reagents on the measurements made. The precision and accuracy of the procedure should then be
assessed by the analysis of either synthetic or 'spiked' samples. In the latter method, the recovery of a
known amount of analyte added to an actual sample matrix is checked. Finally, the procedure selected
should be compared with other procedures available. The following examples will serve to illustrate
some aspects of sampling problems, method selection and procedures.
Evaluation of Methods for the Determination of Fluoride in Water Samples
Example 12.2.1
In selecting a suitable method for the determination of fluoride at low levels (0.1–5 ppm), reference to
analytical texts and the literature reveals that there are a large number of methods available, most of
which are spectrophotometric. A critical appraisal of such methods was required by a laboratory
involved in the routine determination of fluoride in drinking and river waters. Eight methods selected
from the literature are listed in Table 12.1. The distillation, titrimetric and ion-exchange methods (6–8)
were rejected, without further investigation, on the basis of published data, the findings of which are
summarized in the table. The four spectrophotometric methods included three based on the bleaching
effect of fluoride ions on a coloured metal complex (1–3) and one in which a coloured fluoride complex
is formed directly (4). It was decided to investigate these four and the potentiometric method on a
comparative basis in the laboratory. The particular aspects to be considered included precision,
accuracy, sensitivity, interferences, speed, convenience and cost. For the spectrophotometric methods,
temperature effects, the stability of the colour formed and of the reagent solutions were also considered.
All measurements were made using the same set of instruments, and reagents were of analytical grade
wherever possible. Sodium fluoride (dried at 120°C for 2 hours) was used to prepare a 500 ppm
standard fluoride solution which was stored in a polythene container and from which more dilute
solutions were prepared as required. Precision was evaluated statistically, estimates of standard
deviations being based on at least ten replicates at each of two concentration levels within the range of
fluoride expected. One of the most important considerations was the effect of interferences, particularly
those which reduce the concentration of free fluoride ions by complex formation or precipitation, e.g.
hydrogen ions, aluminium, iron and calcium. The three metals in particular are frequently present in
water samples and must be removed or complexed to ensure reliable results. Only in the case of
measurements with the fluoride ion electrode was the problem of interferences easily and reliably
eliminated. This was achieved by using a 'total ionic strength adjustment buffer' consisting of an inert
electrolyte of high ionic strength (e.g. 1 M KCl) to which were added one or more of the following:
sodium acetate, sodium citrate, EDTA, sodium hexametaphosphate (Calgon).