Elucidation of chiral recognition mechanisms of solutes by amylose tris[(s)-α-methylbenzylcarbamate] sorbent

Hung-Wei Tsui, Purdue University

Abstract

Enantioselective separations of chiral molecules are important in various chemical fields, such as pharmaceuticals and agrochemicals industries. Polysaccharide-based sorbents have been widely used in chiral liquid chromatography. The recognition mechanisms which determine their enantioselectivities are not completely understood. In this dissertation, the chiral recognition mechanisms of a widely used commercial sorbent, amylose tris[(S)-alpha-methylbenzylcarbamate], for benzoin (B) enantiomers were first studied. The HPLC data for benzoin with pure n-hexane as the mobile phase have been obtained. The behavior of sorbent-solute-hexane systems can be interpreted by considering only sorbent solute two-component interactions. Infrared (IR) spectra showed evidence of substantial hydrogen bonding (H-bonding) interactions in the pure polymer phase, and additional H-bonding interactions between AS and benzoin. Density Functional Theory (DFT) was used to model the chain-chain and chain-benzoin H-bonding or other interactions. From high performance liquid chromatography (HPLC), and IR data, and DFT and molecular simulations, the observed enantioselectivities were inferred to be due primarily to two strong H-bonds, of the kind (AS) CO ... HO (R-benzoin) and (AS) NH ... OC (R-benzoin) and one strong H-bond (AS) CO ... HO (S-benzoin) for S-benzoin.Three additional solutes containing the same functional group, O=C-C-OH, as benzoin were studied: ethyl lactate (EL), methyl mandelate (MM), and pantolactone (PL). IR, DFT, and molecular simulations lead to a general hypothesis for the chiral recognition mechanism for these solutes. The mechanisms for these systems involve a non-enantioselective strong, or "leading", H-bonding interaction and an enantioselective weaker, or "secondary", H-bonding interaction, which is affected by geometrical restrictions. There is one or more additional interactions which determine the overall enantioselectivity. A new measure of molecular rigidity was developed with MD simulations. The solute with the small rigidity or high flexibility has the lowest enantioselectivities. The adsorption mechanisms of the chiral solutes were also probed macroscopically by using retention factor data for hexane-alcohol modifier solutions and stoichiometric displacement models. The models were used to explain the slope of the plot of the logarithms of the solute retention factor versus the molar concentration of a competitive modifier in an inert solvent. In previous models the slope was inferred to be equal to the total number of the modifier molecules displaced from the sorbent and from the solute-modifier complex upon adsorption of a solute molecule, and were presumed to be generally greater than 1. Nonetheless, for the four chiral solutes studied, with increasing IPA concentration C I0, it was discovered that slopes (B) smaller than 1 were possible, at concentrations from 0.13 to 1.3 M. The slopes were slightly more than 1 at higher concentrations. Such data cannot be explained by any previously available model. To address this problem and make effective use of the data in elucidating the solute-sorbent interactions, five monovalent simple solutes, acetone, cyclohexanone, benzaldehyde, phenylacetaldehyde, and hydrocinnamaldehyde, were chosen for study. The results of IR and DFT simulations showed clear evidence of IPA aggregation with average aggregation number n=3. A new thermodynamic retention model was developed to take into account IPA aggregation. Such aggregation phenomena affect the slopes significantly and lead to a significant reduction in the IPA monomer concentrations, which affects the IPA-sorbent binding, the IPA-solute complexation, and the slope. This discovery and the new models allow an accurate and reliable interpretation of the data in terms of alcohol displacement from the sorbent and the solute, and allow the determination of the number of the interaction sites of the solute with the sorbent. For the above four chiral solutes, a new more complex multivalent retention model is developed. It accounts for alcohol aggregation, multivalent solute adsorption, multivalent solute-alcohol complexation, alcohol adsorption, and solute intra hydrogen-bonding which was also found to be important for these four solutes. The limiting slope LS at a very high ("infinite") IPA concentration is predicted to be equal to the value of (x+y)/n, where y is the average number of the complexation binding sites, x is the adsorption binding sites, and n is the average alcohol aggregation number in hexane solution. The model was found to fit well the HPLC data. The estimated y-values correlate fairly well with the number of the solute functional groups, suggesting that y can be estimated from the inspection of the solutes molecular structures. Moreover, the x-values can be estimated from the values of the limiting slopes and the numbers of solute functional groups. The same values of the binding sites are found for the most R- and S-enantiomers. The results suggest that the effective number of the binding sites are the same for enantiomers of each solute. The binding equilibrium constants were found to be significantly different for the two enantiomers, suggesting that S-enantiomers, which were predicted to be non-H-bonded, simply bind with the sorbent more weakly. Overall, the results of these models provide additional insights and complement the mechanistic studies done for these systems.

Degree

Ph.D.

Advisors

Franses, Purdue University.

Subject Area

Chemistry|Chemical engineering

COinS