Three-way principal component analysis as a tool to evaluate the chemical stability of metal bearing residues from wastewater treatment by the ferrite process

The chemical fractionation patterns of eight metals (Cd, Co, Cu, Cr, Mn, Ni, Pb and Zn) have been determined in 27 metal-bearing residues by using the BCR sequential extraction procedure. The residues were generated as by-products during the optimization of a semi-continuous reactor for metal removal from wastewater based on ferrite synthesis by co-precipitation. The three-dimensional X dataset (samples × metals × fractions) obtained by applying the BCR procedure has been analyzed by multivariate methods: matrix augmentation (MA-PCA) and three-way principal component analysis, 3-PCA (PARAFAC and Tucker3 models). MA-PCA and PARAFAC methods led to two-factor models giving a satisfactory but incomplete picture of the metal fractionation patterns, but the Tucker3 [2,1,2] model allowed to simultaneously describe both the 'pseudo-total' (acid-soluble) contents and the chemical fractionation by means of two non-null interactions g111 and g212 which explain 53.5% and 18.0% of the total variance, respectively. The A-mode loadings of the g212 interaction showed the close relationship between the magnetic character of the solid residues, i.e. the crystalline structure, and the chemical fractionation patterns of the metals resulting from the application of the BCR sequential extraction procedure.