Abstract: A model system with blue dextran, potassium hexacyanoferrate(III) and heparin as analytes was used to generate second-order FIA data. The system incorporated a gel-filtration column (2.5 cm x 16 mm i.d.) packed with Sephadex G-25 Superfine. Photodiode array detection was at 220-594 nm with scanning at 1 s intervals for 75 s, starting 20 s after injection. The tri-linear parallel factor analysis (PARAFAC) algorithm (Smilde et al., J. Chemom., 1994, 8, 21) c ombined with multiple linear regression and tri-linear partial least-squares regression (tri-PLS; Bro, Ibid., 1996, 10, 47) were used for data processing. The prediction abilities of the two methods were satisfactory for two-component mixtures and the detection of unknown interferents. The tri-PLS method was slightly better.

Abstract: Frozen fish was thawed and 10 g portions were homogenized in 10 mL 10 mM Tris buffer of pH 7.8. Homogenates were centrifuged and the supernatant was filtered. Portions of the filtrate were injected in to a Tris buffer carrier of pH 7.8 (0.75 ml/min) and passed through a Sephadex G-25 (20-80 µm) gel filtration column to remove proteins, before injection into a stream of 50 mM phosphate buffer of pH 7.8 (0.75 ml/min). This merged with a similar stream (0.35 ml/min) into which formaldehyde dehydrogenase/NAD had been injected. After reaction at 40°C in a coil (75 cm x 0.5 mm i.d.) photodiode array detection at 340 nm was performed. The method was faster than standard methods and less affected by interferences. The limit of detection was 2.5 mg/l formaldehyde. A calibration graph for up to 50 mg/l is shown. Sample throughput was 10 per hour. The results were compared with those obtained by the Nash test.

Abstract: A FIA system with spectrophotometric diode-array detection was constructed to perform generalized standard addition calibrations. The two line system allowed an undiluted sample stream to be merged with a carrier stream into which the standards were injected. The flow the passed through a knitted coil to the detector cell. The method measured the concentration of the analyte by a single injection and also provided multivariate calibration and outlier detection. The method was demonstrated by determining methyl orange and bromothymol blue in buffered solutions. Samples containing one and two analytes were analyzed as well as samples containing direct interferences. The relative errors were up to 4% for samples containing one analyte and 2-7% for samples containing two analytes. A flow injection analysis system which implements the generalized standard addition method is constructed. The FIA system, which is a model system, is furnished with a photodiode-array spectrophotometer as a detector and is a two-line system one line transports the undiluted sample, which is merged with the second line, into which the standard is injected and dispersed to obtain different amounts of added standard. The performance of the constructed flow system and the partial least-squares solution to the generalized standard addition model is investigated with indicator samples, and an outlier detection technique capable of detecting direct interferences is implemented. By introducing two chasing zones in the standard stream a two analyte calibration space will be spanned, and a generalized standard addition can be performed. Samples containing one analyte and two analytes as well as samples containing direct interferences are analyzed to test the FIA system and the outlier detection method