Molecular Dynamics Investigations of Mobile-Phase Contributions to Analyte Retention and Transport in Reversed-Phase Liquid Chromatography
Loading...
Date
Authors
Publisher
Abstract
This work deals with the influence of the mobile phase on the retention and selectivity of analytes in reversed-phase liquid chromatography (RPLC), the most important mode of high-performance liquid chromatography (HPLC). In RPLC, apolar modified stationary phases and polar mobile phases consisting of a mixture of water (W) and an organic solvent (OS) such as acetonitrile (ACN) or methanol (MeOH) are used to separate moderately polar to apolar analytes. Retention of analytes can be controlled by the elution strength of the mobile phase, which is influenced by the OS type and fraction. Although RPLC has been used successfully for many years, the physicochemical principles of the elution strength and the exact effects of the mobile phase on separations are still not fully understood. Many of the processes that are crucial for retention take place at the molecular level and are difficult to access through experiments. Molecular dynamics (MD) simulations allow the direct investigation of the chromatographic interface, enabling the systematic examination of the influence of the mobile phase on retention. In this work, MD simulations of an RPLC model mesopore filled with W–MeOH and W–ACN mixtures are carried out to investigate the influence of the mobile phase on the structure and properties of the chromatographic interface and how these changes affect the analytes. The aim is to identify the molecular mechanisms by which the mobile phase influences analyte retention and selectivity in RPLC and to determine how analyte retention and surface diffusion affect mass transport in a reconstructed silica monolith.
The first chapter examines how six apolar to moderately polar aromatic analytes (naphthalene, ethylbenzene, benzene, acetophenone, benzyl alcohol, and phenol) behave in terms of their mobility in an RPLC slit-pore model. In addition, OS surface excess adsorption isotherms of MeOH and ACN in W–MeOH and W–ACN mixtures are investigated. For this purpose, MD simulations are used to examine a 10 nm wide silica slit pore with a dimethyl-n-octadecylsilyl (C18) surface modification containing between 2/98 and 99/1 (v/v) W/OS. The shape of the experimental surface excess adsorption isotherms, measured for a typical C18 column, can be reproduced with help of the simulations. Mobile phases with ACN as OS generally generate more OS surface excess at equivalent mole fractions than phases with MeOH. The OS enrichment in the interface region between the mobile and stationary phase is the main cause of OS surface excess. The increased affinity of MeOH for the formation of hydrogen bonds (HB) with W results in an increased W content in the chromatographic interface and explains the lower surface adsorption of MeOH. Thus, the HB properties of the OS prove to be the deciding parameters for the explanation of OS adsorption isotherms in RPLC. Besides surface adsorption, surface diffusion is also investigated, in which analytes exhibit increased diffusion in the interface region between the bonded and mobile phase. The observed mobility gain of analytes decreases with increasing OS content in the mobile phase. Excess of OS in the interface region reduces viscosity and thus facilitates analyte diffusion. A further increase in analyte mobility can be achieved through contacts with flexible chain ends. The highest mobility gain is observed for apolar analytes, particularly in W–MeOH mixtures. Overall, polar solutes show a lower diffusivity and mobility gain due to the formation of HBs with solvent molecules of the mobile phase. The data in this chapter shows how the selection of the OS and its HB properties is significant for the formation of the chromatographic interface and its properties.
In the second chapter, the six analytes introduced in the first chapter and two additional dead time markers (acetone and uracil) are investigated in W–MeOH and W–ACN bulk mixtures with OS fractions between 2 and 99 vol %. MD simulations are used to investigate the solvation behavior of the solutes and the resulting influences on analyte retention. Spatial distribution functions show that the solutes exhibit an anisotropic solvation environment, where hydrophobic groups are preferably solvated by OS and hydrophilic groups are preferably solvated by W. The quantitative description of the solvation environment using the limiting linear preferential solvation 𝛿S,OS0, calculated from Kirkwood–Buff integrals, shows that the dead time markers have values close to zero and that the analytes can be distinguished from them by their clear preferential solvation by OS. The preferential solvation of hydrophobic structural elements and surfaces by OS in W–ACN phases is more pronounced than in W–MeOH phases, due to the lower affinity of ACN to form W–OS HBs. The increased OS content in the solute environment and the therefore more favorable solvation environment can provide insights into the stronger elution strength of ACN compared to MeOH. The ratio of the number of hydrophobic groups to the number of HBs, which a solute forms in 2 vol % OS, gives a first indication of the retention order of analytes and reproduces the experimentally observed retention order to a good degree. Hydrophobic groups emerge as a measure of retentive interactions, whereas the number of HBs estimates the extent to which a solute needs contact with W molecules.
In the third chapter, benzene as the structurally simplest analyte is simulated in W–ACN and W–MeOH mixtures at volume contents ranging from 10 to 90 vol % OS in the RPLC slit-pore from the first chapter. In this chapter, the solvation of the bonded phase and the resulting effects on its conformation and the retention of benzene are investigated. With increasing elution strength of the mobile phase, the OS density in the stationary phase increases and W density in the interface regions decreases and is shifted towards the bulk region. This leads to a redistribution of the C18 and analyte density towards the bulk region. The shift in C18 density is accompanied by several conformational changes in the C18 chains, whereby the number of chains lying parallel to the silica surface decreases and the chains generally exhibit a more stretched conformation. Benzene density profiles in the stationary phase show two distinct peaks, the partitioning peak deep in the bonded-phase region and the adsorption peak in the chromatographic interface. As W density retreats from the chromatographic interface region with increasing elution strength of the mobile phase, the benzene density shifts from the partitioning peak to the adsorption peak, and the adsorption peak shifts slightly towards the bulk region. Spatial distribution functions show that the shift of benzene density leads to a decrease in the average number of benzene contacts with the bonded phase and an increase in the number of contacts with solvent molecules. In W–MeOH systems, the W density the chromatographic interface is higher than in W–ACN systems and benzene needs to occupy regions deeper in the bonded-phase to avoid the W in the interface region. This increases the average number of benzene contacts with the bonded phase. The average number of benzene contacts with the bonded phase is therefore related to retention and reflects the increased retention in W–MeOH systems. Overall, this chapter demonstrated that the general effects influencing retention can be explained by the reaction of analytes and the chromatographic interface to W density, which in turn is dependent on the properties of the OS. This shows how the general effects influencing retention can be explained by the reaction of the analytes and the chromatographic interface to the W density. The distribution of the W density in the slit pore depends on the properties of the OS.
The fourth chapter examines the solute-specific effects that influence and control analyte retention at the molecular level. To this end, the six analytes from the first chapter are examined in the established RPLC slit-pore model in W–MeOH and W–ACN mixtures with an OS content between 10 and 90 vol % using MD simulations. The experimentally observed retention order of the six analytes can be reproduced by the simulations using the stationary-phase averaged number of bonded-phase contacts per analyte molecule. As the number of contacts with the bonded phase decreases, retention decreases and analytes with more contacts generally elute later than those with fewer contacts. The number of contacts depends on the penetration depth of analytes into the stationary phase and on the number of hydrophobic groups. The evasion of W through the hydrophobic groups pushes analytes deeper into the stationary phase, but only as far as the requirements for solvent HBs allows it. Thus, most of the density of the polar analytes is located in the adsorption peak in the stationary phase, leading to lower average number of contacts with the bonded phase. The apolar analytes, on the other hand, have significantly more analyte density in the partitioning peak and the stationary-phase averaged number of contacts with the bonded phase is therefore higher. Through a change in the OS the mobile phase exhibits different HB properties and the solvation of the bonded phase is altered. This allows different selectivities to be achieved in separations, as can be observed, for example with the benzyl alcohol–phenol pair in W–MeOH and W–ACN mixtures. Since the analytes have varying requirements as HB acceptors and donors, they respond differently to changes in the OS, which can result in changes in selectivity. Differences in analyte selectivities can therefore be explained by the solute-specific reaction to changes in the stationary-phase solvation caused by the mobile phase.
The fifth chapter examines the mass transport of the six analytes in a reconstructed silica monolith with a hierarchical pore structure and how it is influenced by the properties of the analytes and the mobile phase. For this purpose, the mesopore and fixed-bed diffusion coefficients of the six analytes are determined using multiscale simulations. The physical reconstructions of the mesopore space are based on measurements from a scanning transmission electron microscope, while those of the macropore space are based on confocal microscopy measurements. The diffusion and density profiles of the analytes are embedded in simplified form in three mesopore networks with average pore diameters dmeso of 12.3, 21.3, and 25.7 nm to determine the effective mesopore diffusion coefficients Dmeso using Brownian dynamics (BD) simulations. Dmeso increases with increasing OS content or with increasing analyte polarity. This can be compensated for by reducing dmeso, with an accompanying reduction in the available bulk volume. The effects of dmeso on Dmeso are thus comparable to those of analyte polarity and the elution strength of the mobile phase. The obtained Dmeso values are transferred to a reconstructed macropore network and the effective bed diffusivities Dbed are simulated using BD simulations. The Dbed values are limited by analyte retention and not by morphological properties of the pore space. Increased analyte retention slows down mass transfer between the mesopores and macropores, causing Dbed to decrease. By linking Dbed to the phase-based retention factor, the B-term of the plate-height equation is determined. In addition to the known decrease in the B-term with increasing flow velocity of the mobile phase, a U-shaped curve is observed when plotting the B-term against the OS fraction of the mobile phase. The B-term is dominated by the retention factor at a low OS fraction and associated high retention, and by Dbed at a high OS fraction and low retention.
Review
Metadata
License
This item has been published with the following license: In Copyright