| Author |
Message |
    
Hyesun Helen Park (Park)
| | Posted on Tuesday, February 13, 2001 - 1:35 pm: | |
I am interested in NIR online applications on a fast moving web. I have encountered quite a few papers dealing with it, but they are mostly result oriented. I am more interested in technical aspects of the applications, for example, how to deal with web fluttering or vibration, or how the speed of the web affects predictions etc. Can any one recommend good sources of reference (either book or articles)?
Thanks.
Hyesun Park |
David Russell (Russell)
| | Posted on Tuesday, February 13, 2001 - 2:16 pm: | |
I don't recall any published information but I'll advance some "obvious" considerations.
The more consistant the orientation the better your measurement will be. You would have to experiment with the scans per spectrum setting to determine the level at which you get useful spectra.
Also a consideration is that the typical process instrument produces about 1 scan per second. If you need 300 to get acceptable signal to noise then control of a "fast" process would be impossible.
However, most measurements of this type are considered to get timely quality measurements (as compared to hours after the fact on grab samples from an off-line lab).
They can work quite well. Long term success requires economic payback considerably in excess of the required on-going cost. |
Tony Davies (Td)
| | Posted on Tuesday, February 20, 2001 - 12:47 am: | |
Have a look at the paper by Peter Hindle in "Leaping Ahead with NIR Spectroscopy", the proceedings of NIR-94 (Australia) pages 372-375.
Tony Davies |
Su-Chin Lo (Suchin)
| | Posted on Wednesday, February 21, 2001 - 8:11 am: | |
Please take look another recent interesting article by O. Berntsson, L.-G. Danielsson and S. Folestad in " Characterization of diffuse reflectance fiber probe sampling on moving solids using a Fourier transform near-infrared spectrometer", Anal. Chim. Acta., 431, p125-131 (2001).
Based on the sample movement at velocities between 0.046 and 0.73 m/s, it was found that sample movement causes features not normally present in the interferograms. However, this artifacts (from interferogram) are present well outside the NIR region. Their conclusion is: FT-NIR (with the present expt set-up) is an appropriate technique for monitoring soild sample moving at low or moderate velocities.
However, according to the author's description: another related article from M. Andersson et al (Anal. Chim. Acta. in press) found that " severe artifacts appeared as a result of a combined effect of spectrometer scanning rate, sample movment and sample heterogeneity" in a dispersive mechanically scanning grating spectrometer. We have to wait to review this article (expt procedures and set-up) in much more detail.
It may be another war between the pros/cons between FT-NIR and grating NIR for an on-line analysis (???). |
hlmark
| | Posted on Wednesday, February 21, 2001 - 9:52 am: | |
I don't have a good "feel" for the relation between the speed of the sample and the scan speed of the interfereometer, but as Dave Russell said, there are some "obvious" possibilities, which are conveniently broken down into three cases (regimes):
1) The sample movement is very slow compared to the scan speed. In this case, each scan is made during a quasi-static period in which the sample is not changing. Then the effect of coadding would be to average over many different "aliquots" of the sample and therefore average out inhomogeneities.
2) The sample movement rate is intermediate. In this case, depending on the particle size and other factors, during a scan there could be regularities in the variability of the sample. These regualrities would be sampled along with the optical variations, and when the FT is performed this would create artifacts in the spectrum, which might not be distinguishable from actual spectral features. This would also depend very strongly on whether the sample movement is synchronized in any way with the scanning control.
3) The sample speed is very fast compared to the scan speed. In this case, different parts of the sample are measured during different parts of a single scan. This would likely introduce a very large noise component into the data.
Of course, these three regimes are not sharply defined. In the intermediate regime especially, you're likely to see effect from BOTH the low-speed and high-speed phenomena as well, at the same time. In fact, I strongly suspect that in any case, you might see all three kinds of effects, the main difference being the degree to which each one shows up.
Howard |
|
Closed: New threads not accepted on this page