Assessing the feasibility of stationary-phase-assisted modulation for two-dimensional liquid-chromatography separations

Two-dimensional liquid chromatography (2DLC) offers great separation power for complex mixtures. The frequently encountered incompatibility of two orthogonal separation systems, however, makes its application complicated. Active-modulation strategies can reduce such incompatibility issues considerably. Stationary-phase-assisted modulation (SPAM) is the most-common of these techniques, but also the least robust due to the major disadvantage that analytes may elute prematurely. The range of liquid chromatography (LC) applications continues to expand towards ever more complex mixtures. Retention modelling is increasingly indispensable to comprehend and develop LC separations. In this research, a tool was designed to assess the feasibility of applying SPAM in 2DLC. Several parameters were investigated to accurately predict isocratic retention of analytes on trap columns under dilution-flow conditions. Model parameters were derived from scanning-gradient experiments performed on analytical columns. The trap-to-trap repeatability was found to be similar to the prediction error. Dead volumes for the trap columns could not be accurately determined through direct experimentation. Instead, they were extrapolated from dead-volume measurements on analytical columns. Several known retention models were evaluated. Better predictions were found using the quadratic model than with the log-linear (“linear-solvent-strength”) model. Steep scanning gradients were found to result in inaccurate predictions. The impact of the dilution flow on the retention of analytes proved less straightforward than anticipated. Under certain conditions dilution with a weaker eluent was found to be counter productive. A tool was developed to quantify the effect of the dilution flow and to predict whether SPAM could be applied in specific situations. For nine different analytes under 36 different sets of conditions and with three different modulation times, the SPAM tool yielded a correct assessment in more than 95% of all cases (less than 5% false positives plus false negatives).