Σpider diagram: a universal and versatile approach for system comparison and classification: application to solvent...

Classification methods based on physico-chemical properties are very useful in analytical chemistry, both for extraction and separation processes. Depending on the number of parameters, several classification approaches can be used: by plotting two- or three-dimensional maps (triangles, cubes, spheres); by calculating comparison values for one system with reference to another one, i.e. the ranking factor F, or the Neue selectivity difference s(2); or with chemometric methods (principal component analysis-PCA or hierarchical cluster analysis-HCA). All these methods display advantages and drawbacks: some of them are limited by the number of studied parameters (e.g. three for triangle or sphere plots); others require a new calculation when changing the reference point (F; s(2)), while for chemometric methods (PCA, HCA), the relationships between the clusters and the physico-chemical properties are not always easily understandable. From previous studies performed in supercritical fluid chromatography for stationary phase classification on the basis of linear solvation energy relationships (LSER) including five parameters, we developed a classification map called the Σpider diagram. This diagram allows plotting in a two-dimensional map the location of varied systems, having as many parameters as the ones required getting a satisfactory classification. It can be three, five, eight, or any number. In the present paper, we apply this diagram, and the calculation mode to obtain this diagram, to different solvent classifications: Snyder triangle, solvatochromic solvent selectivity, Hansen parameters, and also to LSER Abraham descriptors and COSMO-RS parameters. The new figure based on Snyder data does not change the global view of groups, except by the use of corrected data from literature, and allows adding the polarity value onto the map. For the solvatochromic solvent selectivity, it leads to achieve a better view of solvents having no acidic character. For Hansen parameters, the "flattening" of the spherical view down to a single plane could be found easier to use. For COSMO-RS and with Abraham descriptor, a more subtle classification is achieved, mainly due to the use of five parameters instead of three. A strong reversed correlation is established between the Rohrschneider polarity P' and the normalized V (molecular volume) parameter. The study of the location of solvents used for reversed-phase liquid chromatography and the Arizona system for counter-current chromatography is discussed, as well as the replacement of unsafe solvents by greener ones, or the use of these classifications for the study of compound solubility. Besides, this paper also shows the ability to the spider diagram to plot on a single plane three axes from principal component analyses.