Application of two- and three-way principal component analysis to the interpretation of chemical fractionation results obtained by the use of the B.C.R. procedure

This paper describes as two- and three-way principal component analysis (PCA) can be advantageously used to interpret the results originating from the B.C.R. fractionation scheme, an operationally defined speciation procedure used to study the availability and mobility of trace elements and heavy metals present in environmental solid samples. The application of this procedure to find the fractionation of 11 trace elements and heavy metals (Al, As, Cd, Cr, Cu, Mn, Mo, Ni, Pb, V and Zn) in 13 sediments collected at the Mejillones del Sur bay (Antofagasta, Chile) generated a three-dimensional data set X (13 × 11 × 4). Although classical two-way PCA applied to the unfolded Xb matrix (52 × 11) can yield valuable information about the pattern of fractionation of the chemical elements amongst the B.C.R. fractions, a more profound insight can be found when applying three-way PCA procedures, such as Parafac, that take into account the true tridimensional structure of the data set. In this study, Parafac has allowed to classify the sediments according to their associated environmental hazard by means of the A-mode loadings of two significant factors. By plotting these factors into the physical space of the Mejillones del Sur bay, the most hazardous areas of the bay have been located.