Predictive Retention Modeling in Gradient Ion-Pair Chromatography: Effects of Acetonitrile and Ion-Pair Reagent Concentration

Ion-pair reversed-phase liquid chromatography (IP-RPLC) is a useful technique for separating charged analytes, but retention prediction remains challenging due to the combined influence of ion-pair reagent (IPR) concentration and organic solvent content. The aim of this study was twofold: to investigate how acetonitrile (MeCN) content affects the retention of sulfonated analytes at different concentrations of dibutylamine (DBuA), and to evaluate whether retention under gradient elution can be predicted based on isocratic data.Isocratic runs were performed at several MeCN levels (10–30 % w/w) and DBuA concentrations (0–30 mM) to explore how these parameters interactively affect retention. Retention data were also used to estimate surface potential as a function of MeCN content, allowing the construction of a time-resolved model for gradient elution. The model was based on a combination of the linear solvent strength (LSS) theory and the electrostatic retention model proposed by Ståhlberg, in which retention is described as an exponential function of surface potential. Gradient runs were then carried out with constant DBuA concentration while MeCN content increased linearly over time.The model showed reasonable agreement with experimental retention for two of the analytes, but significantly larger deviations were observed for the third. In this case, the analyte’s behavior did not align with the assumptions of the electrostatic model, which made the prediction model sensitive to small variations in surface potential. This reduced the accuracy of the predictions. Despite these limitations, the applied gradient conditions enabled full elution of all analytes within a practical timeframe and improved peak shape compared to isocratic methods.These findings suggest that empirical modeling based on isocratic data may aid gradient method development in IP-RPLC, although further refinement is needed to improve prediction accuracy.