Analysis of linear and cyclic oligomers in polyamide-6 without sample preparation by liquid chromatography using the sandwich injection method. III. Separation mechanism and gradient optimization.

The first six linear and cyclic oligomers of polyamide-6 can be quantitatively determined in the polymer using HPLC with the sandwich injection method and an aqueous acetonitrile gradient. In this final part of the triptych concerning the determination of the oligomers in polyamide-6, the irregular elution behavior of the cyclic monomer compared to the cyclic oligomers was investigated. We also optimized the separation of the involved polyamide oligomers, with respect to gradient steepness, stationary phase, column temperature and mobile phase pH. The irregular elution behavior of the cyclic monomer could be caused by its relatively large exposed/accessible hydrophobic surface, which permits relatively easy penetration into the hydrophobic stationary phase giving extra retention. The dipole moment of the different oligomers was used as a measure for this exposed/accessible hydrophobic area to correlate the retention factors using quantitative structure-retention relationships. We also studied the retention behavior of the polyamide, which is injected each run directly onto the column and modifies the stationary phase. Using a 250-microl post gradient injection zone of formic acid on a 250x3 mm Zorbax SB-C18 column, the polyamide could be effectively removed from the stationary phase after each separation. The linear solvent strength (LSS) model was used to optimize the separation of the first six linear and cyclic oligomers. As the LSS model assumes a linear correlation between the modifier concentration and the logarithm of the retention factor and the cyclic monomer and dimer show extreme curvation of this relation in the eluting region, we investigated different models to predict gradient elution from isocratic data. A direct translation of the isocratic data to gradient retention times did not yield adequate retention times using the LSS model. It was found that the LSS model worked acceptably if gradient retention times were used as input data. Even for fast non-linearly eluting components, an average error of 0.4 resolution units of 4sigma was obtained. Using the LSS model in combination with different column temperatures and mobile phase pH values, a separation of the first six linear and cyclic oligomers was accomplished.