Systematic optimisation and evaluation of on-line, off-line and stop-flow comprehensive hydrophilic interaction chromatography × reversed phase liquid chromatographic analysis of procyanidins, Part I: Theoretical considerations

• A systematic approach for the optimization of HILIC × RP-LC analysis was developed.
• Theoretical findings based on data for HILIC × RP-LC analysis of cocoa procyanidins.
• A detailed comparison of on-line, off-line and stop-flow modes is presented.
• Stop-flow operation is shown to be a promising alternative to off-line LC × LC.

Comprehensive two-dimensional liquid chromatography (LC × LC) provides significantly improved separation for complex real-life samples. LC × LC can be performed in one of three different ways, using on-line, off-line, or stop-flow configurations. We have previously shown how off-line comprehensive hydrophilic interaction chromatography (HILIC) × reversed-phase liquid chromatography (RP-LC) provides a powerful separation system for procyanidins (PCs), one of the most complex fractions of natural phenolics. In the current contribution, a systematic approach for the optimisation and evaluation of each of the LC × LC methodologies is presented using HILIC × RP-LC analysis of PCs as application. Optimisation was performed using the peak capacities of individual one-dimensional separations measured for different gradient times and flow rates and their combination in each of the three LC × LC modes by taking into account the effects of first dimension under-sampling, the degree of orthogonality between the two dimensions and additional band broadening associated with stop-flow analysis. The performance of all three methods is compared in terms of practical peak capacities, analysis times and peak production rates. One-dimensional LC provided the best performance for separations requiring relatively low peak capacities, whereas the on-line LC × LC system was advantageous for required practical peak capacities up to ~600. For higher resolution, the off-line or stop-flow systems should be used. Especially noteworthy is the fact that, due to slow diffusion of PCs, the contribution of stop-flow to first dimension band broadening was negligible for stop-flow times of up to 15 min. In a separate contribution, the experimental verification of the findings of this study will be reported.