Molecular-Dynamics Simulation for Liquid Chromatographic Interactions: Effect of Mobile Phase Composition
High-performance liquid chromatography (HPLC) is one of the most powerful methods used to separate a wide range of mixtures and to determine a quantity of compounds in many samples. Despite the widespread use of HPLC in both analytical and preparative applications, the separation mechanism has not yet been completely clarified. Therefore, method development and optimization are still often based on empirical "trial and error" approaches. In order to avoid such tedious procedures, several groups
... res, several groups have been trying to establish methods to predict retention and to optimize the separation conditions. 1-8 One of their methodologies is to use various database systems which contain the experimentally obtained retention data. 1-4 Another approach is the prediction of retention based on Quantitative Structure Retention Relationships (QSRR) and Linear Solvation Energy Relationships (LSER). 1,     In those approaches, understanding the retention mechanism is the key. Although those works have provided much information about the mechanism, a full understanding of the retention in HPLC is far from being used for retention prediction. In addition, most of those investigations are mainly based on the interaction between the solute and the stationary phase, and some contributions of the mobile phase to the retention mechanism are neglected. Some computational methods have recently been developed to search for the conformation of proteins and biomolecules. It is well known that the molecular dynamics simulation method (MD method) is one of the most powerful tools to investigate the time-series behavior of a molecular system which contains many molecules. The advances of computer technologies have made it possible to easily handle calculations for a huge molecular system. Recently, some studies have applied this approach to investigate the retention mechanism on reversedphase HPLC, 9-16 size-exclusion chromatography (SEC), 17-19 and chiral separation.    In this work, an MD simulation was carried out to investigate the influence of the mobile-phase composition to the retention of solutes in HPLC, to discuss the intermolecular distance between the solute and the stationary phase (n-octadecyldimethylsilyl bonded phase, ODS); the conformations of the ODS stationary phase were determined by theoretical calculations. Experimental Molecular modeling Two types of molecular models were constructed to investigate the contribution of the mobile phase to the retention. One contains only ODS ligands; the another contains ODS ligands, a mobile-phase solvent mixture of methanol and water, and n-propylbenzene as the solute molecule in each molecular system. In this study, the silica gel molecules as the stationary phase support were not contained in the molecular model to reduce the calculation time. Each molecular model is displayed in Fig. 1 . In two systems, the number of ODS molecules and the total solvent molecules are fixed as 3 and 54, due to the limitation of the calculation system which we used, respectively. Periodic boundary conditions with the unit-cell dimension (15 × 15 × 30 Å) were applied to generate a mimic of each molecular model. The ODS ligands are arrayed by about a 7.3 Å interval, which was estimated by a calculation based on the ligand density and the surface area of a commercially available ODS column, Discovery C18 (Supelco, PA, USA). The positions of silicon atoms in the ODS ligands are fixed during the MD simulation. The mobile phase-molecules are randomly positioned and the conformation of ODS chains has 113 A molecular-dynamics simulation method has been applied to investigate the influence of the mobile-phase composition on the retention of solutes in HPLC. The distribution profiles of the distance between two atoms in ODS ligands were constructed to characterize the conformation of ODS ligand molecules. The distinct difference of ODS conformation is observed by comparing molecular models consisting of solvent molecules at each solvent composition. The distribution profiles of the distance between the mobile-phase solvent molecules and ODS ligand molecules were also constructed to characterize the distribution of the solvent molecules at each composition. In all distribution profiles, the difference in the distribution due to a change in the solvent compositions was very clearly found, and the facts seem to be very reasonable. The distribution profiles of the distance between the solute, n-propylbenzene, and the terminal carbon atom in the ODS ligand, and between the solute and the silicon atom in the ODS ligand have been also constructed to see the distribution of the solutes in the separation system. The calculated solute distribution in the ODS-methanol/water system is very consistent with the actual chromatographic retention behaviors.