CORRELATION-ANALYSIS IN LIQUID-CHROMATOGRAPHY OF METAL-CHELATES .3. MULTIDIMENSIONAL MODELS IN REVERSED-PHASE LIQUID-CHROMATOGRAPHY

A number of multiparametric retention models for metal chelates in rev ersed-phase high-performance liquid chromatography (RP-HPLC) were deve loped and compared. The most significant structural descriptors of the chelates and mobile phase parameters were selected with the help of o ne-dimensional correlation analysis performed in this and previously p ublished papers of this series. Log k' values for metal di-n-alkyldith iophosphates used as test solutes were determined using a C18 column a nd dioxane as organic modifier and subjected to multiple regression an alysis. The predictive ability of the resulting multiparametric regres sion equations was evaluated in terms of their statistical significanc e. The most meaningful retention model is described by the linear regr ession equation log k' = 0.716 + (0.236 +/- 0.010)n(C) - (0.040 +/- 0. 018)E(n) - (0.048 +/- 0.003)c + (0.029 +/- 0.010)Z (R = 0.976; S. D. = 0.111), where n(C) is the carbon number, E(n) is the orbital electron egativity of the metal atom, c is the volume concentration of the orga nic modifier and Z is the parameter of proton-donating ability of the mobile phase. The results confirm the modern representations of the se paration mechanism for metal chelates of moderate polarity in RP-HPLC, present rather valuable sets of solute structural descriptors and elu ent parameters to approximate the experimental retention values and op en new possibilities in the application of multivariate statistical me thods to interpret a large number of chromatographic data.