Second-Order Analyte Quantitation under Identical Profiles in One Data Dimension. A Dependency-Adapted Partial Least-Squares/Residual Bilinearization Method

Analyte quantitation can be achieved from second-order data in the presence of uncalibrated components using multivariate calibration methods such as partial least-squares with residual bilinearization. However, the latter fails under conditions of identical profiles for interfering agents and calibrated components in one of the data dimensions. To overcome this problem, a new residual bilinearization procedure for linear dependency is here introduced. Simulated data show that the new model can conveniently handle the studied analytical problem, with a success comparable to multivariate curve resolution-alternating least-squares and also comparable to a version of parallel factor analysis adapted to cope with linear dependencies. The new approach has also been applied to two experimental examples involving the determination of the antibiotic ciprofloxacin in (1) urine samples from lanthanide-sensitized excitation time decay matrixes and (2) serum samples from a novel second-order signal based on the time evolution of chemiluminescence emission. The results indicate good analytical performance of the new procedure toward the analyte in comparison with the classical approaches.