Reversed-phase liquid chromatography system constant database over an extended mobile phase composition range for 25 siloxane-bonded silica-based columns

Abstract

The solvation parameter model is used to characterize the retention properties of 25 siloxane-bonded type-B silica columns under typical reversed-phase liquid chromatography conditions with binary aqueous mobile phases containing 10–70% (v/v) methanol and acetonitrile. To further explore the affect of solvent type on column selectivity complementary information is presented for aqueous phases containing 10–70% (v/v) tetrahydrofuran for two columns. Columns were selected to include examples of all common column packing morphologies for small molecule separations (totally porous particles, superficially porous particles, organic–inorganic hybrid particles, and a monolith) and common topologies octadecylsiloxane-bonded (including phases with a polar-embedded group, mixed-mode phases with a short-chain siloxane-bonded polar functional group, sterically crowded, positive shield, and fluorine-containing phases); octylsiloxane-bonded; and various phenyl-containing stationary phases with different linker arms, pentafluorophenylalkyl and biphenyl groups. The column properties are reported as a system constant database and as system maps for the full range of mobile phase compositions. Selectivity differences are evaluated using principal component factor analysis to classify the columns into selectivity groups and correlation plots of the system constants for compared columns at all mobile phase compositions. The later is shown to be a suitable technique to identify (near) selectivity equivalent columns and varied columns suitable for method development. Contributions to retention from steric resistance and cation-exchange interactions not parameterized in the solvation parameter model are identified for columns and conditions where they may be important. It is shown that the interaction parameters employed in the hydrophobic-subtraction model have little overlap with those of the solvation parameter model and a quantitative comparison of column properties delineated by both models is not possible.