OPTIMIZATION OF 2 FOURIER-TRANSFORM INFRARED LEAST-SQUARES MULTIVARIATE-ANALYSIS METHODS FOR THE SIMULTANEOUS QUANTITATION OF MIXTURES OF 3METAL-CARBONYL-COMPLEXES IN THE PICOMOLE RANGE

In order to develop a simultaneous multi-carbonyl metallo immuno assay (multi-CMIA) associated with Fourier transform infrared spectroscopy as the detection method, we thoroughly studied two multivariate analys is methods for the quantitation of mixtures of metal-carbonyl complexe s from their infrared spectra restricted to the region of the nu(CO) v ibration modes (1850-2200 cm(-1)). These two methods were based on cla ssical least-squares (CLS) and partial least-squares (PLS) algorithms. The influence of some of the parameters involved in the calibration s tage (spectral intervals, nature and number of calibration spectral wa s studied for both multivariate methods. With optimized calibration mo dels, we showed that either artificial or real mixtures of three metal -carbonyl complexes could be simultaneously quantified in the 10-100 p mol range, which is the working area of CMIA experiments, with relativ e errors around 6%.