Polymer adsorption on cellulose crystalline surfaces through efficient free energy calculation

Abstract

This thesis addresses computational challenges in calculating free energy differences in molecular simulations, focusing on polymer and cellulose nanocrystal (CNC) systems. Traditional umbrella sampling (US) techniques are computationally intensive, particularly for complex molecular systems. This work introduces a novel algorithm to optimize US parameter settings, significantly reducing computational time while maintaining high accuracy.

The thesis first examines the adsorption of short-chain methylcellulose (MC) on a CNC surface using molecular dynamics simulations. It is demonstrated that the minimal sampling time required for statistically converged potential of mean force (PMF) results is much longer than typically reported. By employing non-uniform sampling windows and optimized parameters, substantial improvements in sampling efficiency are achieved. Results show that distributing computational resources to regions with steep PMF gradients can yield accurate profiles with reduced computational expense.

Further, interactions of two cellulosic oligomers–methylcellulose (MC) and cellulose (CE)–with three crystallographic CNC surfaces ((100), (110), and (110s)) in the presence of water are explored. The optimized sampling method reveals that CE exhibits higher affinity for the (100) and (110s) surfaces compared to MC, while both oligomers have similar affinity towards the (110) surface. Entropy, particularly from the release of bound water molecules, plays a significant role in the adsorption process.

The study also focuses on the gelation properties of CNC systems combined with various polysaccharide polymers. Eight distinct CNC-oligomer systems are modelled to investigate the effects of molecular features like branching, length, and spatial distribution on adsorption capabilities. Findings confirm that increased branching reduces adsorption affinity, while the proximity of side chains to the backbone enhances it.

Collectively, this thesis advances computational methods for free energy calculations in molecular simulations, providing a deeper understanding of CNC-polymer interactions and enabling the design of CNC-based materials with tailored properties.