12 Automatic Analysers in Oenology 671
12.5.5.3 Matrix Effects
It has been outlined above that matrix effects are due to other compounds present
in the analytical medium which absorb in the mid-IR. It is possible, however, to
circumvent matrix effects. As in expert systems, all that is required is the availability
of the necessary information so that the instrument can, by experience, measure the
impact of the matrix effect on the determination of the target compound. Matrix
effects may thus be eliminated by reference to the calibration value whenever the
analyte occurs in an unknown sample type. At this stage, it is obvious that the greater
the number of different matrices presented to the instrument during calibration,
the more robust the method will be, and less subject to errors arising from matrix
effects. This is one of the strong points of the FTIR method, but nonetheless there
in an inherent disadvantage. The weakness lies in the fact that the more elaborate
of the instrument’s capacity to manage matrix effects, the lower its performance
in terms of sensitivity and precision. Therefore, the operator developing the cali-
bration is obliged to seek the optimum compromise between robustness, sensitivity
and precision. Here again, the approachrapidly becomes extremely complex and
its implementation can only be envisaged ifsignificant means are available for that
purpose. In oenological practice, and the level of understanding permitting, wines
and musts are grouped into five basic classes of matrices:
Unfermented musts
Musts in fermentation
Dry wines or wines containing low sugar concentrations
Liqueur wines
Naturally sweet winesIt is important to point out that matrix effects engendered by different grape
varieties, wine colour or geographical origin are in general of no consequence in
well-executed calibrations.
The complexity of matrix effects has another important consequence: it is impos-
sible to use synthetic matrices or even samples spiked with the analytes, even for the
controls – only natural wines and musts can be applied for this purpose.
12.5.5.4 Reference Values
The quality of a calibration also depends on the quality of the reference samples
used. Glucuronic acid is a particular example; the enzymatic method for this analysis
is not reliable in that it lacks robustness. This problem has not been identified in the
literature and is probably linked to non-identified matrix effects. The first FTIR
calibrations were carried out using the enzymatic method as a reference and, even
though gluconic acid hashighly qualitative absorption characteristics, the results
obtained by FTIR were unreliable. Subsequent calibrations carried out using more
reliable methods such as capillary electrophoresis as a reference technique have
resulted in FTIR analyses of high quality in terms of precision and accuracy.