1
$\begingroup$

I am building a tool which extracts the elemental composition of a light source using its emission spectrum. Therefor, I'm currently writing down the emission spectra of individual elements, but I'm not sure which wavelengths to include and which I should discard. I'm using the NIST database as a reference for the spectra. For example, let's look at the hydrogen spectrum (strong lines in air).

Firstly, there is the question of how to define the span of visible wavelength. I'm seeing a lot of different opinions and have concluded for myself that 385 nm to 750 nm is an acceptable range. But that would mean that I have to use the following wavelengths:

[388.90, 397.01, 410.17, 434.05, 486.13, 656.27] nm

But most of the time I'm seeing the lines at 388.90 nm and 397.01 nm disregarded.

The whole thing gets more complicated when looking at helium where the data differs hugely from the spectral lines I've seen over the internet:

[388.8, 410.2, 434.1, 447.1, 471.3, 486.1, 492.1, 501.5, 504.7, 587.5, 656.3, 667.8, 686.7, 706.5, 728.1]

So basically I'm asking for opinions on which wavelengths I should use and which I shouldn't use. Is the range for which I defined visible light acceptable. Should I discard wavelengths which don't satisfy a certain threshold for their intensity?

Improve this question
$\endgroup$
6
  • $\begingroup$ There are all sorts of complications to consider, even if you restrict attention to, say, 400 to 700 nm, as in the 2D spectrum for hydrogen I acquired here: physics.stackexchange.com/a/768678/313612. Other elements can have far more complicated spectra, e.g., iron, with thousands of spectral lines of widely varying intensities. Can you say more about the goal of your effort? $\endgroup$
    – Ed V
    Commented Oct 15, 2023 at 11:22
  • $\begingroup$ @EdV Basically my idea is the following: take a spectrum of an unknown light source (intensity over wavelength, like the last picture in your post) -> extract the maxima (with larger intensity than some chosen threshold) -> input the list of maxima into some machine learning model which predicts the elemental composition of the light source. The machine learning model should be trained on the emission spectrum data of every element using the data from NIST. $\endgroup$
    – oodani
    Commented Oct 15, 2023 at 11:39
  • $\begingroup$ That helps! Since you say you are soliciting opinions, and that is actually off topic here, your question might get closed. But my opinion, based on what you posted and your comment, is that it is reasonable to use the visible spectral range you stated and just use the NIST strong lines list. Depending on the light sources and excitation conditions, you may or may not get many ion emission lines in addition to the atomic emission lines. Watch out for contaminants, line collisions, etc. Best of success! $\endgroup$
    – Ed V
    Commented Oct 15, 2023 at 12:11
  • $\begingroup$ @EdV thank you, so you are saying I should just use the whole NIST strong line data in my range and not remove any wavelenghts below a certain intensity threshold? $\endgroup$
    – oodani
    Commented Oct 15, 2023 at 12:25
  • $\begingroup$ The NIST strong lines are already the lines above an arbitrary threshold chosen by the NIST people. Neon, for example, has far more lines than just the strong lines (reference on request). So depending on your spectrometer or spectrograph, and excitation conditions, some weaker lines may be seen. I readily observe this using my homemade echelle spectrographs and neon bulbs, neon discharge tubes, etc. Things can be rather variable in terms of relative line intensities and appearance. $\endgroup$
    – Ed V
    Commented Oct 15, 2023 at 12:33

0

Reset to default