinction of smells. The machines I mentioned (expecially NMR and MS) are possibly more sensitive than our nose. They distinguish perfectly between extra-virgin oil and regular oil, or ethyl benzoate and a real banana. And in some applications, they find compounds at a much lower concentration than the one we can smell. $\endgroup$
    – The_Vinz
    Aug 31 '18 at 3:28

I think the major gas sensing technologies are NDIR(cheap), semiconducter type(cheapest), fuel cell type(cheap) and conductive polymers(expensive)


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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