IASIM 2018 Advertisement

Ratio or background correction?

ianm's picture

9. Ratio or background correction?

A number of companies selling spectrometers are now putting an internal beam in the instrument, similar to a double beam. That can be used by either obtaining a background of the fiber optic probe/cell vs the internal "cell" or by ratioing the spectrum against the cell without taking a background - a situation one would encounter in an in-line installation. The question is, what are the advantages and disadvantages of not periodically taking a background against this internal cell.



From: "W. F. McClure" [email protected]


This is a problem as old as spectroscopy itself (see Cowles, Theory of dual wavelength spectrophotometry for turbid samples, 1951). Bottom line to me appears to depend on the stability of the instrument. The extremes are these: (1) Take a background every sample and (2) take only one at the beginning of the run. Where in between depends on how stable the opto-electronics are.


I have had my ups and downs on this one for a while. The convenience of the second fiber are obvious: greater throughput and less programming. The downside is, of course, that the sample path is not truly being compensated for. That is, if any degradation occurs in the sample bundle, the light going through the reference path is nor effected. Thus, in a long process, there would appear to be a baseline drift. The analyst may attribute this drift to a chemical change and not consider the possiblity of, say, a build-up on the fiber optic face or deterioration of the fibers themselves.

While this may be a straw-man, it must be taken into account. Somewhere, possibly after the run, a blank should be run through the measuring fiber. In a similar precaution, periodic cleaning of the probe should be scheduled. During a run, it may be easier to just run a blank in the normal manner. I haven't done much on-line measuring, so these are just my thoughts on the subject. I am anxious to hear what experienced on-liners have to say.