| Because of the limited number of pixels in the spectrometer detector, there is a situation where only a few pixels make up a single spectral peak at the high spectral resolution, resulting in an unsmoothed peak, as shown in the figure below.
The figure shows an example of a drift in the peak position due to the temperature drift of the spectrometer at different temperatures for the same spectral peak. The spectral peaks are not symmetrical because the number of pixels that make up the spectral peaks is small, and the highest pixel of the spectral peak is shifted by one pixel from do to right under the effect of temperature drift.
At this point, we have two ways to evaluate the drift of this peak position. 1. Look at the position of the highest pixel point. Obviously, the peak position is shifted by one pixel, which translates to about 0.15 nm. 2. Fit the peaks before and after the drift and find the peak point of the fit, which is between two pixels. The drift of the fitted peak point is about 0.07nm So which of these two methods is the right one? We believe that the second one is correct and reflects the true drift of the peak position. The reasons are as follows. 1. The high resolution leads to a low sampling rate of the pixels of the spectral peaks, and directly looking at the shape of the pixels together does not reflect the actual shape of the peaks. 2. The spectral peaks are physically closer to the symmetric distribution, and the drift of the peak position does not cause the peak shape to change. 3. By fitting all the pixels that make up the spectral peaks, the peak shape obtained is closer to the physical reality of the peaks.
Therefore, using the fitted peak shape to determine the peak position and evaluate the drift of the spectrometer is the correct method.
In practice, is it all necessary to do so? It depends on the situation.
If 1. The spectral resolution is low, and the sampling rate of the spectral peaks is high, i.e., it takes more pixel points to compose a peak, and these pixel points themselves are connected into a smoother and more symmetrical peak shape, then you can directly find the position of the highest pixel, or between two pixels, and confirm the peak position by visual evaluation.
2. With High spectral resolution and the need to know precisely the peak value, the above fitting method is needed. The fitting can be done by Spline algorithm, or by Gaussian or Lorentzian fitting, and the reader can choose according to the spectral nature of the sample under test. If you don't know how to choose, use Spline. |