| Author | Message | ||
     Eric LALOUM |
Hi, Are the 3 advantages of FT spectroscopy (Jaquinot, Fellgett and Connes) vs dispersive spectroscopy constant whatever the EM region ? For NIR analysis, are the grating instruments doing better than FT ? Does it depend on the detector ? (PbS, InAs...) Does it depend on the application ? Thanks, Eric LALOUM, PhD | ||
| hlmark |
Eric - the throughput and multiplex properties still exist, but the question is whether they still confer advantage to the measurement. The FTIR manufacturers will say "of course they do", but the situation is not so simple. The best discussion of all this is in Griffiths & deHaseth's "Fourier Transform Infrared Spectrometry", WIley (1986) I'll try to summarize the situation: In the mid-IR, where the FT method first became ascendent, the measurement is detector-noise limited, and is indepedent of the energy; in this case the ability to gather more energy is a benefit. In other spectral regions, where the detector noise is not the limiting noise source, that does not automatically follow. If you're in the visible or UV, where photon noise or other similar noise sources are proportional to the square root of the energy, the benefit of gathering more energy washes out. If the noise is scintillation noise, which is directly proportional to the energy, then more energy is a detriment because the noise increases faster than the signal. Also, in some cases, the ability to reduce noise by co-adding interferograms is obviated by too-high a S/N ratio. In order to reduce the noise level by coadding, you have to have some noise entering the A/D converter. If the system is designed for mid-IR, then the extra S/N of near-IR signals will push the noise below the LSB of the A/D converter and you don't get any further benefit (noise reduction) from coadding, unless the A/D design is extended to enough bits, which can be expensive - especially for an A/D with a fast conversion time. So a lot of characteristics interact, and the answer is not simple. I was once asked to do a comparison of FTIR vs dispersive, and to find conditions where the FTIR was better, and I was hard put to do so. If you want to do qualitative analysis via the classic peak-matching methods, then the increased spectral resolution of the FITR can be beneficial. For quantitative analysis, even using the relatively broad resolution of the current crop standard NIR dispersive instruments gives as good results as an FTIR, because the improved S/N of the dispersive makes up for the distortion and lowered absorbance levels of the grating measurement. Howard | ||
| Bruce H. Campbell (Campclan) |
Eric, There is a situation that you should consider when purchasing equipment and that is how easy is it to transfer the calibration to another spectrophotometer. If you are not planning to use more than one instrument, the question is moot, but if you are planning to use more than one, you need to ask the equipment suppliers about calibration transfer. In order to understand the question, you should read the chapter in the second edition of "Handbook of NIR." There are also other parts of the book that you would find of usefulness. But briefly, the people I have talked with say transfers are somewhat easier with FT instruments. Others have added that the older method of transfer of calibration, the slope and intercept method, doesn't always work. I have transfered a calibration from a dispersive instrument to an AOTF one, but it wasn't easy. Bruce | ||
| fernando morgado |
Bruce : The problem to transfer calibration is good point. Using dispersive is necesary create standarization file for each type of matrix. If you have 25 diferent products will be necesary make 25 standarization file. If you use the same sample for standarizated two instrument for all the matrix probably the real result will be not good. For buy a instrument is necesary consider too what will happen with the models if is necesary repair the unit, for example change detector or others parts. Is important ask to the companies who sales instrument if after reparation is necesary adjust the models again or when will be necesary make it. FT have advantage in this point, but about wich instrument use always is important consider the aplication. Fernando Morgado | ||
| dardenne |
Fernando, I run standardisations for 15 years using NIR instruments and one correction is good for all the agricultural dried and ground products. If absorbances of a product (whole grain for exemple) are outside the normal range, then a new std is needed. Several types of grains can be included in 1 standardisation. Pierre | ||
| fernando morgado |
Pierre : I´m accord with you, is possible use the same standarization for similar matrix, but probably you can not use whole grain standarization file for other products, example bone meal. Old software has the capacities for use only one standarization files, but now, new programs include the posibilities to use diferent standarization file for each diferent matrix, the reason is clear, not always one standarization files can be used for all the products. Fernando | ||
| Michel Coene (Michel) |
Eric, You are not very clear on your application. Many parameters will play. If you are e.g. only interested in a small wavelength range, FT will be at a disadvantage. If you work online, speed can become such an issue that only AOTF or Diode Array can work. On the other hand, FT is hard to beat for wavelength precision. Even though calibration transfer is indeed very important, "What do I want to measure?" should still be question nr 1. |