| Author | Message | ||
     Nuno Matos (Nmatos) |
Dear All, I am going to start working with a transflectance probe with NIR. At the moment I have some doubts. It would be great if You could erase those doubts for me: - How to choose an optical path lenght (what happen when I increase or descrease the pathlenght)? - How much of resolution should I use? - How many scans? - Any other thing you might remember. Thank you very much in advance Nuno | ||
| MPDC |
Everything depends on your application. One piece of advice: don't start running to the lab with reference samples straight away. Look at the spectra carefully. If big parts of the spectrum run close to 100% absorption, you will have to decrease pathlength. Depending on which spectrometer you have, there might be gain settings, number of scans etc. If you plot spectra from different samples on top of each other, there should preferably be a nice spread in the graphs, meaning you have a good dynamic use of your detector. Play around with your spectrometer until the different spectra run as little as possible on top of each other. If you can't see trends in your spectra, chances are your calibration will have a hard time as well. | ||
| hlmark |
Nuno - What MPDC states is correct, although only qualitative. If you can, in fact, choose your pathlength so as to optimize it, then there is an exact quantitative answer as well. The idea is to choose the pathlength that will optimize the S/N ratio of your measurement; the optimum value of transflectance is 0.33, corresponding to an absorbance value of 0.48. This is described in Applied Spectroscopy; 56(5), p.633-639 (2002). Strictly speaking, the derivations in that paper are for a situation where Beer's law applies exactly, which may not be the case in transflectance, especially if the sample is scattering. However, the optimum point lies in a relatively shallow and broad valley, so there is a fair amount of leeway as to what the actual absorbance is. A larger issue is that the optimum absorbance will likely occur at only one wavelength. So you will have to modify the pathlength so as to get all the wavelengths of interest as close to the optimum as possible. Howard \o/ /_\ |