Simulation of elution profiles in liquid chromatography - II: Investigation of injection volume overload under gradient elution conditions applied to second dimension separations in two-dimensional liquid chromatography.

An important research direction in the continued development of two-dimensional liquid chromatography (2D-LC) is to improve the detection sensitivity of the method. This is especially important in applications where injection of large volumes of effluent from the first dimension (1 D) column into the second dimension (2 D) column leads to severe2 D peak broadening and peak shape distortion. For example, this is common when coupling two reversed-phase columns and the organic solvent content of the1 D mobile phase overwhelms the2 D column with each injection of1 D effluent, leading to low resolution in the second dimension. In a previous study we validated a simulation approach based on the Craig distribution model and adapted from the work of Czok and Guiochon [1] that enabled accurate simulation of simple isocratic and gradient separations with very small injection volumes, and isocratic separations with mismatched injection and mobile phase solvents [2]. In the present study we have extended this simulation approach to simulate separations relevant to 2D-LC. Specifically, we have focused on simulating2 D separations where gradient elution conditions are used, there is mismatch between the sample solvent and the starting point in the gradient elution program, injection volumes approach or even exceed the dead volume of the2 D column, and the extent of sample loop filling is varied. To validate this simulation we have compared results from simulations and experiments for 101 different conditions, including variation in injection volume (0.4-80μL), loop filling level (25-100%), and degree of mismatch between sample organic solvent and the starting point in the gradient elution program (-20 to +20% ACN). We find that that the simulation is accurate enough (median errors in retention time and peak width of -1.0 and -4.9%, without corrections for extra-column dispersion) to be useful in guiding optimization of 2D-LC separations. However, this requires that real injection profiles obtained from 2D-LC interface valves are used to simulate the introduction of samples into the2 D column. These profiles are highly asymmetric - simulation using simple rectangular pulses leads to peak widths that are far too narrow under many conditions. We believe the simulation approach developed here will be useful for addressing practical questions in the development of 2D-LC methods.