672 M. Dubernet
Fig. 12.3Example of a wine mid-IR spectrum
12.5.6 Principal Steps in Spectral Analysis
An example of a wine IR spectrum is presented in Fig. 12.3.
Based on the data base constructed with reference samples of known values for
the target analytes, chemometric methods enable the most appropriate spectral zones
(usually about 10) to be selected and then a regression model is calculated which is
used to determine the concentration of unknown samples.
12.5.7 The Importance of Quality Control of the Results
It is important to point out that the use of FTIR requires that the laboratory imple-
ments a system to control the quality of the results, and this system ought to be
based on a specific and original strategy. The highly powerful automation of the
system allied to the simplicity of instrument operation constitute a risk in terms of
drift or of results that are either aberrant of unidentified. The quality of results can
only be assured by a very thorough system with strict identification of matrices,
the elimination of results of samples that do not comply with precise definition, the
use of a sufficient number of control samples and perfect data management. Each
laboratory should be aware that the time spent in controlling the quality of results in
FTIR is significantly greater than the analysis time, which is not the case with other
methods used in oenological laboratories.
It is now accepted that laboratories which have implemented the necessary mea-
sures to ensure quality control of results obtain, above and beyond an unequalled
facility of operation, considerable optimisation of laboratory performance, and an
overall quality of analysis that is vastly superior to that offered by other automated
methods. It is strongly recommended that laboratories employing FTIR in rou-
tine analysis retain alternative methodsfor each significant wine parameter to be
analysed.