The rational design of drug sensors and adsorbents for environmental applications

Abstract

While pharmaceuticals are required for human welfare, they are ending up in
water systems through a variety of different pathways and methods, such as human consumption and subsequent excretion. Such contaminations with these
micropollutants is potentially damaging not only for humans but also the general
ecosystem of such areas. However, halting the production or use of pharmaceuticals is neither a reasonable or possible solution. The complete removal of such micropollutants from all waste water systems is a monumental task and very unlikely to occur, modern waste water treatment plants are not well equipped to deal with the variety of these small pharmaceutical residues. Therefore, the focus on minimisation of such residues entering waste water systems will be the primary route to reducing the problem. For example, targeting point sources of these pharmaceutical residues, such as the out flux of waste water from hospitals. One such method of minimisation is the application of pollutant specific Molecularly Imprinted Polymers (MIPs), for both detection and subsequent separation from waste water. MIPs are a class of polymer that contain many cavity sites that are specific to a chosen molecule. MIPs offer this specificity to micropollutants utilising targeted functional design, through various aspects in the prepolymerisation mixture. Such design experimentally is both difficult and time consuming, with many functional monomers, cross-linkers and solvents available for MIP production obtaining the most optimum types, ratios and quantities of such is one of the main challenges of MIP design. With the use of computational modelling, this pre-polymerisation mixture can be analysed and optimum MIP components predicted with both increased speed and accuracy. In this research, the main aim is the investigation of a general computational protocol for the simulation of pre-polymerisation mixtures to aid in the prediction of any MIP component, ultimately optimising the cavity site of any micropollutant specific MIP when produced physically. Molecular dynamics was the primary computational method used, as it offers all-component pre-polymerisation mixtures to be analysed, therefore giving insight into the whole system. The OPLSAA force field was used, which was validated through a variety of density calculations, and logKow determinations. Initially, the developed computational protocol was validated through functional monomer selection for a fluoxetine specific MIP. Good agreement was seen between the predicted and experimental data of identical systems, with itaconic acid highlighted as the optimal choice. This protocol was then used to predict the optimal functional monomer for both norfloxacin and tetracycline, resulting in predictions
of acrylamide and methyl methacrylate (MMA) respectively. The effect of
crosslinker was considered next, utilising the fluoxetine system, with investigations into both quantities/ratio and also type. An ethylene glycol dimethacrylate (EGDMA) crosslinker - template ratio of 40:1 was shown to be optimal, but the correct crosslinker type being trimethylolpropane trimethacrylate (TRIM). The final MIP component, solvent, was investigated using a TNT specific MIP pre-polymerisation mixture. Solvent systems of pure dimethyl sulfoxide (DMSO) and acetonitrile (ACN), along with varying compositions of the two were simulated, and the results compared against TNT specific MIPs which were physically produced and analysed during this research. The simulations identified a pure DMSO system giving rise to the highest interaction of template - functional monomer, however a pure ACN system giving the highest surface area, both computationally and experimentally. Finally, an approach to optimise the developed computational protocol was investigated using the molecular dynamics sampling technique called umbrella sampling. This technique correctly predicted the matching functional monomer and crosslinker ratio of the full molecular dynamics fluoxetine simulation, however at only 2 hours compared to 18 hours.

Date of Award	4 Dec 2023
Original language	English
Awarding Institution	University of Bath
Supervisor	Carmelo Herdes Moreno (Supervisor) & Bernardo Castro Dominguez (Supervisor)