Computer simulations of adsorption and molecular recognition phenomena in molecularly imprinted polymers

Abstract

Molecularly imprinted polymers (MIPs) are a novel, promising family of porous materials with potential applications ranging from separations, chemical sensing and catalysis to drug delivery and artificial immunoassays. The unique feature of these materials is their biomimetic molecular recognition functionality. Molecular recognition is the biological phenomenon of specific, selective and strong association between a substrate and a ligand. In man made MIPs this functionality is implemented via templated synthesis protocol. MIPs are synthesized in the presence of additional template molecules which form complexes with functional monomers in the pre‐polymerization mixture. After polymerization, the template is removed, leaving cavities in the structure which are complementary in shape and interaction patterns to the template molecules. These cavities act as mimics of biological receptors and are able to recognize and rebind template molecules. Although the imprinting concept is simple in principle, synthesis of MIPs with precisely controlled characteristics and performance remains a challenging task. Composition, polymerization conditions, template removal process and application conditions all affect the properties of MIPs. The material is affected at different scales, but crucially at the microscopic level, the number, fidelity and accessibility of binding sites are dependent on all the factors mentioned. The full potential of these materials can only be achieved if researchers can control and optimize the properties of MIPs through detailed understanding of adsorption and molecular recognition processes in these materials. The objective of this work is to, using computer simulations and statistical mechanics; develop a fundamental description of MIP formation and function, and to link morphological features of the model materials to their molecular recognition capabilities. In general, molecular simulations employed in this study should allow easier and more efficient exploration of a vast number of factors influencing the behaviour of MIPs. At the heart of the approach developed in this thesis is a computational strategy that imitates all the stages of MIP formation and function. First, the model simulates the pre‐polymerization mixture, allowing the formation of template‐functional monomer complexes. (This stage is implemented via canonical Monte Carlo simulation). Complexes can have different structures, depending on the chemical nature of template and functional monomer; therefore complexes can have a range of association constants. The distribution of template‐functional monomer complexes also translates into a distribution of binding sites of different specificity after template removal. In the second stage of the process, adsorption simulations (grand canonical Monte Carlo) are performed for a variety of model MIPs prepared to assess the role of various processing conditions such as composition, density and binding sites degeneration. This strategy was first applied to a simplified description of MIP species in order to identify the minimal model capable of molecular recognition and thus shed the light on the very nature of this phenomenon. In the developed model, the molecular species are constructed from hard spheres, featuring small interaction sites on their surfaces. The bond between two interaction sites has the strength and topological features of a typical hydrogen bond. The model exhibits molecular recognition, being able to preferentially adsorb template molecules. The associations between template and functional monomers were analyzed and classified to describe the distribution of binding sites and their heterogeneity. Using this model, several experimental trends typically observed in MIP studies could be explained, such as maximum in the selectivity as a function of monomer concentration. Using this model, we were also able to explore hypothetical, alternative protocols for MIP synthesis in order to improve their performance. These include the use of alternative templates and the post‐synthetic surface modifications of MIPs. The general strategy to modelling MIP, employed in this thesis, was then applied to a more detailed description of MIPs with realistic force field potentials for all the species involved. This more elaborate model is simulated with a combination of molecular dynamics (MD) and Monte Carlo techniques. This detailed model provided a wealth of information on various types of complexes observed in the pre‐polymerization mixture. Specifically, it revealed the relative weight of different interactions in the complex and their role in the binding energy of adsorbates. These simulations also provided the comparison of the relative contribution of different types of interactions (van der Waals, Coulombic) involved in a molecular recognition process. We believe the insights gained in this work will contribute to the development of rational MIP design strategies. In the discussion of the results of the thesis we speculate on how these models can be further developed in order to generate quantitative predictions and what type of systems it would be interesting and important to investigate in the future.