Sources of systematic bias in the measurement of water by the coulometric and volumetric Karl Fischer methods.

The sources of systematic bias in the measurement of moisture by the volumetric and coulometric Karl Fischer methods were assessed. Using water-saturated octanol as a moisture standard, the measurement accuracy of five coulometric instruments was evaluated. Six possible sources of systematic bias were examined: accuracy of the moisture standard, nonadjustable instrumental bias, operator-adjustable instrumental bias, solvent composition, cell design, and sample composition. The published water content of water-saturated octanol was confirmed by the method of standard additions. The nonadjustable instrumental bias consisted of two types, one that was variable and was observed at water levels below 200 micrograms of water, and one that represented a constant percentage of the total water over the tested range from 40 to 400 micrograms of water. The adjustable instrumental parameters, if set incorrectly, may cause a small but significant negative bias (< 7%). Solvent composition as a function of both solvent type and titration vessel design can introduce a negative bias of up to 10%. Finally, the nature of the sample introduces a negative bias as demonstrated by the results obtained from titrating water in four different oil samples. Under optimum conditions, different amounts of water were measured in each oil using different Karl Fischer titration procedures and five different instruments.