3.3. Analytical features and application
The proposed procedure was efficient to separate nitrate from the main interfering species for the analysis of natural waters. Solutions containing humic acid were employed to simulate natural organic matter, and results obtained with (a) and without (b) separation are shown in Fig. 4. While severe interferences were observed in the presence of 10 mg L−1 humic acid without separation, concentrations as high as 200 mg L−1 humic acid did not interfere in the nitrate determination when applying the proposed procedure. Similar effect was observed in the presence of Fe3+ that inter-feres in the direct determination in concentrations lower than 5.0 mg L−1. By admitting signal variations lower than 5%, up to 1.0 mg L−1 NO2−; 200.0 mg L−1 PO43−; 75.0 mg L−1 Cl−; 50.0 mg L−1 SO42− and 15.0 mg L−1 Fe3+ also did not interfere in the determination of 5.00 mg L−1 NO3− by the proposed procedure. These concentrations thresholds for manifestation of interferences allow the application of the procedure for the determination of nitrate in freshwaters [20–22]. Even nitrite that absorbs radiation in the samewavelength range is tolerated in concentration higher than typically found in natural waters. As a consequence of the efficiency of retention and elution of nitrate, it was evaluated that for a 500-_L sample volume, the sensitivity was 40% lower than that attained without separation. However, if necessary, sensitivity can be improved by increasing the sample volume. Absorbance is directly pro-portional to the sample volume at least up to 1500 _L: A = 0.061 + 0.00049 V (_L), r = 0.999. For a 500_L sample volume, linear response was attained within 0.50–25.0 mg L−1, described by the equation A = −0.049 + 0.078C (mg L−1), r = 0.999. The detection limit was estimated as 0.10 mg L−1 at the 99.7% confidence level and the coefficient of variation was 0.7% (n = 20). The sampling rate was estimated as 17 determinations per hour, and 300_mol HClO4 (equivalent to 18_L concentrated HClO4) is consumed per determination. Transient signals for samples and reference solutions are shown in Fig. 5. Results obtained for natural waters samples (Table 2) were in agreement with the reference method [17] at the 95% confidence level. It should be mentioned that errors from 14% to 60 % were observed when these same samples were analyzed by UV spectrophotometry without separation.