• 0

Question:How to neutralize/remove the source's spectra to obtain absorption spectra?

shubham is asking a question about spectrometry: Subscribe to answer questions on this topic

shubham asked on August 17, 2016 06:35
631 | 7 answers | shortlink


Hi Everyone

We have made a CCD detector based spectrometer but are facing a problem while taking the absorption spectra.

We are using tungsten lamp as source and stored its spectrum as Yref. Next we placed a transparent acrylic sheet in front of the source and stored the spectrum as Ysamp. We tried to plot both y1=Yref-Ysamp and y2=Ysamp/Yref.

We should obtain a flat horizontal line for absorption spectra of transparent material (since the absorption is almost equal for all wavelengths). But instead, in both y1 and y2 plots we see the trend of the source's spectrum i.e. a curve similar to the source's spectrum curve.

Source link : https://www.researchgate.net/file.PostFileLoader.html?id=57b40483f7b67ecc35723d12&assetKey=AS%3A395959155609600%401471415427618 Sample spectrum link : https://www.researchgate.net/file.PostFileLoader.html?id=57b40483f7b67ecc35723d12&assetKey=AS%3A395959155609602%401471415427858

Thank you



question:spectrometry


5 Comments

Log in to comment

7 Answers

Hi, CCDs do not have a linear response curve -- they detect some colors (like those in the green range) more than others. So to get a flat line, you have to do a gain correction: there is some work towards this under the #gain-correction tag, esp. work by


I agree with Jeff that those results seem to suggest that whatever full spectrum source is used the same general spectral response is found (more brightness in the middle of the spectral range), and that a good explanation is that the detector is more sensitive in the middle of the range. There are other possible explanations for this pattern (e.g., the diffraction grating), and there are other really interesting patterns in the results: 1) the oscillations in the two traces, and 2) the double series in the difference between the two traces.


Do you think we need to fix the offset of magnitude 1 before division operator? i.e, the values of the spectra should begin from 0 rather than 1.


Hey @shubham, first lets start with what your two spectra look like, I'm not sure what the transparent sheet is for, other than a filter or a medium for comparison, but it looks like just attenuation of the source spectrum. Since this is absorption, your


Hey @shubham, this may help you out also: Non linearity in the photon transfer curve of back-illuminated CCDs has been reported by the authors in previous papers and confirmed by others even though exhibiting excellent signal linearity. Previous studie


Sign up or Login to post an answer to this question.