Flow rate, reaction coils length, reagent concentration were maintained throughout, thus the variable assayed aiming to find better sensitivity was the ratio between sample and reagent aliquots inserted into the analytical path. The volume of the sample slug was maintained (66.4 l), while the volume of the reagent solution was varied yielding the results shown
in Table 3. Results obtained for nitrite were similar to those achieved for nitrate, thus they were not showed. Considering sensitivity as the main parameter better results was achieved when the volume of the reagent solution was 66.4l. These
experiments were carried out by settling 10 sampling cycles, thus the overall volume of the sample zone varied from 960
up to 1660l. The volume of the analytical path comprising the volumes of the reaction coil and flow cells was 940 l. This volume was lower than the volume of sample zone, thus minimizing dispersion effect. In this sense, variation in sensitivity
in the first and second cases could be caused because reagent into the sample bulk was not enough to attain the reaction
stoichiometry and in the last one it could be caused by sample dilution owing to reagent slug was higher than sample
slug. A set of river water was analyzed selecting the operational condition shown in Table 3 (line 3). To allow accuracy assessment the samples were also analyzed employing the usual FIA procedure [29] yielding the results shown in Table 4. A applying the paired t-test for both analytes were found 2.23 while the reference value was 2.57, thus indicating
that no significant difference at 95% confidence level was observed. Other advantages such as low reagent consumption (75g) and reduced waste generation (1.6 ml) per determination and a sampling throughput of 40 determinations per hour were also achieved