A generalised modelling approach to predict retention in surfactant-mediated liquid chromatographic modes

Analysing polar solutes in reversed-phase liquid chromatography (RPLC) is challenging due to their inherently low retention on the non-polar stationary phases typically used with relatively polar mobile phases. Hydrophilic interaction liquid chromatography offers an alternative by combining a highly polar stationary phase with an organic solvent-rich mobile phase to enhance retention. Another promising strategy involves introducing surfactants into the mobile phase to modify the RPLC stationary phase, thereby tuning analyte retention while minimising the use of organic solvent. This work investigates the chromatographic behaviour of various polar compounds, including nucleosides, thiazide diuretics, and sulphonamides, under surfactant-mediated conditions. Retention is modelled as a function of the concentration of the anionic surfactant sodium dodecyl sulphate (SDS), using predictive approaches based on the Sips and Langmuir isotherms, which describe the adsorption of surfactant monomers from the mobile phase onto the stationary phase. The retention models were validated against experimental data obtained across a range of SDS concentrations in the presence of acetonitrile, yielding three distinct chromatographic conditions: hydro-organic (no surfactant), submicellar (surfactant below the critical micelle concentration, CMC), and micellar (surfactant above the CMC). Acetonitrile plays a dual role by increasing the CMC and simultaneously decreasing analyte retention. To address this, the models were extended to incorporate both surfactant and organic solvent concentrations. These generalised models enable fine-tuned control of retention through rational selection of surfactant and organic solvent levels, without requiring prior knowledge of the dominant retention mechanisms in each compositional region (hydro-organic, submicellar, or micellar). This is particularly valuable given the complexity introduced by changes in micellar dynamics and stationary-phase heterogeneity. Experimental data fitted with the model based on the Sips isotherm showed good agreement with predicted values across the entire compositional range, with relative errors between 2.5 and 4.4 %, making the model suitable for optimisation purposes.