EVALUATION OF A PREDICTIVE STEADY-STATE FLOW-INJECTION METHOD ADAPTED TO AN OPEN FLOW TUBE WITH A TRACER

This paper describes the development and evaluation of a new approach to processing data from now-injection systems. In this new approach, data from the leading edges of response peaks are used with appropriate mathematical models and curve-fitting methods to predict the steady-state signal that would be measured if sufficient sample were used to fill the flow tube with undiluted sample. This predictive steady-state data-processing option is adapted to and evaluated for open flow tubes. Predictive steady-state results for triiodide detected amperometrically are compared with results obtained by processing the same data sets by the more conventional peak-height, peak-width and peak-area options as well as directly measured steady-state signals of undiluted samples. All data-processing options yielded linear calibration plots, with deviations from linearity being significantly larger for the peak-width option than for the other options. Effects of changes in sample volume and flow-rate on quantitative results obtained with the different options were compared. Minimum sample volumes required by the steady-state methods are larger than those required for the other methods. However, both measured and predicted steady-state signals were much less dependent on sample volume and flow-rate than peak heights, peak areas or peak widths. Relative error coefficients for the steady-state options were 0.006%/mu l for changes in sample volume and 90%/ml s(-1) (1.5%/ml min(-1)) for changes in now-rate. Improvements in relative error coefficients for changes in sample volume of the predictive option relative to peak-height, peak-width and peak-area options are 55-, 633- and 167-fold, respectively. Improvements for changes in now-rates for detectors that do not depend on flow-rate are 12-, 226-, and 36-fold, respectively (same order as above). The steady-state options are much more rugged than the peak-height, peak-area and peak-width methods.