| Author | Message | ||
     Dr Neil J. Dominy (Neil) |
Scanning NIR's are taking the industry by storm in our part of the world at the moment. However their is a large amount of confusion surrounding its limitations. Is it true that NIR can only usefully predict organic compounds or inorganic molecules associated with organic compunds? A number of users are looking to predict Ca and P in animal feeds from raw material sources such as lime, MCP, DCP and phosphoric acid, is this possible??? | ||
| DJDahm |
In, general the thing that NIR Spectroscopy is most sensitive to is the frequency of vibration of specific bonds in molecules. It is particularly useful because almost every organic molecule with a hydrogen atom has some absorption in the NIR range. However, NIR is certainly sensitive to compounds other than organic compounds. Water is a prime example. There is a secondary sensitivity to species that have no NIR absorption in and of themselves, but which perturb the vibrational (absorption) frequencies of bonds that do. Analyses for inorganic species have been developed that are successful because of this effect. However, as a general rule: the less direct an effect being observed is; the more subject it is to interference. After you have been around NIR Spectroscopy a while, you learn not to say that an analysis is �impossible�. However, there are certainly cases where it is very wise to ask lots of questions lest we are being fooled by what is actually being observed. I think you gave some examples of those cases. | ||
| Dr Neil J. Dominy (Neil) |
I appreciate the reply. With respect to animal feeds it has been described that accurate inorganic predictions are rather a factor of other organic compounds which just happen to have a common ratio e.g in broiler diets protein and Ca follow a set ratio to each other. Thus it is explained that the "accurate" prediction is rather one of protein extrapolation rather than Ca determination. If I tested this by adding lime to a sample would it know the difference? Alternatively if the NIR is assuming Ca as a proportion of ash would it increase the Ca prediction if I were to add salt? | ||
| DJDahm |
Using standards in which there is an accidental correlation is viewed as a problem whenever it occurs. Usually one tries to devise a set of standards that breaks down the correlation, just as you described by adding Ca. If protein is what is being measured, then one should say that, rather than that Ca is being measured. You pose two very good experiments. Adding salt is an example of what might be tried to get at organic/inorganic interactions. However, beware that mixing separate particles of ingredients is not the same as mixing them intimately. [email protected] | ||
| hlmark |
Neil - just from curiosity, which part of the world are you from? Don gave some good advice. In general, you might also want to measure the spectrum of the pure feed additives of interest; some of them may have absorbance bands of their own and therefore behave more like independent constituents, with appreciable specificity. Many inorganics, however, have no NIR absorbance and those are more likely to affect only the scattering properties of a feed sample, which would be relatively non-specific. Measuring the spectra of the pure additives will allow you to tell which is which. There are actually several mechanisms that allow measurement of inorganics. These have different sensitivities and specificities, and apply in different circumstances. Ron Rubinovitz gave a dynamite presentation of this at last Pittcon, and there is a summary of his talk in the Pittcon report that Dave Hopkins and I wrote in the March/April issue of NIR News. Howard \o/ /_\ | ||
| Dr Neil J. Dominy (Neil) |
Thanks Don and Howard for the advice I will let you know the results of the experiments. I am from South Africa. Neil |