Systematic salt effects in the automated determination of nutrients in seawater

Abstract

The Bran + Luebbe TRAACS continuous flow analyser, commonly used for the determination of nutrients in seawater, primarily employs standard methods that have been modified and adapted for automated use. The analyser is fitted with self aligning spectrophotometric flow-cells with parallel optical ends of uniform thickness. As the light beam strikes both ends of the flow-cell perpendicularly, there is no refraction therefore no spurious absorbance caused by changes in the refractive index of the flow-cell contents. Investigation of methods for the determination of ammonia, nitrate, nitrite, orthophosphate, soluble reactive silica and urea, using TRAACS, confirmed that this was the case. Also, for the methods assessed, the absorbance signal was not biased by either sample turbidity or wetting agents. Chemical interference from salt was assessed by analysing nutrient spiked solutions, over the 0–35 ppt salinity range, relative to working calibration solutions prepared using demineralised water. A major salt effect which caused an apparent decrease in concentration, was detected over the analytical working ranges in the ammonia and nitrate methods. The salt effect results from a chemical interference which causes the production of reaction products to vary as a function of salinity. The salt effect is totally unrelated to refraction and produces a concentration error in the ammonia and nitrate methods up to 35 and 25% respectively. The degree of salt effect was shown to be dependent on both the salinity and true nutrient concentration of the sample. Over the salinity range 8.75–35.00 ppt, corrective equations were derived for ammonia and nitrate to express the true nutrient concentration in terms of the observed concentration and the salinity of the solution. Within the range 0–8.75 ppt no correction was required for ammonia wheras the nitrate showed a 3.5% increase in apparent concentration at 8.75 ppt. The orthophosphate, soluble reactive silica, urea and nitrite methods showed no significant salt effect as the coefficient of variation (CV) of the observed nutrient concentrations over the 0–35 ppt salinity range was within the accepted CV or analytical precision of the methods. The ammonia and nitrate corrective equations were applied in assessing the true nutrient content of “depleted” or low nutrient seawater, used in the preparation of working calibration solutions, after analysing relative to working calibration solutions prepared in demineralised water.