Determination of inorganic anions in water by ion chromatography: a collaborative study.

The U.S. Environmental Protection Agency (U.S. EPA) and the American Society for Testing and Materials (ASTM) conducted a joint collaborative study validating an ion chromatographic method for determination fo inorganic anions (U.S. EPA method 300.0A and the equivalent proposed revision to ASTM method D4327). This study was conducted to determine the mean recovery and precision of analyses for bromide, chloride, fluoride, nitrate, nitrite, orthophosphate, and sulfate in reagent water, drinking water, and wastewater. The study design was based on Youden's nonreplicate plan for collaborative tests of analytical methods. The test waters were spiked with the anions at 6 concentration levels, prepared as 3 Youden pairs. The 22 volunteer laboratories were instructed to dilute 10 mL sample concentrate to 100 mL test water. A measured volume of sample (20-200 microL) was injected into an ion chromatograph equipped with a guard column, anion exchange column, and a chemical micromembrane suppression device. The anions were then separated using 1.7 mM sodium bicarbonate and 1.8 mM sodium carbonate, and measured by a conductivity detector. Submitted data were evaluated using U.S. EPA's IMVS computer program, which follows ASTM D2777-86 statistical guidance. U.S. EPA method 300.0A and ASTM method D4327 were judged acceptable for measurement of the above anions (except sulfate) at concentrations ranging from 0.3 to 25 mg/L and sulfate concentrations from 2.9 to 95 mg/L. Mean recoveries for the 7 anions from all matrixes, as estimated from the linear regression equations, ranged from 95 to 104%. At concentrations above 2-6 mg/L for bromide, fluoride, nitrate, nitrite, and orthophosphate, and above 24 mg/L for sulfate, the overall and single-analyst relative standard deviations were less than 10 and 6%, respectively. As concentrations decreased, precision became more variable. The relative standard deviations of results for chloride were slightly higher than the other anions, especially in matrixes with high chloride background. Analysis of Variance (ANOVA) tests at the 95% confidence interval indicated a statistically significant matrix effect for chloride, nitrite, and nitrate analyses in drinking water compared to analyses in reagent water. Because these matrix effects were caused by the spiking process and not the drinking water itself, the ANOVA determination was not considered to be of practical significance.